ECR 2025 Book of Abstracts.

Methods

or Background: We retrospectively identified 35 patients without suspicion of pancreatic pathology scanned on both E ID-CT and PCCT in the period from October 2021 until December 2023. IV co ntrast was given according to patient weight and both a late arteria l phase (LA) and a portovenous phase (PV) was performed. Image quality was rated on a 5-point Likert-scale (from 1=nondiagnostic to 5=optimal). E leven different pancreatic parameters were scored by four radiology consultant s. One reader made quantitative measurements of density and noise. All data analysis was performed with RStudio, version 2022.07.1. Continuo us parameters were compared with a paired t-test and mean image qualit y ratings with a Wilcoxon signed rank test.

Results

or Findings: Image quality was rated significantly higher on PCC T for the pancreatic parenchyma in the LA (3.87 vs 2.77, p<.001), the pancreatic parenchyma in the PV (3.31 vs 2.53, p<.001), pancre atic ducts (2.88 vs 2.62, p<.001), SMA (4.10 vs 2.74, p<.001), celiac axis (4 .04 vs 2.70, p<.001) and portal vein (3.29 vs 2.52, p<.001). Noise levels we re significantly lower with PCCT with a mean reduction of 5.8 HU across all par ameters. DLP was significantly reduced with a 31.8% reduction (p< 0. 01) for the LA and 21.5% (p< 0.01) for the PV.

Conclusion

Image quality was significantly improved for all ev aluated pancreatic and peripancreatic structures with PCCT. Additionally, image noise and radiation dose were significantly reduced. The improved image quality with PCCT could potentially lead to improvements in the currently difficult evaluation of pancreatic diseases.

Limitations

No limitations were identified. Funding for this study: Innovation Fund Denmark, Grant No. 1044-00015B. Ethics committee - additional information: The study was approved by the national ethics committee with the number: nvk22153 38. Author Disclosures: Michael Andersen: Speaker: Different Vendor Seminar s, GE, Philips and Siemens Anne Marie Ewald: Nothing to disclose Henrik S. S. Thomsen: Nothing to disclose Bulat Ibragimov: Nothing to disclose Erik Gudmann Steuble Brandt: Nothing to disclose Yousef Wirenfeldt Nielsen: Nothing to disclose Christoph Felix Müller: Nothing to disclose Development and Validation of Contrast-enhanced CT- based Imaging Intratumor Heterogeneity of Pancreatic Ductal Adeno carcinoma *B. Zhao*, S. Ju; Nanjing/CN ([email protected]) Purpose or Learning Objective: To construct an imaging ITH (IITH) through radiomics methodology to effectively reflect the IT H of PDAC and explore its prognostic value.

Methods

or Background: This study enrolled 961 patients with pathologicall y confirmed PDAC who had undergone preoperative contr ast-enhanced computed tomography (CT) in two cohorts. Firstly, T umor regions of interest were automatically segmented in both arterial and v enous phase images. Radiomics features from these 2 phasea images were extracted based on PyRadiomics. Highly variable radiomic features with median absolute deviation > 1, were selected to assess the ITH. Similarity Ne twork Fusion (SNF) was employed to identify distinct imaging heterogeneity phenotypes in the discovery cohort (Cohort 1), and the identical crit eria were applied to Cohort 2. Kaplan-Meier analysis was utilized to investigate t he association between the identified imaging phenotypes and overall survival (OS).

Results

or Findings: A total of 961 patients (mean age, 63.1 years ± 9.2 standard deviation; 460 men) from 2 cohorts were en rolled. And 3378 radiomics features were extracted for each patient. In the discovery cohort (Cohort 1, n = 637), 241 highly variable heterogene ity-related features were selected to identify IITH, and 283 patients were cl assified into a high-IITH subgroup. Patients with high-IITH in the discovery cohort presented significantly poorer OS compared to those with low IITH (median time, 20.6 vs. 37.1 months, P < 0.001). The identical criteria wer e applied to the validation cohort (Cohort 2, n = 324). Kaplan-Meier analysis a lso confirmed that patients with high-IITH (n =149) had shorter OS (median time , 16.4 vs. 26.3 months, P < 0.001).

Conclusion

We established a noninvasive radiomics method to ev aluate the ITH of PDAC. Furthermore, we demonstrated the progn ostic power of IITH.

Limitations

This study requires further multi-omics validation in the future. Funding for this study: NSFC, No. 82330060, 92059202, 823B2040, 61821002 and 82372024) Ethics committee - additional information: IEC for Clinical Research ofZhongda Hospital, Affliated to Southeast Universi ty Author Disclosures: Ben Zhao: Nothing to disclose Shenghong Ju: Nothing to disclose Interobserver agreement of pancreatic tumor size me asurement before and after neoadjuvant therapy: is MRI as reproducib le as CT? *A. Licha*, C. Touloupas, A. Delpla, A. Pouvelle, M . Zins; Paris/FR Purpose or Learning Objective: Assess inter-observer agreement of pancreatic ductal adenocarcinoma (PDAC) tumor size measurement on CT and MRI, before and after neoadjuvant therapy (NAT)

Methods

or Background: We reviewed all patients with a histological diagnosis of PDAC at Paris Saint-Joseph Hospital, b etween 2010 and 2022, and who underwent CT and MRI, both before and after NAT. Three independent radiologists anonymously evaluated the large axial tumor axis on 2 CT acquisitions and 6 MRI sequences. Inter-observ er agreement was assessed by intra-class correlation coefficients (I CCs) and by LOAM graphs (Bland & Altmann extension for multiple observers).

Results

or Findings: The final population consisted of 50 patients. On C T exams, inter-observer agreement was excellent befor e NAT (ICC of 0.83 [0.73;0.90] at arterial phase and 0.84 [0.74;0.90] at portal-venous phase) and decreased but remained good after NAT (ICC of 0.66 [0.52;0.78] at arterial phase and 0.65 [0.51;0.77] at portal-venous phase). On MRI exams, inter- observer agreement was moderate to good before NAT (best sequence being T1 at arterial phase with ICC of 0.67 [0.53;0.79]) and decreased becoming moderate for all sequences after NAT (best sequence being T1 at late phase with ICC of 0.55 [0.37;0.71]).

Conclusion

Inter-observer agreement of PDAC great axe measurem ent is better on CT than on MRI, and decreases between pre -NAT and post-NAT imaging, both on CT and MRI. These results encourag e to keep on performing technically perfect CT scans, without questioning t he need for pre-operative hepato-pancreatic MRI in non-metastatic patients.

Limitations

This is a monocentric retrospective study. Funding for this study: Inter-observer agreement of tumor size measurement is higher on CT than on MRI, both before and after NAT. Inter-observer agreement of tumor size measurement decreases follo wing NAT, both on CT and MRI. Ethics committee - additional information: This study was approved by our organization's Medical Research Ethics Group. Author Disclosures: Caroline Touloupas: Nothing to disclose Marc Zins: Nothing to disclose Arié Licha: Nothing to disclose Alexandre Delpla: Nothing to disclose Arnaud Pouvelle: Nothing to disclose A Radiomics-Based Model for Predicting Lymph Node M etastasis of Pancreatic Ductal Adenocarcinoma: A Multi-Center St udy *B. Zhao*, S. Ju; Nanjing/CN ([email protected]) Purpose or Learning Objective: To develop a radiomics model to predict lymph node metastasis (LNM) in patients with pancre atic ductal adenocarcinoma (PDAC) and assess its value for clin ical management.

Methods

or Background: Patients with pathologically confirmed PDAC were retrospectively enrolled from four centers and divi ded into a training (n = 192), validation (n = 82), testing (n = 100), and clinica l utilization cohort (n = 163).A radiomics model was constructed based on the arteri al phase of computed Abstract-based Programme 5 Wednesday tomography (CT) for predicting LNM. The areas under the curve (AUCs) were used to compare the performance between the radiomi cs model and other models. Subsequently, Kaplan-Meier analysis was use d to validate the model’s value for prognosis and therapy decisions.

Results

or Findings: A total of 437 patients (mean age, 63.1 years ± 9.2 standard deviation; 253 men) were included. The rad iomics model demonstrated AUCs of 0.84, 0.82, and 0.78 in the tr aining, validation, and testing cohorts, respectively, superior to other mo dels (all P < 0.05). Besides, LNM predicted by the radiomics model was strongly a ssociated with overall survival (OS) (P < 0.001). Kaplan-Meier analysis al so demonstrated that patients with a high risk of LNM had a worse progno sis (all P < 0.05). Furthermore, patients who were dissected with ≥ 15 LNs had a longer OS than those with fewer LNs dissected in the high-risk sub group predicted by the radiomics model in the clinical utilization cohort (P = 0.002).

Conclusion

The radiomics model demonstrated impressive perform ance in predicting LNM and prognosis, indicating its potent ial for. therapy decisions.

Limitations

The model we developed should be validated in a pro spective study. Funding for this study: NSFC, No. 82330060, 92359304, 92059202, 823B2040, 61821002 and 82372024 Ethics committee - additional information: IEC for Clinical Research ofZhongda Hospital, Affliated to Southeast Universi ty Author Disclosures: Ben Zhao: Nothing to disclose Shenghong Ju: Nothing to disclose CT-Based Early Indicators of Severe Pancreatic Fist ula and Hemorrhage After Pancreatoduodenectomy D. Palumbo, *A. Campisi*, V. Andreasi, F. Prato, S. Partelli, D. Tamburrino, M. Falconi, F. De Cobelli; Milan/IT Purpose or Learning Objective: Postoperative pancreatic fistula (POPF) and postpancreatectomy hemorrhage (PPH) are major compl ications following pancreatoduodenectomy (PD). Despite their clinical importance, no tool currently exists to predict their occurrence or sev erity. This study aims to identify radiological characteristics that can aid in the early prediction and stratification of POPF and PPH.

Methods

or Background: We retrospectively reviewed 399 patients who underwent PD at San Raffaele Hospital between Janua ry 2015 and December 2021. Patients included had at least one contrast-e nhanced computed tomography (CE-CT) scan within 14 days post-surgery . Several radiological features were systematically assessed, including pa ncreaticojejunostomy (PJ) dehiscence, PJ defects, fluid collections, perianas tomotic air bubbles, and pancreatic remnant density.

Results

or Findings: Clinically relevant POPF occurred in 230 patients (57.9%), with 185 classified as grade B and 45 as g rade C. PPH occurred in 61 patients (15.3%). PJ dehiscence was significantly a ssociated with clinically relevant POPF (31% vs. 22%, p = 0.035), and PJ defe cts were more extensive in patients with severe POPF (median 7 mm vs. 5 mm, p = 0.001). Fluid collections, particularly above the PJ site, were a lso linked to the development of severe POPF (p < 0.001). Additional markers, suc h as stump pancreatitis, perianastomotic air bubbles, and lower pancreatic r emnant density, were significantly more frequent in severe POPF cases.

Conclusion

These findings support the use of postoperative CT scans to prospectively identify patients at risk of developi ng severe POPF and PPH, allowing for better clinical management.

Limitations

Retrospective study design. Funding for this study: None. Ethics committee - additional information: Ethics committee approval number: 28/INT/2015 Author Disclosures: Domenico Tamburrino: Nothing to disclose Valentina Andreasi: Nothing to disclose Diego Palumbo: Nothing to disclose Massimo Falconi: Nothing to disclose Antonino Campisi: Nothing to disclose Stefano Partelli: Nothing to disclose Francesco De Cobelli: Nothing to disclose Francesco Prato: Nothing to disclose Correlation between celiac axis stenosis and compli cations after pancreatoduodenectomy *Y. Shu*, Y. Dai, J. Wei, Q. Xu; Nanjing/CN ([email protected]) Purpose or Learning Objective: We aimed to explore the correlation between celiac axis stenosis and complications after pancre atoduodenectomy.

Methods

or Background: Patients who underwent pancreatoduodenectomy in our hospital pancreas center from January 2021 t o December 2023 were retrospectively collected. The stenosis rate of cel iac axis was measured on pre-operation arterial phase imaging of routine enh anced CT, and graded the severity of celiac trunk stenosis: no stenosis (< 3 0%), mild stenosis (30%- 50%), and significant stenosis (≥50%). The incidence of postoperative complications was evaluated, and both univariate an d multivariate logistic regression analysis were conducted.

Results

or Findings: A total of 774 patients were included in the study, 205 (26.5%) had celiac axis stenosis: 144 (18.6%) with mild stenosis, and 61 (7.9%) with significant stenosis. Celiac axis steno sis was associated with pancreatic fistula (p<0.001), postoperative bleedin g (p=0.033), and postoperative biliary leakage (p= 0.006). In multiv ariate logistic regression analysis, mild stenosis of the celiac axis was an i ndependent risk factor for postoperative pancreatic fistula (OR 2.81, 95%CI 1. 82-4.33, p<0.001), and significant stenosis of the celiac axis was an inde pendent risk factor for postoperative biliary leakage (OR 4.91, 95%CI 1.27- 19.04 , p=0.021).

Conclusion

Celiac axis stenosis was associated with the risk o f complications after pancreatoduodenectomy. Surgeons may need to p ay attention to the condition of celiac axis stenosis before pancreatod uodenectomy.

Limitations

Retrospective study. Single center. Funding for this study: There was no funding for this study. Ethics committee - additional information: None Author Disclosures: Yuping Shu: Nothing to disclose Qing Xu: Nothing to disclose Yuran Dai: Nothing to disclose Jishu Wei: Nothing to disclose Cystic fluid non-invasive evaluation based on photo n-counting detector CT spectral imaging in patients with pancreatic cys tic lesions *I. Dudás*, B. Lovász, M. Benke, Á. Szücs, P. N. Ka posi-Novák, A. Szijártó, P. Maurovich-Horvat, B. K. Budai; Budapest/HU Purpose or Learning Objective: Differentiation between pancreatic cystic lesions is a challenging task for clinicians. Spect ral imaging via photon- counting detector CT (PCD-CT) scanners allows the r econstruction of virtual monoenergetic images (VMI) enabling the measurement of Hounsfield unit (HU) densities at different keV and the generation of spectral absorption curves. Our study aimed to investigate whether muci nous and non-mucinous pancreatic cystic lesions (PCL) have different spec tral absorption curves that could help the differential diagnostics.

Methods

or Background: Our study included 74 patients with PCLs, 53 patients with mucinous cystic neoplasms and 21 with non-mucinous cystic pancreatic lesions diagnosed based on current pract ice guidelines. The spectral absorption curves were generated from the pancreatic-phase scans. The average densities were measured on the 70keV (H U70keV) and 40keV (HU40keV) virtual monoenergetic images (VMIs), by p lacing 3-3 circular regions of interest in PCL’s cystic component parts and the density differences were calculated (HUdiff(40keV-70keV)). Kruskal-Wall is test with post-hoc Dunn’s test was used for comparing the groups. The discrimination performance was assessed by receiver operating char acteristic (ROC) curve analysis. The intraobserver reproducibility and int erobserver reproducibility were evaluated by the intraclass correlation coeffi cient (ICC).

Results

or Findings: On 70keV VMIs, no significant differences were foun d between the average densities of mucinous and non-m ucinous PCLs cystic components, however, a significant difference was f ound in HUdiff(40keV- 70keV) values (p<0.0001). The diagnostic performanc e of HUdiff(40keV- 70keV) in differentiating between mucinous vs. non- mucinous PCLs had AUCs of 0.92 and 0.92 on the training and test datasets, respectively, with a good interobserver (ICC=0.82) and excellent intraobserve r reproducibility (ICC=0.94).

Conclusion

Spectral absorption curve assessment of cystic comp onents could be a useful additional measurement to facilit ate the non-invasive differential diagnosis between mucinous and non-muc inous pancreatic cystic lesions.

Limitations

This was a single-center study with a retrospective study design. Funding for this study: No funding was received for this study. Ethics committee - additional information: The presented study was approved by the institutional ethics committee of o ur University (SE RKEB 256/2023). Author Disclosures: Attila Szijártó: Nothing to disclose Pál N. Kaposi-Novák: Nothing to disclose Pál Maurovich-Horvat: Nothing to disclose Ákos Szücs: Nothing to disclose Ibolyka Dudás: Nothing to disclose Márton Benke: Nothing to disclose Borbála Lovász: Nothing to disclose Bettina Katalin Budai: Nothing to disclose Abstract-based Programme 6 Wednesday CT-Derived Body Composition Metrics Predict Severit y in Acute Pancreatitis: A Post-Hoc Multicenter Study *R. Z. Borbély*¹, B. Teutsch¹, V. Vass¹, K. Márta¹, B. Erőss¹, A. Vincze², A. Szentesi¹, P. Hegyi¹, N. Faluhelyi²; ¹Budapest/H U, ²Pécs/HU ([email protected]) Purpose or Learning Objective: Accurately predicting the severity of acute pancreatitis (AP) remains a significant clinical ch allenge. CT scans, traditionally used for diagnosis, also provide valuable insights into body composition. However, the prognostic utility of CT-derived body composition metrics has been inconsistently reported. This study aimed to e valuate whether key CT body composition metrics can effectively predict th e severity of AP.

Methods

or Background: A post-hoc analysis was conducted on a multicenter study involving 437 AP patients who und erwent CT scans within the first 24 hours of hospital admission. Measureme nts of visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), an d skeletal muscle area (SMA) were obtained at the third lumbar vertebra le vel. These areas were normalized for patient height to calculate the Visc eral Adipose Tissue Index (VATI), Subcutaneous Adipose Tissue Index (SATI), a nd Skeletal Muscle Index (SMI). Muscle radiodensity in Hounsfield Unit s (HU) was assessed to determine fatty infiltration. The fat-to-muscle vol ume ratio was also calculated. These metrics were analyzed as potential predictors of severe AP using receiver operating characteristic (ROC) curves and area under the curve (AUC) values. Severity was determined using the Modified CT Severity Index (mCTSI). Statistical analyses were performed using IBM SPSS Statistics.

Results

or Findings: The fat-to-muscle ratio demonstrated the highest predictive accuracy for severe AP (AUC = 0.68), fol lowed by VATI (AUC = 0.65). Other indices did not show significant predi ctive potential (AUC < 0.6).

Conclusion

CT-derived body composition metrics, particularly t he fat-to- muscle ratio and VATI, are valuable predictors of s evere acute pancreatitis. Incorporating body composition analysis into routin e CT evaluations may enhance prognostic assessments for AP patients.

Limitations

As a post-hoc analysis of retrospective data, the s tudy may be subject to selection bias, potentially limiting the generalizability of the findings. Funding for this study: Funding for Ruben Zsolt Borbély was supported by the EKÖP-2024-239 New National Excellence Program o f the Ministry for Culture and Innovation from the source of the Natio nal Research Development and Innovation Fund. Funding for Brigitta Teutsch w as provided by the ÚNKP- 22-3 New National Excellence Program of the Ministr y for Innovation and Technology from the source of the National Research , Development and Innovation Fund (to BT - ÚNKP-22-3-IPTE-1693). Cent er costs were covered by the University of Pécs, the Momentum Grant of th e Hungarian Academy of Sciences (LP2014-10/2014), and grants from the Nati onal Research, Development, and Innovation Office (GINOP-2.3.2-15- 2016-00015, KH- 125678). The funders had no influence on the study design, data collection, analysis, or manuscript preparation. Ethics committee - additional information: This post-hoc analysis is based on a study that received ethical approval from the Scientific and Research Ethics Committee of the Medical Research Council (I SRCTN63827758, decision 55961-2/2016/EKU). Author Disclosures: Bálint Erőss: Nothing to disclose Nándor Faluhelyi: Nothing to disclose Péter Hegyi: Nothing to disclose Katalin Márta: Nothing to disclose Ruben Zsolt Borbély: Nothing to disclose Aron Vincze: Nothing to disclose Brigitta Teutsch: Nothing to disclose Vivien Vass: Nothing to disclose Andrea Szentesi: Nothing to disclose Role of incidental pancreatic calcifications on com puted tomography as opportunistic biomarker for chronic pancreatitis *A. Pata*, F. Rizzetto, C. B. Monti, A. Vanzulli; M ilan/IT ([email protected]) Purpose or Learning Objective: To evaluate whether incidentally detected pancreatic calcifications on computed tomography (C T) serve as a reliable biomarker for chronic pancreatitis as defined by cl inical criteria.

Methods

or Background: We retrospectively reviewed CT scans from adult patients between 2014 and 2024, identifying cases w here "pancreatic calcifications" were mentioned in the radiology rep ort. Patients with known history of pancreatitis or pancreatic surgery were excluded. For each patient, we recorded pancreatic size, calcification characte ristics (number, size, and location), and other features of chronic pancreatit is, such as duct dilatation or intraductal calculi. Clinical data, including pain, abdominal symptoms, and risk factors for chronic pancreatitis, were also collect ed.

Results

or Findings: A total of 137 patients with incidental pancreatic calcifications were identified. A small subset had coarse calcifications (n=9, 7%), while the majority had both coarse and punctif orm calcifications (n=102, 74%), with calcification numbers ranging from 7 to 50 in over half of the cases. In 121 patients (88%), at least two pancreatic segm ents were involved, most frequently the head (n=125, 91%) and the body (n=11 2, 81%). When coarse calcifications were present alongside duct dilation , the duct caliber was significantly larger (6.7 mm vs 4.1 mm; p=0.026), w hile the tail was slightly smaller (17 mm vs 19 mm, p=0.018), with no signific ant differences in the size of the head or body (p>0.198). Among patients with available clinical information (n=38), only 3 (11%) reported symptoms such as abdominal pain, diarrhea, or bloating, and none (0%) reported weigh t loss. Elevated alcohol consumption or smoking history was not associated w ith calcification type or location (p>0.186), pancreatic size, or duct dilati on (p>0.317).

Conclusion

Incidentally detected pancreatic calcifications are not a reliable biomarker for screening for chronic pancreatitis.

Limitations

The main study limitation is its retrospective desi gn. Funding for this study: No fundings were received for this study Ethics committee - additional information: Institutional Review Board approved the retrospective data collection in anony mous form Author Disclosures: Francesco Rizzetto: Author: nothing to disclose Caterina Beatrice Monti: Author: nothing to disclos e Angelo Vanzulli: Author: nothing to disclose Annamaria Pata: Author: nothing to disclose The diagnostic potential of unenhanced dual-layer s pectral CT quantitative parameters in diabetic pancreas *L. Ge*, Y. Li, Y. Gao, X. Zhang, X. Yu; Xi An/CN ([email protected]) Purpose or Learning Objective: To investigate pancreas characterizing for type 2 diabetes mellitus (T2DM) using unenhanced du al-layer Spectral CT.

Methods

or Background: This retrospective study included patients who underwent abdominal unenhanced dual-layer spectral CT between March 2023 and April 2024. The patients were divided into T2DM group and control group. Nine regions of interest (ROIs) were drawn (three f or head, three for body and three for tail). Mean attenuation on conventional 1 20-kVp CT images (CTconv), effective atomic number maps (Z-eff), iodine densit y maps (ID), virtual non- contrasted (VNC), and mean attenuation on virtual m onoenergetic images (VMIs) at 40-200keV were measured. The Mann -Whitne y U test was used to compare the differences between the two groups. The receiver operating curve (ROC) was used to evaluate the diagnostic efficacy of the above parameters.

Results

or Findings: A total of 84 patients, including 44 T2DM patients and 40 controls, were evaluated. There was a statistica lly significant difference in the CTconv (46.7 ± 5.9 HU vs. 50.7 ± 4.5 HU, p < 0.01), Z-eff (7.43 ± 0.05 vs. 7.46 ± 0.03, p < 0.01), ID (103.9 ± 0.6 vs. 104.3 ± 0.4, p < 0.01), VNC (39.2 ± 5.5 HU vs. 42.3 ± 4.2 HU, p < 0.01) and VMIs (62.1 ± 9.7 HU vs. 68.6 ± 6.2 HU at 40 keV, p < 0.001) between the T2DM group and th e control group. The AUCs of the CTconv, Z-eff, ID, VNC and 40 keV were 0.722,0.671,0.695, 0.691 and 0.734. Histogram analysis found that the 10th percentile value of the above parameters had higher diagnostic efficiency ( 0.757,0.736,0.702,0.734 and 0.805). The AUC of VMIs increased with decreasi ng monoenergeic levels.

Conclusion

The 10th percentile value of 40 keV was the best in dicator for distinguishing T2DM patients from the controls.

Limitations

Not applicable Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Xiao Yu: Nothing to disclose Yishan Li: Nothing to disclose Xiaoyue Zhang: Nothing to disclose Yanjun Gao: Nothing to disclose Liu Ge: Nothing to disclose Abdominal mapping with MOLLI and T2p-SSFP: can you do without the cardiac trigger? A. Lupi, P. Dardeshi, I. Toniolo, *G. Beggiato*, M. Pizzi, D. Negro, E. Quaia, A. Pepe; Padua/IT ([email protected]) Purpose or Learning Objective: Quantitative abdominal imaging is increasingly emerging in the radiology arena, drawi ng inspiration from cardiac mapping, which is significantly changing patient ma nagement. In fact, cardiac mapping sequences offer the possibility of obtainin g abdominal mapping values, although with long acquisition times, due t o the cardiac trigger. The aim of our study is to evaluate the differences between abdominal mapping values obtained with and without cardiac trigger.

Methods

or Background: Ten patients who were candidates for the abdominal mapping study as part of a research proto col were included. MOLLI and T2p-SSFP sequences were acquired for T1 and T2 mapping, respectively, with (T) and without (NT) cardiac trigger and with simulated trigger (ST, HR 60 bpm). T1/T2 mapping values were extracted using man ually traced regions of Abstract-based Programme 7 Wednesday interest (ROIs) at the hepatic, pancreatic and rena l cortical levels, using the cvi42 software. The results obtained were compared using paired t-tests.

Results

or Findings: Liver T1/T2 mapping showed statistically significan t differences between NT and ST sequences (p=0.003 an d 0.004, respectively), and T2 mapping values turned out to be significantl y different between NT and T acquisition also (p<0.001). Pancreas T1mapping va lues were different between NT vs T (p=0.016) and ST (p=0.037), while T 2mapping values were different between NT and ST only (p=0.029). No diff erences between T and ST sequences were found in liver and pancreas T1/T2 ma pping (p>0.05). Renal T1/T2 mapping did not show significant differences among the three acquisition strategies.

Conclusion

Our results show that cardiac trigger should not be avoided in liver and pancreas mapping with MOLLI and T2p-SSFP sequences, but in order to reduce acquisition time, a simulator could be used. Further analysis on larger sample and with standard ROIs are needed to confirm these data.

Limitations

Sample size Funding for this study: n/a Ethics committee - additional information: Azienda Ospedale Università Padova Author Disclosures: Amalia Lupi: Nothing to disclose Alessia Pepe: Nothing to disclose Giulia Beggiato: Nothing to disclose Marco Pizzi: Nothing to disclose Pajtim Dardeshi: Nothing to disclose Donato Negro: Nothing to disclose Emilio Quaia: Nothing to disclose Irene Toniolo: Nothing to disclose 08:00-09:30 Research Stage 2 Research Presentation Session: Cardiac RPS 103 Applications of cardiac CT Moderator R. Vliegenthart; Groningen/NL ([email protected]) Author Disclosures: Rozemarijn Vliegenthart: Advisory Board: Lifelines, ICAN (Institute for Cardiometabolism and Nutrition); Board Member: ESCR ; Grant Recipient: Siemens Healthineers (institutional research grant) ; Speaker: Siemens Healthineers, Bayer Healthcare, Wiley Long-term exposure to particulate and gaseous air p ollution and coronary atherosclerotic disease assessed by cardia c CT *F. Castillo Aravena*¹, C. Desroche², S. Delaney³, R. Nethery³, P. Thavendiranathan¹, H. Ross¹, K. Hanneman¹; ¹Toro nto, ON/CA, ²Kingston, ON/CA, ³Boston, MA/US ([email protected]) Purpose or Learning Objective: Both fine particulate matter (PM2.5) and nitrogen dioxide (NO2) are associated with cardiova scular mortality. However, the underlying pathophysiological mechanisms are un clear. The purpose of this study was to evaluate the relationship between long-term exposure to these air pollutants and extent of coronary artery disease.

Methods

or Background: Adult patients undergoing cardiac CT between 2012-2023 were retrospectively evaluated. Coronary atherosclerosis was quantified using Agatston coronary artery calcium s cores (CACS). Long-term air pollution exposures were assessed as the averag e of daily direct measurements of PM2.5 and NO2 in the ten-year perio d prior to cardiac CT. Multivariable linear regression models were adjuste d for sex, age, year, distance to monitoring station, and socioeconomic s tatus (neighborhood median household income and employment rate).

Results

or Findings: 11,140 patients were included (52% male, mean age 59±11 years). Median 10-year exposure to PM2.5 was 7.5 (range 4.3–9.2) μg/m3 and NO2 was 13.4 (range 3.2-17.8) parts per bi llion (ppb). Each 1 µg/m3 increase in ten-year PM2.5 exposure was assoc iated with 23.2 higher CACS (β-coefficient 23.2, 95%CI, 5.3-41.0, P=0.011) in una djusted analysis and 19.2 higher CACS (β-coefficient 19.2, 95%CI, 0.7-37.7, P=0.042) in multivariable analysis. Each 1 ppb increase in ten- year NO2 exposure was associated with 5.0 higher CACS (β-coefficient 5.0, 95%CI, 1.9-8.2, P=0.002) in unadjusted analysis; however, this association w as attenuated in multivariable analysis (β-coefficient 1.4, 95%CI, -1.7-4.4, P=0.38).

Conclusion

Higher long-term exposure to fine particulate (PM2. 5) air pollution is associated with higher extent of coronary athero sclerotic disease. The relationship with NO2 was not significant in adjust ed analysis. These results highlight the potential for CT to detect the sequel a of long-term air pollution.

Limitations

CACS does not quantify non-calcified plaque and fur ther study is needed to evaluate relationships with total plaque burden. Funding for this study: None Ethics committee - additional information: University Health Network (CAPCR 24-5344) Author Disclosures: Paaladinesh Thavendiranathan: Nothing to disclose Rachel Nethery: Nothing to disclose Kate Hanneman: Nothing to disclose Heather Ross: Nothing to disclose Chloe Desroche: Nothing to disclose Felipe Castillo Aravena: Nothing to disclose Scott Delaney: Nothing to disclose Video Assisted Informed Consent in Cardiac Imaging: Influence on Patient Anxiety during CT – The VAICICI-trial *R. Gohmann*, S. Mettke, C. F. Lücke, C. D. Kriegho ff, M. Gutberlet; Leipzig/DE ([email protected]) Purpose or Learning Objective: CT is a non-invasive tool for the diagnosis of coronary artery disease (CAD) and preoperative plan ning. However, cardiac CT (cCT) can elicit anxiety, potentially impacting patient compliance and ultimately image quality. This study investigates w hether video-assisted informed consent in cardiac imaging (VAICICI) reduc es patient anxiety during cCT and enhances patient understanding of the exami nation process.

Methods

or Background: This prospective, randomized, controlled trial enrolled 205 patients scheduled for cCT. Patients w ere randomized into three groups: Control (n=69), Video I (n=67), and Video I I (n=69). Video I was an educational video with visuals, subtitles, and voic eover explaining the examination. Video II presented only voiceover and subtitles without visuals. All patients received a standard physician consulta tion. Anxiety and patient satisfaction were measured using visual analog scal es immediately before and after the examination. Statistical analyses include d group-comparisons and multivariate-analysis to examine the influence of d emographic and anamnestic variables.

Results

or Findings: Both Video I and II significantly improved patient understanding, and satisfaction compared to the con trol group (p<0.05). Patients with Video II reported the informed consen t form as more important for understanding than those in the control group (p=0. 023). Satisfaction was higher after watching any video (p=0.020) with sign ificant difference between Video I and II. Anxiety levels did not differ betwe en the groups, though female patients (p=0.008) and those having suspected CAD r eported higher pre- examination-anxiety. Overall, 10 baseline demograph ics were found to be partially explanatory to the response and independe ntly statistically significant, e.g. age and previous CT/MRI-experience.

Conclusion

VAICICI improved patient satisfaction and understan ding of the cCT examination. However, its impact on reported an xiety reduction was limited. The findings suggest that VAICICI enhances the informed consent process. Its influence on compliance and thus image quality remains to be investigated.

Limitations

The Single-center Design And The Relatively High Fa miliarity With Imaging Among The Study Cohort May Limit The G eneralizability Of The Results. Funding for this study: None. Ethics committee - additional information: This Study Was Approved By The Local Ethics Committee (Reference-No.:172/22-ek ). Written Informed Consent Was Waived. Author Disclosures: Robin Gohmann: Nothing to disclose Christian Friedrich Lücke: Nothing to disclose Matthias Gutberlet: Nothing to disclose Christian Dominik Krieghoff: Nothing to disclose Sophia Mettke: Nothing to disclose Impact of Cardiac Computed Tomography Angiography t o prevent major adverse cardiovascular events in patients undergoin g diagnostic work-up for orthotopic liver transplant *F. Santoro*, B. La Delfa, D. Tore, C. Guarnaccia, R. Faletti, G. A. Strazzarino, C. Gaetani, A. Depaoli, P. Fonio; Turin/IT ([email protected]) Purpose or Learning Objective: To evaluate the impact of Cardiac CT Angiography (CCTA) in cardiovascular risk assessmen t to prevent major adverse cardiovascular events (MACE) in patients un dergoing diagnostic work- up before orthotopic liver transplant (OLT). Abstract-based Programme 8 Wednesday

Methods

or Background: Monocentric retrospective study on 140 patients with intermediate to high risk of CAD who underwent CCTA during pre-OLT diagnostic work-up at our Institution from March 20 21 to October 2024. All exams were performed using prospective ECG-gated si ngle heartbeat axial acquisition (0.28 s gantry rotation time, kV and mA set depending on patient BMI, ECG window 40-80% of R-R cycle) with a whole-h eart coverage CT scanner (Revolution CT, GE, USA). 95 patients were classified with non-critical CAD at CCTA. 23 patients with suspicion of critical CAD at CCTA underwent invasive coronary angiography (ICA). Sensitivity, s pecificity, positive predicting value (PPV) and negative predictive value (NPV) wer e calculated.

Results

or Findings: In the 95 patients with negative CCTA no one develo ped MACEs or cardiovascular related complications while on waiting list, nor during or after OLT; 23 patients with critical disease (CA D-RADS 4-5) at CCTA subsequently underwent ICA. In 14 cases stenosis wa s confirmed at ICA and patients underwent percutaneous coronary interventi on (PCI) with Drug Eluting Stent (DES). CCTA in this group of patients had 100 % sensitivity, 91.3% specificity, 100% negative predicting value and 60. 9% positive predicting value.

Conclusion

CCTA has an excellent diagnostic accuracy for cardi ovascular risk stratification in the setting of pre-OLT work- up and it may have a role in preventing MACEs or other cardiovascular events in asymptomatic patients with intermediate to high risk of CAD.

Limitations

Monocentric study, small simple size. Funding for this study: Nothing to discloure Ethics committee - additional information: Not applicable Author Disclosures: Giulio Antonino Strazzarino: Nothing to disclose Alessandro Depaoli: Nothing to disclose Riccardo Faletti: Nothing to disclose Carla Guarnaccia: Nothing to disclose Benedetta La Delfa: Nothing to disclose Federica Santoro: Nothing to disclose Davide Tore: Nothing to disclose Paolo Fonio: Nothing to disclose Clara Gaetani: Nothing to disclose Matters of the heart in stroke - acute phase cardia c CT in intracranial large vessel occlusion stroke for the identificatio n of high-risk imaging biomarkers indicative of a cardioembolic cause *K. Mostafa*, C. Wolf, C. Wünsche, S. Krutmann, M. Both, O. Jansen, H. Seoudy, P. Langguth; Kiel/DE ([email protected]) Purpose or Learning Objective: The purpose of this study is the exploration of imaging biomarkers on acute phase cardiac CT tha t may suggest a cardioembolic etiology in patients with large vesse l occlusion (LVO) stroke in patients without intracardial thrombi or atrial fib rillation in an effort to further specify and weigh the known major and minor cardioe mbolic risk factors.

Methods

or Background: A total of 317 patients with LVO stroke and acute- setting one-step cardiac CT imaging examination wer e retrospectively identified and included in this study. Images were assessed for a total of 15 specific imaging findings according to known minor and major cardioembolic risk factors. Final etiology (TOAST) of LVO stroke was determined by interdisciplinary consensus after full clinical wor kup. Multivariate regression analysis was performed to identify cardiac imaging findings associated with a cardioembolic etiology.

Results

or Findings: Overall, 221 (70%) of the LVO strokes were found to have a cardioembolic etiology by interdisciplinary consensus. After correction for atrial fibrillation and intracardiac thrombi, m ultivariate regression analysis defined dilatative cardiomyopathy (adjusted odds-ra tio (AOR) 37.9), right-to-left shunt (AOR 21.8), valvular implants (AOR 23.7), typ e II and III thrombotic aortic arch (AOR 8.1) and visible myocardial scars (AOR 6.8) as risk factors for a cardioembolic etiology (AUC 0.87, p < 0.05 for al l factors respectively).

Conclusion

In patients with LVO stroke without atrial fibrilla tion or intracardiac thrombi on acute phase cardiac CT imaging, the pres ence of dilatative cardiomyopathy, right-to-left shunt, valvular impla nts, type II and II thrombotic aortic arch and visible myocardial scars on acute p hase CT imaging findings is significantly associated with a cardioembolic strok e etiology.

Limitations

This is a retrospective study with its associated l imitations. The number of patients was low due to the single centre design. Funding for this study: None. Ethics committee - additional information: Ethikkommission der Christian Albrechts Universität Kiel Author Disclosures: Cosima Wünsche: Nothing to disclose Carmen Wolf: Nothing to disclose Patrick Langguth: Nothing to disclose Marcus Both: Nothing to disclose Karim Mostafa: Nothing to disclose Sarah Krutmann: Nothing to disclose Olav Jansen: Nothing to disclose Hatim Seoudy: Nothing to disclose The pericoronary adipose tissue attenuation in CT s trongly depends on kernels and iterative reconstructions C. Lisi¹, *K. Klambauer*², L. J. Moser², V. Mergen² , R. Manka², T. Flohr², M. Eberhard², H. Alkadhi²; ¹Milan/IT, ²Zürich/CH Purpose or Learning Objective: To investigate the influence of kernels and iterative reconstructions on pericoronary adipose t issue (PCAT) attenuation in coronary CT angiography (CCTA)

Methods

or Background: Twenty subjects (16 females; median age 52 years (IQR 48-61)) with atypical chest pain and low risk of coronary artery disease (CAD) who were otherwise healthy and without eviden ce of CAD in photon- counting detector CCTA were included. In each subje ct images were reconstructed with a quantitative smooth (Qr36) and three vascular kernels of increasing sharpness levels (Bv36, Bv44, Bv56). Qua ntum iterative reconstruction (QIR) was either switched-off (QIR o ff) or was used with strengths 2 and 4. The fat-attenuation-index (FAI) of the PCAT surrounding the right coronary artery was calculated in each datase t. Histograms of FAI measurements were created. Intra- and inter-reader agreement were determined. A CT edge-phantom was used to determine the edge-spread- function (ESF) for the same datasets

Results

or Findings: Intra- and inter-reader agreement of FAI were excel lent (ICCs=0.99 and 0.98, respectively). Significant dif ferences in FAI were observed depending on the kernel and iterative reco nstruction strength level (each, p < 0.001), with inter-individual variation up to 34HU. FAI showed also considerable intra-individual variation (average FA I difference 19HU, maximal intra-individual difference 33HU), also depending o n kernels and iterative reconstruction levels. The ESFs showed a reduced ra nge of edge-smoothing with increasing kernel sharpness, causing FAI decre ase. Histogram analyses revealed a narrower peak of PCAT values with increa sing iterative reconstruction levels, causing FAI increase

Conclusion

PCAT attenuation determined with CCTA heavily depen ds on kernels and iterative reconstruction levels both wi thin and across subjects. Standardization of CT reconstruction parameters is mandatory for FAI studies to enable meaningful interpretations

Limitations

Single centre study, including only healthy patient s. Single vendor scanner and software analysis, Funding for this study: No funding was received for this study Ethics committee - additional information: No additional information needed Author Disclosures: Thomas Flohr: Nothing to disclose Costanza Lisi: Nothing to disclose Victor Mergen: Nothing to disclose Matthias Eberhard: Nothing to disclose Lukas Jakob Moser: Nothing to disclose Robert Manka: Nothing to disclose Konstantin Klambauer: Nothing to disclose Hatem Alkadhi: Nothing to disclose Inter-Observer Agreement of the Coronary Artery Dis ease-Reporting and Data System (CAD-RADS) 2.0 *J. H. Lund*, J. Erley, G. Adam, E. Tahir, I. Molwi tz, M. Meyer; Hamburg/DE ([email protected]) Purpose or Learning Objective: In 2022, the Coronary Artery Disease- Reporting and Data System (CAD- RADSTM) was updated to standardize coronary CT angiography (CCTA) reports, requiring h igh inter-observer reproducibility. This study aims to assess the inte r-observer agreement of CAD-RADS 2.0.

Methods

or Background: Patients who underwent CCTA between 2022 and 2024 using a 3rd-generation-dual-source-CT were ind ependently evaluated by three readers with varying levels of experience (1, 3, and 12 years). CAD- RADS 2.0 was used to assess visual grading of plaqu e burden (P1 = mild to P4 = extensive), stenosis degree (CAD-RADS 0 = no s tenosis to CAD-RADS 5 = total occlusion), and modifiers (HRP for high-ris k plaque features, E for exceptions, S for stents, G for grafts, and N for n on-evaluable studies). Inter- observer agreement was measured using intraclass co rrelation coefficients (ICC). Abstract-based Programme 9 Wednesday

Results

or Findings: 100 patients (29% female, age 63 ± 12 years) with a median Agatston score of 267 were included. Observe rs 1, 2, and 3 rated 71%, 60%, and 64% of patients, respectively, as CAD-RADS 3 or above. Inter- observer agreement for plaque burden grading (ICC:0 .92, 95% CI:0.88–0.94, p < 0.001) and stenosis degree (ICC:0.88, 95% CI:0.83 –0.92, p < 0.001) on a per-patient level was excellent. The agreement was also excellent on a per- vessel basis, with the highest for the left anterio r descending artery (ICC:0.90) and the lowest for the left main artery (ICC:0.86). Agreement on modifiers was poor (ICC:0.06, p = 0.410).

Conclusion

CAD-RADS 2.0 demonstrates excellent inter-observer agreement for plaque burden and stenosis grading, but agreeme nt on modifier use is low, likely due to limited use or uncertainty in their a pplication in routine practice.

Limitations

Limited assessment of the modifiers. Experienced ba sed-bias due to limited amount of observer, thus the real-wo rld clinical experience might not be entirely reflected. Funding for this study: Nothing to disclose. Ethics committee - additional information: This study was accepted by the ethics committee of the University Medial Center Ha mburg-Eppendorf (UKE). Author Disclosures: Gerhard Adam: Nothing to disclose Isabel Molwitz: Nothing to disclose Mathias Meyer: Nothing to disclose Jennifer Erley: Nothing to disclose Enver Tahir: Nothing to disclose Jonas H. Lund: Nothing to disclose Preliminary Experience of 60-kVp Tube Voltage Combi ned with Deep Learning Reconstruction Algorithm in Coronary CT An giography X. Wu¹, S. Jiang², Y. Zou², *T. Wang*², G. Zhang², F. Huang¹, P. Liu¹, W. Sun¹, W. He¹; ¹Changsha/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: To explore the clinical value of 60-kVp coronary CT angiography (CCTA) combined with DEep L earning Trained Algorithm (DELTA).

Methods

or Background: Thirty-nine patients (20 male, 58.77 ± 16.26 years, 22.87 ± 3.71 kg/m^2) with suspected coronary artery disease (CAD) were prospectively enrolled. Each underwent both low- (6 0 kVp, 28 ml contrast medium at 2.5 ml/s) and routine-dose CCTA (100 kVp, 44 ml contrast medium at 4.0 ml/s) on a 320-row scanner within 2 weeks. T he routine-dose data were reconstructed using hybrid iterative reconstruction (RD-HIR) and served as the

Reference

standard. Low-dose data were reconstructe d using both HIR (LD- HIR) and DELTA (LD-DELTA). Coronary stenosis in the right coronary artery (RCA), left anterior descending (LAD), and left cir cumflex (LCX) was assessed using CAD-Reporting and Data System (CAD-RADS) scor es. The diagnostic performance of LD-HIR and LD-DELTA in distinguishin g moderate (CAD- RADS<3) to severe (CAD-RADS≥3) stenosis was analyzed via receiver operating characteristic analysis. Signal-noise-rat io (SNR) and contrast-noise- ratio (CNR) on each vessel were also compared.

Results

or Findings: The low-dose CCTA reduced radiation dose by 85.8% compared to the routine-dose acquisition (0.55 ± 0. 09 mSv vs. 3.86 ± 1.25 mSv, p<0.001). In distinguishing moderate to severe stenosis, LD-DELTA demonstrated superior diagnostic performance compar ed to LD-HIR, with area under the curve (AUC) being 1.00 (95% CI, 0.91-1.00 ) versus 0.87 (95% CI, 0.72-0.96) in the RCA, and 1.00 (95% CI, 0.91-1.00) versus 0.78 (95% CI, 0.62-0.90) in the LCX. However, no difference was f ound in the LAD, with both showing an AUC of 0.98 (95% CI, 0.88-1.00). Additio nally, LD-DELTA demonstrated higher SNRs and CNRs compared to LD-HI R (all p<0.001).

Conclusion

The 60-kVp low-dose CCTA acquisition with DELTA sig nificantly reduces radiation dose while maintaining diagnostic performance for assessing coronary stenosis.

Limitations

Not applicable. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was approved by the local institutional review board. Author Disclosures: Peng Liu: Nothing to disclose Tiantian Wang: Investigator: at Central Research In stitute, United Imaging Healthcare Yixuan Zou: Investigator: Central Research Institut e, United Imaging Healthcare Feng Huang: Nothing to disclose Guozhi Zhang: Investigator: Central Research Instit ute, United Imaging Healthcare Weiling He: Nothing to disclose Wenjie Sun: Nothing to disclose Senyang Jiang: Investigator: Central Research Insti tute, United Imaging Healthcare Xi Wu: Nothing to disclose Quantitative assessment of early changes in myocard ial extracellular volume during postoperative adjuvant chemotherapy i n breast cancer patients by dual-layer spectral detector CT *H. Wu*, Y. Huang, W. Deng, Y. Wang, Q. Xiao, Y. Gu ; Shanghai/CN ([email protected]) Purpose or Learning Objective: This study utilized dual-layer spectral detector CT (DLCT) technology to simultaneously ass ess cardiac function and myocardial myocardial extracellular volume (ECV) , characterizing the changes in parameters early during breast cancer therapy.

Methods

or Background: Forty female breast cancer patients who underwent postoperative adjuvant chemotherapy were prospectiv ely enrolled and underwent baseline and 3-month-postchemotherapy car diac CT (CCT) and ultrasound cardiography (UCG). Global ECV of the le ft ventricle (LV) were measured based on an iodine map of the late enhance ment phase of DLCT. Changes in cardiac function parameters and global E CV from baseline to the 3-month follow-up were analyzed. Correlation coeffi cients between the changes in cardiac function parameters and global E CV were calculated.

Results

or Findings: LV ejection fraction by UCG (UCG-LVEF) and by CCT (CCT-LVEF) did not significantly change between bas eline and 3 months. Heart rate (HR) increased over 3 months of follow-u p. After normalization to body surface area (BSA), cardiac output (CCT-CO ind exed) and LV late (active) filling volume (LVLFV indexed) significant ly increased (P<0.01), while LV early (passive) filling volume (LVEFV indexed) a nd LVEFV/LVLFV decreased significantly at the 3-month follow-up (P <0.05). Global ECV were elevated significantly at 3 months (25.4±2.4 vs. 27.3±2.7, P<0.01). Although changes in global ECV were not associated with chan ges in LVEFs, global ECV change were moderately correlated with changes in LV end-diastolic volume / BSA (CCT-LVEDV indexed) (r=0.52, P<0.01), LV stroke volume / BSA (CCT-LVSV indexed) (r=0.56, P<0.01), CCT-CO ind exed (r=0.40, P=0.01) and LVEFV indexed (r=0.41, P<0.01) .

Conclusion

CCT-derived ECV can be used to evaluate myocardial changes in the early stage of chemotherapy before LVEF sign ificantly decreases. The increases in global ECV were not correlated with LV EFs. The changes in myocardial global ECV were moderately correlated wi th cardiac function parameters.

Limitations

The small sample. Funding for this study: None Ethics committee - additional information: The study was approved by the Ethics Committee of Fudan University Shanghai Cance r Center, and written informed consent was obtained from all subjects in the study. Author Disclosures: Honglin Wu: Nothing to disclose Yan Huang: Nothing to disclose Yajia Gu: Nothing to disclose Yu Wang: Nothing to disclose Qin Xiao: Nothing to disclose Weiwei Deng: Nothing to disclose Cardiac calcifications detected on planning CT are major predictors of long-term cardiotoxicity after radiotherapy for bre ast cancer K. B. Dimayuga, *A. Belardo*, L. Perna, A. Fodor, P . Mangili, A. Del Vecchio, N. Di Muzio, C. Fiorino; Milan/IT ([email protected]) Purpose or Learning Objective: Breast cancer (BC) patients undergoing radiotherapy (RT) may experience long-term cardioto xicity. In modern series, delivering low dose to the heart, non-dosimetry pre dictors are emerging. The purpose was to test if cardiac calcifications (CAC) at planning CT, suggested as potential predictors, are associated with long-t erm cardiac events.

Methods

or Background: Planning CT and clinical information of 1172 consecutive patients treated at our hospital (2009- 2017) were available (right:569, left:603). The heart of all patients wa s automatically segmented using a previously validated AI-based tool (MIM Pro tegé & MIM assistant) and the mean heart dose (MHD) was assessed. CAC were au tomatically extracted by applying a home-made, validated, Python script e xtracting the Agatson score (AS) and the CAC overall volume. Their associ ation with the risk of cardiac events was tested by logistic regression, i ncluding the potential combined effect of MHD and available clinical param eters.

Results

or Findings: With a median follow-up of 8 years (range: 5-15), 3 2 patients experienced cardiac events. AS/CAC volumes were the most significant predictors (p<0.0001), with similar per formances. Age, laterality (left/right), concomitant chemotherapy, obesity and hypertension were also significant at univariate analysis. MHD encoded usi ng the best cut-off (1Gy, mostly representing laterality) was also predictive . The best multivariate model combined MHD>1Gy, age and CAC volume (AUC=0.79, p<0 .0001, calibration plot: m=1.506, q=-0.007), being CAC volume the stro ngest predictor (OR: 1.0008/mm3, p<0.0001).

Conclusion

CAC load was the most important factor in cardiac r isk stratification after BC RT in a modern series. Abstract-based Programme 10 Wednesday

Limitations

Events are not recovered from a registry. Then, the risk of missing events is not negligible. Funding for this study: None Ethics committee - additional information: All respected. Author Disclosures: Lucia Perna: Nothing to disclose Nadia Di Muzio: Nothing to disclose Claudio Fiorino: Nothing to disclose Alfonso Belardo: Nothing to disclose Antonella Del Vecchio: Nothing to disclose Paola Mangili: Nothing to disclose Kerby Bjorn Dimayuga: Nothing to disclose Andrei Fodor: Nothing to disclose One-Scan Acquisition of Coronary CT Angiography and CT Aortography Using Photon Counting Detector CT *H. Kato*, S. Araki, S. Nakamura, A. Yamazaki, N. K ato, Y. Ichikawa, H. Sakuma, K. Kitagawa; Tsu/JP Purpose or Learning Objective: For patients with aortic diseases, Coronary CT angiography (CCTA) is performed for preoperative evaluation of coronary artery disease (CAD), often in combination with CT aortography (CTAO) within the same examination. However, the shared imaging f ield in both CCTA and CTAO results in overlapping radiation exposure to t he thoracic region. Recently, a dual-source photon-counting detector CT (PCD-CT) has emerged, with its capability to perform high-pitch helical s canning to enable one-scan acquisition of CCTA and CTAO. This study aimed to c ompare the radiation dose and image quality of one-scan CCTA and CTAO im aging with those of separate CCTA and CTAO imaging.

Methods

or Background: This study included 40 patients who underwent CCTA and CTAO for preoperative CAD screening for ao rtic disease: 22 had separate, 18 had one-scan CCTA and CTAO. A low tube potential of 70 or 90 kVp was used for all patients. CT dose-length produ ct (DLP) data for CCTA and CTAO was collected. Image quality for each coro nary segment and the aorta was assessed using a four-point scale (excell ent, good, fair, non- diagnostic).

Results

or Findings: The mean DLP was significantly lower with the one-s can CCTA and CTAO protocol (160.5±36.5 mGy*cm) compared to the separate scans (716.0±203.9 mGy*cm) (P<0.001). There was no significant difference between the separate and one-scan protocols in the assessment of image quality for CCTA (P=0.17) and CTAO (P=0.92), and in both protocols, over 95% segments/cases in the CCTA and CCAO images were rated as having good or excellent image quality.

Conclusion

PCD-CT, with high-pitch helical scanning and low tu be voltage, enabled one-scan acquisition of CCTA and CTAO, sign ificantly reducing radiation exposure while maintaining high image qua lity.

Limitations

There is no reference standard such as coronary ang iography. Funding for this study: No funding was provised for this study. Ethics committee - additional information: Clinical Research Ethics Review Committee of Mie University Hospital (approval No. H2019-207) Author Disclosures: Kakuya Kitagawa: Nothing to disclose Hajime Sakuma: Nothing to disclose Suguru Araki: Nothing to disclose Yasutaka Ichikawa: Nothing to disclose Noriyuki Kato: Nothing to disclose Hiroaki Kato: Nothing to disclose Akio Yamazaki: Nothing to disclose Satoshi Nakamura: Nothing to disclose 08:00-09:30 Research Stage 3 Research Presentation Session: Oncologic Imaging RPS 116 Staging, metastases and response assessment Moderator R. Perez-Lopez; Barcelona/ES Undifferentiated pleomorphic sarcoma: Building an e ffective multi- parametric MRI (mpMRI) predictive treatment respons e model to replace RECIST *R. F. Valenzuela*, B. Amini, E. Duran-Sierra, J. E . Madewell, M. Antony, C. M. Costelloe, W. Murphy; Houston, TX/US ([email protected]) Purpose or Learning Objective: Undifferentiated pleomorphic sarcoma (UPS) is the largest soft-tissue sarcoma subgroup. Post-t herapeutically, UPS demonstrates hemosiderin deposition, fibrosis, and calcification. This study aimed to establish the clinical value of multiparam etric MRI (mpMRI) for predicting UPS response.

Methods

or Background: An IRB-approved retrospective study included 33 extremity UPS patients with pre-operative mpMRI, in cluding diffusion-weighted imaging (DWI), contrast-enhanced susceptibility-wei ghted imaging (CE-SWI), and perfusion-weighted imaging with dynamic contras t-enhancement (PWI/DCE), and surgical resection February 2021-May 2023. Lesions were visually classified on CE-SWI into one of 6 morphol ogy patterns. On PWI/DCE, lesions were classified into one of 6 patterns, and time-intensity curves (TICs) were classified as types I-V. Patients were divided into three groups based on the percentage of pathology-assessed treatment effe ct (PATE) in the surgical specimen: Responders (>=90% PATE, n=16), partial-re sponders (31-89% PATE, n=10), and non-responders (<=30% PATE, n=7). Receiver operating characteristic (ROC) analysis of classification mod els based on CE-SWI and PWI/DCE patterns and TICs compared responders vs. p artial/non-responders.

Results

or Findings: At post-radiation therapy (PRT), a CE-SWI Complete Ring pattern was observed in 71% of responders (p=7 .71x10-6). On PWI/DCE images, 79% of responders displayed a Capsular patt ern (p=1.49x10-7), and 100% demonstrated a TIC-type II (p=8.32x10-7). RECI ST could not separate responders from partial/non-responders; all demonst rated 100% stability at PRT and pseudoprogression at PC. ROC analysis compa ring responders (n=14) vs. partial/non-responders (n=16) at PRT sho wed that the model combining the PWI/DCE TIC-type II, PWI/DCE Capsular pattern and CE-SWI Complete Ring pattern yielded the highest classific ation performance (AUC=0.99).

Conclusion

mpMRI-derived features can help assess UPS treatmen t response. Observing a pre-operative PWI/DCE TIC-typ e II, PWI/DCE Capsular pattern, and CE-SWI Complete Ring pattern can poten tially predict successfully treated UPS patients with >=90% PATE, outperforming RECIST.

Limitations

Limitations include a small sample (n=33) and manua l, time- consuming tumor VOI segmentation. Funding for this study: The John S. Dunn, Sr. Distinguished Chair in Diagnostic Imaging. M.R Evelyn Hudson Foundation Endowed Professorship. Ethics committee - additional information: Not applicable Author Disclosures: Elvis Duran-Sierra: Nothing to disclose Behrang Amini: Nothing to disclose Colleen M Costelloe: Nothing to disclose Raul Fernando Valenzuela: Nothing to disclose William Murphy: Nothing to disclose Mathew Antony: Nothing to disclose John Edward Madewell: Nothing to disclose Abstract-based Programme 11 Wednesday Redefining radiologic responses in high-risk soft-t issue sarcomas treated with neoadjuvant chemotherapy. Final results of ISG -STS 1001, a randomized clinical trial *A. Vanzulli*, R. Vigorito, C. Buonomenna, P. Verde rio, S. Pasquali, P. G. Casali, C. Morosi, S. Stacchiotti, A. Gronchi ; Milan/IT ([email protected]) Purpose or Learning Objective: We report the results of the pre-planned secondary analysis of radiologic responses (RR) of ISG-STS 1001, a randomized clinical trial comparing anthracycline + ifosfamide (AI) vs. histology-tailored (HT) neoadjuvant chemotherapy fo r primary localized high- risk soft-tissue sarcomas of the extremities/trunk wall.

Methods

or Background: Patients with undifferentiated pleomorphic sarcoma (UPS), leiomyosarcoma (LMS), malignant peripheral n erve sheath tumor, synovial sarcoma or myxoid liposarcoma (MLPS) were randomized, whereas patients with myxofibrosarcoma, pleomorphic liposar coma, pleomorphic rhabdomyosarcoma or unclassified sarcoma were alloc ated in the observational arm (O) and treated with AI. Patients with UPS, LMS or MLPS needing concurrent preoperative radiotherapy were i ncluded in O. We evaluated associations between: Disease-Free Surviv al (DFS)/Overall Survival (OS) and centrally reviewed RR, assessed with RECIS T 1.1 and as percent dimensional variation (D; both dichotomized and con tinuous); DFS/OS and histology; RR and histology.

Results

or Findings: 435 patients were included (287 randomized, 148 observed). The analysis of RR comprised 236 patient s (154 randomized, 82 observed) with measurable disease and available for central review. RECIST best responses were: 28 (11.9%) partial response (P R), 195 (82.6%) stable disease (SD) and 13 (5.5%) progressive disease (PD) . RECIST significantly correlated with DFS (PD-vs-PR: HR 8.18, 95% CI 2.96 -22.58; SD-vs-PR: HR 2.96, 95% CI 1.30-6.75) and OS (PD-vs-PR: HR 12.61, 95% CI 3.40-46.84; SD-vs-PR: HR 4.24, 95% CI 1.34-13.47). The median v alue of D was -1.6%. Patients with D > -1.6% had worse clinical outcomes than those with D < -1.6% (DFS: HR 1.73, 95% CI 1.19-2.50; OS: HR 1.86, 95% C I 1.21-2.86). D in continuous scale inversely correlated with DFS (HR 1.53, 95% CI 1.25-1.87) and OS (HR 1.78, 95% CI 1.41-2.25).

Conclusion

Dimensional variation in continuous scale predicted the proportional efficacy of treatment irrespective of tumor histology.

Limitations

Nothing to disclose. Funding for this study: Pharmamar® provided trabectedin for the HG-MLPS cohort. The study was partially funded through a Eu ropean Union grant (EUROSARC FP7 278472). In addition, the French site s were supported by NETSARC, LYRICAN (LYRICAN [INCA-DGOS-INSERM 12563]) and DEPGYN (RHU4). Ethics committee - additional information: The trial protocol and all amendments were approved by the independent ethics committee at each trial center. Author Disclosures: Sandro Pasquali: Nothing to disclose Paolo Giovanni Casali: Nothing to disclose Paolo Verderio: Nothing to disclose Alessandro Gronchi: Nothing to disclose Andrea Vanzulli: Nothing to disclose Carlo Morosi: Nothing to disclose Ciriaco Buonomenna: Nothing to disclose Raffaella Vigorito: Nothing to disclose Silvia Stacchiotti: Nothing to disclose Can a fast T2-Dixon sequence surpass the time obsta cle of whole-body MRI in the evaluation of skeletal metastases? N. Magdi, *M. Elmansy*, M. Elhawary, A. Sultan; Man soura/EG Purpose or Learning Objective: Our study was conducted to elucidate the role of the T2-Dixon sequence as a rapid alternativ e to the standard Whole- body magnetic resonance imaging (WB-MRI) protocol w ith the assessment of its diagnostic accuracy and comparability to the es tablished methodology.

Methods

or Background: This prospective study included 30 patients with primary solid malignancies who underwent WB-MRI. Th e sequences obtained were T1WI, STIR, and T2-Dixon (fat-only and water-o nly images). Skeletal metastases were evaluated in each sequence. Results were compared between the T1-STIR combination and T2-Dixon fat an d water reconstructions.

Results

or Findings: The sensitivity of fat and water reconstructions fr om a single T2-Dixon in the detection of lytic skeletal metastases was marginally superior to a combination of T1WI and STIR sequence s (0-7%). Detection of mixed lesions demonstrated equally high sensitivity in both protocols. Sclerotic metastases detection in WB-MRI showed low sensitivi ty in both protocols. However, specificity surpassed 95% for all lesion t ypes in both protocols. Overall image quality was favored (in 87-90% of pat ients ) in T2-Dixon images. The overall estimated acquisition timing using T2-D ixon appeared to be approximately half that of the standard T1-STIR com bination.

Conclusion

WB-MRI using T2-Dixon fat and water reconstructions showed similar accuracy to T1WI and STIR combination in th e evaluation of skeletal metastases in patients with primary solid cancers w ith significantly shorter acquisition time.

Limitations

Few skull and humeri lesions with limited assessmen t of the sclerotic lesions due to high fals negative results . Lack of quantitative analysis of signal to noise an d contrast to noise ratio. Funding for this study: No funding Ethics committee - additional information: This study was approved by the Research Ethics Committee of the Faculty of Medicin e at Mansoura University in Egypt on 10 /1 /2023; reference number of approv al: MS.22.12.2250 Author Disclosures: Amina Sultan: Nothing to disclose Noha Magdi: Nothing to disclose Mohammed Elhawary: Nothing to disclose Mostafa Elmansy: Nothing to disclose Impact of reader experience on reader agreement for whole-body MRI staging of oesophageal cancer *P. Chapellier*¹, S. W. Soo², O. Westerland², A. Gr een², S. Gourtsoyianni³, V. Goh²; ¹Lausanne/CH, ²London/UK, ³Athens/GR Purpose or Learning Objective: Whole-body MRI (WB-MRI) may be valuable alternative to standard imaging pathways for stagin g. We evaluated how reader experience impacts agreement for tumour-node -metastasis (TNM) staging of oesophageal cancer.

Methods

or Background: Following ethical approval, prospective patients underwent WB-MRI (T2-weighted, diffusion-weighted, T1-weighted post- contrast) alongside standard imaging (contrast-enha nced CT, 18F-FDG PET/CT ± EUS). WB-MRI was staged using AJCC TNMv8 b y four readers from different countries with different levels of experi ence: subspecialty vs. non- subspecialty trained; in-training vs. staff radiolo gists. Inter-reader agreement was assessed using kappa statistics. For each reade r, agreement with a

Reference

standard of final tumour board stage, sur gical pathology and clinical follow up was obtained.

Results

or Findings: 29/30 (97%) patients had adenocarcinoma; 25/30 (83% ) had ≥T3 stage; 27/30 (90%) had locoregional lymphadenopa thy; 12/30 (40%) were metastatic. 22/30 (74%) received chemotherapy only; 8/30 (27%) had surgery, 63% with neoadjuvant treatment. Compared t o reference standard, agreement for T- and N-stage was highest for the tw o gastrointestinal-trained radiologists (T-stage: κ =0.516, κ =0.824; N-stage: κ =0.434, κ =0.589, respectively). Agreement for M-stage was highest fo r the oncology-trained radiologist (κ =0.795). Detection of lung metastases was limited on MRI. Inter- reader agreement was also highest for gastrointesti nal-trained radiologists (T- stage: κ = 0.624; N-stage: κ = 0.822). Agreement across TNM staging was lowest with the in-training radiologist who had no exposure to WB-MRI staging.

Conclusion

WB-MRI could be valuable alternative for initial TN M staging of oesophageal cancer, but reliable interpretation app ears to be related to subspecialty experience and level of training.

Limitations

WB-MRI was a relatively long acquisition, and not s uit all patients. Nodal disease was not just based on size measuremen t. MRI is limited for some sites eg.lung Funding for this study: This project was supported by the National Institut e for Health and Care Research (MIHR) Biomedical Rese arch Centre at Guy’s & St Thomas’ Hospitals and King’s College London. Ethics committee - additional information: The study was approved by the Research Ethics Committee (IRAS ID 107508, 12/LO/17 54). Author Disclosures: Olwen Westerland: Nothing to disclose Adrian Green: Nothing to disclose Sofia Gourtsoyianni: Nothing to disclose Pauline Chapellier: Nothing to disclose Vicky Goh: Nothing to disclose Suet Woon Soo: Nothing to disclose Observer variability and reproducibility of bone ma rrow metastasis biomarkers on MRI *C. Sattin*¹, C. Pizzi¹, M. Kosmin², W. McGuire³, A . Makris², N. J. Taylor², G. Petralia¹, A. R. R. Padhani³; ¹Milan/IT, ²London /UK, ³Northwood/UK ([email protected]) Purpose or Learning Objective: To document inter- and intra-observer variability and test-retest reproducibility of quan titative MRI metastasis biomarkers.

Methods

or Background: Whole-body MRI was performed in women with bone-predominant metastatic breast cancer on a 1.5T MRI system. Paired reproducibility scans were done in 14 patients afte r repositioning. Observer variability was assessed in 10 patients after a two -week washout period. Two trained radiologists delineated up to 5 bone metast ases per patient. Quantitative biomarkers (lesion sizes, apparent dif fusion coefficient (ADC), relative fat fraction (rFF%) and b-900 s/mm2 signal -to-muscle ratio [SMR]) for each lesion were averaged per patient. Data logn tr ansformed after testing for Abstract-based Programme 12 Wednesday normality. Within-patient coefficient of variation (wCV) and variance ratio were calculated. The repeatability (R-value; mean 95% co nfidence) for a single patient was calculated.

Results

or Findings: Reproducibility analysis (2 radiologist consensus; 14 patients; 41 lesions): Variance ratios were >15 for all biomarkers. Size had wCV of 3.5% and R-value of 9.5%; SMR had wCV of 7.4 % and R-value of 19.8%; ADC had wCV of 2% and R-value of 5.2%; rFF% had wCV of 9.4% and R-value of 24.9%. Inter- and intra-observer variabi lity (for 2 independent radiologists; 10 patients; different lesions) were similar (ICC >0.8) indicating consistent reader performance.

Conclusion

Higher reproducibility test variance ratios and int erclass correlations of clinically relevant biomarkers indi cate the reliability of radiological assessments regardless of the observer . Changes in aggregated lesion size of >3mm, SMR of > 1.4 au, ADC of >53 µm 2/s and rFF% of > 2.7% can be used as bone tumour response biomarkers in w omen with metastatic breast cancer.

Limitations

This abstract represents a preliminary part of a la rger study that aims to evaluate whether changes in quantitative an d qualitative WB-MRI biomarkers could be predictive of outcomes in patie nts with bone-predominant MBC on SACT. Funding for this study: This work was funded by the Paul Strickland Scanner Centre Charity (UK registered charity number 298867 ) and the Fighting Breast Cancer (UK registered charity number 1091882). Ethics committee - additional information: All procedures permorfed in this study (ClinicalTrials.gov identifier: NCT03266744) followed the ethical standards of the National Health Service Health res earch Authority East of England-Cambridge East Research Committee and the 1 964 Helsinki Declaration and its later amendments or comparable ethical standards. Author Disclosures: N. Jane Taylor: Nothing to disclose Michael Kosmin: Nothing to disclose Giuseppe Petralia: Nothing to disclose Will McGuire: Nothing to disclose Caterina Pizzi: Nothing to disclose Anwar R. R. Padhani: Nothing to disclose Caterina Sattin: Nothing to disclose Andreas Makris: Nothing to disclose Spectral CT imaging for assessment of metastases in melanoma patients: Multi-reader evaluation *C. Nelles*, P. Rauen, T. M. Dratsch, D. Maintz, J. Kottlors, N. Große Hokamp, D. Zopfs, T. Persigehl, S. Lennartz; Cologne/DE Purpose or Learning Objective: To investigate the sensitivity, specificity and qualitative assessment of spectral image reconstruc tions for metastases in melanoma patients in a large-scale, multi-reader ev aluation.

Methods

or Background: In total, 308 patients with melanoma, 95 patients with metastases and a control group of 213 patients without metastases, who underwent staging CT of the chest and abdomen on a dual-layer dual-energy CT system (dlDECT) were retrospectively included. C onventional images (CI), iodine overlays (IO) and virtual monoenergetic imag es at 40 keV (VMI40keV) were reconstructed. Six radiologists (three experie nced, three less experienced) evaluated all cases in a CI-based sess ion and a session based on a combination of CI, IO and VMI40keV. Readers we re asked to binarily determine presence of metastases in specific tissue s and to indicate diagnostic certainty and lesion delineation on 5-point Likert scales.

Results

or Findings: Sensitivity for detection of metastases in skeletal muscle and peritoneum was significantly higher for the spe ctral assessment (for skeletal muscle 70% vs. 61%; for peritoneum 76% vs. 62%, both: p < 0.05). For subcutaneous metastases, there was a significan t increase in specificity (92% vs. 89%, p < 0.05), however accompanied with a significant decrease in sensitivity (79% vs. 85%, p < 0.05). Diagnostic cer tainty was rated significantly higher for spectral images than CI for 100% (6/6) o f the assessed tissues, whereas improvements in lesion delineation were not ed for skeletal muscle, subcutaneous tissue and pancreas.

Conclusion

In melanoma patients, the benefit of dlDECT-derived spectral reconstructions depends on the assessed tissue. Whi le assessment of skeletal muscle and peritoneal metastases was significantly improved, low or absent iodine uptake of subcutaneous lesions led to false negatives and a consecutive decrease in sensitivity.

Limitations

-Retrospective, monocenter study design -The result s may not be generalisable to all different DECT platforms Funding for this study: Funding was provided by the German Research Foundation (DFG, FI 773/15-1). Ethics committee - additional information: After reviewing the study design, the institutional review board waived the need for informed patient consent for this retrospective single-center study. Author Disclosures: Simon Lennartz: Speaker: Amboss GmbH Author: Amboss GmbH Nils Große Hokamp: Consultant: Bristol-Myers Squibb ; BeiGene Speaker: Philips Healthcare; Amboss GmbH Research/Grant Supp ort: Philips Healthcare David Zopfs: Research/Grant Support: Philips Health care David Maintz: Nothing to disclose Christian Nelles: Nothing to disclose Philip Rauen: Nothing to disclose Jonathan Kottlors: Nothing to disclose Thomas Markus Dratsch: Nothing to disclose Thorsten Persigehl: Nothing to disclose Accuracy of DOTATATE PET CT versus DOTATATE PET MR in the Assessment of Multifocal Small Bowel Neuroendocrine Neoplasm *A. Keane*¹, H. Takahashi², C. Zhang³, C. Thiels², P. Navin²; ¹Dublin/IE, ²Rochester, MN/US, ³Pheonix, AZ/US ([email protected]) Purpose or Learning Objective: Small bowel neuroendocrine neoplasms (sbNENs) provide challenges in diagnosis and manage ment. Current gold standard treatment is open resection with manual sm all bowel palpation as 45- 50% of sbNEN have synchronous lesions. Advancements in nuclear medicine have improved preoperative assessment, potentially negating manual palpation and allowing for laparoscopic approach. D OTATATE PET CT (DPCT) and PET MR can identify somatostatin receptors, fre quently expressed on sbNENs. We aim to assess the accuracy of DPCT versu s PET MR in determining multifocality of sbNENs.

Methods

or Background: Multicentre retrospective analysis was performed on patients with sbNEN who underwent open small bow el resection between January 2016 to August 2022 and had either preopera tive DPCT or PET MR. Blinded retrospective review of images for small bo wel lesions was performed by two fellowship-trained radiologists and compared to postoperative pathology reports. Final radiology diagnosis was attained by consensus agreement. Descriptive statistics, sensitivity, specificity, p ositive predictive value (PPV), negative predictive value (NPV), accuracy, and area under the receiver operator characteristic curve (AUROC) of DPCT and P ET MR were compared.

Results

or Findings: Seventy-eight patients met inclusion criteria. Fift y-six (71.8%) had preoperative DPCT, twenty-two (28.2%) h ad PET MR. Forty-eight patients (61.5%) had multiple (> 1)sbNENs on final pathology. Thirty-six (75.0%) were identified on preoperative imaging. DP CT and PET MR demonstrated sensitivity of 85.0% and 58.0% respect ively, specificity of 81.0% and 60.0%, PPV of 87.9% and 63.6%, NPV of 77.3% and 54.6%, and accuracy of 83.6% and 59.0% for the presence of multifocal d isease. AUROC was 0.8 for DPCT and 0.5 for PET MR.

Conclusion

DPCT demonstrated higher accuracy in identifying mu ltifocal sbNEN versus PET MR. However, as the NPV is insuffi cient, we conclude that preoperative DPCT should not replace open explorati on.

Limitations

Retrospective study, underpowered due to rarity of sbNENs. Funding for this study: Nil Ethics committee - additional information: IRB (Institutional Review Board) Author Disclosures: Cornelius Thiels: Nothing to disclose Hiroaki Takahashi: Nothing to disclose Anna Keane: Nothing to disclose Chi Zhang: Nothing to disclose Patrick Navin: Nothing to disclose Discord Dilemmas in Lung Cancer Clinical Trials: Na vigating Reader Variability in Response Assessment *H. Beaumont*¹, R. Gill², N. Faye¹, A. Iannessi¹; ¹ Valbonne/FR, ²Boston, MA/US ([email protected]) Purpose or Learning Objective: In lung cancer trials, blinded independent central response assessment with double reads is ch allenging and prone to interobserver variability. We analyzed the patterns of discordance in reporting Progressive Disease (PD) and performed a root cause analysis.

Methods

or Background: We retrospectively analyzed data from five clinical trials evaluating 1932 lung cancer patients treated with targeted and immune therapies, read by 17 central readers. Progressive Disease was defined based on RECIST 1.1 criteria. The RECIST components were the Sum of tumors Diameter (SOD), the unequivocal progression of the non-Target Lesions (NTL) and the detection of New Lesion (NL). We analyzed t he RECIST components according to 1) Concordant/discordant PD detection; 2) Positive Predictive Value (PPV) of declaring PD; 3) Offset versus singl e reader detection.

Results

or Findings: Discordance in PD was observed in 39.2% (675/1718) of patients, with adjudication of PD for 62.5% (422/67 5) based on 44.8% (95%CI: 39.9, 49.8) new lesion detection, 28.3% (95%CI: 24. 0, 33.0) significant increase of SOD and 12.6% (95%CI: 9.5, 16.2) unequi vocal non-target Abstract-based Programme 13 Wednesday progression. For 54.2% of concordant PD, at least o ne reader involved more than one RECIST component. The PPV for increased SO D was 0.59, rising to 0.89 when multiple RECIST components were involved. In 49.2% of discrepant cases, PD was reported with a delay of one cycle in majority of cases (80%). Confirmation rate for NLs in the lungs was lowest ( 40.6%) and new nodal lesions was highest (88.4%).

Conclusion

Discordance among trained central readers in lung c ancer trials is common. New lesion detection is pivotal in the d etection of PD, also the main cause of discordances. Involving multiple RECI ST components improves the reliability of assessments. When relying on NL only, the detection of extra pulmonary lesion is more reliable.

Limitations

No limitations Funding for this study: No funding Ethics committee - additional information: Median Technologies institutional ethics committee approved the study, informed conse nt was not required for this retrospective analysis Author Disclosures: Nathalie Faye: Consultant: Median Technologies Madam Ritu Gill: Nothing to disclose Antoine Iannessi: Employee: Median Technologies Hubert Beaumont: Employee: Median Technologies Detection of abdominal metastases in brain tumor pa tients following ventriculoperitoneal shunting *N. Plakhotina*, A. V. Smirnova, K. Boiko, V. Bikul ov, P. Ivanov; Saint-Petersburg/RU ([email protected]) Purpose or Learning Objective: To develop a diagnostic algorithm and identify imaging patterns for metastatic abdominal lesions in children with brain tumors following shunt placement.

Methods

or Background: Clinical cases of shunt-associated metastasis of primary brain tumors in children, identified throug h CT, MRI, and laparoscopy, with morphological confirmation.

Results

or Findings: Acute abdominal pain developed in children with embryonal tumors (2 patients with medulloblastoma, 1 patient with ATRT) and 1 child with ependymoma in remission, as well as du ring chemotherapy. CT scans revealed multiple nodules of varying sizes wi thin the heterogeneous adipose tissue of the abdominal cavity. MRI identif ied multiple solid isointense tumors on both T1- and T2-WI, exhibiting significan t contrast enhancement and diffusion restriction. Tumor sizes ranged from 3 to 15 mm. A notable feature was the predominant spread in the interintestinal s paces, as well as the subdiaphragmatic and subhepatic regions, which comp licates visualization. In one of the cases, imaging did not yield conclusive results, and metastases were confirmed only at autopsy.

Conclusion

Although shunt-associated metastasis in CNS tumors is extremely rare, it carries a very poor prognosis. D ynamic monitoring of children with brain tumor with a shunt system should include abdominal examinations (ultrasound or MRI). The development of an acute ab domen requires urgent evaluation to rule out tumor presence, including la paroscopy if indicated. Reports of such cases suggest that ventriculoperito neal shunting for occlusive hydrocephalus should be considered only as a last r esort when temporary external drainage is not feasible.

Limitations

N/A Funding for this study: N/A Ethics committee - additional information: N/A Author Disclosures: Konstantin Boiko: Nothing to disclose Vyacheslav Bikulov: Nothing to disclose Nadezhda Plakhotina: Nothing to disclose Pavel Ivanov: Nothing to disclose Alina Vyacheslavovna Smirnova: Nothing to disclose Imaging treatment response in High Grade Serous Ova rian Cancer: Metabolic imaging vs Cell death imaging *M. L. Chia*, K. Brindle; Cambridge/UK Purpose or Learning Objective: Given the poor therapy response noted in advanced stage High Grade Serous Ovarian Cancer (HG SOC) patients and the lack of fast and reliable treatment response mo nitoring methods, there is a need to predict treatment response earlier. This pr oject investigated the potential of metabolic and cell death imaging techn iques to detect early treatment response to standard-of-care chemotherapy in HGSOC patients.

Methods

or Background: HGSOC cells, derived from the ascites of stage 3-4 HGSOC patients, were maintained as patient derived organoids(PDO) and implanted into mice subcutaneously. The resulting t umours were imaged with various imaging techniques. Metabolic imaging techn iques included MRS (hyperpolarized [1-13C]pyruvate metabolism) and PET (measurements of 2- deoxy-2-[fluorine-18]fluoro-D-glucose uptake). Cell death imaging techniques included diffusion-weighted 1H MRI (DWI) and 2H MRS I measurements of [2,3-2H2]fumarate metabolism. PDO 2(carboplatin sen sitive) and PDO 5(Carboplatin resistant) tumour models were treated with i.v. Carboplatin (50mg/kg) or drug vehicle weekly, with imaging at b aseline and weekly thereafter.

Results

or Findings: Both metabolic imaging techniques were successful i n discriminating responding from non-responding tumou rs to Carboplatin before there was a change in tumour volume. The techniques for detecting cell death were not as sensitive for detecting treatment respo nse, which may reflect a slow accumulation of dead cells post treatment, a l ack of knowledge of when the rate of cell death increases post treatment and immune clearance of dead cells.

Conclusion

Imaging with hyperpolarized [1-13C]pyruvate has the potential to be used in the clinic to detect the early treatment response in HGSOC patients.

Limitations

We only tested Carboplatin but other chemotherapies or combination treatment regimens would also be import ant for investigation as they might produce a greater and faster increase in cell death, possibly allowing cell death detection techniques to be more successful. This will be part of future work. Funding for this study: Cancer Research UK Cambridge Institute Core funding Ethics committee - additional information: Na Author Disclosures: Ming Li Chia: Nothing to disclose Kevin Brindle: Nothing to disclose 08:00-09:30 Research Stage 4 Research Presentation Session: Musculoskeletal RPS 110 AI, radiomics and other technologies supporting MSK diagnostics Moderator V. Mascarenhas; Lisbon/PT ([email protected]) AI-Driven SuperResolution reconstruction for high-q uality, fast MR imaging of the lumbar spine: enhanced image clarity for pathology detection *R. Hahnfeldt*¹, R. A. Terzis¹, T. M. Dratsch¹, J. Bremm¹, P. Rauen¹, K. Weiss², D. Maintz¹, G. Bratke¹, A-I. Iuga¹; ¹Cologne/DE, ²H amburg/DE Purpose or Learning Objective: The aim of this study was to investigate whether a 2D MRI lumbar spine protocol with an AI-b ased SuperResolution reconstruction method meets the requirements for cl inical diagnostic purposes.

Methods

or Background: In this retrospective study, 25 patients underwent MRI examinations of the lumbar spine using a 1,5T M RI scanner (Philips Ingenia 1.5T, Best, NL). The MRI protocol included three sagittal sequences (STIR, T1 TSE, T2 TSE), and an axial T2 TSE sequenc e. The images were acquired in both standard and low resolution. Both the clinical standard (Compressed SENSE (CS)) and the new AI-based SuperR esolution reconstruction method (SuperRes-AI) were applied. F our experienced readers (two radiologists and two orthopedic surgeons) eval uated the sequences for pathologies (bone marrow edema, neuroforaminal sten osis, disc herniation).

Results

or Findings: The acquisition time for the clinical standard sequ ences was 11 minutes and 5 seconds. In contrast, the acqu isition time for the low resolution SuperRes-AI sequences was 7 minutes and 37 seconds (31% scan time reduction). A generalized estimating equations (GEE) analysis revealed no significant differences in the sensitivity for d etecting edema between reader groups and reconstruction algorithms (all p>0.99). Bonferroni-corrected post- hoc tests in a GEE analysis revealed significantly higher sensitivity for detecting neuroforaminal stenosis with AI-powered r econstruction compared to conventional algorithms among radiologists (p=0.001 ), with no other significant differences observed.

Conclusion

The new AI-based SuperResolution reconstruction of low- resolution 2D MRI sequences of the lumbar spine all ows for a reduction in acquisition time of approximately 31% without compr omising diagnostic quality, showing significantly higher sensitivity for detect ing neuroforaminal stenosis. The AI-based SuperResolution method improves MRI ef ficiency by significantly reducing scan times without compromising image qual ity, potentially enhancing sensitivity in pathology detection, offering advant ages for patient comfort and clinical workflow.

Limitations

Not applicable. Abstract-based Programme 14 Wednesday Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee notification can be found under the number DRKS00024156 Author Disclosures: Robert Hahnfeldt: Nothing to disclose Andra-Iza Iuga: Nothing to disclose David Maintz: Nothing to disclose Philip Rauen: Nothing to disclose Johannes Bremm: Nothing to disclose Thomas Markus Dratsch: Nothing to disclose Kilian Weiss: Employee: Philips GmbH Market DACH Robert Angelo Terzis: Nothing to disclose Grischa Bratke: Nothing to disclose AI-based Detection of Postoperative Abnormalities F ollowing Lumbar Fusion Surgery in Spine Radiographs M. Kim¹, *J. Song*¹, K. Sung², E. Oh¹; ¹Seoul/KR, ² Los Angeles, CA/US ([email protected]) Purpose or Learning Objective: The purpose of this study is to develop a deep learning-based system to detect postoperative abnormalities in spine radiographs following lumbar fusion surgery. This s ystem aims to assist radiologists by detecting postoperative abnormaliti es.

Methods

or Background: A total of 1,505 spine radiographs from 85 patients who underwent lumbar fusion surgery were collected at a secondary healthcare facility between February 2018 and Janua ry 2022. These radiographs, taken post-operation and during follow -up visits, included anteroposterior, lateral, flexion, and extension vi ews. Annotations for periprosthetic loosening, cage subsidence, and comp ression fracture were performed by a musculoskeletal radiologist, and ver ified with CT scans. The Co-DETR model was trained on a subset of 634 radiog raphs from 74 patients with 726 annotations. The class distribution includ ed 58, 24, and 17 patients yielding 278, 215, and 168 images respectively, wit h each image averaging 1.10 annotations. Initial training was conducted on a public dataset (FracAtlas), followed by transfer learning to enhance detection of postoperative abnormalities. Negative samples were included to bo ost training efficiency, and model performance was evaluated using mean Average Precision (mAP).

Results

or Findings: Periprosthetic loosening achieved an mAP score of 0.601 with 0.5 IoU threshold. The mAP score for eac h class of periprosthetic loosening, cage subsidence, and compression fractur e were 0.565, 0.667, 0.572, respectively.

Conclusion

The study demonstrates the potential of detecting p ostoperative abnormalities in spine radiographs after lumbar fus ion surgery using deep learning. The results indicate a foundational poten tial for enhancing diagnostic capabilities in clinical settings. The potential of this approach to improve early detection of complications could lead to more timel y interventions and better patient outcomes.

Limitations

Further validation is required to optimize its perf ormance, particularly to support radiologists in settings wi th limited access to specialists. Funding for this study: Not applicable Ethics committee - additional information: IRB No. 2022-08-018 Author Disclosures: Jeongmin Song: Nothing to disclose Minjee Kim: Nothing to disclose Kyunghyun Sung: Nothing to disclose Eunsun Oh: Nothing to disclose Post-operative X-rays radiomics-based machine learn ing to predict two- year clinical outcome in patients with lumbar spine arthrodesis *I. C. Pizza*¹, M. Pedullà², S. Fusco², F. Serpi², D. Albano³, C. Messina², S. Gitto², L. M. Sconfienza²; ¹Eboli/IT, ²Milan/IT, ³Cefalu'/IT ([email protected]) Purpose or Learning Objective: The aim of this study is to predict two-year clinical outcome in patients with lumbar spine arth rodesis using machine learning and radiomics based on post-operative X-ra ys.

Methods

or Background: This retrospective study was performed at a tertiar y orthopaedic centre and included 162 patients with l umbar spine arthrodesis, post-operative X-rays available for analysis and mi nimum follow-up of two years. Clinical follow-up was evaluated at two year s using Oswestry Disability Index (ODI): ODI≤20 indicated good clinical outcome (n=90), ODI>20 i ndicated poor clinical outcome (n=72). All X-rays were manua lly segmented by drawing rectangular regions of interest including the arthr odesis and one adjacent non- operated vertebra on both proximal and distal sides . Radiomic features were extracted. After feature selection and class balanc ing, machine learning (three ensembles of Random Forest classifiers) was trained , validated using nested 10-fold cross-validation and tested.

Results

or Findings: After training and cross-validation, in the test da taset machine learning showed ROC-AUC (%) of 74 (majority vote), 72.9** (mean) [confidence interval 69-76.7], accuracy (%) of 68 ( majority vote), 67.7** (mean) [65.9-69.5], sensitivity (%) of 60 (majority vote), 60.6** (mean) [52.7-68.6], specificity (%) of 74 (majority vote), 73.3** (mean ) [67.8-78.9], PPV (%) of 65 (majority vote), 64.6** (mean) [62-67.1], and NPV ( %) of 70 (majority vote), 70** (mean) [67.1-72.9] (*p<0.05, **p<0.005).

Conclusion

Radiomics-based machine learning may assist clinici ans in predicting clinical outcome of patients with lumbar spine arthrodesis based on post-operative X-rays, thus modifying physical reha bilitation and therapeutic strategies accordingly.

Limitations

Retrospective study. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Approved by Local Ethics Committee (RETRORAD protocol) Author Disclosures: Irene Carmen Pizza: Nothing to disclose Stefano Fusco: Nothing to disclose Carmelo Messina: Nothing to disclose Salvatore Gitto: Nothing to disclose Luca Maria Sconfienza: Nothing to disclose Martina Pedullà: Nothing to disclose Francesca Serpi: Nothing to disclose Domenico Albano: Nothing to disclose AI based thoracolumbar and sacral spine fracture de tection for computed tomography *J-B. Pialat*¹, D. Gicquel¹, A-K. Golla², C. Bürger ², C. Lorenz², M. Villien¹, S. Gouttard¹, A. Vlachomitrou¹, T. Klinder²; ¹Lyon/ FR, ²Hamburg/DE ([email protected]) Purpose or Learning Objective: AI algorithms which detect vertebral fractures generate limited classifications which on ly identify vertebral body fracture. We propose a thoracolumbar and sacral spi ne fracture detection algorithm able to identify individual fracture loca tions in both the vertebral body and the posterior arch . It segments the entire spi ne, extracts spine-aligned sub volumes and detects spinal fractures using a convol utional neural network.

Methods

or Background: 195 CT scans from polytraumatized patients were collected in a single-center retrospective clinical study. Dataset was split into training (n=145) and validation (n=50) sets. Accura cy for identification of injury location within the body was assessed in the valida tion set using a Free Response ROC (FROC) curve and performance at the ve rtebral body level was measured using a Receiver Operating Characteris tic (ROC) curve. A subsequent test set including 173 patients ( fractu red N=109, non fractured N= 64) was analyzed with the same algorithm. Performan ce was assessed similarly using FROC curve.

Results

or Findings: The algorithm detected 87.3% of the 775 spinal frac ture locations of the validation set using a false posit ive threshold of 5 per case. It detected 92.4% of the fractured vertebrae. 249 fals e positives were detected, most of which were easily rejected upon review by r adiologists. 26 false negatives were found, most of which were transverse process fractures. There were 7 vertebral body fractures; all were single en dplate stable fractures. In the test set, the algorithm detected 88.6% of the 255 f ractures using a false positive threshold of 5 per case.

Conclusion

We have developed and validated a deep learning alg orithm which determines location of fractures in the whole vertebra with reasonable accuracy.

Limitations

This as to be tested prospectively in routine emerg ency condition to assess the gain in time / sensitivity Funding for this study: Collaborative study between Hospices Civils de Lyon and Philips using GOPI research fundings Ethics committee - additional information: Approved by local ethics commitee Author Disclosures: Tobias Klinder: Employee: Philips Healthcare Marjorie Villien: Employee: Philips Healthcare Christian Lorenz: Employee: Philips Healthcare David Gicquel: Research/Grant Support: Philips Heal thcare Sylvain Gouttard: Nothing to disclose Anna Vlachomitrou: Employee: Philips Healthcare Jean-Baptiste Pialat: Research/Grant Support: Phili ps Heathcare Christian Bürger: Employee: Philips Healthcare Alena-Kathrin Golla: Employee: Philips Healthcare The new frontier of MRI: virtual dissection with 3D PDw sequence. A pilot study on ATiFL anatomy *G. Del Gaudio*¹, G. Vuurberg², M. Dalmau-Pastor³, G. Kerkhoffs⁴, M. Maas⁴; ¹Rome/IT, ²Weesp/NL, ³Barcelona/ES, ⁴Amsterdam/NL ([email protected]) Purpose or Learning Objective: In the literature, there is much conflicting data regarding the anatomy of the anterior tibiofib ular ligament (ATiFL), even in studies with anatomical specimens. Therefore, this study aims to reassess the Abstract-based Programme 15 Wednesday anatomy of this ligament using MRI with a high-reso lution isotropic 3D-PDw sequence.

Methods

or Background: From February to May 2024, 72 MRI scans (3T) of the ankle were performed at Amsterdam UMC. The incl usion criterion was patients over 16 years of age. The exclusion criter ia were: absence of a 3D- PDw scan, ATiFL trauma or surgery, congenital anoma lies, metallic or movement artifacts. The 43 3D-PDw valid scans, allo wed for aligning the planes along the individual bundles of the ligament .

Results

or Findings: The high spatial resolution (0,23mm) of 3DPDw allow ed the identification of three bundles: superior, inte rmediate and inferior. Regarding dimensions the superior is the thickest a nd widest (mean 2.68x9.28mm) and the inferior the longest (mean 15. 45mm). Regarding orientation (axial from the fibula to the tibia) th e superior and inferior have a transverse orientation, while the intermediate is o riented backward. Regarding the shape, they are fanned in 97.7%, 71.7%, and 25. 6% respectively, while they are band-like in the remaining cases. We did n ot identify any anatomical variance regarding the number of bundles.

Conclusion

The use of volumetric isotropic sequences as the 3D -PDw, can be a very useful tool for the anatomical study of l igamentous structures in the absence of available anatomical specimens. Understa nding the exact anatomy of this structure is crucial for managing both acut e and chronic traumatic pathologies (impingement, overuse), especially in y oung patients and athletes.

Limitations

Sample size and lack of anatomical specimen compari son. Funding for this study: No funding. Ethics committee - additional information: This study received a waiver by the ethical committee according to local rules and regulations. Author Disclosures: Giovanni Del Gaudio: Nothing to disclose Gwendolyn Vuurberg: Nothing to disclose Miquel Dalmau-Pastor: Nothing to disclose Ginom M.J. Kerkhoffs: Nothing to disclose Mario Maas: Nothing to disclose MRI biomarker assessment of Duchenne muscular dystr ophy disease progression: a 12-month longitudinal study *Y. Song*, H. Xu, R. Xu, K. Xu; Chengdu/CN ([email protected]) Purpose or Learning Objective: To evaluate the disease progression in patients with Duchenne muscular dystrophy (DMD) by using multi-modal quantitative magnetic resonance imaging (qMRI), and comparing the responsiveness of these imaging indicators with the clinical function scales.

Methods

or Background: 130 DMD patients were enrolled and underwent MRI examination of hip muscles to determine fat fra ction (FF) and longitudinal relaxation time (T1). All participants returned for follow-up at an average of 12 months. According to the baseline North Star Ambula tory Assessment (NSAA) score, all patients were divided into three subgrou ps: mild (76-100 score), moderate (51-75 score) and severe (0-50 score) func tional decline. Standardized response mean (SRM) was used as the re sponsiveness to the disease progression, and the responsiveness of qMRI and clinical function scales to the disease progression in different DMD stages was compared. SRM>0.8 is considered as a high response to disease progression.

Results

or Findings: The overall SRM of MRI biomarkers is higher than th at of the clinical function scales. For mild group, FF of adductors and abductors have higher responsiveness, with SRM of 0.816 and 1 .043, respectively. For moderate group, FF of all muscle groups have a high responsiveness, and the SRM are between 1.004 and 1.606. For severe group, T1 of abductors and FF of all muscle groups have high responsiveness, and SRM are between 0.867 and 1.633. However, the SRM of the clinical functio n scales for patients with different disease stages are all less than 0.8.

Conclusion

The sensitivity of MRI biomarkers to DMD disease pr ogression is higher than that of clinical function scales, espec ially the FF of gluteal muscles is more sensitive to disease progression, and the s ensitivity indicators are different in different disease stages.

Limitations

This study didn‘t discuss whether patients received steroid therapy. Funding for this study: National Natural Science Foundation of China (82271981) Ethics committee - additional information: ChiCTR1800018340 Author Disclosures: Ke Xu: Nothing to disclose Huayan Xu: Nothing to disclose Yu Song: Nothing to disclose Rong Xu: Nothing to disclose Hip Imaging: Radiation-free 3D models based on 3D M RI of the hip joint for children with Slipped capital femoral epiphysis T. D. Lerch, *T. Kaim*, K. Ziebarth, M. K. Meier, J . D. Busch; Bern/CH ([email protected]) Purpose or Learning Objective: Slipped capital femoral epiphyses (SCFE) is a common pediatric hip disease with the risk of ost eoarthritis and impingement deformities, and 3D models could be useful for pati ent-specific analysis. Therefore, magnetic resonance imaging (MRI) bone se gmentation was investigated.

Methods

or Background: A retrospective IRB-approved study involving 23 symptomatic pediatric patients (23 hips) with SCFE was performed. All patients underwent preoperative hip MR with pelvic axial hig h-resolution images (T1 VIBE DIXON images). Slice thickness was 1.2 mm. Mea n age was 12 ± 2 years. All patients underwent surgical treatment. M anual and automatic MRI- based bone segmentation was compared. automatic bon e segmentation was performed by machine learning algorithm, a previous ly used and validated convolutional neural network trained for adult pelv is bone segmentaiton was adapted to pelvis of children.

Results

or Findings: Manual MRI-based bone segmentation was feasible (al l patients, 100%, duration 4-5 hours per case). Dice coefficient was calculated to assess differences between manual and automatic bon e segmentation, Dice coefficient was 82% for the pelvis and 88% for prox imal femur. Precision was 80% for the pelvis and 94% for proximal femur.

Conclusion

MRI-based 3D models were feasible for SCFE patients . Three- dimensional models could be useful for SCFE patient s for preoperative 3D printing and deformity analysis. This could aid for patient-specific diagnosis, treatment decisions, and preoperative planning. MRI -based 3D models are radiation-free and could be used instead of CT-base d 3D models in the future for computer-assisted 3D simulation of surgery.

Limitations

MRI is expensive and access is limited Funding for this study: None Ethics committee - additional information: IRB approval was obtained Author Disclosures: Kai Ziebarth: Nothing to disclose Tilman Kaim: Nothing to disclose Till Dominic Lerch: Nothing to disclose Malin Kristin Meier: Nothing to disclose Jasmin D. Busch: Nothing to disclose Could a single isotropic 3D sequence replace a mult isequence knee MRI in the new era of deep learning reconstruction? *E. Nikolova*¹, J. Kroschke¹, C. Obermüller¹, F. Ze cca², K. Pawlus¹, T. Rauer¹, F. Ensle¹; ¹Zurich/CH, ²Cagliari/IT Purpose or Learning Objective: To assess whether a single isotropic 3D proton-density-weighted fat-saturated (PDFS) sequen ce could replace a standard 2D multisequence MRI protocol for comprehe nsive examination of the knee using deep learning reconstruction (DLR).

Methods

or Background: In this retrospective study, 95 consecutive patient s > 18 years without history of prior knee surgery un dergoing MRI knee examination at the same 1.5 Tesla scanner between M ay 2023 and July 2024 were included. Standard MRI protocol with DLR consi sted of a 3D PDFS sequence and five 2D fast-spin-echo sequences in va rious orientations. Two radiologists separately evaluated the 3D sequence i n all three planes and the 2D sequences, assessing pathologies of bone, cartil age, menisci and ligaments for all joint compartments, and overall i mage quality, diagnostic confidence and artifacts. Wilcoxon signed-rank test was used to compare Likert scale gradings, McNemar’s test for binary gr ades. Interreader agreement was assessed with Cohen’s kappa.

Results

or Findings: There was no significant difference between protoco ls regarding assessment of medial(MC) and lateral comp artment(LC) meniscus, (MC) and patellofemoral(PF) cartilage, medial and l ateral collateral ligament, anterior and posterior cruciate ligament, MC and PF bone marrow edema(BME), and fractures in all compartments (p>0. 05). Significant differences were shown in assessment of LC cartilag e (p=0.002) and LC BME (p=0.04). Image quality and artifacts did not demon strate significant differences. Diagnostic confidence was significantl y higher for the 2D protocol(p=0.023). Interreader agreement overall wa s substantial for the 3D- PDFS(k=0.67) and 2D protocol (k=0.66).

Conclusion

Our results suggests comparable performance between a single 3D-PDFS and a multisequence 2D protocol using DLR f or comprehensive assessment of knee structures, except for LC cartil age and BME. With DLR- powered image enhancement, 3D-PDFS might be able to partly replace 2D sequences for time-efficient knee MRI in the future .

Limitations

Retrospective study design. No arthroscopic referen ce standard. Funding for this study: This research received no financial support. Ethics committee - additional information: Not applicable Abstract-based Programme 16 Wednesday Author Disclosures: Elizabet Nikolova: Nothing to disclose Fabio Zecca: Nothing to disclose Falko Ensle: Nothing to disclose Karolina Pawlus: Nothing to disclose Jonas Kroschke: Nothing to disclose Carina Obermüller: Nothing to disclose Thomas Rauer: Nothing to disclose Assessment of proximal tibial fractures with 3D FRA CTURE (fast field echo resembling a CT using restricted echo-spacing) MRI – Intraindividual comparison with computed tomography *I. Ristow*¹, S. Zhang², C. Riedel¹, A. Lenz¹, M. K rause¹, G. Adam¹, P. Bannas¹, F. O. Henes¹, L. Well¹; ¹Hamburg/DE, ²B est/NL Purpose or Learning Objective: To evaluate the feasibility and diagnostic performance of a 3D FRACTURE (fast field echo resem bling a CT using restricted echo-spacing) MRI sequence for the detec tion and classification of proximal tibial fractures compared with CT.

Methods

or Background: We retrospectively included 126 patients (85 male; 39.6±14.5 years) from two centers following acute k nee injury. Patients underwent knee MRI at 3T including FRACTURE-MRI. Ad ditional CT was performed in patients with tibial fractures (32.5%; n=41) as the reference standard for fracture classification. Two radiologi sts independently evaluated FRACTURE-MRI for the presence of fractures and clas sified them according to AO/OTA, Schatzker, and the 10-segment classificatio n. Diagnostic performance of FRACTURE-MRI was assessed using cros stabulations. Inter- reader agreement was estimated using Krippendorff’s alpha. Image quality was graded on a five-point scale (5=excellent; 1=inadeq uate definition of fracture lines and fracture displacement) and assessed using estimated marginal means.

Results

or Findings: Fractures were detected by FRACTURE-MRI with a sensitivity of 91.5% (83.2–96.5%) and a specificity of 97.1% (93.3–99.0%). Regarding fracture classification, diagnostic perfo rmances were slightly lower, with the 10-segment classification yielding the bes t sensitivity of 85.7% (81.4– 89.3%) and specificity of 97.4% (96.6–98.0%), and t he Schatzker classification yielding the lowest sensitivity of 78.2% (67.4–86.8 %) and specificity of 97.7% (94.1–99.4%). Inter-reader agreement across the who le cohort was excellent (Krippendorff’s alpha 0.89–0.96) and when consideri ng only patients with fractures, good to acceptable (0.48–0.91). Image qu ality was rated good (estimated marginal mean 4.3 (4.1–4.4)).

Conclusion

FRACTURE-MRI is feasible at 3T enabling accurate de lineation of fracture lines for precise diagnosis and classif ication of proximal tibial fractures.

Limitations

Future studies need to address in a comparative int ra-individual setting whether the diagnostic performance of FRACT URE-MRI is better or equivalent to other CT-like bone imaging techniques , such as UTE/ZTE, GRE, or SWI, for fracture detection. Funding for this study: N/A Ethics committee - additional information: The retrospective study was approved by the local institutional review board (Ä rztekammer Hamburg). Author Disclosures: Gerhard Adam: Nothing to disclose Christoph Riedel: Nothing to disclose Matthias Krause: Nothing to disclose Alexander Lenz: Nothing to disclose Peter Bannas: Nothing to disclose Frank Oliver Henes: Nothing to disclose Shuo Zhang: Employee: Philips Inka Ristow: Nothing to disclose Lennart Well: Nothing to disclose qBone: a quantitative software for the semi-automat ed extraction of bone microarchitecture metrics in vivo using Photon-coun ting-detector CT and Artificial Intelligence *A. Ferrero*, J. Thorne, A. O. El Sadaney, K. Rajen dran, C. Mccollough, F. Baffour; Rochester, MN/US ([email protected]) Purpose or Learning Objective: Pathologies affecting bone health impact both mineral density (vBMD) and morphometric charac teristics (thickness (Th) and spacing (Sp)) of trabecular (Tb) and cortical ( Ct) bone. This work introduces a semi-automated software, qBone, which quantifies bone morphometry from in vivo CT scans of the extremitie s and the vertebral spine.

Methods

or Background: protocols were optimized for extremity and spine exams using a commercial photon-counting-detector ( PCD) CT system. A dedicated CNN algorithm was trained to reduce image noise of the spine CT exams while maintaining high resolution details. Ad aptive segmentation algorithms automatically delineated Ct and Tb compa rtments allowing quantification of Th, Sp and vBMD for each. To vali date the software’s accuracy, 10 cadaveric wrists were scanned on HRpQC T and PCD-CT. A 3D- printed bone model (Ct.Th=2mm, Tb.Th=0.3mm and Tb.S p=0.75mm) was used to assess the CNN denoising performance across different patient sizes. Finally, qBone was applied in vivo to multiple pros pective cohorts for wrist and spine.

Results

or Findings: optimized PCD-CT protocols for wrist (70kV, 12mGy, <0.1mSv) and spine (120kV, 40mGy, 8mSv) yielded <0. 15mm in-plane resolution. Validation with cadaveric wrists and th e 3D-printed bone model demonstrated excellent agreement in Ct.Th and Tb.Sp metrics. CNN denoising significantly improved trabecular morphometry accur acy in the spine for small and medium patient sizes. In vivo measurements (wri st, N=50; spine, N=14) for each metric (Tb.Th=0.3-0.45mm, Tb.Sp=0.6-1.05mm, Ct .Th=0.5-1.58mm, Ct.vBMD=450-600mg/cm3) were consistent with literat ure values.

Conclusion

qBone facilitates semi-automated quantification of bone morphometry from high resolution CT data, providing a comprehensive assessment of bone health in vivo beyond traditiona l mineral density.

Limitations

Comparisons with microCT are needed to validate met rics for vertebral bones. Additionally, qBone does not lever age spectral information to estimate vBMD in the spine. Funding for this study: NIH R21ar084126-01a1 Ethics committee - additional information: IRB 23_005308, PI: Baffour Author Disclosures: Ahmed O. El Sadaney: Nothing to disclose Francis Baffour: Nothing to disclose Andrea Ferrero: Nothing to disclose Jamison Thorne: Nothing to disclose Kishore Rajendran: Nothing to disclose Cynthia Mccollough: Nothing to disclose 10:00-11:00 Research Stage 1 Research Presentation Session: Head and Neck RPS 208 Key insights in soft tissue neck imaging Moderator E. Loney; Halifax/UK ([email protected]) Author Disclosures: Elizabeth Loney: Consultant: DMC Healthcare Ltd Detection of MRI edema patterns in patients with ac ute neck infections: a prospective blinded multidisciplinary and multice nter interobserver human performance evaluation *J-P. T. Vierula*, J. Velhonoja, A. Sirén, J. Nurmi nen, M. J. Nyman, K. Mattila, J. Hirvonen; Turku/FI Purpose or Learning Objective: In patients with acute neck infections, MRI shows reactive edema patterns that predict disease severity: retropharyngeal edema (RPE) and mediastinal edema (ME). How well ra diologists and clinicians with diverse backgrounds and neck MRI ex perience can detect these edema patterns is unknown.

Methods

or Background: This prospective, blinded, multidisciplinary, multicenter interobserver study evaluated human per formance in detecting RPE and ME from axial in-phase and water T2-weighte d Dixon images. Readers (N=28, including radiologists, neuroradiolo gists, radiology residents, head and neck surgeons) from all five university ho spitals in Finland were briefly trained and assessed the presence of RPE an d ME (yes/no) and rated their confidence (1-5) blinded to clinical data. Em ergency MRI images were obtained from 60 patients with acute neck infection s. Ten patients appeared twice to assess intraobserver variability. Sensitiv ity, specificity, accuracy, and interobserver agreement were assessed.

Results

or Findings: Overall sensitivity, specificity, and accuracy were 0.89, 0.81, and 0.85 for RPE and 0.85, 0.81, and 0.82 for ME. ME accuracy correlated with confidence (p=0.002), whereas RPE a ccuracy did not (p=0.580). Radiologists achieved higher RPE sensiti vity (p=0.01), RPE accuracy (p=0.04), ME sensitivity (p=0.01), and ME accuracy (p=0.02) than clinicians, whereas other group comparisons were no t significant. High confidence was found for RPE (4.3) and ME (4.1). Ov erall, interobserver kappa was 0.61 (substantial) for RPE and 0.52 (moderate) for ME, with radiologists showing higher agreement than clinicians. Median in traobserver accuracy was 90% for both RPE and ME. Abstract-based Programme 17 Wednesday

Conclusion

We show high diagnostic accuracy and substantial in terobserver agreement for detecting clinically relevant reactiv e edema patterns on MRI in patients with acute neck infections. These results encourage using these biomarkers in clinical practice.

Limitations

Limited availability of emergency MRI. Funding for this study: This study was financially supported by the Sigrid Jusélius Foundation. The funders had no role in stu dy design, data collection and analysis, decision to publish, or preparation o f the manuscript. Ethics committee - additional information: A waiver for patient consent was not sought because it is not required by the nation al legislature for retrospective studies of existing data. Author Disclosures: Aapo Sirén: Nothing to disclose Jarno Velhonoja: Nothing to disclose Jussi Hirvonen: Nothing to disclose Jari-Pekka Tapani Vierula: Nothing to disclose Mikko Juhani Nyman: Nothing to disclose Kimmo Mattila: Nothing to disclose Janne Nurminen: Nothing to disclose Enhanced Survival Prediction in Nasopharyngeal Carc inoma Through Integrated Peritumoral and Intratumoral Radiomics *S. Khongwirotphan*, A. Prayongrat, S. Kitpanit, D. Kannarunimit, C. Chakkabat, V. Shotelersuk, S. Sriswasdi, C. Lert butsayanukul, Y. Rakvongthai; Bangkok/TH ([email protected]) Purpose or Learning Objective: Accurately prediction of overall survival (OS) is essential for optimizing treatment strategies in nasopharyngeal carcinoma (NPC), thereby improving patient outcomes. This stu dy aimed to enhance OS prediction by integrating radiomic features from bo th intra- and peritumoral areas, offering a novel biomarker approach beyond t raditional clinical features.

Methods

or Background: We analyzed 251 NPC patients treated with chemoradiotherapy between 2010 and 2019, all follow ed for a minimum of three years. Radiomics features were extracted from the gross tumor volume (GTV) contours and a 3-mm peritumoral area of the p re-treatment CT images using PyRadiomics v3.0.1. The robustness and predic tive power of radiomic features were assessed by intraclass correlation an d univariate Cox regression. Selected radiomics features were combin ed with clinical data (age, gender, T-stage, N-stage) for a multivariate analys is. Cox regression models were optimized with recursive feature elimination a nd 20 repetitions of five-fold cross-validation, reserving 20% of dataset for mode l testing.

Results

or Findings: Addition of peritumoral radiomic features significa ntly (P < 0.05) improved survival predictions (C-index: 0.7 87±0.067 validation; 0.669 test), over intratumoral only (C-index: 0.755±0.063 validation; 0.626 test set). Integrating clinical data with intra- and peritumor al radiomics yielded the best model, with a C-index of 0.832±0.052 in validation and 0.727 in test set, that outperformed (P < 0.05) the model with clinical and intratumoral features (C- index of 0.769±0.066 and 0.705, respectively). The baseline clinical model yielded a C-index of 0.703±0.100 in validation and 0.618 in test set.

Conclusion

Integrating radiomic features from both intra- and peritumoral areas significantly improved OS predictions in NPC, surpassing traditional approaches that utilize only clinical and intratumo ral radiomics. This could lead to more personalized treatment strategies, potentia lly improving patient outcomes.

Limitations

External validation is recommended for this single- center retrospective study before clinical use. Funding for this study: This research project is supported by National Research Council of Thailand (NRCT) and grants for development of new faculty staff, Ratchadaphiseksomphot Fund, Chulalon gkorn University Ethics committee - additional information: The study received ethics approval from the Institutional Review Board of the Faculty of Medicine, Chulalongkorn University, Thailand (IRB no. 0630/66 ). Author Disclosures: Yothin Rakvongthai: Nothing to disclose Sira Sriswasdi: Nothing to disclose Anussara Prayongrat: Nothing to disclose Danita Kannarunimit: Nothing to disclose Chakkapong Chakkabat: Nothing to disclose Vorasuk Shotelersuk: Nothing to disclose Chawalit Lertbutsayanukul: Nothing to disclose Sararas Khongwirotphan: Nothing to disclose Sarin Kitpanit: Nothing to disclose Predictive potential of dynamic contrast-enhanced M RI and plasma- derived angiogenic factors for response to concurre nt chemoradiotherapy in human papillomavirus-negative oropharyngeal cancer *A. Longo*, P. Hudler, P. Strojan, G. Plavc, L. Ume k, K. Surlan Popovic; Ljubljana/SI ([email protected]) Purpose or Learning Objective: Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) can assess tumour vascu larity, which depends on the process of angiogenesis and affects tumour r esponse to treatment. Our study explored the associations between DCE-MRI par ameters and the expression of plasma angiogenic factors in human pa pilloma virus (HPV)- negative oropharyngeal cancer, as well as their pre dictive value for response to concurrent chemoradiotherapy (cCRT).

Methods

or Background: 25 patients with locally advanced HPV-negative oropharyngeal carcinoma were prospectively enrolled in the study. DCE-MRI and blood plasma sampling were conducted before cCR T, after receiving a radiation dose of 20 Gy, and after the completion o f cCRT. Perfusion parameters ktrans, kep, Ve, initial area under the curve (iAUC) and plasma expression levels of angiogenic factors (vascular e ndothelial growth factor [VEGF], connective tissue growth factor [CTGF], pla telet-derived growth factor [PDGF]-AB, angiogenin [ANG], endostatin [END] and t hrombospondin-1 [THBS1]) were measured at each time-point. Patients were stratified into responders and non-responders based on clinical eva luation. Differences and correlations between measures were used to generate prognostic models for response prediction.

Results

or Findings: Higher perfusion parameter ktrans and higher plasma VEGF levels successfully discriminated responders f rom non-responders across all measured time-points, whereas higher iAU C and higher plasma PDGF-AB levels were also discriminative at selected time points. Using early intra-treatment measurements of ktrans and VEGF, a predictive model was created with cut-off values of 0.259 min-1 for ktra ns and 62.5 pg/mL for plasma VEGF.

Conclusion

Early intra-treatment DCE-MRI parameter ktrans and plasma VEGF levels may be valuable early predictors of res ponse to cCRT in HPV- negative oropharyngeal cancer.

Limitations

The small sample size and absence of healthy contro ls for angiogenic factors limit our findings. Perfusion pa rameter values in DCE-MRI can vary with different post-processing software, s o comparisons between studies should be made with caution. Funding for this study: This research was funded by the Slovenian Research and Innovation Agency (ARIS), grant number P3-0307. Ethics committee - additional information: The study was approved by the National Medical Ethics Committee of the Republic o f Slovenia (No. 0120- 247/2019/4, 12 June 2019) and the Committee for Med ical Ethics of the Institute of Oncology Ljubljana (OI: 28.5.2019, ERI DEK-0064/2019). Written informed consent was obtained from all patients. Author Disclosures: Lan Umek: Nothing to disclose Gaber Plavc: Nothing to disclose Katarina Surlan Popovic: Nothing to disclose Petra Hudler: Nothing to disclose Primož Strojan: Nothing to disclose Alja Longo: Nothing to disclose The impact of quantification of circulating tumor H PV-DNA in the clinical and surgical management of patients with oropharyng eal squamous cell carcinoma *S. Ruggiero*¹, S. Lucchese², V. Dolcetti¹, S. Marz i¹, A. Vidiri¹; ¹Rome/IT, ²Naples/IT ([email protected]) Purpose or Learning Objective: To conduct a multifactorial assessment in patients with oropharyngeal squamous cell carcinoma (OPSCC) that includes: 1) clinical characteristics; 2) detection of the am ount of circulating tumor HPV- DNA in plasma (ctHPVDNA); 3) MRI-based volumetric a nalysis of both the primary tumor and lateral cervical lymph node metas tases.

Methods

or Background: A prospective study was conducted on patients with OPSCC. As controls, patients with suspected HPV-neg ative OPSCC were used. Both the primary tumor and lateral cervical l ymph node metastases were manually delineated, slice by slice, by two expert head and neck radiologists on T2-weighted axial images at diagnosis. In patien ts with multiple lymph node metastases, the total volume was obtained by summin g the volume of each lymph node. This measurement was correlated with th e number of ctHPVDNA copies using the Mann-Whitney test. Abstract-based Programme 18 Wednesday

Results

or Findings: A total of 95 patients were included, of which 58 ( 61%) were p16+/HPV16+, 3 (3%) were p16-/HPV33+, 23 (24%) were p16-/HPV-, 1 (1%) was p16-/HPV16+, and 10 (10%) were p16+/HPV-. No association was found between the number of ctHPVDNA copies and the primary tumor volume. However, significant correlations emerged w ith the volume of lymph node metastases (Rho = 0.42, p = 0.004) and with th e combined volume of the lymph node metastases and primary tumor (Rho = 0.51 , p < 0.001).

Conclusion

ctHPVDNA is a promising biomarker that could potent ially eliminate the need for solid biopsy for diagnosis. The data demonstrate an excellent correlation between p16+/HPV DNA testing, ctHPVDNA, and the volume of the primary tumor and lymph node metastas es. This suggests that liquid biopsy could be useful in identifying the su bgroup of patients with better oncological outcomes.

Limitations

Principal limitation of the study is the number of enrolled patients, but it was statistically appropriate. Funding for this study: Partially supported by LILT 2020-21 Program and MAECI-Call for Joint Project Proposals Italy-Brazil # BR22GR03. Ethics committee - additional information: All enrolled patients signed an informed consent to the protocol approved by the IR CCS Regina Elena National Cancer Institute, Istituti Fisioterapici O spitalieri, Institu- tional Review Board (RS1647/22). Author Disclosures: Sergio Ruggiero: Nothing to disclose Simona Marzi: Nothing to disclose Sonia Lucchese: Nothing to disclose Vincenzo Dolcetti: Nothing to disclose Antonello Vidiri: Nothing to disclose The utility of intraoral ultrasonography in differe ntial diagnosis of benign and malignant oral mucosal lesions *R. Abdalla-Aslan*¹, D. E. Gaitini¹, A. Rchmiel¹, G . Merhav¹, S. Akrish¹, M. Javitt², D. Shilo¹, O. Emodi¹, N. Beck-Razi¹; ¹H aifa/IL, ²Miami/US ([email protected]) Purpose or Learning Objective: Simple yet reliable methods for differentiating between benign and malignant soft t issue tumours of the oral cavity are currently lacking. Our primary aim was t o assess the correlation between pre-operative intraoral ultrasound (US) var iables and malignancy of oral lesions.

Methods

or Background: This is a cross-sectional prospective study of consecutive patients over the age of 18 years from both genders, with a clinical diagnosis of a soft tissue lesion in the oral mucos a. Within a 2-weeks interval, patients who are scheduled for biopsy and histopath ological examination underwent high-resolution intraoral US obtained usi ng a 7-15 MHz-L15-7io- linear-ultrasound-transducer-‘hockey stick-probe’ o n a Philips-Epiq-5, 7 machines (Philips Medical, Netherlands). Sonographi c variables included: size in 3-dimensions, echogenicity, presence of cystic a reas, presence of calcifications, margins and vascularity. The sonogr aphic findings were compared with histopathology.

Results

or Findings: Full data was available for 52 patients with 53 tum ors. Included were 24 females and 28 males, with a mean age 60.11±16.7 years [range 18-90]. Following histopathological results, 22 patients with 23 tumors were diagnosed with squamous cell carcinoma (SCC), 3 patients with dysplasia (1 mild, 1 moderate and 1 severe), and th e remaining 26 patients with benign lesions. Sonographic variables of maximal diameter (a cutoff of 12 mm with 0.76 sensitivity and 0.77 specificity), margins (ill-defined, p<0.001) and vascularity (high or type III/IV, p=0.002) proved to be significantly correlated to SCC group, compared to benign lesions group.

Conclusion

The utility of intraoral US in the differential dia gnosis of benign and SCC tumors in the oral cavity is demonstrated b y this prospective clinical study, using sonographic features of maximal diamet er, margins and vascularity.

Limitations

Small sample size and operator dependent technique. Funding for this study: None Ethics committee - additional information: Ehics committee of Rambam Health Care Campus reference RMB-19-0596. Author Disclosures: Adi Rchmiel: Nothing to disclose Sharon Akrish: Nothing to disclose Marcia Javitt: Nothing to disclose Omri Emodi: Nothing to disclose Goni Merhav: Nothing to disclose Diana E. Gaitini: Nothing to disclose Dekel Shilo: Nothing to disclose Nira Beck-Razi: Nothing to disclose Ragda Abdalla-Aslan: Nothing to disclose Neoadjuvant radiochemotherapy in patients with loca lly advanced oral- cavity tumour: Response-predictive radiological ima ging features *I. Burck*, A. Gleich, R. Winkelmann, M. Fleischman n, E. Herrmann, J-E. Scholtz, P. Thönissen, T. Vogl, D. Pinto Dos S antos; Frankfurt/DE Purpose or Learning Objective: To explore radiological MR imaging features to predict response to neoadjuvant radiochemotherap y in patients with locally advanced oral-cavity tumour.

Methods

or Background: We included 30 patients (15 women, mean age 60±10 years) with oral cavity cancer (stage IVa) who underwent neoadjuvant radiochemotherapy (RTX) before surgery. MRI scans w ere performed before RTX, 15 days after its initiation and preoperativel y. Two radiologists retrospectively evaluated the images for overall tu mour signal intensity (SI), SI change over time, and tumour extent using a Likert scale. Quantitative analysis was performed for the absolute SI of the tumor in A DC-, DWI-, and T2- weighted sequences normalized to the spinal cord. T umour volume (TV) was calculated manually in a contrast-enhanced T1 seque nce. Differences and ratios of ADC, DWI and T2-SI and TV were calculated . Patients with a stage pT1 or T0 were classified as responders, all others as non-responders.

Results

or Findings: ADC-SIs at 2nd and 3rd MRI differed significantly between responders and non-responders (p = 0.010 an d p = 0.013), as did the ratio between baseline and preoperative DWI-SIs (p = 0.041) and the difference between normalized baseline and preopera tive ADC-SIs (p = 0.049). Non-responders showed an increase in TV at the 2nd MRI, while responders showed a significant decrease in TV, so the calcula ted percentage decrease and ratio are significant markers of response progr ession.

Conclusion

Diffusion weighted imaging parameters as well as tu mour volumetry may predict response to neoadjuvant radio chemotherapy in oral cavity cancer and may be benefical for image guided treatment potentially be used to guide treatment or extent of surgery in the se patients.

Limitations

Limitations of this study include small sample size , single-center and retrospective study design. Funding for this study: None. Ethics committee - additional information: Ethics approval was obtained by the institutional review board (approvals number 20 8/12). Author Disclosures: Ria Winkelmann: Nothing to disclose Thomas Vogl: Nothing to disclose Jan-Erik Scholtz: Nothing to disclose Daniel Pinto Dos Santos: Nothing to disclose Philipp Thönissen: Nothing to disclose Alexander Gleich: Nothing to disclose Eva Herrmann: Nothing to disclose Maximilian Fleischmann: Nothing to disclose Iris Burck: Nothing to disclose Impact of Deep Learning-Based Image Reconstructions in Head and Neck MRI *F. Albisinni*¹, C. Carbone¹, C. Zacchi¹, M. Ravane lli¹, D. Farina¹, B. Van Deberge²; ¹Brescia/IT, ²Leuven/BE ([email protected]) Purpose or Learning Objective: This study aimed to assess the impact of deep learning (DL)-based reconstructions on T2 imag e quality in head and neck MRI. Additionally, potential time savings from using DL in various sequences were evaluated.

Methods

or Background: Three sequences were compared: (A) TSE T2 without DL, with three signal averages and an acqui sition time of 2'35''; (B) TSE T2 with DL at intermediate strength, with two s ignal averages and an acquisition time of 1'25''; and (C) TSE T2 with DL at maximal strength, with one signal average and acquisition time of 43''. Images from 52 patients were randomly and blindly evaluated by three radiologist s with varying levels of experience using MR scanners from different vendors . For each patient, two images were analyzed at the level of the nasopharyn x and oral cavity. Four categories were assessed: overall image quality, ar tifacts, edge sharpness, and noise, each rated on a 3-point Likert scale. Ad ditionally, radiologists were tasked with identifying the correct sequence for ea ch image. The two main endpoints evaluated were inter-rater reproducibilit y and comparison of the sequences

Results

or Findings: A total of 636 images were rated by three radiologi sts. Inter-rated reproducibility was poor across all cat egories. The sequences type was correctly identified in only 44% cases. The ove rall quality scores for sequence A, B and C were 2.63, 2.52, and 2.52 respe ctively (p=0.055). Artifacts scores were 2.61, 2.48, and 2.55 (p=0.07) ; edge sharpness scores were 2.57, 2.39, and 2.43 (p=0.02, with significant differences between sequences A and B); and noise scores were 2.51, 2.4 1, and 2.41 (p=0.1)

Conclusion

The performances of the three sequences were simila r overall. DL-based sequences for head and neck MRI were shown effective and offered significant time savings, enabling potential ultra- fast imaging protocols

Limitations

Small sample Funding for this study: No funding was provided for this study Abstract-based Programme 19 Wednesday Ethics committee - additional information: Not applicable Author Disclosures: Davide Farina: Nothing to disclose Baptiste Van Deberge: Nothing to disclose Flavia Albisinni: Nothing to disclose Marco Ravanelli: Nothing to disclose Chiara Zacchi: Nothing to disclose Carmela Carbone: Nothing to disclose 10:00-11:00 Research Stage 2 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 205 Meta-level topics in AI: cost-effectiveness, non-interpretive use-cases and evidence Moderator E. Neri; Pisa/IT ([email protected]) Early health technology assessment for an artificia l intelligence tool to detect incidental pulmonary embolisms on computed t omography *E. H. M. Kemper*¹, K. Redekop¹, F. Vos², M. Ijzerm an¹, M. P. A. Starmans¹, J. J. Visser¹; ¹Rotterdam/NL, ²Delft/NL ([email protected]) Purpose or Learning Objective: Incidental pulmonary embolisms (IPE) on computed tomography (CT) are missed in up to 70% of cases. While artificial intelligence (AI) tools for IPE detection exist, an evaluation on if and how these tools can provide actual value, e.g., fit patients and end-users needs (i.e., radiologists), have never been performed. The aim o f this early health technology assessment (eHTA) is to determine the re quirements for a value- based AI tool for IPE detection on CT.

Methods

or Background: A comprehensive eHTA process for radiology-AI was proposed and conducted for IPE. A literature se arch, structured interviews, focus group, and evaluation meetings we re performed with the identified stakeholders to define criteria and scen arios for a multiple criteria decision analysis (MCDA). A representative survey w as developed and circulated to weigh the importance of the criteria and assess performances of four possible AI designs. MCDA analysis on the surv ey help quantify the value requirements.

Results

or Findings: Consultations with radiologists, treating physician s, patients, radiology technologists, AI specialists, legal experts, and ethicists resulted in 14 sub-criteria and five main criteria; patient impact, model performance, physician support, environmental impac t, and costs. Preliminary outcomes indicate that a short follow-up time for d iagnosing IPE is more important than a high sensitivity for IPE detection .

Conclusion

A value-based AI tool for IPE detection should be f ocused on triage to reduce the impact of the diagnosis of IPE on the patient, mainly because delay of diagnosis can result in progressio n of the IPE and preventable stress for the patient, while an improv ed detection rate is considered to result in significant overtreatment.

Limitations

The scope of this analysis has been within Europe. Outcomes might not be applicable elsewhere. Funding for this study: E.H.M.K., K.R., M.P.A.S., F.V, and J.J.V. acknowledge funding by LSH-TKI (Health~Holland Dutc h Top Sector Life Sciences and Health) 23024 Ethics committee - additional information: No applicable Author Disclosures: Maarten Ijzerman: Nothing to disclose Ken Redekop: Nothing to disclose Erik Hermanus Marcellinus Kemper: Nothing to disclo se Martijn Pieter Anton Starmans: Nothing to disclose Jacob Johannes Visser: Advisory Board: Contextflow Frans Vos: Nothing to disclose Potential costs and benefits of AI for fracture det ection in cervical spine CT scans at hospital level *G. Van Den Wittenboer*¹, I. M. Nijholt¹, M. Maas², M. F. Boomsma¹; ¹Zwolle/NL, ²Amsterdam/NL Purpose or Learning Objective: Aim of this study was to assess healthcare costs at the hospital level for patients screened f or cervical spine (CS) fractures using CT, and to estimate costs and benefits of inc orporating artificial intelligence (AI) to detect CS fractures in clinica l practice.

Methods

or Background: Diagnostic accuracy of on-duty radiologists and AI in detecting CS fractures on CT scans from a retros pective database (n=2321, ≥18 years, 2007-2014) was compared with a reference standard. Healthcare costs for patients were inventoried up to 7 months after their emergency department visit. Total and average costs per patie nt based on the radiologist diagnosis were calculated for four categories: true positive, true negative, false positive, and false negative. Finally, a scenario-a nalysis was conducted to estimate the diagnostic accuracy of radiologists co mbined with AI, and the corresponding total healthcare costs per diagnostic category.

Results

or Findings: Radiologists identified 193 out of 219 scans with fractures and 2085 out of 2102 scans without fractu res, whereas AI identified 177 out of 219 fractures and 2065 out of 2102 scans without fractures. AI identified 23 fractures missed by the radiologists and correctly classified 16 non-fracture scans that had been misclassified as f ractures by the radiologists. This resulted in a potential sensitivity of 216/219 (98%) and specificity of 2101/2102 (>99%) for the combined radiologist-AI ap proach. On average, €5,978 less was spent per missed fracture. The tota l cost for the AI-assisted scenario was €61,132 (0.3%) higher than for radiolo gists alone.

Conclusion

In this scenario-analysis, the use of AI appears to increase hospital costs by 0.3% due to more accurate diagnos es. A next step could be to complement these results with non- hospital cost s and quality-adjusted life years to further investigate the cost-effectiveness of this AI.

Limitations

No limitations were identified. Funding for this study: The radiology department of the Isala received a grant from AIDOC Medical to have a third party (THI NC, Utrecht, the Netherlands) that is specialized in early health te chnology assessments, perform the analyses for this study. AIDOC medical had no role in the data analyses itself. Neither AIDOC Medical nor THINC ha d a role in data collection or drafting of the abstract. Ethics committee - additional information: The study uses retrospective data. Author Disclosures: Martijn Franklin Boomsma: Other: The Department of Radiology, Isala, has established a strategic partnership with Aidoc Medi cal. However, Aidoc Medical had no decisive role in data collection, da ta analysis nor data interpretation. Ingrid M. Nijholt: Other: The Department of Radiolo gy, Isala, has established a strategic partnership with Aidoc Medical. However, Aidoc Medical had no decisive role in data collection, data analysis nor data interpretation. Mario Maas: Nothing to disclose Gaby Van Den Wittenboer: Other: The Department of R adiology, Isala, has established a strategic partnership with Aidoc Medi cal. However, Aidoc Medical had no decisive role in data collection, da ta analysis nor data interpretation. Cost-effectiveness of AI-assisted digital mammograp hy – results from a Swedish model-based analysis *P. Gialias*, J. Lyth, M. Kristoffersen Wiberg, T. Bjerner, M. Husberg, L. Bernfort, H. Gustafsson, L-Å. Levin; Linköping/S E ([email protected]) Purpose or Learning Objective: To evaluate the cost-effectiveness of AI- assisted biennial digital mammography (AI-DM) in co mparison to conventional digital mammography (cDM) with double reading of sc reening mammograms (screening interval ages 40-74).

Methods

or Background: We used a Markov decision analytic model with a life-time horizon. The analysis was conducted from a healthcare perspective. Model parameters were based on Swedish registry dat a and published randomized AI-DM studies. The model estimates the c osts and quality- adjusted life-years (QALYs) related to mammography and breast cancer. Mammography-related costs were collected from the u niversity hospital in Linköping. Stage-specific cancer cost,QALY-weights were obtained from the literature. Scenario analyses were performed with d ifferent screening strategies.

Results

or Findings: Per 1000 individuals AI-DM gained 10.8 QALYs compared to cDM. The costs per 1000 individuals wer e USD 3,752,278 and USD 3,816,443 for AI-DM and cDM, respectively. AI-D M resulted in a cost saving of USD 64 165 which makes it a dominant stra tegy. The isolated screening costs were slightly higher in the used AI -DM setting, USD 597, but this was offset by reduced lifetime costs of cancer treatment. A screening strategy with AI plus one radiologist for all exami nations saves USD 9128 Abstract-based Programme 20 Wednesday screening costs compared to cDM, however the QALYs gained were decreased to 8.8.

Conclusion

AI-DM is cost saving in our setting and generates m ore quality- adjusted life-years. One of the add-on benefits is the possibility to free radiological time to other clinical work. These ben efits could be further improved by changing the AI-DM triaging strategy.

Limitations

We based AI parameters in the model mainly on two S wedish randomized trials and cancer data from the populati on-based cancer registry from Sweden. However, cost data are highly dependen t on the Swedish health care system and the generalizability to other healt h care systems might be limited. Funding for this study: None Ethics committee - additional information: Not ethics committee approval was need for this study Author Disclosures: Maria Kristoffersen Wiberg: Nothing to disclose Magnus Husberg: Nothing to disclose Tomas Bjerner: Nothing to disclose Håkan Gustafsson: Nothing to disclose Johan Lyth: Nothing to disclose Lars Bernfort: Nothing to disclose Lars-Åke Levin: Nothing to disclose Pantelis Gialias: Nothing to disclose AI Tools to Reduce Claims and Compensation Payments of Missed Fractures on Radiographs: A Potential Game Changer? M. Tordjman¹, L. Gracia¹, E. Guillo¹, R. Amar¹, J. Ventre¹, *N-E. Regnard*², R. Y. Carlier¹, J-L. Marmorat¹, J-D. Laredo¹; ¹Pari s/FR, ²Lieusaint/FR ([email protected]) Purpose or Learning Objective: To evaluate the potential of BoneView, an AI tool for fracture detection on radiographs, in clai ms files of missed fractures which led to financial compensation.

Methods

or Background: This retrospective study included all the files of patients who submitted a claim and had financial co mpensation for missed fractures on radiographs from January 2013 to Decem ber 2019 in the 38 university hospitals of the Greater Paris area Hosp itals (APHP, France). Of the 29 patients who claimed files, 26 were finally incl uded (3 were not available in the system). For each patient with a claim, 5 patie nts with radiographs from the same anatomical areas (with or without fracture) we re included from consecutive patients who had radiographs at a unive rsity hospital in 2022. Two readers (one fellow in musculoskeletal radiology an d one expert radiologist in musculoskeletal imaging with more than 20 years of experience) read the radiographs, blinded from which patients had missed fractures.

Results

or Findings: 156 patients were included (26 patients with missed fractures and 130 « control » patients). The AI sof tware was able to detect 80.7% of fractures (21/26) for the patients who fil ed claims for missed fractures. The sensitivity of readers was also impr oved with AI for these patients: the junior reader had a sensitivity of 61 .5% without AI and 69.2% with AI and the expert reader had a sensitivity of 73.1% without AI and 84.6% with AI. The total of potentially avoided financial comp ensation would have been 265.314 euros.

Conclusion

The sensitivity of the two readers is improved with AI in a cohort of patients with missed fractures who submitted cla ims and had financial compensations. AI was able to detect most of these fractures.

Limitations

A limitation was the small number of claims files. Funding for this study: There was no funding for this study. Ethics committee - additional information: Not applicable Author Disclosures: Raphael Amar: Nothing to disclose Nor-Eddine Regnard: Founder: Chief Medical Officer of Gleamer Jean-Luc Marmorat: Nothing to disclose Mickael Tordjman: Nothing to disclose Jean-Denis Laredo: Employee: Gleamer Robert Yves Carlier: Nothing to disclose Jeanne Ventre: Employee: Gleamer Enora Guillo: Nothing to disclose Laure Gracia: Nothing to disclose Overlooked and underpowered: a meta-research study addressing sample size in radiomics research *J. Zhong*¹, J. Lu², Y. Xing¹, Y. Hu¹, D. Ding¹, X. Liu¹, S. Dai¹, H. Zhang¹, W. Yao¹; ¹Shanghai/CN, ²Stanford, CA/US ([email protected]) Purpose or Learning Objective: To investigate how studies determine the sample size when developing radiomics models, and w hether it is sufficient.

Methods

or Background: We identified radiomics studies published from January to December 2023 on seven leading peer-revi ewed radiological journals owned by European Society of Radiology and Radiological Society of North America. We reviewed the sample size justific ation methods, and actual sample size used. We calculated the minimum sample size according to 3 criteria proposed by Riley et al, and compared the estimated and the actual sample size used. We investigated which characteris tics factors were associated with the sufficient sample size.

Results

or Findings: We included 116 studies. 11/116 studies justified t he sample size, in which 6/11 performed a priori sampl e size calculation. The mean ± standard deviation (SD), median (first and third quartile, Q1, Q3) of total sample size of models are 451 ± 871, 223 (130 , 463), and those of sample size for training are 292 ± 676, 150 (90, 288). The mean ± SD, median (Q1, Q3) of difference between the total sample siz e and minimum sample size according to Riley et al criterion 3 are 120 ± 888, -100 (-216, 183), and those of difference between the sample size for training and minimum sample size according to Riley et al all 3 criteria are -386 ± 1264, -268 (-427, -157). The model testing method and specialty of topic were as sociated with sufficient sample size.

Conclusion

Radiomics models are often designed without sample size justification, as a consequence many models are too small to avoid overfitting, noise, and outliers. It should be encouraged to jus tify, perform and report sample size calculations when developing radiomics models.

Limitations

The limitation of the study is limited number of le ading peer- reviewed radiological journals. Funding for this study: Funding was provided by National Natural Science Foundation of China (82302183, 82471935, 82271934), Yangfan Project of Science and Technology Commission of Shanghai Munic ipality (22YF1442400), Research Found of Health Commission of Changing District, Shanghai Municipality (2023QN01), Laboratory Open F und of Key Technology and Materials in Minimally Invasive Spine Surgery ( 2024JZWC-ZDA03, 2024JZWC-YBA07), and Research Fund of Tongren Hospi tal, Shanghai Jiao Tong University School of Medicine (TRKYRC-XX202204 , TRYJ2021JC06, TRYXJH18, TRYXJH28). Ethics committee - additional information: The study is a meta-research study with a protocol available on OSF (https://osf .io/pbukc/), and no human participants or animals were included in the study. Author Disclosures: Defang Ding: Nothing to disclose Yue Xing: Nothing to disclose Huan Zhang: Nothing to disclose Jingyu Zhong: Board Member: Dr. Jingyu Zhong acknow ledges his position as a member of the Musculoskeletal section of the Scie ntific Editorial Board of European Radiology, a member of Scientific Editoria l Board of BMC Medical Imaging, and a guest editor of the collection “AI i n radiology: revolutionizing medical imaging and interpretation” of BMC Artifici al Intelligence. Junjie Lu: Nothing to disclose Xianwei Liu: Nothing to disclose Yangfan Hu: Nothing to disclose Weiwu Yao: Nothing to disclose Shun Dai: Nothing to disclose Evolution of commercially available artificial inte lligence in radiology: a follow-up on peer-reviewed evidence of 179 produc ts N. Antonissen¹, *I. B. Houben*², O. Tryfonos³, M. D e Rooij¹, K. G. Van Leeuwen⁴; ¹Nijmegen/NL, ²Zwolle/NL, ³Amsterdam/NL, ⁴De Bilt/NL Purpose or Learning Objective: To investigate changes in peer-reviewed evidence on commercially available radiologic artif icial intelligence (AI) products from 2020 to 2023.

Methods

or Background: A comprehensive review of the literature published between January 2015 and March 2023 of CE certified radiological AI products (according to www.healthairegister.com) was perform ed. Complying with the previous systematic review, this follow-up study ca tegorized the publications according to the hierarchical model of efficacy: fr om technical and diagnostic accuracy (levels 1 and 2) to impacts on clinical de cision-making and patient outcomes (level 3-5) or socio-economic impact (leve l 6).

Results

or Findings: By March 2023, 91 vendors were identified, offering a total of 179 products, with 120 of these (67%) havi ng peer-reviewed evidence, compared to 36% in 2020. In 2023, there were 662 pu blications on these 120 products, compared to 237 publications on 36 produc ts in 2020. An increase (22 to 25%) was found in publications focusing on t echnical or potential clinical efficacy. The majority of publications described th e diagnostic accuracy of the product (level 2), although relatively showing a de crease (55 to 52%). For the higher levels of efficacy (level 3-6) the respectiv e contribution to the total remained the same as 2020 (23%).

Conclusion

While there is an increase in the amount of publica tions validating AI products, the majority of publication s continue to describe the lower levels of efficacy. This suggests that even t hough the field has been maturing, we still have limited knowledge and evide nce of the clinical impact of AI products in radiology.

Limitations

Several products have a high number of publications , which may cause them to be overrepresented in the total. Funding for this study: No funding was received for this study. Abstract-based Programme 21 Wednesday Ethics committee - additional information: Not applicable Author Disclosures: Ignas Bernardus Houben: Nothing to disclose Kicky Gerhilde Van Leeuwen: Owner: Romion Health Fo under: Health AI Register Noa Antonissen: Nothing to disclose Olga Tryfonos: Nothing to disclose Maarten De Rooij: Research/Grant Support: Siemens H ealthineers Flexible Deep Learning MR Image Enhancement with Pe rformance Monitoring Z. Zhou, C. Arnold, H. Gandhi, P. Gulaka, A. Shanka ranarayanan, *S. Pasumarthi Venkata*; Menlo Park, CA/US ([email protected]) Purpose or Learning Objective: Deep learning (DL) MR image enhancement allows scan time reduction while maintaining the di agnostic quality. However, its performance may deteriorate over time. This stu dy aims to develop an adaptive image enhancement DL model and investigate s a non-reference- based metric without human annotation for performan ce monitoring.

Methods

or Background: A single DL model with a ConvNeXt backbone was trained on 3027 paired MR data. High-quality images were enhanced by a commercial algorithm as targets. Low-quality input images were acquired with various acceleration methods (0-80%) for model to l earn adaptive enhancement. The trained DL model was evaluated on another diverse set of 205 cases. Line profiles and region-of-interests (R OIs) were manually labeled for each case. The slope/gradient was extracted fro m line profiles to measure image sharpness, and signal-to-noise ratio (SNR) wa s derived from ROIs to evaluate noise level. In addition, gradient entropy (GE) as a non-reference- based metric (lower GE higher quality) was compared with line/ROI based metrics.

Results

or Findings: Compared to inputs, over 90% of model outputs achieved 45% SNR increase and 8% sharpness increase . On average, SNR and sharpness were improved by 73% and 27%, respect ively. GE measured on outputs was reduced by 0.5% for 95% of test case s. For test cases with >0.5% GE reduction, the Pearson correlation of the relative change between GE and SNR is -0.333 (p < 0.05), and between GE and sharpness is 0.214 (p < 0.05), showing a weak but significant correlation between GE and annotated image quality (IQ) metrics.

Conclusion

The developed DL model can adaptively improve IQ su pporting flexible protocol acceleration. Its strong denoisin g also enables MR scans with higher acceleration/resolution. In addition, gradie nt entropy can be simply deployed for performance monitoring and mitigate th e risk of mis-interpretation.

Limitations

Not applicable Funding for this study: NIH SBIR grant (R44MH135725) Ethics committee - additional information: Not applicable Author Disclosures: Srivathsa Pasumarthi Venkata: Employee: Subtle Medi cal Inc Campbell Arnold: Employee: Subtle Medical Inc Praveen Gulaka: Employee: Subtle Medical Inc Ajit Shankaranarayanan: Employee: Subtle Medical In c Harsh Gandhi: Employee: Subtle Medical Inc Zechen Zhou: Employee: Subtle Medical Inc Radiologist-Guided Active Learning for Medical Imag e Segmentation: Moving Beyond the Dice Score to Clinically Relevant Targets *B. Föllmer*, V. Serafimoski, K. Schulze, F. Biavat i, M. Bosserdt, M. Dewey; Berlin/DE ([email protected]) Purpose or Learning Objective: Deep learning models for medical image segmentation typically require extensive pixel-wise annotations, which are costly and time-consuming. Active learning can miti gate this challenge by labeling only the most informative (i.e., uncertain ) cases in multiple annotation and training rounds. However, conventional active l earning methods do not account for clinically relevant segmentation target s. This study introduces a radiologist-in-the-loop approach for targeted activ e learning, to optimize model performance beyond standard metrics like the Dice s core, focusing on clinically significant segmentation objectives.

Methods

or Background: We propose a targeted active learning framework consisting of four iterative steps: (1) Automated i dentification of uncertain cases for review by the radiologist, (2) Radiologis t selection of cases relevant to predefined clinical segmentation targets, (3) Co mbined selection of uncertain and clinically relevant cases, and (4) Ef ficient partial annotation and model retraining. We applied this approach to multi -class segmentation of coronary arteries using the SCCT 18-segment model, evaluating it on CTAs from 300 patients of the DISCHARGE (NCT02400229) an d CAD-Man trials. Initial model training was conducted using standard active learning, followed by targeted active learning with three predefined obje ctives: (1) segmentation of rare vessels (e.g., Ramus Intermedius), (2) segment ation of thin vessels (e.g., R-PDA, R-PLB), and (3) segmentation of heavily calc ified segments.

Results

or Findings: Our framework demonstrated improved segmentation performance and time-efficiency over standard activ e learning for the three predefined targets (rare vessels, thin vessels, and calcified segments.

Conclusion

The proposed targeted active learning framework ena bles more time-efficient, radiologist-guided model training f ocused on clinically relevant segmentation targets, improving performance beyond conventional accuracy metrics like the Dice score.

Limitations

This framework was evaluated exclusively on coronar y artery segmentation in cardiac CT, with only three segment ation targets considered. Broader validation is needed for other anatomical s tructures and imaging modalities. Funding for this study: This work was funded by the German Research Foundation through the graduate program BIOQIC (GRK 2260, project-ID: 289347353) and the DISCHARGE project (603266-2, HEA LTH-2012.2.4.-2) funded by the FP7 Program of the European Commissio n. Ethics committee - additional information: This study does not require any approval of the ethics committee. Author Disclosures: Kenrick Schulze: Nothing to disclose Marc Dewey: Other: Hands-on cardiac CT courses (www .ct-kurs.de) Other: European Society of Radiology (ESR) Publications Ch air (2022-2025); the opinions expressed in this abstract/presentation ar e the author’s own and do not represent the view of ESR Patent Holder: Patent on fractal analysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10,991,109, approved) Author: Cardiac CT (Springer Nature) Equipment Support Recipient: Siemens, General Elect ric, Philips, Canon Grant Recipient: EU (EC-GA 603266 in HEALTH.2013.2.4.2-2) DFG (DE 1361/14-1, DE 1361/18-1, BIOQIC GRK 2260/1, Radiomics DE 1361/ 19-1 [428222922] and 20-1 [428223139] in SPP 2177/1), GUIDE-IT (DE 1 361/24-1), Berlin University Alliance (GC_SC_PC 27), Berlin Institute of Health (Digital Health Accelerator). Federico Biavati: Nothing to disclose Maria Bosserdt: Research/Grant Support: Received fu nding from EU-FP7 Framework Program (DISCHARGE EU FP EC-GA 603266). Vladimir Serafimoski: Nothing to disclose Bernhard Föllmer: Nothing to disclose 10:00-11:00 Research Stage 3 Research Presentation Session: Vascular RPS 215 Advances in peripheral imaging Moderator E. Claus; Leuven/BE A Novel Human Amputated Limb Model for Advancing Pe ripheral Artery Disease Research and Device Testing *J. Csőre*, A. Crichton, B. Benfor, C. Karmonik, T. L. Roy ; Houston, TX/US ([email protected]) Purpose or Learning Objective: Traditional animal models often fail to capture the complexity of peripheral artery disease (PAD) lesions, leading to a gap between preclinical and clinical research in pe rcutaneous vascular interventions (PVI). To address this, we developed a human amputated limb model combined with a proprietary MRI-histology pro tocol for detailed plaque characterization and simulation of PVI procedures, assessing lesion-specific device impact on the vessel wall.

Methods

or Background: Amputated limbs from end-stage PAD patients were scanned using 3T or 7T MRI, incorporating Ultr ashort Echo Time and T2- weighted sequences to differentiate hard (collagen/ calcium) and soft (fat/thrombus/smooth muscle) plaque components. PVI procedures were simulated in a hybrid operating room, targeting ide ntified lesions. Device testing included balloon angioplasty, lithotripsy, atherectomy, drug-coated balloons, and novel wires/catheters. Vessel impact was evaluated intraprocedurally using intravascular ultrasound, f ollowed by post-procedure micro-CT and 9.4T MRI. Histopathological analysis w as performed with Movat’s and H&E stains.

Results

or Findings: A total of 70 amputated limbs were collected, yield ing 133 target lesions and 2500 histologic cross-sectio ns. Key findings include: 1. Validation of the MRI protocol and human amputated limb model. 2. Successful testing of vessel preparation devices, showing plaq ue disruption and dissection in calcified lesions. 3. Correlation of chronic tot al occlusion crossing success Abstract-based Programme 22 Wednesday with pre-intervention MRI histology scoring. 4. Ide ntification of calcified lesions as barriers to effective drug delivery. 5. Collabor ation with industry for device development and testing.

Conclusion

This human cadaveric model offers a unique platform for PAD research, providing detailed insights into plaque m orphology and PVI device performance. By correlating plaque characteristics with procedural outcomes, it enables precise device testing and fosters innovati on in vascular interventions. This model informs clinical decision-making, enhanc es new technology design, and guides personalized treatment strategies for PA D patients.

Limitations

Single-center study, small cohort Funding for this study: Jerold B. Katz Academy of Translational Science (project ID 15790002, recipient: Trisha Roy); Ameri can Heart Association Transformational Award (project ID: 17590004, recip ient: Trisha Roy); National Institutes of Health Research Project grant (R01) ( award ID: R01HL174587, recipient: Trisha Roy) Ethics committee - additional information: This study was approved by the Institutional Review Board under study ID PRO000272 58. Author Disclosures: Alexander Crichton: Nothing to disclose Bright Benfor: Nothing to disclose Judit Csőre: Nothing to disclose Christof Karmonik: Nothing to disclose Trisha L. Roy: Research/Grant Support: Baylis Medic al Technologies, Boston Scientific, Light Matter Interaction Founder: Magel lan Biomedical Inc. Application and significance of precise CTA scannin g technology in the assessment of lower extremity arterial diseases *J. Xing*, H. Yu, L. Zhu; Shang Hai/CN ([email protected]) Purpose or Learning Objective: Objective: This study seeks to investigate the differences in image quality and radiation expo sure between an advanced precision scanning technique and traditional scanni ng methods in 320-slice computed tomography angiography (CTA) of the lower limb

Methods

or Background: Methods: A cohort of 89 patients with suspected lower limb a arteryial disease, who underwent CTA e xamination at our institution, were randomly allocated to either grou p A or group B. In group A,low-dose testing was first used. TS was obtained by subtracting the time to peak of the dorsalis pedis artery T2 and the time t o peak of the main abdominal artery T1;the formal scan began at T1 + 4 seconds,and the scan was completed after adjusting the entire acquisitio n time to TS by the variable pitch method.Group B was scanned using standard pit ch.The statistical analysis involved the assessment of image quality,r adiation dose, and contrast agent dosage.

Results

or Findings: Results:Both subjective and objective evaluations demonstrated superior image quality of lower extrem ity artery in group A(all P<0.001).The subjective score for group A demonstra ted a significant 21% increase compared to that of group B, particularly in the assessment of ankle and dorsum images (4.32±0.79 vs. 3.57±0.94).In terms of patients' radiation dose and contrast agent dosage, group A exhibited a 16.23% reduction in radiation dose and a 12.28% reduction in contrast a gent dosage compared to group B, respectively (both P< 0.001).

Conclusion

Conclusion: The implementation of VHP technology in lower extremity artery CTA scanning facilitates enhanced visualization of distal blood vessels and improves overall image quality, meanwhi le effectively reducing radiation exposure and contrast agent consumption, which presents substantial clinical value.

Limitations

The sample size is relatively small and warrants ex pansion for further validation of the derived conclusions. Funding for this study: Young Scientists Fund of the National Natural Science Foundation of China (82302188) Ethics committee - additional information: No:2019tjdx123 Author Disclosures: Jun Xing: Nothing to disclose Hong Yu: Nothing to disclose Lin Zhu: Nothing to disclose Run-off CT angiography with a patient-tailored post -trigger delay: Optimized scan timing compared with a fixed delay *K. Qi*, J. Liu; Zhengzhou/CN ([email protected]) Purpose or Learning Objective: To validate the feasibility of using bolus tracking with a patient-tailored post-trigger delay (PTD) in run-off CTA and to compare image quality with that using a fixed PTD.

Methods

or Background: Participants undergoing run-off CTA with bolus tracking were prospectively assigned at random, coh ort A comprised 30 participants with a fixed 10-second PTD and cohort B comprised 30 participants with a patient-tailored PTD. The atten uation of abdominal and lower limb arteries was measured in 11 different an atomical positions in one leg and divided into four vascular segments accordi ng to the anatomical location: aortoiliac, femoropopliteal, tibioperonea l and foot. The mean attenuation, SNR and CNR of each vessel segment wer e calculated. Two readers rated subjective image quality. Two-way ana lysis of variance was used to assess the mean attenuation of four vascular seg ments. Sidak's multiple comparison was used to determine differences in att enuation between the two cohorts and at each anatomical location within each cohort. Mann-Whitney test was used to determine SNR and CNR between two group s, while the chi- square test compared subjective image quality score s.

Results

or Findings: Cohort B using new bolus tracking algorithm predict s the mean patient-tailored PTD of 12 ± 1.8 s. The demographic and frequency of PAD revealed no statistically significant differenc es. Cohort B showed greater attenuation of tibioperoneal (432±76 HU vs 364±69HU, p=0.001) and foot (369±79 HU vs 281±77HU, p=0.001) segments. SNR (p<0.002), CNR(p<0.002) and subjective image quality (excellen t or good image quality, 96.7% vs 74.3%, p=0.038) were higher in cohort B th an in the fixed cohort.

Conclusion

Bolus tracking with a patient-tailored PTD provides reliable scan timing, resulting in improved image quality and opt imized vessel opacification in run-off CTA .

Limitations

Further research is needed on the relationship betw een frequency of PAD and PTD. Funding for this study: The Science and Technology Research Project of Henan Provincial Health Commission (No. 21210231014 2) Ethics committee - additional information: Ethics Committee of Zhengzhou University Author Disclosures: Jie Liu: Nothing to disclose Ke Qi: Nothing to disclose Shear Wave Elastography in Differentiating Acute an d Subacute Thrombosis of Dialysis Arteriovenous Fistulas *Ö. Altun*, A. Dablan, M. Sam Özdemir, M. Karagülle , M. Cingöz, M. F. Arslan; Istanbul/TR ([email protected]) Purpose or Learning Objective: To evaluate the utility of shear wave elastography (SWE) in distinguishing between acute and subacute thrombi in thrombosed dialysis arteriovenous fistulas (AVFs).

Methods

or Background: This retrospective study analyzed 32 dialysis patients with thrombosed AVFs treated between June 2022 and June 2024. All patients underwent Doppler ultrasound and SWE to de termine thrombus characteristics. Based on ultrasound findings and c linical history, patients were categorized into acute or subacute thrombus groups. Thrombus stiffness was quantified using SWE in terms of average, median, a nd maximum kilopascal (kPa) values.

Results

or Findings: The study included 16 patients with acute thrombi a nd 16 with subacute thrombi. SWE measurements revealed significantly higher stiffness values in subacute thrombi compared to ac ute thrombi (p < 0.001). A strong positive correlation was observed between th rombus age and SWE- derived kPa values (average: r = 0.770, median: r = 0.727, maximum: r = 0.835). Receiver operating characteristic (ROC) ana lysis demonstrated SWE’s high accuracy in differentiating thrombus age, with an optimal average cut-off value of 31.7 kPa, resulting in a sensitivity of 90 .5% and specificity of 73.9%.

Conclusion

Shear wave elastography shows promise as a non-inva sive tool for differentiating between acute and subacute thro mbi in thrombosed AVFs, aiding in personalized treatment planning for dialy sis patients.

Limitations

This study's patient numbers was not much. Funding for this study: This study was not supported by any funding. Ethics committee - additional information: All procedures performed in studies involving human participants were in accord ance with the ethical standarts of the instituional and/or national resea rch commitee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standarts. Author Disclosures: Merve Sam Özdemir: Nothing to disclose Mehmet Karagülle: Nothing to disclose Mustafa Fatih Arslan: Nothing to disclose Ömer Altun: Nothing to disclose Mehmet Cingöz: Nothing to disclose Ali Dablan: Nothing to disclose MRI-Histology as a Predictive Tool for Crossing Fai lure in Below-the- Knee Peripheral Arterial Disease *J. Csőre*, A. Crichton, C. Karmonik, B. Benfor, T. L. Roy ; Houston, TX/US ([email protected]) Purpose or Learning Objective: Recent randomized trials have challenged the traditional endovascular-first approach for tre ating below-the-knee arterial disease. Scoring systems like TASC and GLASS overlo ok lesion composition and morphology, which influence peripheral vascular intervention (PVI) Abstract-based Programme 23 Wednesday success. Conventional imaging provides limited plaq ue composition insight, while ultrashort echo time (UTE) MRI can distinguis h between soft (e.g., fibrous tissue, thrombus) and hard (e.g., calcification, de nse collagen) plaque components. This study aimed to assess if MRI-histo logy could better predict lesion crossing failure compared to GLASS and TASC scoring.

Methods

or Background: Amputated limbs were collected from patients with chronic limb-threatening ischemia (CLTI) and scanne d ex-vivo on a 3T MRI using UTE and T2w contrasts. Lesions were classifie d as 'hard' if >50% of the lumen was occluded by calcium or dense collagen bas ed on the MRI. The distribution of hard components (eccentric, concent ric, central), lumen stenosis caused by hard/soft components, and collagen densit y were recorded. Ex-vivo PVIs were carried out in a hybrid operating room an d TASC and GLASS scoring was performed.

Results

or Findings: Seventeen patients yielded 29 target lesions, 76% (22/29) of which were classified as ‘hard.’ Of thes e, 45% had a collagen- dominated composition. 'Hard' lesions showed a sign ificantly higher crossing failure rate compared to 'soft' lesions (95% vs. 14 %, p<.001). MRI scoring of 'hard' lesions was strongly associated with crossin g success (p<.001), outperforming TASC and GLASS scoring (p=.062 and p= .112, respectively). Total vessel occlusion was not predictive of failur e (p=0.64). Most crossing failures (64%) occurred in centrally distributed 'h ard' lesions, though this was not significant.

Conclusion

This MRI-histology scoring system identifies plaque composition as a predictor of PVI failure, outperforming TASC a nd GLASS scoring, and highlighting MRI's potential role in preoperative a ssessment and device selection for CLTI patients.

Limitations

Single-center study, small cohort Funding for this study: Jerold B. Katz Academy of Translational Science (project ID 15790002, recipient: Trisha Roy); Ameri can Heart Association Transformational Award (project ID: 17590004, recip ient: Trisha Roy); National Institutes of Health Research Project grant (R01) ( award ID: R01HL174587, recipient: Trisha Roy) Ethics committee - additional information: This study was approved by the Institutional Review Board under study ID PRO000272 58. Author Disclosures: Alexander Crichton: Nothing to disclose Bright Benfor: Nothing to disclose Judit Csőre: Nothing to disclose Christof Karmonik: Nothing to disclose Trisha L. Roy: Nothing to disclose Assessment of left renal vein areas ratios on CT-ph lebography as surrogate parameter for pressure gradient in pelvic congestion syndrome *T. Nemirovskaya*, R. Bredikhin, R. Akhmetzyanov, E . Fomina, D. Ryabinina, A. Yaglova; Kazan/RU ([email protected]) Purpose or Learning Objective: Pelvic venous diseases are considered widespread problem. Among etiological factors compr ession of left renal vein (LRV) between aorta and superior mesenteric artery (SMA), so-called nutcracker syndrome, is considered primary for symp toms evolvement. Diagnostic workflow includes selective phlebography , but CT-phlebography is gaining popularity. Primary issue is applicability of CT-phlebography results concerning surgical correction selection. Study obj ective was parameters assessment that could be accounted surrogate charac teristics of pressure gradient in the left renal vein.

Methods

or Background: Prospective assessment of CT-phlebography in patients with pelvic congestion syndrome was perfor med. Expiratory CT scanning was performed with 130 s delay after Iodin ated contrast medium administration. Following measurements were perform ed: maximum diameter of pelvic veins, gonadal veins diameter, aorta/SMA angle. Three LRV areas were outlined orthogonally projected to gonadal tri butary, aorta/SMA angle, inferior vena cava сonflux (IVC) with areas ratio calculation. LRV narr owing extension was also measured.

Results

or Findings: 74 patients underwent IVC and tributaries delayed C T- angiography from 2022 to 2024. Mean age 37 years, 1 5 male, 59 female. All had pelvic veins enlargement with associated sympto ms according to ultrasound examination and history. No severe devel opment anomalies was observed except 3 cases of retroaortal LRV. Twelve cases were accompanied with May-Turner variant. Lineal regression assessme nt displayed correlation between aorta/SMA angle and maximum/minimum LRV are as ratio with p < 0.001, also with LRV narrowing extension (p = 0.003 ). May-Turner variant was contributing factor with more severe pelvic veins e nlargement (p < 0.001). Correlation was found with pressure gradient accord ing to direct phlebography, however only 19 patients underwent it.

Conclusion

CT-phlebography may be considered as supportive dia gnostic modality to direct phlebography for selection candi dates for surgical correction of pelvic congestion syndrome with LRV compression.

Limitations

Retrospective direct phlebography data collection w ith incomplete cohort coverage. Funding for this study: No funding Ethics committee - additional information: Institutional ethics board of Interregional Clinical Diagnostic Center Author Disclosures: Rustem Akhmetzyanov: Nothing to disclose Alina Yaglova: Nothing to disclose Daria Ryabinina: Nothing to disclose Elena Fomina: Nothing to disclose Tatiana Nemirovskaya: Nothing to disclose Roman Bredikhin: Nothing to disclose Evaluation of low-dose upper extremity CTA with art ificial intelligence iterative reconstruction for hemodialysis arteriove nous fistula/graft: Image quality and diagnostic value of stenosis dete ction *B. Shou*¹, J. Li², Y. Zou², W. Zhang¹, G. Zhang², X. Hu¹, F. Jiang¹, H. Hu¹; ¹Hangzhou, Zhejiang/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: To assess the image quality and diagnostic value of artificial intelligence iterative reconstr uction (AIIR) in low-dose upper extremity CT angiography for hemodialysis arteriove nous fistula or graft (AVF/G).

Methods

or Background: A total of 56 patients with suspected or known AVF/G dysfunction were prospectively enrolled and w ere randomly divided into two groups: routine-dose group (RD-group, n=28) and low-dose group (LD- group, n=28). RD-group employed a routine CTA proto col (tube voltage: 100kVp; contrast dosage: 1.0ml/kg) with hybrid iter ative reconstruction, while LD-group used the low-dose protocol (tube voltage: 80kVp; contrast dosage: 0.6ml/kg) with AIIR. Two radiologists independently scored the overall image quality using a 4-point scale (1=poor; 4=excellent) . Area under the curve (AUC), accuracy, sensitivity, and specificity of tw o groups for detecting significant (>50%) stenosis were calculated on a pr e-segment basis, using digital subtraction angiography (DSA) as the refere nce standard. Signal-to- noise ratio (SNR) and contrast-to-noise ratio (CNR) in fistula were also analyzed.

Results

or Findings: No significant differences in demographics characte ristic were observed between the two groups (all p>0.05). The radiation dose and contrast dosage in LD-group were reduced by 53% (22 4.56mGyxcm vs. 479.24mGyxcm) and 42% (36mL vs. 63mL), respectively , compared to the RD- group. The mean subjective scores between the RD-gr oup and LD-group showed no significant difference (3.86±0.36 vs. 3.68±0.48, p=0.12). The AUC, accuracy, sensitivity, and specificity were 0.91, 9 0% (47/52 segments), 92%, and 99% for RD-group and were 0.94, 90% (47/52 segm ents), 100%, and 98% on a pre-segment basis for LD-group. In fistula, SN R and CNR of LD-group were 130% and 140% higher than those of RD-group, r espectively (both p<0.001).

Conclusion

Low-dose CTA with AIIR provides superior image qual ity and maintains high accuracy for detecting stenosis in A VF/G, while significantly reducing radiation dose and contrast dosage.

Limitations

N/A Funding for this study: N/A Ethics committee - additional information: This study was approved by the local ethics Committee at the University Hospital. Author Disclosures: Yixuan Zou: Nothing to disclose Xi Hu: Nothing to disclose Hongjie Hu: Nothing to disclose Guozhi Zhang: Nothing to disclose Beili Shou: Nothing to disclose Feng Jiang: Nothing to disclose Wenming Zhang: Nothing to disclose Jing Li: Nothing to disclose Abstract-based Programme 24 Wednesday 10:00-11:00 Research Stage 4 Research Presentation Session: Abdominal and Gastrointestinal RPS 201 What's new in biliary diseases? Moderator J.-H. Yoon; Seoul/KR ([email protected]) Multicenter validation of the DiStrict score, a nov el classification and prognostic score for individuals with primary scler osing cholangitis (PSC) *A. Grigoriadis*¹, S. G. Hamma¹, G. Kemmerich², J. S. Nayagam³, K. Horsthuis⁴, M-C. Londoño⁵, D. Assis⁶, S. Charanjeet⁷, A. Bergquist¹; ¹Stockholm/SE, ²Oslo/NO, ³London/UK, ⁴Amsterdam/NL, ⁵Barcelona/ES, ⁶Connecticut, CT/US, ⁷New Haven, CT/US ([email protected]) Purpose or Learning Objective: To validate the reproducibility and prognostic value of the DiStrict-score in a multicenter intern ational cohort.

Methods

or Background: The DiStrict-score is an MRCP-based classification of the severity of ductal changes (ranging from 1 t o 8) and is reproducible and associated with liver-transplantation and liver-rel ated death. However, it has not been validated. For this retrospective multicen ter study with participation of eight international PSC-centers, hepatologists from each center retrieved data of consecutive adult PSC individuals (MRCP, demogra phics, liver-tests, PSC diagnosis date, hepatobiliary cancer development, l iver- transplantation, death, and cause of death). Two radiologists from each cen ter applied the DiStrict- score independently to the patients of their center . Cases of disagreement were resolved in consensus. Interreader agreement w as assessed for each pair of radiologists with the intraclass correlatio n coefficient (ICC), with a two- way random-effects model, absolute-agreement, and s ingle-measurement. The association of the DiStrict-score with outcomes (tr ansplant-free survival, development of hepatobiliary malignancy) was assess ed with Cox-regression. Survival estimates were calculated with Kaplan-Meie r curves and the curves were compared with the log-rank test.

Results

or Findings: 415 patients (271 males, 248 with ulcerative coliti s) with median diagnosis age of 39 years, were included. Du ring a median follow-up of 84 months 101 patients developed outcomes (liver-tr ansplantation; n=78, liver- related death; n=10, hepatobiliary cancer; n=13). T he interreader agreement ranged between 0.61 and 0.91 for the different cent ers. Patients with high DiStrict-scores (5–8) had a higher risk of developi ng outcomes compared to patients with low scores (1–4) (log-rank test; p=0. 0008) with a hazard ratio of 1,98 (95%CI; 1.32-2.96, p=0.001).

Conclusion

The DiStrict-sore is reproducible with good interre ader agreement and is associated with transplant-free survival and development of hepatobiliary malignancy.

Limitations

The limitations of the study are its retrospective design and the non-standardized MRCP acquisition technique. Funding for this study: Funding was provided by Medical Diagnostics Karolinska. Ethics committee - additional information: Each center obtained ethical approval by local ethical committees . Author Disclosures: Singh Charanjeet: Nothing to disclose Jeremy Shanika Nayagam: Nothing to disclose Aristeidis Grigoriadis: Speaker: Have received inst itutional honoraria by JANSSEN-CILAG AB Annika Bergquist: Nothing to disclose Gunter Kemmerich: Nothing to disclose Karin Horsthuis: Nothing to disclose Stefan Gmail Hamma: Nothing to disclose David Assis: Nothing to disclose María-Carlota Londoño: Nothing to disclose Prevalence, prognostic value, and interreader agree ment of high-grade strictures in individuals with primary sclerosing c holangitis (PSC) *A. Grigoriadis*, S. G. Hamma, A. Bergquist; Stockh olm/SE ([email protected]) Purpose or Learning Objective: To assess the prevalence and reproducibility of the evaluation of the presence of high-grade str ictures (HGS) in MRCP, in PSC individuals. Moreover, to assess the predictive value of HGS for development of hepatobiliary malignancy, liver-tran splantation, and liver- related death.

Methods

or Background: AASLD and EASL have introduced in their guidelines for PSC the term HGS defined as a strict ure seen in MRCP with >75% reduction of the lumen of common and/or biliar y ducts. However, the prevalence, reproducibility of their detection, and their potential value for predicting outcomes have not been assessed. Two rad iologists independently assessed the presence of HGS in MRCPs of 203 indivi duals with PSC recruited at Karolinska University Hospital in the SUPRIM study between 2012 and 2015. MRCP, demographic, clinical-laboratory an d outcome data (hepatobiliary malignancy, liver-transplantation, a nd liver-related death) were retrieved for all patients. Interreader agreement o f the evaluation of HGS was calculated with the intraclass correlation coeffici ent (ICC) using a two-way random-effects model, single-measurement, and absol ute-agreement. The association of HGS with outcomes was assessed with Cox-regression. Survival estimates were calculated with Kaplan-Meier curves and the curves were compared with the log-rank test.

Results

or Findings: After exclusion, 171 patients (103 males, 95 with ulcerative colitis) with a median diagnosis age of 40 years were included. During a median follow-up of 124 months 49 patients developed outcomes (liver-transplantation=36, liver-related death=5, h epatobiliary malignancy=8). 80 patients (47%) had HGS. The agreement was modera te with ICC=0.72 (95%CI; 0.64-0.78). Patients with HGS had a higher risk to develop outcomes (p=0.01) with a hazard-ratio of 2.08 (95%CI; 1.17-3 .71).

Conclusion

HGS are common, can be identified with acceptable reproducibility, and are associated with outcomes.

Limitations

The limitations of the study are its retrospective design and that no intrareader agreement analysis was performed. Funding for this study: Funding was provided by Medical Diagnostics Karolinska. Ethics committee - additional information: The study was approved by the Swedish ethical review authority (2011/824-31/2, 20 18/1111-32, 2018/1494- 31/3). Author Disclosures: Aristeidis Grigoriadis: Speaker: Received Instituti onal honoraria from JANSSEN-CILAG AB Annika Bergquist: Nothing to disclose Stefan Gmail Hamma: Nothing to disclose Deep Learning of Preoperative Gadoxetic Acid-Enhanc ed MRI for Prediction of Perineural Invasion in Intrahepatic C holangiocarcinoma *X. Zhou*¹, J. Hu², S-T. Feng¹; ¹Guangzhou/CN, ²Bei jing/CN Purpose or Learning Objective: To preoperatively predict the Perineural invasion (PNI) in intrahepatic cholangiocarcinoma ( ICC) on gadoxetic acid (EOB)-enhanced MRI, the deep learning with clinical model based fusion model was developed and evaluated.

Methods

or Background: A total of 165 patients with pathologically diagnos ed ICC who underwent preoperative EOB-enhanced MRI wer e retrospectively recruited from two independent centers (center1, tr aining set, n = 115; validation set, n = 14; internal test set, n = 15; center 2, external test set, n = 21). The medmanba was used to extract image feature s on the pre-contrast, arterial phase, portal venous phase, and hepatobili ary phase of MRI. These features combined with clinical factors (such as Ne utrophils, lymphocytes, and serum tumor markers), and classified by a linear la yer. For comparison, a DL model was constructed by removing clinical factors, and a clinical model was established by the random forest selection on the c linical features. Model performance was evaluated with the area under the r eceiver operating characteristic curve (AUC). Gradient-weighted class activation mapping (Grad- CAM) heatmaps were used to show the focus area in p redicting PNI.

Results

or Findings: The PNI positive rate was 42.4% (61/144) in center 1 and 28.6% (6/21) in center 2. On the internal test and external test set, the combined model showed the highest AUC of 0.944 and 0.844. The DL model achieved the moderate AUC of 0.926 and 0.833. The p erformance of clinical model is relatively low, with AUCs of 0.852 and 0.7 11. Grad-CAM showed the DL model focused on the solid component of the tumo r, especially the margin area.

Conclusion

MRI based DL model can accurately predict PNI-posit ive ICC, and the tumor margin area may have important indica tions for the model.

Limitations

Retrospective study; limited sample size. Funding for this study: National Natural Science Foundation of China (82271958) Ethics committee - additional information: The Institutional Review Board of The First Affiliated Hospital, Sun Yat-sen Universi ty(approval number: [2023]014) Author Disclosures: Shi-Ting Feng: Nothing to disclose Jing Hu: Nothing to disclose Xiaoqi Zhou: Nothing to disclose Abstract-based Programme 25 Wednesday Preoperative prediction of IDH1-mutation and perine ural invasion in intrahepatic cholangiocarcinoma based on Gd-EOB-DTP A-enhanced MRI and MRI-derived habitats X. Zhou, M. Chen, *S-T. Feng*; Guangzhou/CN ([email protected]) Purpose or Learning Objective: To preoperatively predict isocitrate dehydrogenase 1 (IDH1) mutation and perineural inva sion (PNI) of intrahepatic cholangiocarcinoma (ICC) based on the Gd-EOB-DTPA-e nhanced MRI and MRI-derived habitat imaging to improve the reliabil ity and interpretability.

Methods

or Background: A total of 129 ICC patients with Gd-EOB-DTPA- enhanced MRI before resection between 2018 and 2024 were collected, and randomly assigned to training set and the test set in a ratio of 7:3. IDH1 mutation and PNI status were assessed on pathologic tissue slides. Clinical information and MRI features were qualitatively and quantitatively collected. Matchable tumors in the pre- and post-enhancement T 1 mapping images were manually outlined for habitat analysis and divided into five habitats based on kmeans clustering (Habitat 1-5). A combined nomogra m model was constructed based on clinical features, MRI feature s and habitat fraction. The diagnostic accuracy was evaluated using the area un der the receiver operating characteristic curves (AUCs).

Results

or Findings: The IDH1 nomogram model consists of age, T2 central brightness, liver ADC value, tumor T1 value reducti on rate and percent of Habitat 4, with AUCs of 0.926 and 0.924 in the trai ning and validation sets. The PNI nomogram model consists of CEA, tumor location, intrahepatic bile duct dilation and percent of Habitat 1, with AUCs of 0.8 54 and 0.896 in the training and validation sets. By mapping the habitats to mul ti-sequence MRI, Habitat1 is located predominantly at the edge of the tumor, with signals suggestive of a parenchymal component representing the aggressive e dge of the tumour. Habitat 4 is located intratumorally, with signals s uggestive of an intratumoral fibrotic area with little tumor component.

Conclusion

MRI and habitat imaging can noninvasively and preop eratively determining the IDH1 mutation and PNI of ICC with g ood accuracy and interpretability.

Limitations

Retrospective study, limited sample size. Funding for this study: National Natural Science Foundation of China (82271958) Ethics committee - additional information: The Institutional Review Board of The First Affiliated Hospital, Sun Yat-sen Universi ty(approval number: [2023]014) Author Disclosures: Shi-Ting Feng: Nothing to disclose Xiaoqi Zhou: Nothing to disclose Meicheng Chen: Nothing to disclose Distinguishing Bile Sludge from Physiological Bile Concentration on Abdominal MRI: Key MRI Features and Diagnostic Accu racy *K. Kadirli*, A. Usta, S. Sahin, A. Cantürk, S. Özk an, H. Mutlu; Istanbul/TR ([email protected]) Purpose or Learning Objective: Routine abdominal MRI often shows signal changes in the gallbladder on T1- and T2-weighted i mages, aside from gallstones. Differentiating between bile sludge and physiological bile concentration is key for accurate diagnosis, but li terature shows overlapping signal characteristics with limited guidance. This study aims to identify specific MRI findings to distinguish between bile sludge and physiological bile concentration for improved diagnosis and clinical m anagement.

Methods

or Background: This retrospective study included patients with non - stone signal changes on upper abdominal MRI from Ja nuary 2022 to April 2024, who had follow-up ultrasounds within one mont h. The study involved 42 patients in the sludge group and 44 in the non-slud ge group according to ultrasound findings. Key parameters assessed were l ayering, fluid-fluid levels on T2, T1 signal characteristics, and signal loss o n out-of-phase (OOP) sequences. Quantitative measures included the T2 ab normal signal/spleen ratio, ADC values, and the normal-bile/abnormal-sig nal ratios on T2 and ADC maps. Multivariate regression was performed on para meters with p < 0.05, and the model's diagnostic performance was evaluated.

Results

or Findings: T1-weighted hypointensity and fluid-fluid levels we re strongly associated with bile sludge (p < 0.001), w hile layering and OOP signal loss associated with normal bile(p<0,001). The best multivariate model, using T1W, fluid levels, and OOP signal loss, showed 83.7 2% specificity, 72.22% sensitivity, and 78.48% accuracy, with an AUC of 0. 889.

Conclusion

An abnormal signal in the gallbladder lumen on MRI that does not exhibit OOP signal loss but shows fluid-fluid l evels and T1-weighted hypointensity may be indicative of bile sludge.

Limitations

This study is limited by its retrospective design, which may introduce selection bias. Additionally, ultrasound was used as the gold standard, biochemical evaluation of the bile was no t conducted. Funding for this study: No funding Ethics committee - additional information: Ethics committee approval was obtained from the relevant institution. Author Disclosures: Anıl Usta: Nothing to disclose Kenan Kadirli: Nothing to disclose Suat Özkan: Nothing to disclose Ali Cantürk: Nothing to disclose Hakan Mutlu: Nothing to disclose Serdar Sahin: Nothing to disclose 11:30-12:30 Research Stage 1 Research Presentation Session: Neuro RPS 311 Illuminating the brain: neuroimaging insights into epilepsy and neuroinflammation Moderator S. Gerevini; Cremona/IT Expanding Language Assessment in Epilepsy Patients through the Individual Functional Connectome *R. Stepponat*, M. Berger, L. Schäfer, M. S. Yildir im, J. Leinkauf, F. Fischmeister, S. Bonelli, G. Kasprian; Vienna/AT ([email protected]) Purpose or Learning Objective: Pre-surgical language evaluation has remained largely unchanged for years, underscoring the need for improved methods. Given that epilepsy and language both oper ate as network functions, analyzing the individual functional connectome may offer greater clinical precision, enabling more accurate predictions and b etter-informed surgical decisions for enhanced patient care.

Methods

or Background: Language is predominantly left-lateralized, and fMRI is commonly used for preoperative evaluation o f lateralization by calculating the lateralization index (LI). However, a survey by Benjamin et al. highlights significant uncertainties in fMRI result s among clinicians. In this study, 46 patients with temporal lobe epilepsy and 25 healthy controls underwent preoperative fMRI. Data preprocessing was performed with fMRIPrep, and correlation maps were created using t he CONN toolbox. Seed- based connectivity (SBC) analysis of core language areas, based on Fredarenko et al. 2024, was conducted, and a LI bas ed on graph measurements (degree) was calculated. This was comp ared to traditional LI analysis and neuropsychological data, with all anal yses conducted in native space.

Results

or Findings: Results indicated that the connectivity-based lateralization index (LI) provided a more consisten t measure of language lateralization compared to standard LI methods, ali gning better with neuropsychological assessments. Patients with lesio nal epilepsy showed greater variability in lateralization compared to M R-negative patients, while healthy controls exhibited stronger left-lateraliza tion as expected. The use of seed-based connectivity (SBC) analysis enhanced the detection of individual differences in language network organization, under scoring its potential clinical value in preoperative assessment.

Conclusion

The initial findings suggest that connectivity-base d LI may improve the accuracy of preoperative language asses sments. This approach offers a promising enhancement over traditional LI methods, providing more reliable insights for surgical planning and potenti ally leading to better predictions of post-surgical cognitive outcomes.

Limitations

-No neuropsychological data for the controls. -Retrospective data. Funding for this study: This study has been conducted as part of a PhD- thesis at the medical university of Vienna. Nothing to disclose. Ethics committee - additional information: EK-Number: 1141/2023 Author Disclosures: Mehmet Salih Yildirim: Nothing to disclose Laurin Schäfer: Nothing to disclose Joel Leinkauf: Nothing to disclose Florian Fischmeister: Nothing to disclose Marc Berger: Nothing to disclose Silvia Bonelli: Nothing to disclose Radheshyam Stepponat: Nothing to disclose Gregor Kasprian: Nothing to disclose Abstract-based Programme 26 Wednesday Temporal lobe epilepsy with isolated amygdala enlar gement: anatomo- electro-clinical features and long-term outcome M. Ferro¹, J. Ramos², *F. M. Doniselli*³, G. Didato ³; ¹Lisbon/PT, ²Gaia/PT, ³Milan/IT ([email protected]) Purpose or Learning Objective: This study focuses on the radiological characteristics of patients with temporal lobe epil epsy and isolated amygdala enlargement (TLE-AE).We aim to assess the imaging f indings, particularly in relation to amygdala size and signal alterations, a nd their correlation with clinical and pathological data to guide better diag nostic and therapeutic decisions.

Methods

or Background: We conducted a retrospective analysis of 143 brain MRI scans from adult patients at a tertiary neurolo gy center. Forty-one patients with TLE-AE were selected. Imaging was reviewed by two neuroradiologists for amygdala size and T2-hyperintensity. Fluorodeoxyglu cose-PET (FDG-PET) data were also included for a subgroup of patients. Amygdala signal alterations were quantified and correlated with clinical, neuro physiological, and pathological findings. Statistical analyses assesse d relationships between imaging features and clinical outcomes.

Results

or Findings: Out of 41 patients, 32% had bilateral amygdala enlargement, while the remainder had unilateral fin dings. T2-hyperintensity was noted in 65.9% of cases, significantly correlat ing with amygdala enlargement. FDG-PET showed temporomesial hypometab olism in 64% of patients, further supporting the epileptogenic invo lvement of the amygdala.Surgical pathology confirmed various under lying etiologies, including gliosis, low-grade tumors, and inflammatory infiltr ates. Post-surgical outcomes were favorable, with 70.6% of operated patients bei ng seizure-free at last follow-up.

Conclusion

Radiologically, amygdala enlargement with or withou t T2- hyperintensity is a significant marker of TLE, ofte n associated with underlying structural abnormalities or inflammation. FDG-PET i s a valuable adjunct for identifying hypometabolic regions corresponding to AE, supporting its role in the epileptic network.MRI findings, combined with s urgical evaluation, can improve clinical outcomes in TLE-AE patients, parti cularly those with drug- resistant epilepsy.

Limitations

The study's retrospective nature and variability in MRI protocols over time limit the consistency of radiological fin dings. Additionally, not all patients underwent advanced imaging techniques, suc h as PET or follow-up MRIs, reducing the ability to track longitudinal ch anges. Funding for this study: None. Ethics committee - additional information: Retrospective study. Author Disclosures: Fabio Martino Doniselli: Nothing to disclose Margarida Ferro: Nothing to disclose Joao Ramos: Nothing to disclose Giuseppe Didato: Nothing to disclose The relationship of glutamate and glutamine and met abolic profiling in focal epilepsy using 7T CRT-FID-MRSI *S. Chambers*, H. Shayeste, P. Lazen, L. Haider, E. Pataraia, G. Kasprian, W. Bogner, S. Trattnig, G. Hangel; Vienna/AT ([email protected]) Purpose or Learning Objective: Identifying epileptogenic foci is essential in therapy-planning and predictive for post-operative seizure freedom in epilepsy. In this work we present a novel MRSI technique (CRT -FID) at 7T, allowing for ultra-high resolution whole-brain maps in focal epi lepsy. We offer a qualitative analysis of its feasibility in identifying and char acterizing metabolic alterations over multiple pathologies.

Methods

or Background: Following informed written consent, forty-two patients with focal epilepsy (16-52 years, 21 femal es/21 males) underwent a 3D-MRSI protocol in 15min with a 3.4 mm isotropic r esolution at 7T using a 32Rx/1Tx-coil. Data processing involved spectral qu antification and ratio mapping of Glu, Gln, Ins, tCho, tCr and NAA normali zed to NAA and tCr. Furthermore, the concentration estimates of Glu and Gln were qualitatively assessed in seizure onset zones.

Results

or Findings: Though we could find no consistent metabolic patter n throughout pathologies, ratio maps showed high dete ctability of lesions, identifying alterations in seizure onset zones in 7 8,6% when normalized to NAA. This detection rate was reduced to 71,2% when normalized to creatine. Of the assessed ratios, Ins/tNAA and Cho/tNAA showe d highest stability across patients, whereas Glu/tNAA and Gln/tNAA show ed varying directionality of changes. Assessing these changes further in rela tion to clinical parameters such as the seizure frequency, we found a trend for relative increases of Gln/Glu in patients with low seizure frequencies an d the inverse for high seizure frequencies.

Conclusion

7T CRT-FID-MRSI shows promising results in identify ing metabolic alterations in patients suffering from fo cal epilepsy, even in the absence of structural lesions. Furthermore, this or ientational study may point to an altered glutamate/glutamine cycle in epilepsy, p otentially the result of altered energy metabolism pathways in chronic epile psy.

Limitations

This study is limited by the small cohort size and qualitative nature of the analysis. Funding for this study: This research was funded by the FWF grant 10.55776/KLI1121, of the Mayor of the Federal Capit al Vienna (Project Number 21186). Ethics committee - additional information: Ethic committee number: EK 1039/2020 Author Disclosures: Wolfgang Bogner: Nothing to disclose Philipp Lazen: Nothing to disclose Stefanie Chambers: Nothing to disclose Lukas Haider: Nothing to disclose Ekatarina Pataraia: Nothing to disclose Siegfried Trattnig: Nothing to disclose Gilbert Hangel: Nothing to disclose Gregor Kasprian: Nothing to disclose Haniye Shayeste: Nothing to disclose Multiparametric MR-based assessment supports the in flammatory nature of symptomatic CSF HIV Escape S. Capelli¹, A. Caroli¹, G. Pezzetti², F. Ferretti³ , R. Vercesi⁴, P. Cinque⁴, *S. Gerevini*²; ¹Ranica/IT, ²Bergamo/IT, ³London/UK , ⁴Milan/IT Purpose or Learning Objective: Symptomatic cerebrospinal fluid (CSF) HIV escape is an infrequent but severe condition occurr ing in persons living with HIV (PLWH) undergoing combination antiretroviral th erapy (cART). It is characterized by HIV-RNA in the CSF despite being u ndetectable in plasma. Similarly to HIV encephalitis in cART-untreated PLW H (HIV-ENC), HIV CSF escape (HIV-ESC) is accompanied by neurological imp airments and brain MRI alterations. This study aimed to investigate the ne uroimaging features of HIV- ESC in comparison to HIV-ENC and neuro-asymptomatic controls.

Methods

or Background: Brain structural and microstructural alterations we re quantified in: HIV-ESC (n=12), HIV-ENC (n=10), neur o-asymptomatic PLWH (n=11) and neuro-asymptomatic HIV-negative controls (n=12). The quantitative analysis included measurements of normalized FLAIR signal intensity, Apparent Diffusion Coefficient (ADC) from diffusion -weighted MRI, and brain tissue volumes from T1-weighted MRI.

Results

or Findings: Both HIV-ESC and HIV-ENC demonstrated significantly higher FLAIR signal intensity in white matter (WM), elevated ADC values in both white and gray matter (GM) and reduced GM volu mes as compared to neuro-asymptomatic controls, while the HIV-ESC grou p had higher WM volumes compared to HIV-ENC. In the HIV-ESC group, GM ADC values were negatively correlated with nadir CD4+ T-cell counts , while GM volume showed a positive correlation. In contrast, in HIV-ENC, WM ADC, FLAIR signal intensity, and WM volume all positively correlated with nadir and current CD4+ T-cell counts.

Conclusion

WM hyperintensities and increased ADC values in HIV -ESC and HIV-ENC reflect active WM damage, while reduced GM volumes are indicative of long-term brain atrophy. However, the higher WM volume in HIV-ESC suggests persistent inflammation. These findings, a long with their correlation to laboratory data, support the hypothesis that inflam mation is the primary mechanism of brain damage in HIV-ESC.

Limitations

Retrospective study with limited patient sample and lack of a standardized MRI protocol across all subjects. Funding for this study: National Institutes of Health (NIH); University of California, San Francisco (UCSF) Ethics committee - additional information: The study was approved by “Comitato Etico IRCCS Ospedale San Raffaele” on 15/ 12/2016 (ref. 235/2015) Author Disclosures: Serena Capelli: Nothing to disclose Giulio Pezzetti: Nothing to disclose Anna Caroli: Nothing to disclose Francesca Ferretti: Nothing to disclose Riccardo Vercesi: Nothing to disclose Paola Cinque: Nothing to disclose Simonetta Gerevini: Nothing to disclose Unravelling PIRA brain atrophy pattern and progress ion *T. Fakhreddine*¹, A. Tamanti¹, C. Salvatore², D. C alderaro¹, S. Ziccardi¹, M. Calabrese¹, A. Fattorini¹, I. Castiglioni², F. B . B. Pizzini¹; ¹Verona/IT, ²Milan/IT ([email protected]) Purpose or Learning Objective: Progression independent of relapse activity (PIRA) is the primary factor contributing to irreve rsible disability accumulation in relapsing multiple sclerosis (MS). This study ai med to investigate differences in brain volumes through Magnetic Resonance Imaging (MRI). Abstract-based Programme 27 Wednesday

Methods

or Background: We conducted a retrospective study with MS patients with and without cognitive PIRA who perfor med T1-weighted 3D MRI study (at baseline and at 24-months follow-up) at t he Verona University Hospital. The TRACE4AD™ medical-device software (De epTrace Technologies, Italy) performed automatic structural brain segmentation from MR scans into 53 regions and calculated correspondi ng total volumes and asymmetry indices, defined as the percentage differ ence between brain region volumes on each side. Atrophy progression was measu red as the percentage difference in volumes at follow-up versus baseline. A statistical comparison of volumes at baseline and atrophy progression was per formed using a two- sample t-test between MS patients with and without PIRA.

Results

or Findings: We collected 95 MRIs from 40 patients with PIRA (26 with follow-up) and 55 patients without PIRA (43 wi th follow-up). At baseline, the difference in asymmetry indices of total white matter, gray-matter occipital- lobe volume and parieto-occipital cortex volume wer e statistically significant (p<0.05): PIRA patients showed more right-side atro phy in the white matter volume and less in the right occipital lobe and par ieto-occipital cortex than non- PIRA patients. Atrophy progression was statisticall y different (p<0.05) for the gray-matter volume of the right temporal lobe, bein g faster in patients with PIRA than in those without (-3.1% vs -1.3%).

Conclusion

These results indicate that atrophy progression may be faster in MS patients with PIRA in regions related to early d ementia and non-verbal language functions.

Limitations

Cohort size and statistical power show potential fo r improvement. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is retrospective. Author Disclosures: Davide Calderaro: Nothing to disclose Isabella Castiglioni: Nothing to disclose Agnese Tamanti: Nothing to disclose Christian Salvatore: Nothing to disclose Tommaso Fakhreddine: Nothing to disclose Massimiliano Calabrese: Nothing to disclose Francesca Benedetta Benedetta Pizzini: Nothing to d isclose Anna Fattorini: Nothing to disclose Stefano Ziccardi: Nothing to disclose Mean Upper Cervical Cord Area (MUCCA) in MOGAD comp ared to MS, NMOSD and healthy controls *E. Lotan*¹, V. Anania², T. Billiet², I. Kister¹, I. Lotan³; ¹New York, NY/US, ²Leuven/BE, ³Petach Tikva/IL ([email protected]) Purpose or Learning Objective: Relatively little is known about how mean upper cervical cord area (MUCCA) changes in MOGAD c ompared to MS, NMOSD, and healthy controls (HC). We aim to assess MUCCA values in MOGAD as compared to MS, NMOSD, and HC.

Methods

or Background: We retrospectively reviewed the NYU Multiple Sclerosis Care Center database to identify all adul t MOGAD patients with available brain MRI performed in stable remission a nd compared them with NMOSD and MS patients and HC. Cross-sectional MUCCA from T1 brain MRIs was quantified using icobrain ms+ (version 5.1 5.0) and normalized for head size. A linear modeling analysis was used to e valuate the impact of covariates on cross-sectional MUCCA. The covariates were age, T1 slice- thickness, sex, and group. Post hoc testing was con ducted using estimated marginal means (EMMEANS) to evaluate group differen ces while controlling for covariates.

Results

or Findings: 20 MOGAD patients, 37 AQP4+ NMOSD patients, 40 MS patients, and 31 HC were included in the analysi s. Age, sex, and group showed significant effects on MUCCA measurements. T he EMMEANS of MUCCA values were lower for the NMOSD group (86.1+/ -1.5), followed by MOGAD (89.3+/-1.7), MS (90.3+/-1.2), and HC (91.6+/ -1.5). Pairwise comparison between groups showed no statistically s ignificant differences between the MOGAD and other groups. In contrast, a statistically significant difference between the NMOSD and HC groups and a tr end towards significance between the NMOSD and MS groups were o bserved.

Conclusion

Our proof of concept study shows the feasibility of calculating cervical volume loss from routine brain MRI using a utomated software. While greater spinal cord tissue loss is evident in NMOSD , the degree of spinal cord tissue loss in MOGAD is lower and not significantly different compared to MS and HC. Additional analyses on a larger cohort are underway.

Limitations

N/A Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee notification can be found under study number i20-01556 Author Disclosures: Eyal Lotan: Nothing to disclose Vincenzo Anania: Employee: icometrix Thibo Billiet: Employee: icometrix Ilya Kister: Nothing to disclose Itay Lotan: Nothing to disclose Medial lemniscus as a diagnostic marker: differenti ating multiple sclerosis from small vessel disease *W. H. E. Hamed*, D. Werring, D. S Lynch, R. Jäger, T. A. Yousry; London/UK ([email protected]) Purpose or Learning Objective: Evaluate the diagnostic utility of medial lemniscus (ML) in differentiating between multiple sclerosis (MS) and small vessel disease(SVD).

Methods

or Background: Distinguishing between MS and SVD remains a significant challenge in the elderly due to their o verlapping clinical presentations, emphasising the necessity for reliab le, non-invasive neuroimaging markers such as ML. A retrospective st udy analysed 270 MRI scans(100 MS, 170 SVD). SVD subtypes included arter iosclerotic(50), cerebral amyloid angiopathy(CAA)(50), mixed(50), and genetic (20). The signal intensity of ML in the pontine tegmentum was assessed visuall y on T2-w and FLAIR images. Statistical analysis included univariable t ests to identify differences between MS and SVD, followed by multivariable logis tic regression to determine independent predictors of ML involvement.

Results

or Findings: SVD patients were significantly older than MS patie nts (mean age:68±13vs43±10 years, p<0.001) and had lower female predominance (44.7%vs64%). MS patients had no ML in volvement(87%) or unilateral involvement(13%) with no bilateral cases , while 38% of SVD patients had bilateral involvement(p<0.001). Among SVD categ ories, ML involvement was most frequent in genetic(80%), followed by mixe d(64%) and arteriosclerotic(46%). The least affected was CAA(2 0%). In MS cohort, ML hyperintensity was associated with higher lesion lo ad(p<0.001) but not the disease duration(p=0.4). Among SVD subtypes, the mu ltivariable analysis revealed that lesion load(p<0.001) and hypertension (p=0.002) were the strongest predictors of ML involvement, followed by age(p=0.007), diabetes(p=0.023), and hypercholesterolemia(p=0.048 ). Factors such as alcohol, gender, and smoking were not significant p redictors.

Conclusion

Our study establishes ML as a reliable radiological marker for distinguishing MS from SVD, particularly when there is bilateral involvement, evidenced by a statistically significant presence i n SVD and a notable absence in MS. Moreover, the variation in ML involvement ac ross SVD subtypes, specifically its minimal presence in pure CAA, sugg ests its potential role in differentiating specific SVD categories.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: A retrospective study Author Disclosures: Tarek A. Yousry: Nothing to disclose Weaam Hamed Elsayed Hamed: Nothing to disclose Rolf Jäger: Nothing to disclose David Werring: Nothing to disclose David S Lynch: Nothing to disclose Brain Disconnection and Atrophy Assessment Multiple Sclerosis Converters *S. Hannoun*, S. Ghazal, L. Halawi, C. Al-Dahouk, N . El-Ayoubi, S. Khoury; Beirut/LB ([email protected]) Purpose or Learning Objective: Differentiating patients who convert from relapsing-remitting Multiple sclerosis (RRMS) to se condary progressive MS (SPMS) remains a critical challenge, as early ident ification of converters can significantly impact treatment strategies. This stu dy explores specific brain regions associated with disconnection probabilities and volume reductions, aiming to identify potential MRI biomarkers predict ive of conversion. This study aimed to investigate whether distinct patterns of W M disconnection and regional brain atrophy are associated with RRMS con version to SPMS.

Methods

or Background: We retrospectively analyzed 47 RRMS patients (17 Converters and 30 non-converters) who underwent bas eline and follow-up MRI scans approximately 1.5 years apart. Mixed-effects models evaluated the interaction between conversion status (converters v s. non-converters) and time across various brain regions, focusing on disconnec tion probabilities and volumetric changes measured using the Vol2Brain too l.

Results

or Findings: Converters exhibited significant disconnection in k ey white matter tracts, including the uncinate fascicu lus, corticobulbar tract, superior longitudinal fasciculus, and cingulum para hippocampal parietal. These disruptions are linked to cognitive, emotional, and motor functions. Additionally, grey matter atrophy was more pronounced in converte rs, particularly in the precentral gyrus, temporal lobe, and thalamus. Lesi on burden and volume, especially in juxtacortical areas, were greater in converters, with increased third ventricle volume indicating more severe brain atrophy.

Conclusion

Specific patterns of white matter disconnection and regional brain atrophy are associated with conversion from RRMS to SPMS. These MRI biomarkers provide valuable insights into disease p rogression and offer potential therapeutic targets. Further validation i n larger cohorts is needed to integrate these findings into clinical practice. Abstract-based Programme 28 Wednesday

Limitations

A limitation of our study is its relatively small s ample size and retrospective nature, which may limit the generaliz ability of the findings. Funding for this study: No funding. Ethics committee - additional information: This study was approved by the Institutional Review Board (IRB), and all participa nts provided informed consent. Author Disclosures: Samia Khoury: Nothing to disclose Lean Halawi: Nothing to disclose Salem Hannoun: Nothing to disclose Nabil El-Ayoubi: Nothing to disclose Sola Ghazal: Nothing to disclose Cezar Al-Dahouk: Nothing to disclose 11:30-12:30 Research Stage 2 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 305 Career, workforce issues and radiologist visibility Moderator F. Mankertz; Greifswald/DE Radiological discrepancy review: a novel, customisa ble, cloud-based technology to implement the REALM paradigm *P. Brennan*¹, Y. Hughes-Roberts², J. Richenberg², I. Francis², M. Suleiman¹; ¹Sydney/AU, ²West Sussex/UK ([email protected]) Purpose or Learning Objective: To develop a geographically-limitless REALM infrastructure that facilitates an optimised and reflective educational activity.

Methods

or Background: Reviewing radiological discrepancies is a well- known activity that promotes diagnostic excellence, encourages reflection and minimises future errors. Formalisation of discrepan cy reviews can be seen with the Royal College of Radiology REALM (Radiology Eve nts and Learning Meetings) program where radiologists in the UK can engage both as authors and recipients. However, to promote a clinically-re alistic REALM activity, comprehensive radiologic interactions with full res olution anonymised images should occur in a geographically limitless way. The infrastructures to support this, are elusive. An existing technology DxCARES w as modified for this activity. This incorporated: multi-modality viewing capabilities and 3D and multi- planar reconstructions (MPR): advanced AI-powered a nonymization and masking engines to automatically remove sensitive d ata while maintaining diagnostic integrity; a user-friendly interface for users with different levels of expertise; a web-based architecture supporting high performance DICOM streaming and real time interaction with large data sets.

Results

or Findings: We have built a new technology which allows clinici ans to upload REALM cases from PACs systems and distrib ute across a geographically-limitless health enterprise to limit less recipients. Customisable authorship of each case is available so that each c reator can demand from recipients the type of case-specific interaction or reflection that is required, regardless of image or pathology-type. The multifun ctional cloud-based viewer allows the examination and manipulation of all case s as would occur with a primary diagnostic workstation. All REALM outputs a nd interactions are available for review and CPD accumulation. The new technology is currently being implemented across clinical centres.

Conclusion

This new technical innovation should promote widesp read clinically-realistic REALM engagement by simplifyin g time/cost implications. The potential of radiological discrepancy review sh ould be maximised.

Limitations

N/A Funding for this study: N/A Ethics committee - additional information: N/A Author Disclosures: Jonathan Richenberg: Nothing to disclose Patrick Brennan: Founder: DetectedX Pty Ltd Ian Francis: Nothing to disclose Ynyr Hughes-Roberts: Nothing to disclose Moayyad Suleiman: Founder: DetectedX Pty Ltd Sex differences in inappropriate imaging requests: Insights from the Medical Imaging Decision And Support (MIDAS) trial *S. Dijk*¹, C. Wollny², T. Kroencke², M. G. M. Huni nk¹; ¹Rotterdam/NL, ²Augsburg/DE ([email protected]) Purpose or Learning Objective: We analyzed sex-related disparities in inappropriate imaging requests using data from the Medical Imaging Decision And Support (MIDAS) trial.

Methods

or Background: This study analyzed baseline data from the MIDAS trial, a multi-center cluster randomized trial cond ucted at three German academic hospitals. The study population encompasse d all imaging requests submitted to the 26 participating departments via t he computerized order entry system during a 15-month period. Imaging appropriat eness was assessed using the ESR iGuide, a clinical decision support s ystem (CDSS). Requests were categorized as inappropriate if imaging was de emed unlikely to be indicated or if the potential risks outweighed the benefits for the patient. Chi- square tests were employed to compare the proportio n of inappropriate imaging requests between men and women, with a sign ificance level of 0.01. Secondary analyses explored differences in inapprop riate requests by age group and exam type, with a Bonferroni correction a pplied to account for multiple testing.

Results

or Findings: Women had more inappropriate imaging requests (7.32%) than men (6.08%; χ² = 37.176, p < 0.001, OR 1.22 [95%CI 1.12-1.33]). This disparity was particularly evident in the 25-6 5 age group and for MRI examinations. Further research is needed to explore the underlying causes of this discrepancy, including potential differences i n physician awareness of guidelines, adherence to guidelines, or limitations in the CDSSs ability to account for female-specific factors.

Conclusion

In our study clinicians were 22% more likely to req uest inappropriate imaging for women than for men across nearly all modalities and age-groups. While the absolute percentage-point dif ferences were small (1.24%), the disparity warrants further investigati on

Limitations

Our analysis judges each request independently, wit hout considering the cumulative impact of requests per i ndividual, underuse, or time between symptom onset and imaging. Our data only di stinguished male/female sexes. Funding for this study: The MIDAS study was funded by the German Innovation Fund (reference: Förderkennzeichen 01VSF 18008). Ethics committee - additional information: Approval from the Medical Ethics Review Committee was obtained under protocol number s 20-069 (Augsburg), B 238/21 (Kiel), 20-318 (Lübeck) and 2020-15125 (Ma inz). The trial is registered in the ClinicalTrials.gov register under registration number NCT05490290. Author Disclosures: Thomas Kroencke: Nothing to disclose Claudia Wollny: Nothing to disclose Myriam G. M. Hunink: Nothing to disclose Stijntje Dijk: Nothing to disclose Multicenter and multimodality evaluation of radiolo gical workload and development of a benchmarking metric *P. Dankerl*, J. Lang, A. Glaser, H. P. Beyer, M. F orsting; Dortmund/DE ([email protected]) Purpose or Learning Objective: The increasing complexity and volume of radiological examinations have led to growing conce rns about radiologists' workload, diagnostic efficiency and accuracy. The a im of this study was to evaluate and compare radiological workload in a mul ticenter and multimodality analysis while creating a benchmarking metric.

Methods

or Background: Over 100 days radiologists’ reading times for all examinations and modalities across 34 centers have been collected utilizing RIS-export, while grouping these into 67 different body regions. In order to sort, retrieve and evaluate the various output formats fr om the RIS-data, a uniform nomenclature was introduced and all examinations re ceived these additional unique identifying labels. For benchmarking, report ing times were translated into relative value units – as defined by us as the mean reporting time of all X- ray exams and termed RADPoints.

Results

or Findings: We examined 290.748 examinations and found significant variations in average reporting times a cross modalities and body region, e.g. abdominal MRI 14.83min and CT 13.59min . Reporting times varied considerably, with the highest average times observ ed in complex examinations such as cardiac MRI (26.04min) and CT (16.59min). Conversely, regions like the fingers showed much lower averages , e.g. 10.95min for MR and 9.83min for CT. We found our relative value uni t time of one RADPoint to equivalent to 2.23min which further served as commo n devisor in order to allocate specific RADPoints to all body region spec ific examinations.

Conclusion

The findings underscore the need for targeted workl oad management strategies in radiology departments, par ticularly when high- complexity cases are in the mix. Benchmarking repor ting times across modalities and body regions by the means of present ed RADPoints provides a Abstract-based Programme 29 Wednesday critical reference for optimizing radiologist workl oad, potentially leading to enhanced diagnostic accuracy and efficiency.

Limitations

None Funding for this study: None Ethics committee - additional information: Retrospective evaluation and blinding of patinet as well as doctor identyfiers m akes this not applicable Author Disclosures: Michael Forsting: Employee: Evidia Peter Dankerl: Employee: Evidia Haemi Phaedra Beyer: Employee: Evidia Jochen Lang: Employee: Evidia Andrzej Glaser: Employee: Evidia Assessing the perceived impact of ESOR training pro grams on radiologists' professional development *J. Gregory*¹, M. L. Kofoed-Ottesen², B. Lindlbauer ², C. Loewe², V. Vilgrain¹; ¹Clichy/FR, ²Vienna/AT ([email protected]) Purpose or Learning Objective: This study evaluates the perceived impact of European School of Radiology (ESOR) training progra ms on radiologists' professional development.

Methods

or Background: A cross-sectional survey targeted alumni who participated in ESOR fellowships from 2011 to 2023. The survey included questions on demographics, professional background, ESOR program details, and career impact. Data were collected via a web-ba sed questionnaire and analyzed using descriptive statistics and thematic analysis.

Results

or Findings: A total of 190 participants responded, with a media n age of 31 years (range 29-33), and 54% were female. Mos t worked in public healthcare (62%) and were involved in academic acti vities (24%). Fellowship types included Visiting Scholarship Program (44%), Bracco Fellowship (32%), and Exchange Program for Fellowships (25%). The maj ority (59%) reported the fellowship helped them reach their current position , and 35% noted it upgraded their CV. Significant application of learned skills was reported by 69%. Ongoing cooperation with former tutors was maintained by 54 %. Financial support was crucial, with 41% stating they could not have compl eted the training without it, 33% considering it very important, and 13% deeming it important. Participants rated the impact on clinical skills with a median s core of 9 out of 10. Other areas of impact included research skills (median sc ore 7), subspecialization (median score 9), exposure to diverse practices (me dian score 9), networking opportunities (median score 10), and personal and p rofessional growth (median score 10).

Conclusion

ESOR training programs significantly enhance radiol ogists' professional development through comprehensive supp ort, high-quality training, and substantial financial aid, ensuring p articipants are well-equipped for career advancement.

Limitations

This study has limitations, including reliance on s elf-reported data, potential recall-bias, and a 20% response rat e. The survey may not fully represent all ESOR alumni, and program heterogeneit y could influence the generalizability of results. Funding for this study: None Ethics committee - additional information: Given the nature of the survey involving professional feedback without sensitive p ersonal data, ethical approval was not required. However, all participant s were informed about the purpose of the survey and the anonymous handling of their data. Author Disclosures: Jules Gregory: Nothing to disclose Brigitte Lindlbauer: Board Member: European School of Radiology Office Valérie Vilgrain: Board Member: European School of Radiology Office Christian Loewe: Board Member: European School of R adiology Office Mathias Lange Kofoed-Ottesen: Board Member: Europea n School of Radiology Office Expanded AI learning: AI as a tool for human learni ng S. Faghani¹, C. Tiegs-Heiden², M. Moassefi², G. Pow ell², M. Ringler², B. J. Erickson², *N. Rhodes*²; ¹Minneapolis, MN/US, ²Rochester, MN/US ([email protected]) Purpose or Learning Objective: To use artificial intelligence (AI) as a teaching tool to identify new imaging findings and improve the radiologist’s ability to recognize subtle imaging findings withou t additional AI assistance.

Methods

or Background: We studied the learning of a new task by humans using a deep learning (DL) model that can identify sex differences from frontal knee radiographs with high accuracy. We then ascert ained imaging features via occlusion interpretation maps (“heat maps”) to help human readers improve their ability to perform this task. Three human rea ders were tasked to classify 50 frontal knee radiographs into male and female se x. They were then informed of the patient’s sex and were given the as sociated AI-derived “heat maps” for subsequent study. After two weeks, the gr oup was retested with a new set of 50 radiographs.

Results

or Findings: The DL model categorized sex with 0.96 accuracy. Th e average accuracy of the 3 human readers was initial ly 0.59 (range: 0.56-0.66). After study of AI-derived “heat maps” and associate d radiographs, the average accuracy of the human readers increased to 0.80 (ra nge: 0.76-0.84), a statistically significant improvement (p=0.0270).

Conclusion

We believe this improvement serves as a proof of co ncept for the use of AI as a tool for discovery science to advanc e human learning, an idea that we have not seen advanced in the radiology lit erature.

Limitations

1. This is an education item that does not fall wel l into an AI or MSK section. 2. Understanding how AI algorithms wor k is limited. 3. Our study is small. Large enough to prove an improvement afte r training, but too small to dissect more. Funding for this study: None. Ethics committee - additional information: N/A Author Disclosures: Bradley James Erickson: Nothing to disclose Mana Moassefi: Nothing to disclose Garret Powell: Nothing to disclose Shahriar Faghani: Nothing to disclose Christin Tiegs-Heiden: Nothing to disclose Michael Ringler: Nothing to disclose Nicholas Rhodes: Nothing to disclose From observation to interpretation: elevating repor ting skills in radiology *S. Ramirez*; Bogota/CO ([email protected]) Purpose or Learning Objective: The preparation of a radiological report involves integrating knowledge, skills, and trainin g. While many articles offer recommendations for quality reports, there is a lac k of literature on teaching strategies for efficiently constructing these repor ts. Despite this, some residents and radiologists have developed this comp etency over time, revealing a gap in understanding how residents acqu ire the critical thinking skills needed. This work aims to determine how radi ology residents learn to create their reports.

Methods

or Background: A qualitative multi-case study was conducted involving three cases represented by residents of a radiology residency program, the program's faculty, and a group of spec ialists in high-quality report production. The study included document reviews of the syllabus, semi- structured interviews, and a focus group to capture experiences, perspectives, and learning processes, as well as the factors and strategies that influenced the development of the knowledge and skills needed for creating radiological reports.

Results

or Findings: Information was collected on concepts, perceptions, beliefs, emotions, interactions, and experiences to construct coherent meanings about the topic. Common patterns and chall enges in learning to create radiological reports were identified. Partic ipants described the hidden curriculum as as the main component of the educatio nal process through which residents learn to write reports. Preparing p reliminary reports was regarded as the most valuable pedagogical activity for developing critical thinking skills. All residents agreed on the need t o design educational activities focused on learning to prepare radiological reports .

Conclusion

This study represents, to our knowledge, the first research on how the critical thinking skills necessary for crea ting a radiological report are acquired. By understanding how a person effectively acquires a skill, teaching strategies can be designed and adapted to maximize meaningful learning.

Limitations

None Funding for this study: No Ethics committee - additional information: Ethics committee Fundación Universitaria Sanitas Author Disclosures: Sandra Ramirez: Nothing to disclose The use of Instagram in medical education: gender d ifferences and student Satisfaction and gender differences compare d to traditional learning platforms (Blackboard) *M. Alvarez García*, M. E. Pueyo, P. Boldó, C. Urta sun Iriarte, A. Ezponda Casajus, P. Chico, J. Pueyo Villoslada; Pamplona/ES ([email protected]) Purpose or Learning Objective: This study aimed to explore the student satisfaction with Instagram as an educational tool in radiology education for fourth-year medical students, compared to a traditi onal learning platform.

Methods

or Background: The study focused on gender differences and overall preferences in platform usage among fourth- year medical students (64% female). Both Instagram and the traditional pl atform were used to provide multiple-choice test questions with images based on radiology class content. Students chose their preferred platform for complet ing quizzes, and their experiences were assessed via anonymous satisfactio n surveys. Statistical analysis explored gender-based differences in platf orm usage, preferences, Abstract-based Programme 30 Wednesday and perceived learning benefits. The Mann-Whitney U test was applied to analyze the results.

Results

or Findings: A total of 65.8% of participants chose Instagram as their preferred platform for answering clinical cases. No significant gender differences were observed in overall platform choic e. The majority (71%) of students expressed moderate to high satisfaction wi th Instagram as a complementary learning tool. A significant gender d ifference (p=0.019) was found in the reasons for preferring Instagram: wome n favored its user-friendly interface, while men prioritized its speed. No sign ificant differences were found between genders regarding Instagram's usefulness fo r personal study or whether it caused distractions (p=0.31). Additional ly, 97.1% of students preferred Instagram over other social networks.

Conclusion

Instagram was generally preferred over traditional learning platforms and other social media platforms for radi ology education, with a moderate to high level of student satisfaction.

Limitations

Certain students do not use Instagram. There might be missing data. Instagram tool “Stories” only allows users (s tudents) to see the content during a period of 24h. Funding for this study: None Ethics committee - additional information: This study has no ethical implications Author Disclosures: Pablo Chico: Author: Study design, Reviewer Cesar Urtasun Iriarte: Author: Data analysis, revie wer María Elena Pueyo: Author: Data analysis Patricia Boldó: Author: Data analysis Jesus Pueyo Villoslada: Author: Study design, data analysis, reviewer Miguel Alvarez García: Author: Study design, data c ollection, data analysis, reviewer. Ana Ezponda Casajus: Author: Reviewer Stock photos lead to inaccurate representations of radiologists by the media *L. Hartog*¹, J. M. L. Bosmans², J. Bouziotis², P. M. Parizel³, A. Snoeckx²; ¹Antwerp/BE, ²Edegem/BE, ³Perth/AU ([email protected]) Purpose or Learning Objective: Value-based radiology requires thorough understanding by patients of the role of radiologis ts in healthcare. News media frequently enlighten content on radiology with seem ingly outdated images from stock photo databases. We conducted a critical eval uation of the ability of stock photos in three major databases to reliably r epresent contemporary radiology.

Methods

or Background: We collected two hundred consecutive photos from three major commercial databases, using the keyword ‘radiologist’. Each image was evaluated for overall quality, image desc ription, descriptive elements, displayed body regions, diagnostic or int erventional context, depicted modality, inclusion of medical professiona ls, and accurate display on the viewing device.

Results

or Findings: Just 6% of the stock photos received a high overall quality score. Radiographs were the most frequently depicted modality (52%), followed by MRI (21%) and CT (19%), with a focus pr imarily on diagnostic imaging (99%). On 4% of the stock photos, images we re displayed on a PACS workstation, while 66% showed a radiologist holding up films ‘in the air’, and 10% on an X-ray viewbox. In 48% of the photos, the lead person was shown wearing a stethoscope. In all, only 9% of the stock photos presented a realistic view of a radiologist’s job.

Conclusion

The vast majority of stock photos fail to accuratel y portray the contemporary role of radiologists, contributing to gross misinterpretation of our profession by the general public.

Limitations

The evaluation was done by a radiology resident but reviewed by a consultant radiologist. To our knowledge, this ki nd of evaluation of an essential source of information for the public has never been undertaken before. Funding for this study: No funding was received for this study. Ethics committee - additional information: No patient data was used for this study. Author Disclosures: Annemiek Snoeckx: Nothing to disclose Paul M. Parizel: Nothing to disclose Jason Bouziotis: Nothing to disclose Jan Maria Lodewijk Bosmans: Nothing to disclose Laura Hartog: Nothing to disclose 11:30-12:30 Research Stage 3 Research Presentation Session: Interventional Radiology RPS 309 Image-guided tissue biopsies and thyroid interventions Moderator G. T. Yusuf; London/UK ([email protected]) Author Disclosures: Gibran Timothy Yusuf: Speaker: Siemens, Samsung, GE , Terumo, Bracco Diagnostic accuracy of core needle biopsy in patien ts with lymphoproliferative disorders: an optimized protoco l in 478 patients *P. Marra*, L. Dulcetta, R. Muglia, F. S. Carbone, A. Weber, S. Ferrari, A. Rambaldi, P. A. Erba, S. Sironi; Bergamo/IT ([email protected]) Purpose or Learning Objective: Surgical excision biopsy of lymph nodes stands as the gold standard for histological charac terization of lymphoproliferative disorders (LD). However, contem porary clinical practice increasingly leans towards core needle biopsy (CNB) . This study seeks to explore the factors influencing the diagnostic yiel d of CNB in LD.

Methods

or Background: This unicentric retrospective study presents data from patients referred for suspicion of new or rela psing LD. All patients underwent image-guided CNB of the target lesion bas ed on CT/PET findings. The primary endpoint was the diagnostic outcome, co mparing the ability to achieve a definitive diagnosis according to interna tional guidelines with CNB versus the necessity for subsequent excisional biop sy.

Results

or Findings: We enrolled 478 consecutive patients undergoing CNB , categorized into two cohorts. Cohort A comprised pa tients who underwent CNB using 18-20G full-core Menghini needles, with a median macroscopic fragment dimension of 1 cm. Cohort B included patie nts who underwent CNB with 16-18G semiautomatic guillotine needles, with a median macroscopic fragment dimension of 1.5 cm. In cohort A, the rate s of diagnostic and non- diagnostic (or non-sufficiently detailed) CNBs were 95 (73%) versus 35 (27%), respectively. In cohort B, these rates were 299 (86 %) versus 49 (14%).

Conclusion

The type and size of the needle used for CNB, as we ll as the histologic variant of LD, emerged as factors influe ncing diagnostic yield and accuracy. Given the swiftness of CNB compared to su rgical excision, optimizing this technique could streamline the diag nostic and therapeutic workflow for patients with suspected LD.

Limitations

Retrospective study; lack of control group undergoi ng surgery Funding for this study: None Ethics committee - additional information: Comitato Etico di Bergamo - Lymphoid Cancer Registry (NCT03131531) Author Disclosures: Sandro Sironi: Nothing to disclose Alessandro Rambaldi: Nothing to disclose Paola Anna Erba: Nothing to disclose Paolo Marra: Nothing to disclose Francesco Saverio Carbone: Nothing to disclose Riccardo Muglia: Nothing to disclose Alessandra Weber: Nothing to disclose Silvia Ferrari: Nothing to disclose Ludovico Dulcetta: Nothing to disclose The interplay of time and angle with the incidence of Pneumothorax in a CT-guided Lung Biopsy *N. Maalouf*¹, M. Abou Mrad¹, R. Benayed¹, R. A. Pu gliesi², J. C. Apitzsch¹; ¹Pforzheim/DE, ²Stuttgart/DE ([email protected]) Purpose or Learning Objective: This study evaluates the relationship between the needle-pleura angle and the duration of needle traversal (NTD) through lung tissue during CT-guided lung biopsies, and their impact on the incidence of pneumothorax.

Methods

or Background: 96 patients (54 m, 42 f, median age: 71 years) underwent CT-guided lung biopsies between January 2 020 and March 2024. Procedures were performed using a semi-automatic 18 G needle and a 17G trocar. The minimum delta (δmin) was calculated as the absolute difference between a 90° angle and the measured angles to the pleura and correlated Abstract-based Programme 31 Wednesday with pneumothorax occurrence. NTD was recorded from needle puncture to retraction. Patients with immediate intraprocedural pneumothorax were excluded. A multivariate analysis compared four pat ient groups, categorized by δmin (greater or less than ten degrees) and NTD (les s than or more than six minutes).

Results

or Findings: 96 biopsies were performed, with six patients exclu ded. Pneumothorax occurred in 22 out of 90 procedures. S ix of these 22 patients had δmin greater than 10 ° and NTD over six minutes; fiv e had δmin less than ten degrees and NTD under six minutes. In contrast, 68 patients did not experience pneumothorax. Of these, 29 had δmin less than ten degrees with NTD under six minutes, while 20 had δmin less than ten degrees with NTD over six minutes. Pneumothorax occurred in 14.7% of patients with δmin less than ten degrees and NTD under six minutes, compare d to 33.3% with δmin greater than ten degrees and NTD over six minutes.

Conclusion

Optimizing the needle-pleura angle and minimizing n eedle traversal duration during CT-guided lung biopsies c an reduce pneumothorax risk.

Limitations

A relatively small number of patients, as it was a single-center observational study. Funding for this study: No funding was received for this study. Ethics committee - additional information: Approved by the local ethics committee (F-2021-038) Author Disclosures: Rosa Alba Pugliesi: Nothing to disclose Jonas Christoph Apitzsch: Nothing to disclose Mazen Abou Mrad: Nothing to disclose Roua Benayed: Nothing to disclose Nour Maalouf: Nothing to disclose Sensitivity and specificity of FNAC with ROSE of lu ng lesions: a single- center experience on 643 patients *C. Verde*, L. Tarotto, S. Stilo, R. D'Angelo, V. S toia, V. Sanna, N. Martucci, L. Arenare, F. Fiore; Naples/IT ([email protected]) Purpose or Learning Objective: The purpose of this study was to evaluate sensitivity and specificity of FNAC with ROSE in th e diagnosis of lung lesions.

Methods

or Background: The study was conducted at the INT of Naples “Foundation Pascale” between 2013 and 2017. CT, CBC T FNAC was performed on 643 patients, of which 195 subsequentl y underwent surgical resection. Exclusion criteria were: platelet count ( 1.5). Unenhanced TC scans or CBCT were performed pre-biop sy. A 18 G coaxial needle is used and a thinner needle (23-22G) is ins erted into the lesion (2-3 times, if necessary). FNAC is associated with the e xtemporaneous examination (ROSE), which evaluates the adequacy of the sample. Pneumothorax is the most frequent complication but it is asintomatic in most cases and only 7 % of cases requests pleural drainage. No evidence of oth er major complications.

Results

or Findings: Comparing surgical cytological and histological sam ples, positive cytological samples are neoplastic in 99.3 % of cases (152/153) and non-neoplastic in only 0.6% of cases (1/153). The s ensitivity of FNAC is 86.8%, the positive predictive value 99.3% and the specificity 75%.

Conclusion

FNAC is a reference diagnostic tool in the characte rization of lung lesions for the purpose of target therapy and immunotherapy. It is an effective procedure with high sensitivity and speci ficity and low complication rate. Extemporaneous testing increases sensitivity, reducing the number of inadequate samples and false negatives.

Limitations

Single center study. Funding for this study: No funding was received for this study. Ethics committee - additional information: No information provided by the submitter. Author Disclosures: Vincenzo Stoia: Nothing to disclose Nicola Martucci: Nothing to disclose Francesco Fiore: Nothing to disclose Caterina Verde: Nothing to disclose Roberto D'Angelo: Nothing to disclose Veronica Sanna: Nothing to disclose Luca Tarotto: Nothing to disclose Laura Arenare: Nothing to disclose Salvatore Stilo: Nothing to disclose Targeted Fine-Needle Aspiration of Thyroid Nodules Guided by Shear Wave Elastography: A Novel Diagnostic Approach *M. Khaleghi*¹, A. Aziz Ahari²; ¹Tehran/IR, ²Boston , MA/US ([email protected]) Purpose or Learning Objective: The aim of this study is to investigate target thyroid nodules based on shear wave elastography.

Methods

or Background: Suspicious nodules requiring fine-needle aspiration (FNA) can sometimes be heterogeneous. Even though t hey may appear homogeneous on ultrasound at times, they are hetero geneous on shear wave elastography examination. Sampling from highly susp icious areas in shear wave elastography evaluation can lead to decreased inadequate samples and increased detection of malignant thyroid diseases. In the present study, patients referred for thyroid nodule sampling under go elastography assessment and are included in the study if they ex hibit heterogeneity. They then undergo thyroid FNA under ultrasound guidance using a 22-gauge spinal needle with stylet.

Results

or Findings: As of today, five patients have been examined. In t hese patients, inadequate or inappropriate samples were not observed. Suspicious areas in elastography in these patients had high in dices above 80 kPa, while non-suspicious areas were below 40 kPa. All these p atients were diagnosed with papillary thyroid cancer.

Conclusion

Thyroid nodules have always been a significant chal lenge. Shear wave elastography presents a new criterion for bett er diagnosis of suspicious nodules. Target thyroid FNA as a new concept should receive attention.

Limitations

One of the limitations of this study is the necessi ty of a highly skilled radiologist who can accurately sample high- risk areas. Funding for this study: No Funding is received. Ethics committee - additional information: This research has been reviewed and approved by the National Ethics Committee under the number IR.IUMS.FMD.REC.1399.177 Author Disclosures: Mohammadreza Khaleghi: Nothing to disclose Alireza Aziz Ahari: Nothing to disclose Comparison of radiofrequency ablation and microwave ablation in the treatment of benign thyroid nodules *P. Glielmo*, G. Mauri, D. Albano, S. Gitto, S. Fus co, L. M. Sconfienza; Milan/IT Purpose or Learning Objective: Radiofrequency ablation (RFA) and microwave ablation (MWA) are minimally invasive tec hniques used to treat benign thyroid tumours. This study aims to compare the efficacy and safety of these two methods.

Methods

or Background: We retrospectively evaluated all patients with benign thyroid nodules treated with either RFA or M WA at our Istitution between January 2021 and December 2021. The primary outcomes assessed were the volume reduction rate (VRR) of the ablated areas at 1, 6, 12 and 24 months, procedure duration, and complication rates.

Results

or Findings: A total of 56 patients were enrolled, 35 treated wi th RFA and 21 with MWA. At 1 month, the VRR was 57% in the MWA group and 48% in the RFA group (p=0.045). At 6 months, both group s showed a VRR of 72%. At 12 months, the VRR was 75% in the MWA group and 76% in the RFA group and at two years of 76% in MWA group and 75% in RFA group. Both techniques achieved significant volume reduction wi th no major complications reported.

Conclusion

Both RFA and MWA are effective and safe for treatin g benign thyroid nodules. MWA demonstrated a higher initial VRR at 1 month, while both techniques achieved similar efficacy at 6, 12 and 24 months. These findings support the use of either method as viable non-surgical alternatives for patients seeking treatment options for benign thyro id nodules.

Limitations

Limitations of this study include its retrospective design, relatively small sample size, and absence of a control group f or direct comparison. Funding for this study: None Ethics committee - additional information: “This study was approved by the Ethical Committee of IRCCS Ospedale Galeazzi - Sant 'Ambrogio, and all participants provided informed consent in accordanc e with the Declaration of Helsinki.” Author Disclosures: Pierluigi Glielmo: Nothing to disclose Stefano Fusco: Nothing to disclose Salvatore Gitto: Nothing to disclose Luca Maria Sconfienza: Nothing to disclose Giovanni Mauri: Nothing to disclose Domenico Albano: Nothing to disclose Image-guided thermal ablation as an alternative to surgery for papillary thyroid microcarcinoma, a 7-year experience G. Mauri, *P. Glielmo*, D. Albano, S. Gitto, S. Fus co, L. M. Sconfienza; Milan/IT Purpose or Learning Objective: Thermal ablation has emerged as a minimally invasive treatment option for thyroid mic rocarcinomas, offering an alternative to surgery. This study presents a 7-yea r experience (2018-2024) in treating thyroid microcarcinomas with thermal ablat ion.

Methods

or Background: We retrospectively evaluated all patients who underwent US-guided thermal ablation for thyroid mi crocarcinomas between January 2018 and September 2024. We assessed the te chnical efficacy of the procedure, complications, and local or distant recu rrences. Abstract-based Programme 32 Wednesday

Results

or Findings: Of the 59 patients referred for evaluation, 5 were deemed unsuitable for thermal ablation, and 1 opted for surgery. The remaining 53 patients (40 females, 13 males; mean a ge 51.4 ± 7.7 years) underwent thermal ablation successfully. The proced ure was well-tolerated with no major adverse events. The follow-up period ranged from 71 to 2 months (mean 36 months). No local or distant recurr ences occurred.

Conclusion

Thermal ablation is a safe and effective treatment for thyroid microcarcinomas, providing a viable alternative to surgery. This 7-year experience supports its role as a standard treatmen t option for thyroid microcarcinomas, demonstrating excellent outcomes i n terms of safety and efficacy in tumour control.

Limitations

Limitations of this study include its retrospective design, relatively small sample size, and absence of a control group f or direct comparison. Funding for this study: None Ethics committee - additional information: This study was approved by the Ethical Committee of IRCCS Ospedale Galeazzi - Sant 'Ambrogio, and all participants provided informed consent in accordanc e with the Declaration of Helsinki. Author Disclosures: Pierluigi Glielmo: Nothing to disclose Stefano Fusco: Nothing to disclose Salvatore Gitto: Nothing to disclose Luca Maria Sconfienza: Nothing to disclose Giovanni Mauri: Nothing to disclose Domenico Albano: Nothing to disclose Transarterial Embolization Outperforms Radiofrequen cy Ablation for Thyroid Goiters Exceeding 100 mL: A Study on Effica cy and Safety W-C. Lin, *Y. J. Lee*, C-K. Wang, A-N. Lin, Y-S. Ch en, C. Y. Lee, P-L. Chiang, C. Y. Lu; Kaohsiung City/TW ([email protected]) Purpose or Learning Objective: This study aims to compare the efficacy and safety of radiofrequency ablation (RFA) and transar terial embolization (TAE) in managing large benign thyroid nodules (BTNs), parti cularly those exceeding 100 mL, where established guidelines are limited.

Methods

or Background: This retrospective multicenter study, conducted from January 2018 to May 2022, included 70 patients with a total of 76 large BTNs. Of these, 53 underwent RFA and 17 underwent T AE. Nodules were categorized by initial volume ( 100 mL) and diameter (9 cm). Treatment efficacy was evaluate d using the volume reduction rate (VRR) at 6 months. Complications, as well as improvements in symptom and cosmetic scores, were documented and an alyzed.

Results

or Findings: At 6 months, TAE demonstrated a significantly highe r mean VRR than RFA (p = 0.007), especially for nodules larger than 100 mL (TAE: 63.34% vs. RFA: 49.71%; p = 0.035). The complication rate in the TAE group (5.88%) was lower than that in the RFA group, where transient hoarseness and hematoma were common complications. Both treatments resulted in significant improvements in symptom and cosmetic scores (p < 0.001), with TAE providing greater improvements in larger nodules.

Conclusion

TAE is more effective and has fewer complications t han RFA for the treatment of large BTNs exceeding 100 mL. These findings suggest that TAE may serve as a minimally invasive alternative t o surgery for patients with large thyroid nodules. Further prospective studies are necessary to develop size-specific guidelines for selecting between RFA and TAE.

Limitations

Our study's limitations include the absence of gros s pathological confirmation, retrospective design, short-term foll ow-up, and variations in TAE protocols. Funding for this study: No funding was received for this article. Ethics committee - additional information: Chang Gung Medical Foundation Institutional Review Board IRB No.: 202401138B0 Author Disclosures: Yun Ju Lee: Nothing to disclose Cheng-Kang Wang: Nothing to disclose Wei-Che Lin: Nothing to disclose Pi-Ling Chiang: Nothing to disclose Chia Yin Lu: Nothing to disclose Yueh-Sheng Chen: Nothing to disclose An-Ni Lin: Nothing to disclose Chih Ying Lee: Nothing to disclose 11:30-12:30 Research Stage 4 Research Presentation Session: Musculoskeletal RPS 310 Selected applications of MSK ultrasound Moderator F. Zaottini; Genoa/IT ([email protected]) Incidence of Palmar Fibromatosis Nodules Following Carpal Tunnel Release: A Prospective High-Resolution Ultrasound S tudy *S. A. Jengojan*¹, A. Piacentini¹, F. P. Papa¹, V. König¹, D. Albano², Ž. Snoj³, G. Ivanac⁴, G. Bodner¹; ¹Vienna/AT, ²Milan/IT, ³Ljubljana/SI, ⁴Zagreb/HR ([email protected]) Purpose or Learning Objective: Palmar fibromatosis (PF), also known as Dupuytren’s disease (DD), is characterized by fibro us nodules and thickening of the palmar fascia, potentially leading to functi onal impairment of the hand. While its etiology remains unclear, we have observe d over and over again in our daily routine that surgical procedures, such as carpal tunnel syndrome (CTS) release may trigger the development of fibrom atosis in the palmar Fascia. In this study we investigated whether patie nts who undergo CTS surgery are more likely to develop palmar fibromato sis nodules, compared to healthy individuals and those with non-operated CTS , using high-resolution ultrasound (HRUS) imaging.

Methods

or Background: In this prospective study, we examined 100 patients who had previously undergone CTS surgery, screening for palmar fibromatosis nodules with HRUS. We further evaluated two control groups: 50 healthy volunteers and 30 patients diagnosed with CTS who h ad not yet had surgery. The prevalence of PF nodules across these groups wa s documented and compared.

Results

or Findings: Out of the 100 patients who had CTS surgery, 35% (n=35) were found to have PF nodules. In contrast, only 1 individual in the group of healthy volunteers had nodules 2% (n=1). N one of the patients with non-operated CTS showed evidence of nodules. These findings suggest a highly significant increase in PF among post-CTS su rgery patients compared to both healthy controls and those with non-operate d CTS.

Conclusion

Our findings suggest that CTS release surgery may b e linked to a higher incidence of DD. This raises important quest ions about the role of surgical intervention in the development of fibroma tosis.

Limitations

A limitation of our study is the lack of preoperati ve imaging, which could not exclude the possibility that the palmar f ibromatosis nodules already existed before CTS surgery. Funding for this study: No funding was needed Ethics committee - additional information: The study was approved by the Institutional Review Board. (vote Number 2028/2024) approved the study. Author Disclosures: Alessio Piacentini: Nothing to disclose Gerd Bodner: Nothing to disclose Francesco Pio Papa: Nothing to disclose Suren Armeni Jengojan: Nothing to disclose Gordana Ivanac: Nothing to disclose Žiga Snoj: Nothing to disclose Viktoria König: Nothing to disclose Domenico Albano: Nothing to disclose The effect of soft tissue compression on shear wave velocity of peripheral nerves J. Peterca, Ž. Snoj, *G. Omejec*; Ljubljana/SI ([email protected]) Purpose or Learning Objective: The objective was to determine the effect of soft tissue compression applied by the US probe on peripheral nerve shear wave velocity (SWV) measurements

Methods

or Background: Lower and upper limbs of 20 fresh cadavers were dissected at hip and shoulder region, strapped to i nox tray and emerged into the 37°C water bath. The US probe was mounted on ro botic arm Yaskawa MH6 (Yaskawa Electric, Fukuoka, Japan). Median nerv e at wrist and forearm, and sciatic nerve just proximal to tibial and fibul ar nerve branching were scanned longitudinally using US Resona R9 (Mindray Bio-medical Electronics, Shenzhen, China) and L14-3WU transducer. The compre ssion of the soft tissue applied by US probe was gradually increased from no contact to 10mm (wrist), 20mm (forearm), and 30mm (posterior thigh) in successive steps of 1, Abstract-based Programme 33 Wednesday 2, 3, 5, 10, 15, 20, 25 and 30mm. After every succe ssive step, SWV was measured three times and median was used for statis tical analysis.

Results

or Findings: Average distances between the US probe and peripheral nerve at baseline were 5.4mm for median nerve at the wrist, 18.9mm for median nerve at the forearm, and 24.8mm for sciatic nerve. For median nerve at wrist, significant changes in SWVs were found for compression distances of ≥2mm, and for median nerve at forearm for compression distances of ≥15mm (p <0.05). No changes in SWV were found for sciatic nerve.

Conclusion

The influence of soft tissue compression on SWV mea surements depends on the thickness of the soft tissue between the probe and peripheral nerve. No or only minor pressure is advised when me asuring SWV of superficial peripheral nerves and more pressure can be applied when measuring SWV of deeper peripheral nerves.

Limitations

Possible small changes in ROI position after roboti c arm movement Funding for this study: The authors report grants from The Republic of Slovenia Research Agency (Grant No. P3-0338 and J3- 4507) Ethics committee - additional information: The study was approved by the National Medical Ethics Committee of Slovenia (Nr. 0120-25/2023/3). Author Disclosures: Gregor Omejec: Nothing to disclose Jošt Peterca: Nothing to disclose Žiga Snoj: Nothing to disclose Dominance differences in hamstring stiffness among athletes: Insights from ultrasound shear wave elastography *B. S. Alvarez De Sierra*, P. Nieto; Madrid/ES Purpose or Learning Objective: Hamstring injuries are prevalent among football players, necessitating a deeper understand ing of muscle stiffness and its implications. This study investigates the diffe rences in hamstring stiffness between dominant and non-dominant legs in athletes using Shear Wave Elastography (SWE).

Methods

or Background: A retrospective study was conducted involving 30 football players (15 males, 15 females). Ultrasound SWE was used to measure shear wave velocities (SWV) in the semitendinosus, semimembranosus, and biceps femoris muscles. Data were analyzed based on leg dominance and sex, with statistical significance set at p < 0.05.

Results

or Findings: Significant differences were found in semitendinosu s stiffness between sexes, particularly in the non-do minant leg (p = 0.02). Males exhibited higher SWV in the semitendinosus mu scle compared to females, with mean SWV of 3.8 m/s (dominant) and 4. 1 m/s (non-dominant) for males versus 2.9 m/s (dominant) and 2.8 m/s (non-do minant) for females. No significant SWV differences were observed in the se mimembranosus and biceps femoris muscles between sexes or based on le g dominance.

Conclusion

SWE effectively identifies dominance-related differ ences in hamstring stiffness, offering valuable insights for optimizing athletic performance and reducing injury risks. Future resea rch should explore the implications of these findings on long-term injury prevention and rehabilitation strategies.

Limitations

The sample size was small (30 players), the exact d egree of stiffness remains uncertain without a histological correlation serving as a standard of reference for the hamstrings. Other fac tors such as muscle geometry, fiber orientation, and intramuscular pres sure may influence SWV. Funding for this study: No funding Ethics committee - additional information: The study was approved by the Clinical Research Ethics Committee. CEIC 2024.232 Author Disclosures: Patricia Nieto: Nothing to disclose Beatriz Sierra Alvarez De Sierra: Nothing to disclo se Ultrasound-Guided Interphalangeal Injection (US-IPI ) of Mucoid Cysts: Technical Notes and Clinical Efficacy E. Faiella, *E. Vergantino*, D. Santucci, A. Bruno, G. Pacella, R. F. Grasso; Rome/IT ([email protected]) Purpose or Learning Objective: Digital mucous cysts (DMCs) are common soft tissue tumors affecting interphalangeal joints . Various treatment options exist, with surgical excision being the standard. U ltrasound-guided cortisone injection into the distal interphalangeal (US-IPI) joint has been proposed as a therapeutic alternative. This study aims to assess the technical success and clinical efficacy of US-IPI in terms of swelling re solution and pain control.

Methods

or Background: This study assessed corticosteroid infiltration for mucoid cysts in interphalangeal joints. Fifty-two p atients (16 males, 36 females; median age 53, range 45-73) were treated b etween January 2020 and March 2023. Inclusion criteria included joint s welling, growth, and chronic pain >3 months; infections were excluded. The ultra sound-guided procedure used a 26-gauge needle and Triamcinolone acetonide (Kenacort). Pain and swelling were evaluated via the Numeric Rating Scal e (NRS) at 2 weeks, 1 month, 3 months, and 6 months. Statistical analysis was performed using SPSS (v.22)

Results

or Findings: The study assessed pain relief from ultrasound-guid ed corticosteroid infiltration for mucoid cysts. NRS s cores decreased significantly from a median of 6.8 pre-procedure to 3.2 at two we eks (p < 0.01) and to 0 by one month (p < 0.01). Eighty-three percent of patie nts responded positively; 17% experienced persistent pain (average NRS 6.2). A secondary infiltration reduced their scores to 3.5 at two weeks and 0 by o ne month (p < 0.01). For joint swelling, 68% had a 50% reduction within one month, with complete resolution by three months. No recurrences were obs erved at the six-month follow-up.

Conclusion

Ultrasound-guided injection for digital mucous cyst s offers effective pain relief and aesthetic improvement. Fu ture studies should assess long-term outcomes and compare efficacy with other treatments.

Limitations

The lack of extended follow-up beyond one year post -treatment limits the assessment of long-term outcomes. Funding for this study: No funding. Ethics committee - additional information: The study was conducted according to the guidelines of the Declaration of H elsinki. Ethical review and approval were waived for this study due to its retr ospective nature. Author Disclosures: Rosario Francesco Grasso: Nothing to disclose Amalia Bruno: Nothing to disclose Eliodoro Faiella: Nothing to disclose Elva Vergantino: Nothing to disclose Giuseppina Pacella: Nothing to disclose Domiziana Santucci: Nothing to disclose Clinical and Radiological Outcomes of Ultrasound Gu ided Closed- Circuit Irrigation of Calcific Tendinitis of the Shoulder: a prospective study *A. De Grandis*¹, C. D'Alessandro¹, G. Sussan¹, A. Crimì¹, D. Coraci¹, S. Masiero¹, R. Ragazzi², E. Quaia¹, F. Crimì¹; ¹Pa dova/IT, ²Venice/IT ([email protected]) Purpose or Learning Objective: Ultrasound-guided percutaneous treatments are a recognized and effective option for calcific tendinopathy of the shoulder. In this study, we enhanced the standard double-need le technique with a closed-circuit irrigation system and evaluated the clinical/radiological outcomes of the procedure.

Methods

or Background: We prospectively enrolled 24 patients (14 females; median age 54years, IQR:50-62) with painful calcifi c tendinopathy of the shoulder between October 2023 and March 2024. All p atients had a calcification >5 mm treated with ultrasound-guided closed-circuit irrigation, and the procedure duration was recorded. Ultrasound, ra diography evaluation, and OSS and SPADI clinical questionnaires were administ ered before and 3 months after the procedure.

Results

or Findings: After the procedure, there was a significant reduct ion in the size of the calcifications (12 mm, IQR:10-20 mm vs. 5.5 mm, IQR:2-10 mm; p=0.0001). The median duration of the procedure was 41 minutes (IQR:39-45 minutes). After the procedure, none of the patients experienced infections while two developed bursitis. There was a significa nt improvement in the OSS score 3 months after the procedure (16.5, IQR:10-23 vs. 32, IQR:36-45.5; p<0.0001) and a significant reduction in SPADI scor es: pain (88, IQR:74-95 before vs. 13, IQR:4-24; p<0.0001), disability (72, IQR:60-90 before vs. 8, IQR:4-20; p<0.0001), and total score (78, IQR:66-91 before vs. 11, IQR:4-20; p<0.0001). The improvement of SPADI total score was higher, although not significantly (p=0.2891), compared to a 2015 review .

Conclusion

The closed-circuit double-needle barbotage for calc ific tendinopathy of the shoulder is an effective treatm ent that improves both shoulder pain and function with a very low risk of short-term complications.

Limitations

No limitations were identified. Funding for this study: No founding were received for this study. Ethics committee - additional information: CET-ACEV code: 471n/AO/24 Author Disclosures: Giovanni Sussan: Nothing to disclose Carlo D'Alessandro: Nothing to disclose Daniele Coraci: Nothing to disclose Roberto Ragazzi: Nothing to disclose Filippo Crimì: Nothing to disclose Alberto Crimì: Nothing to disclose Emilio Quaia: Nothing to disclose Andrea De Grandis: Nothing to disclose Stefano Masiero: Nothing to disclose Abstract-based Programme 34 Wednesday Real-Time Elastosonography of the Achilles Tendon a nd Plantar Fascia: Predictive Insights for Diabetic Foot Ulcers V. Burulday¹, S. Ceylan Durmaz², A. Gungunes³, *A. Tezcan*¹; ¹Malatya/TR, ²Ankara/TR, ³Kirikkale/TR ([email protected]) Purpose or Learning Objective: This study aims to evaluate the plantar fascia and Achilles tendon in patients with diabeti c foot ulcers using two elastosonography methods: strain elastography (SE) and shear wave elastography (SWE). The goal is to identify biomech anical changes in these structures, which may predict the risk of diabetic foot ulcer development. This study is the first in the literature to evaluate bo th the Achilles tendon and plantar fascia together using both SE and SWE metho ds.

Methods

or Background: Twenty-five patients with type 2 diabetes and diabetic foot ulcers, along with 30 healthy individ uals, were evaluated. B-mode ultrasound, SE, and SWE methods were used to assess the thickness and stiffness of the Achilles tendon and plantar fascia . Diabetic foot ulcers were staged, and measurements were performed by a single experienced radiologist. The study compared patients with diabe tic ulcers to those without, analyzing changes in tissue stiffness and structure .

Results

or Findings: The study found significant increases in Achilles t endon thickness and stiffness in diabetic foot patients c ompared to the control group (p<0.0001). Plantar fascia stiffness was also signi ficantly higher in diabetic patients (p<0.0001). Subgroup analysis revealed tha t patients with foot ulcers had even greater Achilles tendon and plantar fascia stiffness than those without ulcers. No significant difference in planta r fascia thickness was observed between groups (p=0.539), but stiffness ch anges were evident.

Conclusion

The elastosonographic evaluation of the Achilles te ndon and plantar fascia provides valuable insights into the biomechanical changes in diabetic foot patients. Both SE and SWE are complem entary methods that may help predict the development of diabetic foot ulcer s. Early detection of stiffness and tissue changes through these methods can be cru cial in preventing ulcer formation.

Limitations

Operator dependence, lack of MRI comparison, small sample size, absence of reliability assessment Funding for this study: None Ethics committee - additional information: Decision number 15/03 dated 01.10.2018 Author Disclosures: Alperen Tezcan: Nothing to disclose Askin Gungunes: Nothing to disclose Veysel Burulday: Nothing to disclose Senay Ceylan Durmaz: Nothing to disclose 11:30-12:30 Room G1 Research Presentation Session: Radiographers RPS 314 Innovative imaging practices and patient- centred care: radiographers at the forefront of diagnostic excellence Moderators C. Messina; Milan/IT H. Precht; Middelfart/DK ([email protected]) Author Disclosures: Carmelo Messina: Grant Recipient: Bracco Imaging It alia, Echolight Analysis of a focus group survey on person-centred care (PCC) practices in Nuclear Medicine: insights from the European Con gress of Radiology 2024 *P. S. Costa*¹, D. Fonseca Ribeiro², M. Champendal³ , S. Murphy⁴, C. Baun⁵, C. Andersson⁶, A. Karangelis⁷, K. Borg Grima⁸, A. Geão⁹; ¹Porto/PT, ²London/UK, ³Lausanne/CH, ⁴Dublin/IE, ⁵Odense/DK, ⁶Uppsala/SE, ⁷Patra/GR, ⁸Naxxar/MT, ⁹Lisbon/PT ([email protected]) Purpose or Learning Objective: This study aimed to gather insights into the understanding, implementation, and challenges of Pe rson-Centered Care (PCC) practices among Nuclear Medicine professional s in Europe. Additionally, the focus group validated the questio ns used in the survey, serving as the ground work for a forthcoming Europe an-wide study involving Nuclear Medicine Radiographers.

Methods

or Background: A focus group survey was conducted during the European Congress of Radiology 2024. The participan ts included Radiographers and Nuclear Medicine Radiographers/Te chnologists from various European countries. The survey covered demo graphics, professional background, understanding of PCC, its application i n clinical settings, and factors aiding or hindering its implementation. The focus group also provided some informal feedback to validate the survey quest ionnaire for future use.

Results

or Findings: Thirty-two participants participated in this focus group and contributed to the preliminary survey, with n=5 ; 45% Radiographers and n=6; 55% NM Radiographers/Technologists . Most part icipants had between 1- 10 years of experience. Participants were from Alba nia (9%), Italy (27%), Denmark (36%), Malta (9%), and Portugal (3%). Repor ted Key factors aiding PCC implementation included good communication (45% ), empathy (40%), and sensitivity to patient characteristics (36%). T he reported barriers to the implementation of PCC included burnout (82%), compl exity of procedures (36%), and time constraints (27%). Additionally, th e focus group provided feedback on the questions used while aiding to impr ove the survey set-up.

Conclusion

The findings revealed a strong recognition of the i mportance of PCC among Nuclear Medicine professionals, but also significant challenges such as burnout and time constraints which could hi nder its implementation. Recommendations included, amongst others, the use o f the validated questionnaire to gain a broader understanding of PC C practices in Nuclear Medicine across a wider spectrum of European countr ies.

Limitations

The small sample size and the limited geographical spread of the participants. Funding for this study: No funding was received for this study. Ethics committee - additional information: Not applicablw Author Disclosures: Daniela Fonseca Ribeiro: Nothing to disclose Karen Borg Grima: Nothing to disclose Apostolos Karangelis: Nothing to disclose Ana Geão: Nothing to disclose Camilla Andersson: Nothing to disclose Christina Baun: Nothing to disclose Pedro Silva Costa: Nothing to disclose Shauna Murphy: Nothing to disclose Mélanie Champendal: Nothing to disclose Comparison of Virtual Monoenergetic Images of a cli nical Photon- Counting-Detector CT with a state-of-the-art Energy -integrated-Detector CT in Dual-Energy Mode *M. Scheweder*¹, D. Ecker¹, E. Wurzinger², A. Gratz er², P. Kullnig²; ¹Linz/AT, ²Graz/AT ([email protected]) Purpose or Learning Objective: Photon-counting detector computed tomography (PCD-CT) is a promising novel technique for clinical CT, with new opportunities for image optimization while reducing radiation dose compared to conventional energy-integrated detector CT (EID-CT) . Therefore, a PCD-CT and an EID-CT were compared to assess the image qua lity on spectral data and image reconstructions. The goal was to explore technical potentials for clinical practice.

Methods

or Background: A whole-body phantom was scanned on an EID-CT in Dual-Energy (DE) Mode and a PCD-CT with similar CTDIvol. Virtual monoenergetic images (VMI) were processed at 16 keV levels (40-190 keV) and different reconstructions. Signal-to-noise Rati o (SNR) and contrast-to- noise ratio (CNR) ROIs were evaluated from liver eq uivalent tissue for each level and reconstruction. Mann-Whitney U test was u sed to compare image quality. Besides, a dose-reduced PCD-CT scan was co mpared to the EID-CT scan.

Results

or Findings: PCD-CT-VMI data show significantly higher (all p<.0 5) SNR and CNR (all p<.001) than EID-CT across all keV levels and reconstruction methods. However, SNR and CNR highly depended on the reconstruction method and keV level. PCD-CT SNR/CNR exceeded ≥80 keV (SNR: +7% to +502% / CNR: +3% to +801%). A 40% dose -reduced PCD-CT scan provided higher dose-normalized SNR against th e EID-CT scan at 100- 190 keV (+35% to +272%). However, at 40-90 keV, PCD -CT had lower SNR (- 53% to -1%).

Conclusion

PCD-CT-VMI demonstrate higher SNR/CNR capabilities compared to EID-CT in DE-Mode. This advantage can b e used to optimize scan setups regarding radiation exposure. Few studi es have compared the VMI data of PCD-CT and EID-CT for this purpose. How ever, in addition to these promising objective results, subjective image assessment by radiologists is necessary to clarify diagnostic accuracy.

Limitations

This project is a phantom study and requires furthe r clinical research to apply findings in practice. Funding for this study: This project was internally funded by the Universit y of Applied Sciences for Health Professions Upper Austr ia. Project-Number: P- 2002-003 Abstract-based Programme 35 Wednesday Ethics committee - additional information: The project was evaluated by the core team of the Institutional Review Board of the University of Applied Sciences for Health Professions Upper Austria, with the following conclusion: "There are no objections to the execution of this s tudy in its current form". IRB- Number.: A-2022-018 Author Disclosures: Peter Kullnig: Nothing to disclose Alexander Gratzer: Nothing to disclose Eric Wurzinger: Nothing to disclose Mario Scheweder: Nothing to disclose Dominik Ecker: Author: DE is an employee at Siemens Healthineers. During the course of this work, he was a master’s student at the University of Applied Sciences for Health Professions Upper Austria and w orked on the project. Radiographer-operated urgent diagnostic imaging in hybrid mobile stroke unit: a pilot technical study in challenging conditions *D. Pakizer*¹, A. Chalánková², R. Líčeník³; ¹Ostrava/CZ, ²Olomouc/CZ, ³Peterborough/UK ([email protected]) Purpose or Learning Objective: Diagnostic assessment shift and treatment at emergency site proved beneficial for acute strok e or other neurological patients by using hybrid-mobile stroke unit (h-MSU) ambulance with mobile computed tomography (CT), portable ultrasound (US), and telemedicine onboard. We aimed to determine feasibility, safety, and efficacy of radiographer-operated h-MSU CT/US in advanced preho spital work-up for patients with acute neuroemergencies in challenging geographical/weather conditions.

Methods

or Background: In our pilot prospective open-label cohort study, h - MSU was available constantly for 19 consecutive day s (November/December 2023) in Czechia mountain/rural region. Patients we re examined by CT intracranially in standby ambulance. Extracranial c arotid US underwent selected stable patients with time of transport >30 min to stroke center (ambulance on move). The h-MSU CT/US efficacy was c ompared with standard in-hospital modalities; feasibility and sa fety were assessed.

Results

or Findings: Of 46 patients, 37 brain CT, 1 intracranial CT angiography, and 8 extracranial carotid US examinat ions were conducted. CT helped find contraindications for thrombolysis in 3 patients; 6 patients received the treatment. Of 106 CT scans, only 3% of full exa minations had to be repeated. Mobile-CT mean radiation dose was only sl ightly higher compared to in-hospital CT, findings and image quality were sim ilar. Low temperatures and uneven mountainous terrain were responsible for 4% of repeat CT scans. Good-quality US images were achieved and no hemodyn amically significant stenosis was found but approximately 15min extracra nial carotid examination was needed. Moreover, 2 US-guided cannulations were performed.

Conclusion

Mobile-CT and portable US onboard of h-MSU were saf e, feasible, and effective modalities for neuroemergen cy patients and proved to be beneficial for time reduction and faster treatme nt decision-making in challenging geographical/weather conditions.

Limitations

Older age of mobile-CT/US, CT contrast agent only f or study second-half, low patient number examined by US, and lack of Doppler US. Funding for this study: None. Ethics committee - additional information: Tomas Bata Hospital Zlin Ethics Committee (approval 2023-66) Author Disclosures: Anežka Chalánková: Nothing to disclose Radim Líčeník: Nothing to disclose David Pakizer: Nothing to disclose Optimisation in CT using tin filtration: A systemat ic Review *A. Bellizzi*, J. L. Portelli, P. Bezzina, G. Galea , F. Zarb; Msida/MT ([email protected]) Purpose or Learning Objective: To identify which non-contrast CT examinations benefit from image quality and radiati on dose optimisation using tin filtration (TF) and which optimisation strategy is best suited for this purpose.

Methods

or Background: The review was registered in PROSPERO, and used PICO to create the research question, and excl usion/inclusion criteria. From the research question, MeSH search terms were obtained and inputted into five electronic databases: PUBMED, Scopus, CIN AHL complete, Cochrane Central Register of Controlled Trials and Health & Medical Collection. Studies identified from the search were loaded into Covidence and reviewed by a team of 3 experts using PRISMA guidel ines. The Joanna Briggs Institute (JBI) critical appraisal tool was used to evaluate the studies while data extraction was performed using a self-designed vali dated tool.

Results

or Findings: From the retrieved 1479 studies, 410 were found to be duplicates leaving 1069 studies for title/abstract screening. Subsequently, 130 studies were included for full text-review, with a final 84 studies included for evaluation. TF was used to optimise scanning in 14 protocols. Scan parameters used in conjunction with TF as an optimi sation strategy were: iterative reconstruction (IR) level, tube voltage ( kV), pitch, rotation time and

Reference

mAs. Use of TF achieved a significant dos e reduction ranging from 17-95% in all protocols.

Conclusion

TF is an efficient dose reduction technique in non- contrast CT examinations, but has limitations meriting consider ation in terms of objective image quality. These limitations could potentially be solved by varying the IR level, however more studies are needed for this to be confirmed.

Limitations

Six studies could not be retrieved. A meta-analysis could not be conducted due to the heterogeneity of the studies. Paediatrics and CT protocols using IVCM were excluded. Funding for this study: No funding was received for this study. Ethics committee - additional information: None required - systematic review. Author Disclosures: Gabriel Galea: Nothing to disclose Jonathan Loui Portelli: Nothing to disclose Francis Zarb: Nothing to disclose Andrea Bellizzi: Nothing to disclose Paul Bezzina: Nothing to disclose The use of Ultra-Low Dose Computed Tomography in th e diagnosis of Suspected Physical Abuse in paediatric patients - A phantom study E. K. Mahon, A. A. Mohammed, A. England, R. Young, *N. Moore*, G. A. Curran, M. F. Mcentee; Cork/IE ([email protected]) Purpose or Learning Objective: Suspected physical abuse (SPA) poses a global threat to children, particularly those under the age of two. Current practices utilise conventional radiography skeletal surveys (SSs) to help with the diagnosis and management of SPA. The recent exp loration of ultra-low dose CT (ULDCT) could improve diagnostic accuracy o f SPA whilst minimising radiation exposure and thus is the focus of this st udy.

Methods

or Background: CT datasets were acquired on a paediatric phantom using ULDCT (DLP=1.49 mGycm2) and standard-dose (DL P=22.92 mGycm2) CT protocols. Participants (radiographers and radio logists) subjectively scored the image quality (IQ) of both protocols using a 5- point Likert scale. Mann- Whitney U tests assessed significant differences in IQ between protocols. Participants also estimated radiation dose differen ces and confidence in diagnosing SPA using the CT datasets provided.

Results

or Findings: Responses from 46 participants were included. Data were categorised into four anatomical areas; head/n eck, thorax, abdomen/pelvis and extremities. IQ scores were cons istently higher for STD when compared to ULDCT (H&N 2.7 vs 2.1; Thorax 2.8 vs. 2.2; Abdo/Pelvis 2.8 vs. 2.0 and Extremities 2.9 vs. 2.2; p<0.05). I mportantly, the ULDCT protocol scored highest in “Borderline acceptable” for all regions. Interestingly, 45 (97.8%) participants underestimated the dose dif ference (1.49 versus 22.92 mGycm2) between the two protocols. The ULDCT protoc ol scored a total of 38% (88/230) confidence for the diagnosis of SPA, w hereas the STD protocol scored a total of 70% (161/230) confidence.

Conclusion

Despite the IQ difference between protocols, ULDCT results are promising. Results may suggest that with adjustment s to the ULD protocol 'optimisation' CT could contribute further SPA diag nosis but further research, including clinical studies, are needed.

Limitations

Phantom-based study. Funding for this study: None. Ethics committee - additional information: Medical School Social Research Ethics Committee - University College Cork Author Disclosures: Mark F. Mcentee: Nothing to disclose Niamh Moore: Nothing to disclose Rena Young: Nothing to disclose Ahmed Abdulahad Mohammed: Nothing to disclose Gráinne Alison Curran: Nothing to disclose Andrew England: Advisory Board: RoClub Research/Gra nt Support: GE Healthcare, Organon Pharma Ltd. Board Member: EFRS Eimear Kate Mahon: Nothing to disclose Deep-learning based image reconstruction in body CT imaging: What is the real gain? A quantitative study *D. Delarbre*¹, E. Maturana¹, M. Scheffler¹, J. L. Navarro Quirante¹, D. Racine²; ¹Geneva/CH, ²Lausanne/CH ([email protected]) Purpose or Learning Objective: Compare the performance between deep learning-based Advanced intelligent Clear-IQ Engine (AICE), iterative Adaptive Iterative Dose Reduction (AIDR-3D) and filtered bac k projection (FBP) computed tomography (CT) image reconstruction algor ithms in terms of image texture, low-contrast lesion detectability, and dos e reduction potential.

Methods

or Background: An abdominal anthropomorphic phantom was scanned at five computed tomography dose index (CTD Ivol) level settings: Abstract-based Programme 36 Wednesday 10.3, 6.4, 3.3, 2.4, and 1.9 mGy. Images were recon structed using AICE, kernel “Body Sharp”, 1mm slice thickness, then AIDR -3D, kernel “FC08”, 1mm and 2mm slice thicknesses, and classic FBP includin g quantum denoising system (QDS+) reconstruction. Noise and contrast-de pendent spatial resolution were assessed through noise power spectr a (NPS) and target transfer functions (TTF). Texture similarity of the se algorithms was evaluated using peak frequency difference (PFD) and root mean square deviation (RMSD). Lesion detectability was quantified using a non-prewhitening (NPW) observer model with eye filter. The area under the curve (AUC) and receiver operating characteristic (ROC) served as the figure of merit (FOM). Dose reduction potential for AICE 1 mm, compared to AIDR -3D with 2mm slice thickness, was calculated to achieve equivalent AUC values.

Results

or Findings: At higher dose levels, AIDR-3D better preserved FBP - like noise texture. At lower doses, this difference diminished. The PFD for AIDR-3D ranged from -0.05 to -0.14, while for AICE it ranged from -0.13 to - 0.17. RMSD values followed a similar trend. AICE co nsistently achieved higher AUC values than AIDR-3D with 1mm slice thickness, w ith an increasing difference as dose decreased. AICE especially demon strated a radiation dose reduction potential of up to 45% compared to AIDR-3 D with 2mm slice thickness.

Conclusion

AICE provides equivalent low-contrast lesion detect ability at significantly reduced radiation doses compared to A IDR-3D, without adversely affecting noise texture.

Limitations

Constant kV, phantom morpho-type M Funding for this study: 0 Ethics committee - additional information: None Author Disclosures: Damien Racine: Nothing to disclose Jose Luis Navarro Quirante: Nothing to disclose David Delarbre: Nothing to disclose Enrique Maturana: Nothing to disclose Max Scheffler: Nothing to disclose The effect of simulated reduced temporal resolution and motion artefacts on CT-derived cardiac left ventricular ejection fra ction *M. W. Kusk*¹, O. Gerke², S. Hess², S. J. Foley³; ¹ Esbjerg/DK, ²Odense C/DK, ³Dublin/IE ([email protected]) Purpose or Learning Objective: To test whether mid-range CT-scanners, with low temporal resolution (TR) can reliably meas ure left ventricular ejection fraction, (LVEF), compared to high-end cardiac scan ners.

Methods

or Background: 77 low-dose functional CT datasets, from a 3rd generation DSCT scanner, with 66 msec (TR) , recons tructed at 5% intervals of the entire cardiac cycle were used Cardiac MRI serv ed as gold standard for classifying patients with potential heart failure ( LVEF below 50%). Reduced TR dataset were simulated by temporal averaging betwee n adjacent phases, using a MATLAB script. Furthermore, in 25 artefact-free d atasets, we inserted simulated discontinuity artefacts of varying magnit ude and location. LVEF was measured on a clinical workstation using standard ( ST) and Blood Volume (BV) modes. Absolute LVEF was compared between orig inal and simulated images with Bland-Altman plots and t-tests, while c orrelation between effective TR was examined wiht Spearman rank-correlation.

Results

or Findings: For BV-mode, LVEF was not significantly different between original and reduced TR images, (p=0.88) bu t significantly lower in ST-mode by 2.8% (p<0.01.) However, no patients were reclassified according to the 50% threshold. There was significant negativ e correlation between LVEF and effective TR in the ST, but not in BV mode. Mot ion artefacts increased the widths of 95% limits of agreement to 5.6% in ST mod e vs 2.8% in BV mode.

Conclusion

Halving effective TR did not affect LVEF in the BV- mode . This mode also provided lower dispersion in the presence of motion artafacts, making it the recommended measurement method. Mid-r ange scanners appear suitable for LVEF measurement, e.g. in conju nction with standard Chest-Abdomen-Pelvis scans.

Limitations

The main limitation is the lack of a true reference standard with superior temporal, and similar spatial resolution, thus making the effect of the two factors hard to disentangle. Funding for this study: The study was funded by: The Esbjerg Fund, Research Fund of Danish Radiographers Association, Karola Jørgensen Fund Ethics committee - additional information: Regional Author Disclosures: Martin Weber Kusk: Nothing to disclose Søren Hess: Nothing to disclose Oke Gerke: Nothing to disclose Shane J Foley: Nothing to disclose Patient-tailored contrast optimization in coronary CT angiography on Photon Counting CT *T. Busselot*, P. Giordano, W. Coudyzer, H. Bosmans , S. Dymarkowski; Leuven/BE ([email protected]) Purpose or Learning Objective: Standardized injection protocols are often integrated in coronary CTA (cCTA), yet failing to a ccount for patient variability. This study aims to introduce an evidence based inje ction protocol, achieving a target enhancement (expressed in terms of HU) in th e overall patient population.

Methods

or Background: In a first retrospective study, 162 cCTA patients, scanned on a photon-counting CT, were included. Inj ection parameters, demographic data and coronary enhancement were retr ospectively collected and HU enhancement was measured in the aorta ascend ens and proximal RCA using circular ROIs. Using the principle that i odine concentrations and HU correlate linearly, ideal iodine delivery rates (IDR) that would have provided a HU target of 500 in 55keV mono-energetic images wer e calculated from the HU of the original scans. Linear regression analysi s was performed with the different demographic parameters and their ideal ID R, to obtain an evidence based IDR. Secondly, 62 patients were prospectively recruited in an IRB- approved study and injected with newly proposed IDR s, wherefrom volume was calculated. To obtain lowest volumes iodine-sal ine dilution ratios were used. A minimal total volume of 60ml and fixed inje ction duration of 17s was defined to ensure complete coronary enhancement. Co ronary enhancement was measured in the same ROIs as retrospectively.

Results

or Findings: Ideal IDRs correlated best with bodyweight adjusted for Deurenberg’s fat formula (Pearson r=0.69), with IDR =0.0116x+0.4556 Mean coronary enhancement was 530±99HU (median: 539 and IQR:462 – 599), agreeing with the target HU of 500. Mean injected i odine volume was 51ml (median:51ml and IQR:45–57ml). All low iodine volum es could be administered under the same duration and injection rates (from 3 .5–3.8ml/s).

Conclusion

A new method for injection protocol optimization, b ased on linear regression and integrating iodine-saline dilutions, was introduced and achieved the preset target HU enhancement.

Limitations

Mono-centric study. Funding for this study: AI-POD project. HORIZON Action Grant Budget-Based. Grant number 101 080302. Ethics committee - additional information: UZ/KU Leuven ethics, internal number S58042 Author Disclosures: Walter Coudyzer: Nothing to disclose Hilde Bosmans: Nothing to disclose Tim Busselot: Nothing to disclose Pierpaolo Giordano: Nothing to disclose Steven Dymarkowski: Nothing to disclose 13:00-14:30 Research Stage 1 Research Presentation Session: Breast RPS 402 Advances in breast imaging: innovations shaping the future of cancer care Moderator R. Woitek; Vienna/AT ([email protected]) 18F-fluoroestradiol hybrid imaging in clinical mana gement of breast carcinoma *J. Ferda*, E. Ferdova, T. Barakova, S. Vokurka; Pl zen/CZ ([email protected]) Purpose or Learning Objective: 18F-fluoroestradiol is a novel radiopharmaceutical useful in breast carcinoma, the indications in clinical scenarios are under development. The purpose of the study is to assess the clinical impact of the imaging of the breast carcin oma with estrogen-positive receptors (ER+) using 18F-fluoroestradiol (18F-FES) PET/CT or PET/MRI according to its impact to the treatment decision m aking. The study is concerned in the different preference of PET/CT and PET/MRI in the staging and restaging.

Methods

or Background: 40 patients with estrogen positive breast carcinoma underwent the hybrid imaging after intravenous appl ication of 18F-FES, in 25 cases it was used PET/CT, in 15 cases PE/MRI. The r adiopharmaceutical was Abstract-based Programme 37 Wednesday injected in the activity of 2,5 MBA/kg and the imag ing was performed the imaging. In 10 patient PET/MRI was used as restagin g method, PET/MRI was performed in the 5 cases of the staging before surg ery with targeted full diagnostic MRI imaging of the breast in prone posit ion, followed by the trunk imaging in supine position, in five to seven steps. All PET/MRI were performed with the gadolinium contrast material, the imaging included brain imaging in T1 STARVIBE. PET/CT was performed using the continuous PET acquisition after CT with the administration of the iodinated contras t material, in 5 cases was performed in staging, in 20 cases in rest-aging

Results

or Findings: The most important information was detection of ER+ metastases when 18F-FDG-PET was negative (12x) - in cluding brain and liver metastases, the persistent ER+ of the metastases (7 x), staging of the disease (10x), the loss of the ER (4x) and the negative fin ding for metastases (2), no added information was found in 5 examinations.

Conclusion

18F-FES-PET provided the important clinical informa tion to treatment strategy, PET/MRI improves the imaging of brain and liver.

Limitations

Small sample Funding for this study: No Ethics committee - additional information: according to the Helsinky declaration Author Disclosures: Eva Ferdova: Nothing to disclose Tana Barakova: Nothing to disclose Samuel Vokurka: Nothing to disclose Jirí Ferda: Nothing to disclose Conspicuity as new CEDM descriptor: likelihood of M alignancy and Relationship With Breast Tumor Receptor Status L. Corradini, *D. Ballerini*, A. Bonanomi, E. D'Asc oli, G. Della Pepa, C. De Berardinis, E. Ancona, C. Depretto, G. P. Sca perrotta; Milan/IT ([email protected]) Purpose or Learning Objective: Lesion conspicuity, defined as the "degree of enhancement" relative to the background, was the fo cus of this retrospective monocentric study, which aimed to explore its corre lation with malignancy of lesions, histology, receptor profile and grading in breast cancer patients.

Methods

or Background: Two breast radiologist and one radiology resident evaluated all CEDM performed in our oncological cen ter from January 2023 to April 2024, assigning degrees of conspicuity to bre ast lesions, and evaluating a possible correlation with Ki-67 values (≤ 20% or > 20%), HER-2 status, estrogen (ER) and progesterone (PGR) receptor posit ivity, molecular subtype, and histological grade. Statistical analysis employ ed the Cramer’s V test.

Results

or Findings: Out of 352 patients included (median age=54, IQR=18 ), 100 were excluded due to chemotherapy controls and 53 due to B3 lesion. In the 199 remaining patients we observed a moderate t o strong association between conspicuity and ER expression (V=0.534) and Ki-67 value (V=0.36). A moderate association was found between conspicuity and PGR expression (V=0.31). No significant correlation was noted betw een conspicuity and histological grade (V=0.184) or HER2 status (V=0.2) .

Conclusion

Conspicuity, a recently incorporated descriptor in CEDM BI- RADS lexicon, was validated by our findings, which are in line with the initial evidence in the literature.

Limitations

Retrospective monocentric study with limited number of patients. Conspicuity is a subjective descriptor, potentially introducing variability in the data and affecting the findings. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Gianmarco Della Pepa: Nothing to disclose Claudia De Berardinis: Nothing to disclose Elisa D'Ascoli: Nothing to disclose Alice Bonanomi: Nothing to disclose Daniela Ballerini: Nothing to disclose Eleonora Ancona: Nothing to disclose Gianfranco Paride Scaperrotta: Nothing to disclose Catherine Depretto: Nothing to disclose Lisa Corradini: Nothing to disclose A novel metabolic MRI method for malignant breast t umors detection M. Rivlin¹, *R. Sivan Hoffmann*², V. Hadar², S. Suk hotnik², N. E. Weisenberg², O. Shmain-Naydenov², M. Zaiss³, S. Weinmüller³, G. Navon¹; ¹Tel Aviv/IL, ²Kfar Saba/IL, ³Erlangen/DE ([email protected]) Purpose or Learning Objective: Molecular imaging with 18F- fluorodeoxyglucose positron emission tomography (18 FDG-PET) is a powerful and well-established tool in breast cancer manageme nt, as increased glucose uptake is a known cancer hallmark.However, it carri es the risk of repeated radiation exposure. We have recently discovered tha t glucosamine (GlcN), a non-toxic, biocompatible glucose-based material can be detected using a unique MRI contrast mechanism termed chemical excha nge saturation transfer (CEST). CEST has emerged as an attractive molecular imaging approach for providing valuable metabolic information. Our goal is to develop an innovative molecular imaging modality based on CEST-MRI of Glc N to visualize and measure breast tumors while also distinguishing bet ween benign and malignant tumors.

Methods

or Background: Breast cancer patients and control group were scanned using the CEST-MRI pulse sequence on a 3T s canner (VIDA, Siemens, Germany) equipped with breast coil. The pr otocol included CEST scans before and two hours after drinking a solutio n of GlcN (184 mg/kg). The data were evaluated using magnetization transfer as ymmetry ratio (MTRasym) and area under curve (AUC) analysis.

Results

or Findings: GlcN treatment resulted in higher CEST values in tu mor regions of interest (ROIs), with maximum net MTRasy m signal (at 2 ppm) of 6.3±2.6% and averaged AUC (2-5 ppm) increase ratio of 3.3±2.1% (N=7). Yet, no significant GlcN CEST signal enhancement was det ected in healthy volunteers (N=9). GlcN CEST signal values were high ly correlated with the BI- RADS category.

Conclusion

These findings suggest that the GlcN CEST MRI techn ique can detect breast cancer while also providing molecular -level diagnostic tools for discriminating between benign and malignant breast tumors. These promissing

Results

pave the way to a future metabolic imaging of additional diseases.

Limitations

Due to the size of the study group we were not able to clarify into sub-groups. Funding for this study: Funding by the ISF No. 1689/18 Ethics committee - additional information: approval number MMC0201-21 Author Disclosures: Olga Shmain-Naydenov: Nothing to disclose Noemi Edith Weisenberg: Nothing to disclose Moritz Zaiss: Nothing to disclose Rotem Sivan Hoffmann: Nothing to disclose Stephanie Sukhotnik: Nothing to disclose Simon Weinmüller: Nothing to disclose Vivian Hadar: Nothing to disclose Michal Rivlin: Nothing to disclose Gil Navon: Nothing to disclose Artificial intelligence in digital mammography and serial breast tomosynthesis for neoadjuvant breast cancer treatme nt response prediction *D. Förnvik*¹, S. Zackrisson¹, I. Skarping²; ¹Malmö /SE, ²Lund/SE ([email protected]) Purpose or Learning Objective: To predict neoadjuvant chemotherapy (NACT) treatment response by applying artificial in telligence (AI) to digital mammography (DM) and serial breast tomosynthesis (D BT) images. Explainable AI (XAI) for enhanced clinical credibil ity is explored.

Methods

or Background: NACT for early-stage breast cancer (BC) has recently become an attractive approach to patients who are eligible for chemotherapy. MRI is the imaging modality of choice for evaluating tumor response but not as readily available as DM and lat ely DBT. Nevertheless, predicting residual cancer, as assessed by radiolog ist, after NACT using imaging is challenging; thus, AI could be an altern ative. Between 2005/2014 - 2019, 453 (DM) and 149 (DBT) patients, respectively , at Skane University Hospital, Sweden, comprised the cohorts. Two deep l earning architectures (DM: ResNet24, DBT: backbone 3D ResNet) applied to images from both the cancer and contralateral healthy breasts acquired a t three time points: pre- NACT (DM and DBT), mid-NACT (DBT) and post-NACT (DB T) were used to predict pathological complete response (pCR). For D BT, GradCAM was used to produce saliency maps to obtain insights into th e model-based decisions.

Results

or Findings: The DM and the DBT AI models predicted pCR as represented by the area under the ROC curve of 0.71 (95% CI: 0.53–0.90; p = 0.035) and 0.83 (95% CI: 0.63–1.00; p = 0.008), res pectively. The spatial correlation of saliency maps for DBT volumes from t he same patient but at different time points was high, likely indicating t hat the model focuses on the same areas during decision-making.

Conclusion

The DBT model demonstrates a high discriminative pe rformance for predicting pCR/non-pCR, possibly outperforming radiologists' assessment.

Limitations

Availability of larger datasets and inclusion of cl inicopathological variables would permit more comprehensive training of the models and more rigorous evaluation of their prediction performance for future patients. Funding for this study: Swedish Breast Cancer Group (BRO), Allmänna Sjukhusets i Malmö Stiftelse för bekämpande av canc er, and the Governmental Funding of Clinical Research within th e National Health Services. Ethics committee - additional information: Regional Ethics Committee in Lund, Sweden (committee reference numbers: 2014/13, 2014/521, and 2016/521, and 2021-05637-02). Abstract-based Programme 38 Wednesday Author Disclosures: Daniel Förnvik: Nothing to disclose Ida Skarping: Nothing to disclose Sophia Zackrisson: Nothing to disclose Image Quality and Diagnostic Values of Diffusion-We ighted Breast MRI: A Comparison of Single-Shot EPI with Deep Learning Reconstruction and Multi-Shot EPI with Simultaneous Multislice *H. S. Ahn*, S. H. Kim, M. J. Hong, H-S. Lee; Seoul /KR Purpose or Learning Objective: To evaluate the image quality and diagnostic value of single-shot echo-planar imaging (ss-EPI) w ith deep learning reconstruction (DLR) versus simultaneous multi-slic e echo-planar imaging (SMS rs-EPI) in breast MRI.

Methods

or Background: This study included 77 cases of breast cancer from 74 patients who underwent preoperative breast MRI. All patients underwent breast MRI that included the ss-EPI sequence combin ed with post-processing using DLR, as well as the SMS rs-EPI sequence. Two radiologists independently assessed qualitative image quality fa ctors and determined their preferences, while the cancer detection rate (CDR) was calculated. Quantitative analysis included measurements of appa rent diffusion coefficient (ADC), signal-to-noise ratio (SNR), contrast-to-noi se ratio (CNR), and lesion contrast, including phantom measurements.

Results

or Findings: Regarding qualitative image quality parameters, ss- EPI with DLR demonstrated significantly higher scores i n fat suppression and overall image quality as assessed by both radiologi sts, with comparable scores in artifact presence and lesion conspicuity to the SMS rs-EPI sequence. The CDR showed no significant difference between the tw o sequences. Both radiologists preferred ss-EPI with DLR (Reader 1: 7 8.4%, Reader 2: 89.2%). For quantitative parameters, ss-EPI with DLR exhibi ted significantly higher CNR (p = 0.002) and lesion contrast (p < 0.001), wh ile ADC and SNR values were comparable. In phantom measurements, mean ADC was lower for ss-EPI with DLR (DLR: 1.08 ± 0.58 vs. SMS: 1.12 ± 0.59, p = 0.007), but SNR was not significantly different (DLR: 607.45 ± 346.1 and SMS: 630.03 ± 624.51, p = 0.911). The acquisition time was shorter for ss-EPI with DLR (2:06 min) compared to SMS rs-EPI (3:29 min).

Conclusion

ss-EPI with DLR provided superior image quality and greater reader preference compared to SMS rs-EPI.

Limitations

This is a retrospective study which performed at si ngle center. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Hyun-Soo Lee: Nothing to disclose Min Ji Hong: Nothing to disclose Sung Hun Kim: Nothing to disclose Hye Shin Ahn: Nothing to disclose Dual Imaging Power: CT and Contrast-Enhanced Mammog raphy (CEM) for Advanced Detection of Metastatic Breast Cancer *M. Balbino*¹, M. Montatore², F. Masino³, F. A. Car pagnano⁴, G. Capuano², G. Guglielmi⁵; ¹Triggiano/IT, ²Barletta/IT, ³Bari/IT, ⁴Foggia/IT, ⁵Andria/IT ([email protected]) Purpose or Learning Objective: This study, one of the first in Italy, aims to evaluate the efficacy of performing CT and CEM cons ecutively using the same contrast medium in a single imaging session. The go als include reducing the amount of contrast agent injected into oncological patients and enhancing the detection of metastases in various districts throug h CT, while also providing a precise diagnosis of breast extension and identifyi ng additional foci within the breasts through CEM.

Methods

or Background: A cohort of female patients with confirmed primary breast cancer and suspected metastatic disease were enrolled. Each patient underwent a CT scan followed immediately by a CEM u sing the same contrast medium. The CT was performed to identify visceral m etastases, while the CEM targeted the detection of additional breast lesions and regional lymph node involvement. Both imaging modalities utilized iodin e-based contrast agents, administered intravenously. The diagnostic outcomes were compared with those from conventional imaging techniques, includi ng standard mammography, ultrasound, and MRI.

Results

or Findings: The combined CT and CEM approach demonstrated a higher sensitivity and specificity in detecting met astatic sites compared to traditional imaging methods. In particular, CEM rev ealed additional lesions in the breast and regional lymph nodes that were not i dentified by CT alone. The concurrent use of the same contrast medium was foun d to be safe and well- tolerated, with no significant increase in adverse reactions. The integrated imaging protocol provided comprehensive anatomical and functional information, leading to more accurate staging and b etter-informed treatment decisions.

Conclusion

The integration of CT and CEM using the same contra st agent offers a promising advancement in the diagnostic im aging of metastatic breast cancer. This combined approach enhances the detecti on of metastatic lesions, providing a more comprehensive assessment of diseas e spread.

Limitations

No Funding for this study: No Ethics committee - additional information: No Author Disclosures: Marina Balbino: Nothing to disclose Giuseppe Guglielmi: Nothing to disclose Francesca Anna Carpagnano: Nothing to disclose Manuela Montatore: Nothing to disclose Federica Masino: Nothing to disclose Giulia Capuano: Nothing to disclose Novel and robust approach to breast density predict ion: utilising the Tree-Structured Parzen Estimator algorithm-driven t ransfer learning approach *M. Bobowicz*, M. Kosno, K. P. Brzozowski, M. Rygus ik; Gdańsk/PL ([email protected]) Purpose or Learning Objective: The breast density visual assessment in mammography is a subjective process prone to errors but impacting diagnostic decisions. To overcome this problem, we developed a robust and reliable AI model that employs convolutional neural network-bas ed transfer learning, specifically ResNet, DenseNet, and EfficientNet arc hitectures, to predict breast density. Our research benefits from the Tree-struct ured Parzen Estimator (TPE) algorithm, an advanced tool for hyperparamete r optimisation.

Methods

or Background: A dataset of 2101 digital MLO mammography images performed at the Medical University of Gdans k from 2014 to 2022 was selected for analysis. The images were acquired usi ng various devices from SIEMENS, GE HEALTHCARE, and HOLOGIC to ensure a hig h degree of image characteristics variability. The dataset was divided into 80% training and 20% validation sets. ResNet, DenseNet, and Efficien tNet architectures were trained using the TPE algorithm. The assembly model comprises three five-fold cross-validated convolutional networks.

Results

or Findings: An ensemble model resulted in good performance metrics: AUC-ROC (0.99), accuracy (0.91), F1-score (0.91), and recall values (0.90) for the test dataset. The TPE algorithm faci litates the development of high-performance models on a relatively small datas et, eliminating the need for image segmentation to extract the skin and pectoral muscle opacities, which is challenging to implement and often burdened with si gnificant errors.

Conclusion

Our methodology enables the straightforward trainin g of a robust model that can provide highly precise breast densit y predictions, reducing and automating the burden of required density reporting . Furthermore, our findings demonstrate the efficacy of advanced hyperparameter numerical optimisation

Methods

in enhancing the efficiency of transfer dee p learning models in the context of breast density prediction.

Limitations

The study's limitations are its relatively small da taset, single- centre design, and lack of external validation. Funding for this study: Funding was provided by the European Union’s Horizon 2020 research and innovation programme unde r grant agreement No 952103 (EuCanImage project) and was co-funded by th e Digital Europe programme under grant agreement No 101100633 (EUCAI M project). Ethics committee - additional information: This retrospective study uses fully anonymised data from the EuCanImage project u nder the global ethics committee agreement for MUG. Author Disclosures: Maciej Bobowicz: Nothing to disclose Michał Kosno: Nothing to disclose Marlena Rygusik: Nothing to disclose Krystian Paweł Brzozowski: Nothing to disclose Abstract-based Programme 39 Wednesday 13:00-14:30 Research Stage 2 Research Presentation Session: Genitourinary RPS 407 What's new in prostate imaging: advances and emerging techniques Moderator P. A. Bonaffini; Bergamo/IT ([email protected]) Advancing prostate cancer imaging: a comparative an alysis of MET- RADS-P and PCWG3 in the assessment of mCRPC *L. Russo*¹, S. Bottazzi¹, O. Longoria², G. Avesani ¹, S. J. Withey², L. D'Erme¹, E. Sala¹, D-M. Koh², N. Tunariu²; ¹Rome/IT, ²London /UK Purpose or Learning Objective: Treatment response assessment in metastatic castration-resistant prostate cancer (mC RPC) is critical because the Prostate Cancer Working Group 3 (PCWG3) criteria ha ve notable limitations. The METastasis Reporting and Data System for Prosta te Cancer (MET-RADS- P) provides standardised guidelines using whole-bod y MRI (WBMRI). Our main aim was to compare MET-RADS-P and PCWG3 criteria fo r disease progression categorization in mCRPC, as well as the prognostic value of MET- RADS-P for progression-free survival (PFS) and over all survival (OS).

Methods

or Background: A cohort of 201 mCRPC patients treated at The Royal Marsden Hospital between January 2013 and Feb ruary 2024 was retrospectively included. All patients underwent WB MRI, CT and BS at each time point. CT and BS were interpreted according to PCWG3 and WBMRI according to MET-RADS-P. Concordance between MET-RA DS-P and PCWG3 in disease progression categorization was assessed overall, in bone and soft- tissue only. PFS and OS were evaluated using Kaplan -Meier survival curves with log-rank test comparisons.

Results

or Findings: Overall, 64.5% of time points (302/468) were concor dant between MET-RADS-P and PCWG3 criteria, with MET-RAD S-P detecting progression earlier in 31.8% (149/468). Discrepanci es were more pronounced in bone metastases, where MET-RADS-P identified pro gression in 55.1% of cases classified as non-progressive disease by PCWG 3. Progressing patients by MET-RADS-P had significantly worse PFS: median 2 .7 months versus 4.2 months by PCWG3 (p<0.001). The median OS was 12.5 m onths for progressing patients by MET-RADS-P at 12-week asses sment compared to 19.8 months for those stable or responding (p<0.001 ).

Conclusion

MET-RADS-P allowed for earlier progression detectio n compared with PCWG3, particularly in bone metastases, potent ially permitting earlier therapeutic interventions. MET-RADS-P also demonstr ated strong predictive value for PFS and OS, suggesting its potential role as an imaging biomarker in future clinical trials.

Limitations

Retrospective design and lack of cost and availabil ity comparative analysis. Funding for this study: This study represents independent research funded by the National Institute for Health and Care Resea rch (NIHR) Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, London, and by the Ro yal Marsden Cancer Charity, and Cancer Research UK (CRUK) National Can cer Imaging Trials Accelerator (NCITA) and Prostate Cancer UK. The vie ws expressed are those of the author(s) and not necessarily those of the N IHR or the Department of Health and Social Care. This work uses data provide d by patients and collected by the NHS as part of their care and supp ort. Ethics committee - additional information: The study was approved by the Institutional Ethics Committee (reference no. 21/LO /0605). Author Disclosures: Samuel Joseph Withey: Nothing to disclose Giacomo Avesani: Nothing to disclose Dow-Mu Koh: Nothing to disclose Nina Tunariu: Nothing to disclose Ossian Longoria: Nothing to disclose Silvia Bottazzi: Nothing to disclose Evis Sala: Nothing to disclose Luca D'Erme: Nothing to disclose Luca Russo: Nothing to disclose Analysis of biopsy strategy in young men with suspi cious PSA in a prostate cancer screening setting – data from the P ROBASE trial *M. Boschheidgen*¹, R. Al-Monajjed¹, J. P. Radtke¹, H-P. Schlemmer², G. Antoch¹, L. Schimmöller¹, P. Albers¹; ¹Düsseldor f/DE, ²Heidelberg/DE ([email protected]) Purpose or Learning Objective: To analyze the performance of targeted (TB) and systematic (SB) MRI/US fusion-guided prostate b iopsy within the prospective PROBASE prostate cancer (PC) screening trial.

Methods

or Background: Men aged forty-five from the general population were invited to screening. Those with confirmed pro state-specific antigen (PSA) levels of 3 ng/ml or higher were offered an M RI and were referred to MRI/US-guided biopsy. Biopsies were performed in ev ery participant unrespective of the MRI result. Targeted and system atic biopsies with software-based fusion techniques were offered. The primary endpoint of this analysis was the PC detection in either TB or SB.

Results

or Findings: A total of 554 men (median age 50 (range 44-54), median PSA level 4.1 ng/ml) were analyzed who under went an MRI followed by MRI/US-guided biopsy. Of 217 PC diagnosed, 198 ( 91%) and 140 (65%) were detected by SB and TB, respectively. 64 of 217 PC (29%) were low grade (ISUP 1). 40 significant tumors were found exclusiv ely by SB (26%), while 9 significant tumors were only diagnosed with TB (6%) . SB detected significantly more low-grade cancer compared to TB (p<0.001). Can cer detection rate was 20% for PIRADS 1-2, 26% for PIRADS 3, 59% for PIRAD S 4, and 92% for PIRADS 5.

Conclusion

In young men and in the setting presented here, sys tematic biopsy in addition to targeted MRI/US guided fusion biopsy still appears to be justified to adequately detect PC. Most PC (71%) we re clinically significant. Performing only TB in young men without SB faces th e risk of missing a significant number of csPC even if it simultaneousl y diagnoses fewer low-grade carcinomas.

Limitations

MRI did not influence clinical decision-making; exp erience in MRI reading and the image quality differed widely at th e time of initiation Funding for this study: Deutsche Krebshilfe Ethics committee - additional information: The study was approved by local ethics committee. Author Disclosures: Jan Philipp Radtke: Nothing to disclose Peter Albers: Nothing to disclose Rouvier Al-Monajjed: Nothing to disclose Matthias Boschheidgen: Nothing to disclose Lars Schimmöller: Nothing to disclose Heinz-Peter Schlemmer: Nothing to disclose Gerald Antoch: Nothing to disclose Is there an MRI phenotype for the cribiform pattern of Prostate Cancer? M. D. M. Palma, D. Freire Maia Vieira, T. A. Leite De Lima, J. Nather, F. Chahud, R. B. Reis, *V. F. Muglia*; Ribeirao Pre to/BR ([email protected]) Purpose or Learning Objective: Cribriform pattern (CP) is a distinct histological feature present in various neoplasms, characterized by cohesive tumour cells surrounding circular spaces, creating a "Swiss cheese" appearance. In prostate cancer (PCa), CP is one of the four architectural subtypes of the Gleason 4 pattern and has been link ed to worse outcomes compared to other morphologies. This study aimed to determine if CP presents distinct features on multiparametric magnetic reson ance imaging (MRI).

Methods

or Background: In this retrospective, single-centre study, we identified PCa cases with CP from 2016 to 2023, wit h MRI conducted within four months of histological diagnosis. Patients wit hout CP but with equivalent Gleason grades and risk stratification were include d as controls in a ratio of up to 1.5:1. Two radiologists, with over 5 and the oth er with 7 years of prostate imaging experience, evaluated lesion size, form, lo cation, prostate volume, mean apparent diffusion coefficient (ADC) values, a nd post-contrast kinetic curves. Clinical staging, prostate-specific antigen (PSA), and PSA density (dPSA) were also reviewed.

Results

or Findings: The study included 42 patients with CP and 72 witho ut CP. No significant differences were found between t he groups regarding PSA (p=0.43), dPSA (p=0.37), lesion size (p=0.33), loca tion (p=0.65), mean ADC (p=0.21), or kinetic curve type (p=0.75). Significa nt differences were observed for age (p<0.0001), prostate volume (p=0.001), and PI-RADS category (p=0.05). In univariate logistic regression, age an d PI-RADS score were independent predictors of CP presence, but only age remained significant in multivariate analysis (p=0.0001).

Conclusion

Cribrifrom pattern in PCa is more common in older m en with larger prostates and higher PI-RADS scores. However , no specific morphological or functional MRI parameters were ass ociated with the presence of this pattern.

Limitations

Single-centre, retrospective study. Funding for this study: None Abstract-based Programme 40 Wednesday Ethics committee - additional information: Our Institutional Review Board approved the research with a waiver for informed co nsent due to the retrospective nature. Author Disclosures: Fernando Chahud: Nothing to disclose David Freire Maia Vieira: Nothing to disclose Julio Nather: Nothing to disclose Matheus De Moraes Palma: Nothing to disclose Valdair Francisco Muglia: Nothing to disclose Rodolfo B. Reis: Nothing to disclose Thalyne Aparecida Leite De Lima: Nothing to disclos e The effect of prostate volume - does PSA density al ways work? *S. Durmaz*¹, S-C. J. Wu², K-L. Lee², A. Shakur², I . Caglič², T. Barrett²; ¹Istanbul/TR, ²Cambridge/UK ([email protected]) Purpose or Learning Objective: To evaluate the impact of PSA density (PSAd) on the probability of detecting clinically s ignificant prostate cancer (csPCa) across different prostate volume ranges and PI-RADS scores.

Methods

or Background: 2097 patients undergoing multiparametric MRI (mpMRI) for suspected PCa were included. 738/2097 ( 35.2%) had PCa, and 566/2097 (27%) had csPCa (Gleason ≥3+4), patients were classified as negative after biopsy (n=299) or were diagnosed wit h clinically insignificant PCa (n=172) or having a negative mpMRI and completi ng at least one-year follow-up without developing PCa (n=1060). Single-v ariable logistic regression analyses were conducted to assess the impact of PSA d on the probability of csPCa within the different prostate volume ranges ( 80 mL) and PI-RADS groups.

Results

or Findings: The median age, PSA, PSAd, and prostate volume was 66 years (IQR: 61-72), 5.6 ng/mL (IQR:4.05-8.05), 0 .10 ng/mL/mL (IQR:0.07- O.15), and 56 mL (IQR:39-80), respectively. Logisti c regression at a PSAd of 0.15 ng/mL/mL showed the probability of csPCa decre ased with increasing prostate volume: 80 mL (18%). At the same PSAd level, the probability o f csPCa increased with increasing PI-RADS score: PIRADS 1-2 (3%), PI-RADS 3 (26%), PI-RADS 4 (63%), PI-RADS 5 (70%). Regardless of the PSAd leve l, the risk of csPCa in patients with PI-RADS 4-5 lesions was always >20%.

Conclusion

When using PSAd to assist in the decision to perfor m MRI, or to biopsy patients with PI-RADS 1-3 scores, caution sh ould be exercised in those with larger volume prostates, as the lower PSAd can provide false reassurance.

Limitations

Retrospective design. All patients underwent mpMRI and prostate biopsy at a single tertiary referral center with ex tensive experience in prostate MRI and biopsy. Funding for this study: No funding was received for this study. Ethics committee - additional information: This is a retrospective study. Author Disclosures: Iztok Caglič: Nothing to disclose Amreen Shakur: Nothing to disclose Shun-Chin Jim Wu: Nothing to disclose Selahattin Durmaz: Nothing to disclose Kang-Lung Lee: Nothing to disclose Tristan Barrett: Nothing to disclose A transformer-based deep learning model for early p rediction of biochemical recurrence after radical prostatectomy using pretreatment mpMRI *F. Li*¹, L. Zhuo¹, L. Yue¹, L. Juan¹, L. Wang¹, R. Liu¹, F. Wang², Y. Xiang³; ¹Mianyang City/CN, ²Luzhou/CN, ³Leshan/CN ([email protected]) Purpose or Learning Objective: The purpose of this study is to develop and verify a deep learning model using preoperative mul ti-parameter MRI images to predict BCR risk after radical prostatectomy.

Methods

or Background: Patients after radical surgery at 4 centers between August 2013 and September 2021 were retrospectively included with the endpoint outcome of 3-year BCR (two consecutive spe cific antigen [PSA] levels > 0.2 ng/mL [0.2µg/L]). A transformer-based DL model was used to predict BCR after radical surgery using 3D tumor im ages, a clinical model was constructed by multivariate logistic regression, Ka plan-Meier plots were used for estimating recurrence-free survival, and finall y, pre- and post-surgical Capra models, a clinical model, a multi-instance model, a nd a transformer model, Multimodal Combine model were compared to assess th e performance of predicting BCR.

Results

or Findings: A total of 582 patients (median age 70 years, (IQR 44-89 years) with a median follow-up of 43 months (IQR, 2 9-71 months) were randomized into a training group (n=249 ), an inter nal test set (n=107), an external test set 1 (n=189), and an external test s et 2 (n=37).The AUC of the Transformer model in the 0.92 in the internal test set, 0.84 in the external test set 1, and 0.82 in the external test set 2, and the multimodal Combine model further improves the performance, respectively, wit h 0.94 (95% CI. 0.885 - 0.992), 0.94 (95% CI, 0.900 - 0.969), and 0.83 (95% CI, 0.693 - 0.965), and early recurrence-free survival and overall survival could be better risk-stratified and predicted using the Combine model.

Conclusion

A transformer-based DL model for predicting BCR aft er radical surgery was developed and internally and externally validated, and the joint model is better and expected to guide individualize d treatment.

Limitations

Not applicable. Funding for this study: No funding was provided for this study Ethics committee - additional information: Ethics （2024）014-1 Author Disclosures: Lu Wang: Nothing to disclose Ruishan Liu: Nothing to disclose Fan Li: Nothing to disclose Fei Wang: Nothing to disclose Liao Juan: Nothing to disclose Ye Xiang: Nothing to disclose Lv Yue: Nothing to disclose Lihua Zhuo: Nothing to disclose Short MRI sequence suitable for re-identification o f prostate lesion during in-bore biopsy? *C. Peter*, A. Schaudinn, C. Ehrengut, T. Franz, L- C. Horn, N. Linder, J-U. Stolzenburg, H. Busse, T. Denecke; Leipzig/DE ([email protected]) Purpose or Learning Objective: To evaluate the image quality of a rapid intraprocedural balanced steady-state free precessi on (b-SSFP) sequence for re-identification of prostate lesions during transr ectal in-bore biopsies in comparison with that of a T2-weighted

Methods

or Background: In this retrospective study, 127 patients with 140 PI- RADS ≥ 3 (version 2.1) lesions based on multiparametric 3 T MRI (mpMRI) underwent transrectal in-bore biopsies. b-SSFP imag es were acquired at 1.5T for interventional guidance. Two radiologists (R1: 11 years and R2: 2 years of mpMRI experience) independently rated the image qua lity of both b-SSFP (acquisition time: 11-15 seconds) and diagnostic T2 -weighted TSE (3T, acquisition time: 4-5 minutes) sequences using a 4- point scale (3: good, 2: acceptable, 1: poor, 0: impossible for lesion ident ification). Recognition rates (RR) were calculated as the percentage of cases wit h sufficient image quality (scores of 3 or 2). Subgroup analyses were performe d by zonal location (PZ/TZ), lesion size (</≥ 0.5 mL), and PI RADS score (3/4-5). Differences between readers and sequences were analysed using M cNemar's test (p < 0.05).

Results

or Findings: The RR for the T2-weighted reference sequence was 98% for both radiologists, with subgroups ranging f rom 94% to 99%. For b- SSFP, the RR was 87% for R1 and 81% for R2, with su bgroups ranging from 75% (PI-RADS 3) to 92-93% (PI-RADS 4/5, large lesio ns). No significant differences were found between readers. RR differen ces between sequences were statistically significant, except for TZ and l arge lesions rated by R1.

Conclusion

b-SSFP showed only moderately lower RR than the ver y high RR of T2-weighted reference images, especially for the experienced reader R1. Given its much shorter acquisition time, b-SSFP of discernible lesions therefore has the potential to reduce biopsy times, particula rly for large or (highly) suspicious lesions (PI-RADS 4-5).

Limitations

Retrospective; single-center. Funding for this study: None Ethics committee - additional information: Ethics committee was consulted, written informed consent was obtained from particip ating patients Author Disclosures: Nicolas Linder: Nothing to disclose Timm Denecke: Nothing to disclose Harald Busse: Nothing to disclose Lars-Christian Horn: Nothing to disclose Constantin Ehrengut: Nothing to disclose Alexander Schaudinn: Nothing to disclose Jens-Uwe Stolzenburg: Nothing to disclose Toni Franz: Nothing to disclose Christian Peter: Nothing to disclose Abstract-based Programme 41 Wednesday Diagnostic assessment of early DWI changes after Si ngle-Dose Ablative Radiation Therapy for localized prostate cancer *P. N. Franco*, C. R. G. L. O. M. Talei Franzesi, C . Maino, R. Corso, D. Ippolito; Monza/IT ([email protected]) Purpose or Learning Objective: To investigate the diagnostic value of diffusion-weighted (DWI) MRI early changes, 1 hour after treatment, in patients with organ-confined unfavourable prostate cancer (P Ca) treated with Single- Dose Ablative Radiation Therapy (SDART), in compari son with biochemical markers.

Methods

or Background: Twenty-four patients with intermediate unfavourable or high-risk localized PCa treated with SDART (21 G y on the entire prostate with boost up to 24 Gy on the focal lesion) associa ted with hormone therapy were prospectively enrolled. Each patient was exami ned with a 3T scanner four times: (1) 1-2 weeks before RT (t0) for treatm ent planning; (2) 1 hour after treatment (t1); (3) 3 months after treatment (t2); (4) 2 years after treatment (t3). Regions of interest (ROIs) were plotted on apparent diffusion coefficient (ADC) maps and T2-HR sequences on lesions, benign periphe ral zone, and the entire prostate gland. Patients’ laboratory data (PSA and testosterone) was collected.

Results

or Findings: ADC values significantly increased in neoplastic le sions at t1, t2 and t3 (+22%, +43% and +53%, respectively ). Conversely, no significant changes were observed in the benign per ipheral zone and the entire prostate gland. On T2 sequences, signal intensity p rogressively decreased in the benign peripheral zone (t1: -1%; t2: -33%; t3: -42%) and in the entire prostate gland (t1: -6%; t2: -24%; t3: -31%), while no significant changes were observed in lesions. All patients except one had a complete biochemical response.

Conclusion

The study findings showed high diagnostic value of DWI and a good correlation between early (t1) changes in ADC values after SDART and later (t2 and t3) tumour response (both biochemical and imaging) in patients with unfavourable PCa. Early DWI changes can repres ent a useful parameter to evaluate treatment response and predict patients ’ outcomes.

Limitations

Small sample size; associated hormone therapy. Funding for this study: None Ethics committee - additional information: The local ethics committee formally approved this study. Author Disclosures: Cesare Maino: Nothing to disclose Cammillo Roberto Giovanni Leopoldo Oreste Massimili ano Talei Franzesi: Nothing to disclose Rocco Corso: Nothing to disclose Paolo Niccolò Franco: Nothing to disclose Davide Ippolito: Nothing to disclose Maximal radial distance as a new parameter for pred icting extraprostatic extension of prostate cancer on multiparametric mag netic resonance imaging: a histo-radiological study *F. Porões*, A. Nobile, L. Widmer, J. A. Vidal, J. Di Vincenzo, H. Najberg, J. M. M. Froehlich, C. Reischauer, H. Thoeny; Fribo urg/CH ([email protected]) Purpose or Learning Objective: We introduce a new parameter for predicting extraprostatic extension (EPE) on multiparametric m agnetic resonance imaging (mpMRI): the maximal radial distance (maxRA DD). It corresponds to the largest diameter of a prostate cancer focus (PC F) perpendicular to a contact with the prostate pseudocapsule. We compare accuracy and reproducibility of maxRADD with the previously prop osed maximal capsular contact length (maxCCL) for predicting EPE.

Methods

or Background: We retrospectively and consecutively included 81 patients undergoing prostate mpMRI between October 2018 and December 2020, followed by radical prostatectomy. One uropat hologist with 8 years of experience collected for each PCF: location, maxCCL , maxRADD, and presence/absence of EPE. Four radiologists with 0, 2, 3, and 6 years of experience in prostate mpMRI determined maxRADD and maxCCL on mpMRI for each PCF twice in separate readings. Accuracy i n predicting EPE was assessed using the area under the curve (AUC), with the pathologic findings as the gold standard. Inter-/intra-reader agreement we re assessed using intraclass correlation coefficients (ICCs) and Cron bach’s alpha.

Results

or Findings: On histolpathology, there was no significant differ ence in the accuracy of predicting EPE between maxRADD and maxCCL (AUCmaxRADD = 0.92, AUCmaxCCL = 0.91, p = 0.28). Pe arson correlation showed a strong correlation of both parameters dete rmined on mpMRI with their histopathological counterparts (>0.7), with t he exception of maxCCL assessed by the reader w/o experience in prostate m pMRI (0.54). On mpMRI, inter-reader agreement was significantly higher for maxRADD (ICCmaxRADD = 0.96, ICCmaxCCL = 0.94, p = 0.046) and intra-read er agreement was higher but did not reach significance (average alphamaxRAD D = 0.95, average alphamaxCCL = 0.92, p = 0.31).

Conclusion

MaxRADD permits assessing EPE with good accuracy an d shows higher reproducibility compared with maxCCL.

Limitations

No significant limitation. Funding for this study: This study has received funding by the Swiss Nation al Science Foundation (Grant/Award Number: 32003B_1762 29/1) and the HFR Research GRANT (2352). Ethics committee - additional information: The study was approved by our institutional ethics committee (CER-VD). The ethics committee notification can be found under the project-ID 2020-01859. Author Disclosures: Harriet Thoeny: Advisory Board: Guerbet Jana Di Vincenzo: Nothing to disclose Hugo Najberg: Nothing to disclose Johannes Malte Maria Froehlich: Consultant: Guerbet Antoine Nobile: Nothing to disclose Fabio Porões: Nothing to disclose Lucien Widmer: Nothing to disclose Julian Alexis Vidal: Nothing to disclose Carolin Reischauer: Nothing to disclose Impact of Centrally Reviewed PI-QUAL v2 Scores on t he Diagnostic Performance of Prostate MRI *G. Brembilla*, D. Cannoletta, M. Cosenza, F. Pelle grino, M. E. Porzi, L. Quarta, A. Stabile, A. Briganti, F. De Cobelli; Milan/IT Purpose or Learning Objective: To assess the impact of image quality, defined by PI-QUAL v2 scores, on the diagnostic yie ld of prostate MRI in centrally reviewed scans.

Methods

or Background: We retrospectively identified consecutive patients who underwent MRI-targeted and systematic biopsies at our Institution (January 2023 - June 2024), with MRI performed exte rnally. All the external MRI scans were centrally reviewed by an experienced uro-radiologist, who assigned PI-QUAL v2 and PI-RADS v2.1 scores. We ass essed the proportion of PI-RADS 3 lesions and the detection rate of clin ically significant prostate cancer (csPCa), stratified by PI-QUAL v2 scores, be fore and after central revision. Histopathological results from the biopsi es were used as the

Reference

standard.

Results

or Findings: A total of 151 consecutive patients were included i n the analysis. 37/151 (24%) of the MRI scans were scored PI-QUAL 1, 72/151 (48%) PI-QUAL 2, and 42/151 (38%) PI-QUAL 3. Based on original reports, the overall proportion of PI-RADS 3 scans was 34/151 (2 3%). In PI-QUAL 1-2 vs 3 scans, the proportion of PI-RADS 3 in was 27% vs 12 %, respectively; the csPCa detection rate was 48% vs 62%, respectively. The reclassification rate of PI-RADS scores at central review was 64/151 (42% ), and was higher for PI- QUAL 1-2 scans (47%) than for PI-QUAL 3 scans (31%) . After central revision, the overall proportion of PI-RADS 3 in PI-QUAL 1-2 vs 3 was 19% vs 5%, respectively; the detection rate of csPCa was 58% v s 78%, respectively.

Conclusion

Lower prostate MRI image quality, as defined by the PI-QUAL v2 scoring system, is associated with a higher proport ion of equivocal scans (PI- RADS 3) and a reduced csPCa detection rate in centr ally revised MRI scans.

Limitations

Small sample size, only one radiologist for review Funding for this study: None Ethics committee - additional information: IRB approved Author Disclosures: Maria Elena Porzi: Nothing to disclose Donato Cannoletta: Nothing to disclose Michele Cosenza: Nothing to disclose Francesco Pellegrino: Nothing to disclose Alberto Briganti: Nothing to disclose Armando Stabile: Nothing to disclose Leonardo Quarta: Nothing to disclose Francesco De Cobelli: Nothing to disclose Giorgio Brembilla: Nothing to disclose Prostate Volume Assessment on MRI: Comparison of fu ll manual segmentation to PIRADS-based approximation in 2 pla nes and its influence on PSA-density *J. Uhlig*, L. Biggemann, C. Louizi, A. Uhlig; Gött ingen/DE ([email protected]) Purpose or Learning Objective: To evaluate differences in prostate volume quantification on MRI comparing full manual segment ation and PIRADS-based approximation in 2 planes.

Methods

or Background: Patients imaged with 3T mpMRI (Siemens VIDA) for suspected prostate cancer between 2021-2023 wer e included. PSA measurements were obtained at the time of mpMRI or extracted from patients records up to 3 months prior. Manual segmentation o f the prostate was performed on all axial T2w slices serving as refere nce standard. Prostate volume was approximated using 3 measurements on T2w sagittal and axial planes according to the PIRADSv2.1 recommendations. Prostate volumes from manual segmentation and approximation were compared and the influence on PSA-density quantified using different cut-off valu es. Abstract-based Programme 42 Wednesday

Results

or Findings: n=331 patients were included (mean age 67 ± 7 years ) with a mean PSA value of 8.1 ± 5.7 ng/ml. Mean prostate volume using manual segmentation was 63.3cc (± 33cc), with n=142 patients having a volume of 151c c: n=8. The mean absolute difference of prostate volume using approx imation vs. segmentation was 9.1cc (±9.3cc, p=0.01). In general, smaller pro state volumes were overestimated, and larger volumes underestimated by approximation. Using a PSA-density cut-off <0.1 ng/ml/cc, the approximatio n method yielded an accuracy = 88%. Using a PSA-density cut-off <0.15 n g/ml/cc, the approximation method yielded an accuracy = 90%.

Conclusion

Using PIRADSv2.1-based approximation of prostate vo lume on mpMRI yields a statistically significant difference when compared to full manual segmentation. These differences have a relevant eff ect on PSA-density calculation with potential impact on downstream pat ient management.

Limitations

Patients were recruited in only one tertiary center and imaged on one MRI scanner, which could limit the generalizabi lity of presented results. Funding for this study: This study received no funding. Ethics committee - additional information: Ethics committee of the University Medical Center Goettingen Author Disclosures: Annemarie Uhlig: Nothing to disclose Johannes Uhlig: Nothing to disclose Lorenz Biggemann: Nothing to disclose Chiheb Louizi: Nothing to disclose MRI without contrast media injection for prostate c ancer screening:

Results

from Prostate Cancer Secondary Screening in Sapienza (PROSA) *E. Messina*, A. Borrelli, L. Laschena, S. Lucciola , M. Pecoraro, V. Panebianco; Rome/IT ([email protected]) Purpose or Learning Objective: PROSA is a randomized MRI-based screening protocol, investigating the role of MRI w ithout contrast media injection (bi-parametric MRI, bpMRI) as secondary p revention test for prostate cancer (PCa) early diagnosis, comparing MRI with PS A-test. PROSA aims to investigate the efficiency of this screening protoc ol, both in terms of diagnostic accuracy, and cost-effectiveness.

Methods

or Background: 590 men aged 49 to 69 years were enrolled and blindly randomized into two different arms: (A) Men underwent bpMRI regardless of their PSA values; (B) Men with increa sed PSA were directed to bpMRI, while those with normal PSA were not. Men sc reened positive on MRI were directed to MR-directed targeted biopsy. To ev aluate the efficiency of the protocol we calculated the experimental event rate (EER), control event rate (CER), absolute risk reduction (ARR), number needed to treat (NNT). Health Technology Assessment analysis was implemented to e valuate the cost- effectiveness. The cost/effectiveness ratio is calc ulated as follows: Delta Costs/ Delta effectiveness = (CA–CB)/(EA-EB).

Results

or Findings: 289 men were randomized on Arm A and among them 15 clinically significant PCa (csPCa) were detected ; 291 men were randomized on Arm B, with 6 csPCa detected (p=0.04). On arm A, 8 men diagnosed with csPCa (53.3%) presented normal PSA levels. Consider ing the efficiency of the screening protocol, EER was 5.23%, CER 2.06%, ARR 3 .17%, and NNT 31.6. Therefore 32 interventions (in this study MRIs) are needed to find one event (in this study one csPCa). The final cost/effectiveness ratio resulted to be € 3.562,61 for the diagnosis of one csPCa.

Conclusion

Prostate MRI without contrast media injection showe d promising

Results

compared to the use of PSA analysis alone a s a screening tool, both in terms of efficiency and cost-effectiveness.

Limitations

Single center Funding for this study: No Ethics committee - additional information: CE Approved Author Disclosures: Sara Lucciola: Nothing to disclose Valeria Panebianco: Nothing to disclose Antonella Borrelli: Nothing to disclose Emanuele Messina: Nothing to disclose Martina Pecoraro: Nothing to disclose Ludovica Laschena: Nothing to disclose 13:00-14:30 Research Stage 3 Research Presentation Session: Neuro RPS 411 Aging brain and neurodegeneration imaging Moderator F. Barkhof; Amsterdam/NL ([email protected]) Author Disclosures: Frederik Barkhof: Advisory Board: Combinostics, Sco ttish Brain Sciences, Alzheimer Europe, Merck; Author: Clinical Neuroradi ology - The ESNR textbook; Consultant: Roche, Celltrion, Rewind Ther apeutics, Bracco; Founder: Queen Square Analytics; Grant Recipient: Roche, UK MS Society, Biogen, Merck, ADDI; Other: DSMB member Prothena, EISAI Connecting the Dots: Linking White Matter Hyperinte nsity Patterns to Longitudinal Cognitive Changes in Aging *M. M. Courtney*, R. A. Kenny, J. F. Meaney, C. De Looze; Dublin/IE ([email protected]) Purpose or Learning Objective: White matter hyperintensities (WMHs) are known to correlate with cognitive decline, stroke, and dementia. Previous research has explored the independent effects of wh ite matter macrostructure, microstructure, and spatial distribution on cogniti ve function, yet a comprehensive analysis combining elements is limite d. We provide a comprehensive WMH analysis to assess their associat ion with cognitive decline over a six-year period.

Methods

or Background: Data was obtained from The Irish Longitudinal Study on Ageing (TILDA), including MRI scans and co gnitive performance scores from 497 community-dwelling older adults. WM Hs were segmented using Lesion Prediction Algorithm, analysed using E xplore DTI for diffusion metrics. Statistical Analysis performed in R-Studio . Linear mixed effect models used to assess relationship between lesion phenotyp es and cognitive decline, adjusting for demographic and health-related variab les.

Results

or Findings: 11,933 WMHs analysed. Average 24 lesions per subjec t. Average lesion volume 263mm3. Average lesion FA 0.2 9, MD 1.10. K-means clustering identified 3 primary WMH phenotypes. Dee p WMHs associated with older age and 2 or more cardiovascular risk factors (p<0.001 respectively). Higher volume lesions were associated with cardiova scular risk factors (p<0.001), smoking (p<0.01) and older age (p<0.001) . High-volume, low FA lesions in both deep (p=0.5) and periventricular (p =0.04) white matter exhibited accelerated cognitive decline over six years. Incre ased number of periventricular lesions was associated with cogniti ve decline (p<0.01).

Conclusion

WMHs manifest diverse phenotypes associated with co gnitive decline. High-volume, low FA lesions in both perive ntricular and deep white matter are predictive of cognitive deterioration. I dentifying WMH phenotypes may inform early intervention strategies and improv e patient outcomes by targeting individuals at higher risk of cognitive d ecline.

Limitations

Cross-sectional MRI data implies findings are assoc iative and correlate with longitudinal findings, but correlati on does not equal causation. Further investigation with serial MRI would provide more reliable data. Funding for this study: The Irish Longitudinal Study on Ageing is funded by the Irish Department of Health, the Atlantic Philan thropies and Irish Life. Ethics committee - additional information: This study was approved by the Trinity College Faculty of Health Sciences Research Ethics Committee, Dublin, Ireland. Protocols conformed with the Declaration o f Helsinki. Signed informed consent was obtained from all respondents prior to participation. Additional ethics approval was received for the magnetic reson ance imaging (MRI) sub- study from the St James’s Hospital/Adelaide and Mea th Hospital, Inc. National Children’s Hospital, Tallaght (SJH/AMNCH) Research Ethic Committee, Dublin, Ireland. Those attending for MRI also completed an additional MRI-specific consent form. (De Looze et al) Author Disclosures: Michael Michael Courtney: Nothing to disclose Céline De Looze: Nothing to disclose Rose Anne Kenny: Nothing to disclose James F Meaney: Nothing to disclose Abstract-based Programme 43 Wednesday Normal Aging-Related Brain Morphological Connectivi ty Network Linked to Multiple Neurological Diseases *L. Yuna*; Beijing/CN ([email protected]) Purpose or Learning Objective: Disentangling the complex interaction between aging and various aging-related neurologica l diseases at the brain network level.

Methods

or Background: We presented a unified analysis framework to identify aging-related morphological connectivity n etworks (MCNs) and determined their clinical relevance in various neur ological diseases (including mild cognitive impairment, Alzheimer's disease, Par kinson’s disease, small vessel disease multiple sclerosis and multiple scle rosis). First, individual MCNs in the HC group were constructed and further decomp osed into distinct subnetworks using linked independent component anal ysis. Aging-related subnetworks were defined as those significantly ass ociated with age. The aging-related subnetworks were spatially correlated with disease-related MCN disruptions. Further, the regression coefficients o f the aging-related subnetworks were calculated for each patient’s MCN using linear regression. The regression coefficients were then correlated wi th various clinical variables within each disease group to assess the clinical si gnificance of the aging- related subnetworks. Finally, a series of annotated biological maps were utilized to advance the biological interpretation o f the identified aging-related subnetworks.

Results

or Findings: We first identified three aging-related subnetworks , including the perceptual-limbic subnetwork, attenti on-somatomotor subnetwork, and somatomotor-predominant subnetwork, that exhibited distinct aging trajectories. Normal aging interacted with va rious neurological diseases, exhibiting both transdiagnostic and diagnosis-speci fic patterns at the brain network level. The aging-related subnetworks were c losely related to cognitive and physical performance in patients. Biological co rrelation analysis revealed that glucose metabolism and several neurotransmitte rs, such as cannabinoids and dopamine, played critical roles in aging-relate d subnetworks.

Conclusion

This study elucidated the network mechanisms underl ying the complex interactions between aging and neurological diseases, offering insights that could improve clinical management and therapy development by evaluating aging effects.

Limitations

This study is limited by an uneven sample distribut ion, variability in disease durations, and the absence of longitudin al research Funding for this study: We demonstrated how normal aging interacted with various neurological diseases at the brain network level, both transdiagnostically and diagnosis-specifically. The identified aging-related subnetworks might serve as imaging markers to disti nguish normal aging effects from disease-specific mechanisms, thereby i mproving disease monitoring and management. Ethics committee - additional information: Beijing Tiantan Hospital, Capital Medical University, Beijing, China; No. KY-2019-050 -02 Author Disclosures: Li Yuna: Nothing to disclose A comparative evaluation of four commercially avail able artificial intelligence software solutions for brain volumetry and lesion segmentation in dementia *G. Di Cerbo*, G. Saltarelli, A. Innocenzi, M. Cell a, C. De Felici, F. Bruno, A. Splendiani, E. Di Cesare; L'Aquila/IT ([email protected]) Purpose or Learning Objective: The purpose of this study is to compare the operating features and analysis outputs of four dif ferent commercially available software for brain volumetric analysis.

Methods

or Background: We analyzed consecutive brain MRI scan of 32 patients (25 males, aged between 50 and 90 years) e valuated in a singles Institution for cognitive decline. All MRI examinat ions were performed on 3T scanner (GE MR750w.), including a volumetric T1 GRE sequence (slice 1 mm, TR 8.5, frequency FoV 25.6, phase FoV 0.8). MRI dat a were analyzed through four different dedicated softwares (S1, S2, S3, S4) after quality check by an experienced neuroradiologist. Volumetric output dat a of brain segmentation and volume for frontal, temporal, parietal, occipit al lobes, hippocampus, and lateral ventricles, were collected and compared.

Results

or Findings: The results revealed no significant consensus among the four artificial intelligence software applicati ons in measuring various brain areas. S1-S2 showed non statistically significant o utput values in all brain regions. S1-S3 showed statistically significant dif ferences in frontal and parietal lobe, lateral ventricles and hippocampus. S1-S4 sho wed statistically significant differences in frontal parietal and occipital lobe, and lateral ventricles. S2-S3 showed statistically significant differences in tem poral and occipital lobes. S3- S4 showed statistically significant differences in parietal and occipital lobes, hippocampus and lateral ventricles.

Conclusion

Although AI software are becoming increasingly popu lar in clinical practice, the findings indicate a low degr ee of concordance among the four applications evaluated in this study. Therefor e, clinicians integrating these tools into routine practice should be aware of the limited result interchangeability across different software platfo rms and consider their use as complementary aids rather than substitutes for clin ical expertise.

Limitations

Small sample size Funding for this study: None Ethics committee - additional information: Local IRB Author Disclosures: Alessandra Splendiani: Nothing to disclose Gaspare Saltarelli: Nothing to disclose Antonio Innocenzi: Nothing to disclose Ernesto Di Cesare: Nothing to disclose Giovanni Di Cerbo: Nothing to disclose Marco Cella: Nothing to disclose Claudia De Felici: Nothing to disclose Federico Bruno: Nothing to disclose Glymphatic dysfunction mediates the impact of tau p athology on neurodegeneration in cognitively unimpaired individ uals and prodromal Alzheimer’s Disease X. Xu, B. Zhang, *Z. Zhu*; Nanjing/CN Purpose or Learning Objective: To elucidate the pathological mechanism of glymphatic system dysfunction are associated with r egional tau deposition and tau-mediated neurodegeneration across the preclinic al and prodromal stage of the AD continuum.

Methods

or Background: Cognitively normal (CN) controls (n=94), individual s with mild cognitive impairment (MCI; n = 83), and t hose with significant memory concern (SMC; n =84) were included from the Alzheimer's Disease Neuroimaging Initiative. Tau pathology was measured by positron emission tomography, the glymphatic activity assessed by dif fusion tensor image analysis along the perivascular space (DTI-ALPS), a nd neurodegeneration reflected by hippocampal volume. Mediation analysis was used to study the possible pathways.

Results

or Findings: ALPS was significantly associated with tau and tau- mediated neurodegeneration, especially in parahippo campal gyrus. The relationship between glymphatic function and neurod egeneration was mediated by tau pathological deposition (indirect e ffect: 0.012, 95%CI [0.001— 0.029]) rather than ALPS index mediated the relatio nship between tau and neurodegeneration (indirect effect: 0.212, 95%CI [- 0.013,0.0002]). The relationship between glymphatic dysfunction and cog nitive decline were fully mediated by tau deposition and neurodegeneration in preclinical AD.

Conclusion

Tau deposition in specific region may mediate the r elationship of glymphatic dysfunction and neurodegeneration, which contribute to cognitive decline in the preclinic AD stage, facilitating the development of therapeutics targeting tau protein and glymphatic dysfunction in AD.

Limitations

Firstly, the cross-sectional design employed in thi s study limits us to explore the causal relationships or investigate longitudinal changes over time, future longitudinal studies could provide fur ther. Secondly, the ALPS index is mainly used to measure the function of the subcortical glymphatic system. Future research should continue to use othe r methods, such as BOLD-CSF coupling measurements, to validate the fun ction of the subcortical glymphatic system in AD. Funding for this study: This work was supported by the National Science and Technology Innovation 2030 -- Major program of "Bra in Science and Brain-Like Research" (2022ZD0211800); the National Natural Sci ence Foundation of China (82271965, 81971596, 82001793); the Fundament al Research Funds for the Central Universities, Nanjing University (2 020-021414380462); the Key Scientific Research Project of Jiangsu Health Commi ttee (K2019025); Special Funded Project of Nanjing Drum Tower Hospital (No. RC2022-023), Development Plan (Social Development) Project of Ji angsu Province (No. BE2022679). China Postdoctoral Science Foundation ( 2023M741648). The National Natural Science Foundation of China (82302 172); The funders had no role in the study design, data collection and analy sis, decision to publish, or preparation of the manuscript. Ethics committee - additional information: ADNI Ethics committee Author Disclosures: Bing Zhang: Nothing to disclose Zhengyang Zhu: Nothing to disclose Xinru Xu: Nothing to disclose The Impact of Temporal Muscle Thickness as an Indic ator of Sarcopenia on Clinical Status in Parkinson's Disease B. Atalay, K. Erincik, M. B. Doğan, *M. Gezgin*, H. Yıldız, F. B. Ozdilek; Istanbul/TR ([email protected]) Purpose or Learning Objective: To assess the impact of temporal muscle thickness (TMT), as an indicator of sarcopenia on c ognitive status and medication dosage in patients with Parkinson's dise ase (PD). Abstract-based Programme 44 Wednesday

Methods

or Background: A total of 54 patients with PD and 46 healthy controls were retrospectively analyzed. Brain MR im ages from both groups were reviewed by two radiologists who independently measured the right and left temporal muscle thickness using T1-weighted ax ial images. Clinical assessments included the Unified Parkinson’s Diseas e Rating Scale (UPDRS), Mini-Mental State Examination (MMSE), Hoehn and Yah r Scale, L-dopa equivalent daily dose (LEDD), and disease duration, collected by a neurologist. Interobserver agreement was evaluated using the int raclass correlation coefficient. The relationship between TMT and clini cal data was analyzed using Spearman’s correlation.

Results

or Findings: In the PD group, 33.3% of patients were female, compared to 50% in the control group. Interobserver agreement for TMT measurements was excellent. No significant differen ce in TMT was observed between the PD and control groups (p=0.16, p=0.34). A weak but statistically significant correlation was found between TMT, LEDD , and disease duration, while no correlation was found between TMT and UPDR S or Hoehn and Yahr scores. A weak but significant correlation was obse rved between TMT and MMSE scores.

Conclusion

Sarcopenia, characterized by muscle mass loss, shar es contributing factors with Parkinson’s disease. TMT, a reliable marker of muscle mass, can be measured on brain MRIs to assess sarco penia risk in PD patients. Our findings suggest that TMT correlates with cognitive function and medication dosage in PD patients, making it a valua ble tool for early sarcopenia detection and management in clinical set tings.

Limitations

Lack of patient follow-up and inability to assess t he impact of muscle mass increase on medication dosage. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee notification can be found under the number 2023/0912. Author Disclosures: Merve Gezgin: Nothing to disclose Hüseyin Yıldız: Nothing to disclose Kendal Erincik: Nothing to disclose Fatma Betül Ozdilek: Nothing to disclose Başak Atalay: Nothing to disclose Mahmut Bilal Doğan: Nothing to disclose Can a Coronal Swallow Tail Cleft Sign Increase the Confidence of Visualization of the Nigrosome-1 Layer of the Subst antia Nigra in Normal Subjects and Those with Parkinsonism? *S. Rajan*, J. S. Chatha, H. Mahajan; New Delhi/IN ([email protected]) Purpose or Learning Objective: To assess whether reconstruction of phase images from susceptibility-weighted imaging (SWI) i n a coronal plane enhances the visualization confidence of the nigros ome-1 layer in the substantia nigra. The coronal Swallow Tail cleft si gn may serve as a valuable indicator for this structure.

Methods

or Background: A retrospective review was conducted on 433 consecutive MR brain scans acquired using a routine protocol for various indications. The axial phase images of SWI were ana lyzed in Phase 1 by two radiologists with 9 and 22 years of experience. The y evaluated the visibility of the nigrosome-1 layer below the level of the red nu cleus using a 5-point Likert scale (1 = very difficult to 5 = easily seen). In P hase 2, coronal reconstructed images of the phase SWI were assessed perpendicular ly to the substantia nigra in posterior sections for the presence of the cleft sign, also scored with the 5-point Likert scale.

Results

or Findings: In Phase 1, the Likert scores of 1 and 2 were simil ar (6% and 9%, respectively). There was a marginal decreas e in score 3 (from 4.97% to 4.04%). Scores of 4 decreased significantly from 16.97% in Phase 1 to 7.96% in Phase 2, while the score of 5 increased su bstantially from 62.01% in Phase 1 to 72.97% in Phase 2. Statistical analysis demonstrated that the addition of the coronal Swallow Tail cleft sign sig nificantly enhanced confidence in visualizing the nigrosome-1 layer (p < 0.01).

Conclusion

The coronal Swallow Tail cleft sign significantly i mproves the confidence of visualization of the nigrosome-1 laye r in both normal subjects and those with Parkinsonism. This finding underscor es the utility of coronal reconstructions in enhancing diagnostic accuracy in neuroimaging.

Limitations

Correlation with nuclear scans or clinical history was not done Funding for this study: None Ethics committee - additional information: Restrospective Author Disclosures: Jagneet Singh Chatha: Nothing to disclose Sriram Rajan: Nothing to disclose Harsh Mahajan: Nothing to disclose Machine learning approach effectively discriminates between Parkinson’s disease and progressive supranuclear palsy: multi-l evel indices of rs- fMRI *W. Cheng*; Nanchang/CN ([email protected]) Purpose or Learning Objective: Parkinson’s disease (PD) and progressive supranuclear palsy (PSP) present similar clinical s ymptoms, but their treatment options and clinical prognosis differ significantly . Therefore, we aimed to discriminate between PD and PSP based on multi-leve l indices of rs-fMRI via the machine learning approach.

Methods

or Background: A total of 58 PD and 52 PSP patients were prospectively enrolled in this study. Participants were randomly allocated to a training set and a validation set in a 7:3 ratio. V arious resting-state functional magnetic resonance imaging (rs-fMRI) indices were e xtracted, followed by a comprehensive feature screening for each index. We constructed fifteen distinct combinations of indices and selected four machine learning algorithms for model development. Subsequently, different vali dation templates were employed to assess the classification results and i nvestigate the relationship between the most significant features and clinical assessment scales.

Results

or Findings: The classification performance of logistic regressi on (LR) and support vector machine (SVM) models, based on multiple index combinations, was significantly superior to that of other machine learning models and combinations when utilizing automatic an atomical labeling (AAL) templates. This has been verified across different templates.

Conclusion

The utilization of multiple rs-fMRI indices signifi cantly enhances the performance of machine learning models and can effectively achieve the automatic identification of PD and PSP at the indiv idual level.

Limitations

Only the rs-fMRI index was used in this study, and DTI-related microstructure data was not included. Funding for this study: This study was supported by the National Natural Science Foundation of China (82160331), Jiangxi Pro vince Double Thousand Talent Plan (jxsq2023201039). This project is imple mented by the Jiangxi Clinical Research Center for Medical Imaging (20223 BCG74001), and Jiangxi Province Key Laboratory for Precision Pathology and Intelligent Diagnosis (2024SSY06281). Ethics committee - additional information: This study was approved by the Medical Ethics Committee of the First Affiliated Ho spital of Nanchang University (approval number: IIT2022124). Author Disclosures: Weiling Cheng: Nothing to disclose Exploring the Relationship Between Body Composition and Brain Morphology in Aging: A Focus on Thigh Muscle Mass a nd Subcutaneous Fat as Predictors of Cortical Thickness in Healthy Older Adults *M. Sarkinaite*¹, U. Lukoseviciute¹, N. Masiulis¹, S. Lukoševičius¹, O. Levin², R. Gleiznienė¹; ¹Kaunas/LT, ²Leuven/BE ([email protected]) Purpose or Learning Objective: Subcutaneous fat accumulation has been linked to adverse brain health outcomes. This study examines the relationship between thigh muscle mass, subcutaneous fat distrib ution, and brain structure in elderly adults. It explores how body composition affects cortical thickness in brain regions linked to cognitive function.

Methods

or Background: Fifty-four healthy elderly individuals underwent imaging of the right thigh and brain using a 3T Sie mens Avanto MRI system. Muscle and subcutaneous fat cross-sectional areas ( CSA) were measured at 50% and 20% of thigh length, with the muscle-to-fat ratio calculated at the 50% mark. Cortical thickness was assessed through brain volumetric analysis using Freesurfer 7.4.1 software.

Results

or Findings: Significant positive correlations (r ≥ 0.2, p ≤ 0.05) were found between the muscle-to-fat ratio and cortical thickness in the left cerebellum, cuneus, and transverse temporal cortex, in the right entorhinal cortex, inferior temporal cortex, postcentral gyrus , superior temporal cortex, and the banks of the superior temporal sulcus (BANK SSTS). Additionally, significant negative correlations (r ≤ -0.2, p ≤ 0.05) were observed between subcutaneous fat CSA at 50% of thigh length and cor tical thickness in the left cuneus and entorhinal cortex, in the right BANKSSTS , postcentral gyrus, and superior temporal cortex. Additionally, subcutaneou s fat CSA at 20% of thigh length was inversely correlated with cortical thick ness in the left cuneus and right BANKSSTS (r ≤ -0.2, p ≤ 0.05).

Conclusion

Our findings demonstrate that increased thigh muscl e mass correlates with greater cortical thickness in cogni tive regions, while elevated subcutaneous fat is linked to reduced thickness. Th ese results highlight the role of body composition in maintaining brain healt h in the elderly and underscore the importance of muscle mass in mitigat ing age-related cortical decline.

Limitations

With only 54 participants, the study's findings may lack generalizability. Funding for this study: Supported by the Research Council of Lithuania (grant number P-MIP-22-217). Abstract-based Programme 45 Wednesday Ethics committee - additional information: Approved by the Kaunas region Medical Ethics Committee for Biomedical Research (N o. BE-2-22). Author Disclosures: Milda Sarkinaite: Nothing to disclose Nerijus Masiulis: Nothing to disclose Saulius Lukoševičius: Nothing to disclose Rymantė Gleiznienė: Nothing to disclose Urte Lukoseviciute: Nothing to disclose Oron Levin: Nothing to disclose Age-related hearing loss may be associated with sma ll vessel disease. Peak skeletonized mean diffusivity and TBSS study B. Genç, *I. C. Koc*, K. Aslan; Samsun/TR Purpose or Learning Objective: Peak skeletonized mean diffusivity (PSMD) is being proposed as a novel biomarker for small ve ssel disease. Tract-Based Spatial Statistics (TBSS) is a well-established DTI analysis method that enables the fully automated detection of microstruc tural changes in white matter. The aim of this study is to investigate whi te matter changes in patients with age-related hearing loss using both PSMD and T BSS.

Methods

or Background: All individuals from the OpenNeuro hearing loss connectome dataset were included in the study (http s://openneuro.org/ datasets/ds005026). The dataset consisted of 52 hea ring loss (HL) patients and 30 healthy controls. From the DTI images in the dataset, FA, MD, RD, and AD were obtained using FSL with preprocessing steps including TOPUP and eddy. The standard TBSS procedure was applied to in vestigate group differences in FA, MD, RD, and AD. The standard PSM D method with histogram analysis was used to calculate the differ ence between the 95th and 5th percentile MD values (https://www.psmd-marker.c om/). A comparison between the groups was made.

Results

or Findings: The TBSS analysis did not show any differences between the groups for any of the DTI parameters. T he PSMD value in the HL group was 225,21±23,50 x 10^-6 mm²/s, while in the control group it was 214,53±24,01 x 10^-6 mm²/s, showing a statistically significant increase in the hearing loss group (p=0,039).

Conclusion

Our findings suggest that small vessel disease may underlie the pathophysiology in patients with age-related hearin g loss. To our knowledge this is the first study to show the relationship be tween age-related hearing loss and small vessel disease.

Limitations

Since numerical data of hearing tests were not avai lable, the correlation between PSMD data and hearing test coul d not be analyzed. Funding for this study: No funding was received for this study. Ethics committee - additional information: OpenNeuro permits the use of patient data under the CC0 license. Ethical approva l has already been obtained by the "University of Salerno." Therefore, no separate ethical approval has been obtained from our institution. Author Disclosures: Bariş Genç: Nothing to disclose Kerim Aslan: Nothing to disclose Irem Ceren Koc: Nothing to disclose Glymphatic Dysfunction Correlate with Spatial Navig ation Deficits in Subjective Cognitive Decline: Insights from 5.0T MR I and Plasma Biomarkers Analysis F. Chen¹, B. Zhang¹, Q. Chen¹, X. Fan², L. Zou², Y. Li², G. Cheng², *G. Danni*¹; ¹Nanjing/CN, ²Shenzhen/CN ([email protected]) Purpose or Learning Objective: To assess the feasibility of analysis along the perivascular space (ALPS) using DTI obtained fr om 5.0 Tesla MR, assess glymphatic system function in SCD and its correlati on with spatial navigation abilities and plasma biomarkers.

Methods

or Background: Glymphatic dysfunction is implicated in cognitive impairment associated with AD. Spatial navigation i mpairments are among the earliest manifestations in individuals with SCD. Ho wever, the relationship between glymphatic function and spatial navigation remains poorly understood. Between May 2023 and January 2024, 62 SCD patients and 62 matched controls underwent high-resolution DTI on 5.0T MR s canner, spatial navigation behavioral tests, cognitive assessments, and Simoa plasma biomarker analyses. The ALPS index reflecting glymphatic acti vity was calculated by a ratio of the diffusivities along the x-axis in the projection and association neural fibers to the diffusivities perpendicular to them a nd compared according to the groups with use of multivariate analysis of varianc e. Inter-reproducibility of ALPS index among 5.0TMR and 3.0TMR scanners was eva luated using consistency interclass correlation coefficient. Pea rson correlation analysis was used to assess the relationship between cognitive p erformance, spatial navigation performance, plasma biomarkers, and the ALPS index.

Results

or Findings: The ALPS on 5.0TMR and ALPS on 3.0TMR showed strong consistency and correlation. SCD patients ha d significantly lower ALPS index on 5.0TMR and higher navigation errors compar ed to controls. The ALPS index was positively correlated with spatial n avigation, cognitive performance, and memory performance, and negatively correlated with plasma pTau217 levels.

Conclusion

The glymphatic function is impaired in SCD at the p reclinical AD stage, which may represent one of the physiological mechanisms leading to deficits in spatial navigation abilities. DTI-ALPS on 5.0T MR may serve as a sensitive neuroimaging biomarker for the preclinica l stage of AD.

Limitations

Our study is a small-sample cross-sectional study. Funding for this study: National Science and Technology Innovation 2030 -- Major program of "Brain Science and Brain-Like Rese arch" (2022ZD0211800) Ethics committee - additional information: The Research Ethics Committees of Nanjing Drum Tower Hospital, the Shen zhen Institute of Advanced Technology of the Chinese Academy of Scien ces, and Peking University Shenzhen Hospital. Author Disclosures: Qian Chen: Nothing to disclose Ye Li: Nothing to disclose Xiang Fan: Nothing to disclose Guanxun Cheng: Nothing to disclose Bing Zhang: Nothing to disclose Ge Danni: Nothing to disclose Futao Chen: Nothing to disclose Lixian Zou: Nothing to disclose Static and Dynamic Functional Connectivity Alternat ions of Medial and Lateral Entorhinal Cortex with Subjective Cognitive Decline *G. Danni*, Z. Bing; Nanjing/CN ([email protected]) Purpose or Learning Objective: To investigate the static functional connectivity (sFC) and dynamic functional connectiv ity (dFC) of medial entorhinal cortex (MEC) and lateral entorhinal cort ex (LEC) in individuals with subjective cognitive decline (SCD) and the associat ions with cognitive performance, spatial navigation and olfactory memor y.

Methods

or Background: Seventy-seven control subjects and 106 SCD individuals were enrolled, and neuropsychological e valuations, 2D computerized spatial navigation test, olfactory mem ory test and resting-state functional magnetic resonance imaging (rs-fMRI) wer e collected. Bilateral MEC and LEC were selected as seeds to investigate alter nations of the volumes, sFC and dFC.

Results

or Findings: Compared to control subjects, SCD individuals exhib ited decreased sFC between bilateral LEC and visual netw ork, between right LEC and left posterior cingulate gyrus and sensory moto r network, and between right MEC and left hippocampus, visual network and sensory motor network. The dFC between right LEC and right triangular part of inferior frontal gyrus (IFGtriang) decreased, while dFC between left MEC a nd right putamen, and between right MEC and right middle temporal gyrus i ncreased. In SCD group, volumes of bilateral MEC were positively correlated with spatial navigation ability, and sFC between bilateral LEC and visual n etwork was positively correlated with olfactory memory. The dFC between r ight LEC and right IFGtriang was correlated positively with global cog nitive performance. The combination of sFC and dFC as biomarkers to identif y SCD showed an area under curve of 92.1%.

Conclusion

There were functional alternations of EC subregions in SCD individuals, and we demonstrated the association be tween LEC and spatial navigation, and MEC and olfactory memory. The combi nation of sFC and dFC may be a new neuroimaging biomarker for the early d iagnosis of AD.

Limitations

The study lacked genetic and biomarker data. Also, we didn't have follow-up data to track pathological progressi on. Funding for this study: National Science and Technology Innovation 2030_Major program of "Brain Science and Brain-Like Research" (No. 2022ZD0211800) Ethics committee - additional information: Nanjing Drum Tower Hospital Ethics Committee Author Disclosures: Ge Danni: Author: Department of Radiology, the Affi liated Drum Tower Hospital of Nanjing University Medical School Zhang Bing: Nothing to disclose Automated cerebral microhemorrhage detection on T2* GRE for Alzheimer’s disease screening S. Van Eyndhoven, *R. Magalhaes*, R. Khan, T. V. Ph an, A. Liseune, A. Brys, D. M. Sima, J. Verheyden, A. Ribbens; Leuven/BE ([email protected]) Purpose or Learning Objective: Develop a robust automated deep learning- based method for assessment of cerebral microhemorr hages on T2* gradient- echo (GRE) images.

Methods

or Background: Hypointensities on GRE images can be indicative of cerebral microhemorrhages, and serve as exclusionar y criteria for anti-amyloid Abstract-based Programme 46 Wednesday therapies for Alzheimer’s disease, as they are link ed to increased risk of intracerebral hemorrhage. Automated detection tools could greatly aid radiologists in the challenging task of accurately quantifying these findings. A deep learning model was developed to detect microhe morrhages on cross- sectional GRE images. Training was done using 247 2 D GRE images and accompanying manual microhemorrhage annotations fro m the EMERGE clinical trial (NCT02484547). Detection accuracy wa s evaluated on a stratified subset of cases (N=600) of the Alzheimer’s Disease Neuroimaging Initiative, where up to 10 microhemorrhages were annotated on a 2D GRE sequence by experts. This is representative of the population t hat would be screened via MRI before administration of anti-amyloid treatment , for which assessment is more challenging relative to cases with a large num ber of microhemorrhages. For each case, we evaluated the F1 score, i.e., the harmonic mean between detection sensitivity and positive predictive value , and the absolute error between the number of microhemorrhages according to the expert ground truth and the automated count.

Results

or Findings: The median F1 score of the trained model was 0.67, and its median absolute count error was 1 microhemorrha ge.

Conclusion

Detection of microhemorrhages on 2D GRE images is a challenging task, which is gaining importance with the advent of novel anti- amyloid treatments that may lead to hemorrhagic sid e effects. Using a robust, validated AI tool, as described here, can assist ra diologists in microhemorrhage detection, providing value especial ly in sparse MR images, though expert assessment remains vital.

Limitations

None. Funding for this study: N/A Ethics committee - additional information: N/A Author Disclosures: Arne Brys: Employee: icometrix Ricardo Magalhaes: Employee: icometrix Annemie Ribbens: Employee: icometrix Shareholder: i cometrix Rafay Khan: Employee: icometrix Diana M. Sima: Employee: icometrix Thanh Vân Phan: Employee: icometrix Simon Van Eyndhoven: Employee: icometrix Arno Liseune: Employee: icometrix Jan Verheyden: Shareholder: icometrix Employee: ico metrix 13:00-14:30 Research Stage 4 Research Presentation Session: Abdominal and Gastrointestinal RPS 401 Estimation of liver fat and stiffness with imaging Moderator J. M. Lee; Seoul/KR ([email protected]) The Diagnostic Accuracy of Quantitative Ultrasound with Fat Fraction parameter for the Assessment of Hepatic Steatosis i n Patients with Metabolic-Dysfunction Associated Fatty Liver Diseas e R. Cannella, A. A. Blandino, A. Tulone, S. Petta, * T. V. Bartolotta*; Palermo/IT ([email protected]) Purpose or Learning Objective: To investigate the performance of quantitative ultrasound with US Fat Fraction (USFF) in patients with metabolic dysfunction-associated steatotic liver disease (MAS LD).

Methods

or Background: This study included consecutive patients with MASLD who prospectively underwent MRI and quantitat ive ultrasound on the same day. MRIs were acquired on a 3T scanner and th e fat fraction quantification was obtained with mDixonQuant sequen ce. Quantitative ultrasound consisted of Tissue Attenuation Imaging (TAI), Tissue Scatter- distribution Imaging (TSI), and US Fat Fraction (US FF) acquired by two operators (a radiologist and a radiology resident) to evaluate the inter-reader reliability. Spearman's rank-order correlation was calculated between USFF and MRI fat fraction. The diagnostic performance wa s investigated with the area under the receiver operating characteristics c urve (AUC), sensitivity and specificity according to the optimal cutoff.

Results

or Findings: Fifty-nine patients (40 males, median age of 60 yea rs) were enrolled. Among them, 47 (79.7%) had grade ≥1 steatosis and 23 (39.0%) had grade≥2 steatosis. There was a high positive correlation between USFF and MRI fat fraction (rho: 0.877, p10.0 had a sensitivity 94.1% and a specificity of 100% for grade≥1 steatosis. For the diagnosis of grade≥2 steatosis, the AUCs were 0.958 (95%CI: 0.777, 0.999) for TAI, 0.858 (95%CI: 0.757, 0.916) for TSI , and 0.958 (95%CI: 0.777, 0.999) for USFF. An USFF >11.8 had a sensitivity 10 0% and a specificity of 83.3% for grade≥2 steatosis. The reproducibility of USFF was excell ent (ICC of 0.98; 95%CI: 0.95, 0.98).

Conclusion

USFF provides an excellent performance and reproduc ibility for the quantification of hepatic steatosis in patients with MASLD.

Limitations

Lack of liver biopsy. Funding for this study: Study supported by Samsung. Ethics committee - additional information: All participants provided written informed consent. Author Disclosures: Adele Tulone: Nothing to disclose Antonino Andrea Blandino: Nothing to disclose Roberto Cannella: Research/Grant Support: Co-fundin g by the European Union - FESR or FSE, PON Research and Innovation 20 14-2020 - DM 1062/2021; research collaboration with Siemens Heal thineers. Other: Support for attending meetings from Bracco and Bayer Salvatore Petta: Nothing to disclose Tommaso Vincenzo Bartolotta: Nothing to disclose Multiparametric ultrasound evaluation of hepatic fi brosis and steatosis in patients treated with bempedoic acid: a comparative study on 100 patients *G. Daccordi*, P. Sacco, F. Lazzeretti, G. De Filip po, C. Tucci, M. A. Mazzei; Siena/IT ([email protected]) Purpose or Learning Objective: This study evaluates the effectiveness of bempedoic acid in improving hepatic steatosis using multiparametric ultrasound and serum lipid levels, comparing result s with a control group.

Methods

or Background: Background: Hepatic steatosis is associated with dyslipidemia and liver dysfunction. Multiparametric ultrasound, using techniques like shear-wave elastography and attenua tion imaging (ATI), offers a non-invasive assessment of liver fibrosis and ste atosis. Bempedoic acid is a potential therapy to reduce liver fat and improve l ipid profiles. Materials and

Methods

We enrolled 100 patients with hepatic stea tosis: 50 received bempedoic acid therapy, and 50 served as controls. Ultrasound was performed using the Canon Aplio i800, evaluating fibrosis via shear-wave elastography, and steatosis using ATI and dispersion imaging. Two operators with different experience levels performed the exams at baseline a nd after treatment. Serum cholesterol, triglycerides, and liver enzymes were also monitored.

Results

or Findings: Significant improvements in fibrosis, steatosis, an d dispersion were observed in the treatment group (p 0.05), highlighting the utility of multiparametric ultrasound for follow-up. Good inte r-operator agreement was noted for ATI (ICC = 0.96), with slight discordance in fibrosis evaluation (ICC = 0.88), likely due to experience differences.

Conclusion

Multiparametric ultrasound plays a fundamental role for the non- invasive follow-up of hepatic fibrosis and steatosi s during therapy. This approach is a valuable tool for assessing patient r esponse, aiding in informed clinical decisions.

Limitations

Ongoing single-center study Funding for this study: Not addictional funding Ethics committee - additional information: University of Siena Author Disclosures: Maria Antonietta Mazzei: Nothing to disclose Cristina Tucci: Nothing to disclose Francesco Lazzeretti: Nothing to disclose Palmino Sacco: Nothing to disclose Giorgia Daccordi: Nothing to disclose Giovanna De Filippo: Nothing to disclose Functional Liver Imaging Score: A biomarker for pre diction of acute-on- chronic liver failure *A. Kristic*, L. Balcar, A. Ba-Ssalamah, T. Reiberg er, M. Mandorfer, N. Bastati-Huber, R. Ambros, L. Beer, S. Pötter-Lan g; Vienna/AT Purpose or Learning Objective: The Functional Liver Imaging Score (FLIS) derived from gadoxetic acid (GA)-enhanced MRI is a prognostic biomarker in patients with advanced chronic liver disease (ACLD) . The aim of this study was to investigate whether FLIS, as well as quantitativ e imaging parameters, including the relative liver enhancement (RLE), rel ative enhancement ratio of the biliary system (REB), and liver-to-portal vein contrast ratio (LPC), can predict acute-on-chronic liver failure (ACLF; a syn drome defined by extrahepatic organ failure and high short-term mort ality) in patients with an acute decompensation (AD) of cirrhosis, i.e., the m ain at-risk population. Abstract-based Programme 47 Wednesday

Methods

or Background: We included 322 ACLD patients with GA-MRI- derived, semi-quantitative FLIS, in whom the RLE, R EB, and LPC were also computed by two independent radiologists. Patients were stratified into stable ACLD (compensated or non-hospitalized/electively ho spitalized/non-liver- related hospitalized decompensated ACLD patients) a nd acutely decompensated (AD) patients (non-elective liver-rel ated hospitalization). The predictive values of semi-quantitative FLIS and qua ntitative GA-MRI parameters for ACLF development were investigated b y Cox regression analyses.

Results

or Findings: The FLIS was lower in AD (vs. stable ACLD) patients . Furthermore, low FLIS was an independent risk facto r for ACLF development/liver-related death in AD patients (adj usted hazard ratio [aHR]: 2.26; 95%CI: 1.08-4.71; P=0.03), as well as in clin ically stable ACLD patients (aHR: 2.35; 95%CI: 1.21-4.55; P=0.01). Conversely, while RLE, REB, and LPC distinguished between AD and clinically stable ACLD patients (P<0.001), they failed to predict ACLF/liver-related death.

Conclusion

The FLIS is a simple prognostic imaging biomarker i n AD patients in whom ACLF risk stratification is important to id entify patients who may benefit from intensified monitoring or timely liver transplant evaluation.

Limitations

The retrospective design could have led to a select ion bias; however, the study allowed for a long clinical foll ow-up and a high number of endpoints. Funding for this study: No funding was received for this study. Ethics committee - additional information: Received approval EK 2023/2017. Author Disclosures: Antonia Kristic: Nothing to disclose Sarah Pötter-Lang: Nothing to disclose Lorenz Balcar: Speaker: received speaker fees from Chiesi, and Gilead. Raphael Ambros: Nothing to disclose Mattias Mandorfer: Advisory Board: served as a spea ker and/or consultant and/or advisory board member for AbbVie, Collective Acumen, Echosens, Gilead, Ipsen, Takeda, and W. L. Gore & Associates. Other: received travel support from AbbVie and Gilead. Consultant: served as a speaker and/or consultant and/or advisory board member for AbbVie, Collective Acumen, Echosens, Gilead, Ipsen, Takeda, and W. L. Gore & A ssociates. Grant Recipient: received grants from Echosens. Speaker: served as a speaker and/or consultant and/or advisory board member for AbbVie, Collective Acumen, Echosens, Gilead, Ipsen, Takeda, and W. L. Gore & Associates. Lucian Beer: Speaker: received speaker fees from Ta keda, and Lilly. Ahmed Ba-Ssalamah: Speaker: received honoraria for lectures and a consultancy from Bayer without relation to the pres ent article. Consultant: received honoraria for lectures and a consultancy f rom Bayer without relation to the present article. Nina Bastati-Huber: Nothing to disclose Thomas Reiberger: Speaker: speaking honoraria from AbbVie, Gilead, W. L. Gore & Associates, Intercept, Roche, and MSD. Grant Recipient: received grant support from AbbVie, Boehringer-Ingelheim, Gi lead, Intercept, MSD, Myr Pharmaceuticals, Philips Healthcare, Pliant, Siemen s, and W. L. Gore & Associates. Other: and travel support from AbbVie, Boehringer-Ingelheim, Gilead, and Roche. Consultant: consulting/advisory board fees from AbbVie, Bayer, Boehringer-Ingelheim, Gilead, Intercept, MSD , and Siemens; and travel support from AbbVie, Boehringer-Ingelheim, Gilead, and Roche. Advisory Board: consulting/advisory board fees from AbbVie, Bayer, Boehringer- Ingelheim, Gilead, Intercept, MSD, and Siemens. A compound model improves the accuracy of ultrasoun d-estimated fat- fraction *P. N. Kaposi-Novák*, B. Zsély, Z. Zsombor, M. Hims el, V. Bérczi, G. Györi, P. Maurovich-Horvat, A. D. Rónaszéki; Budapest/HU Purpose or Learning Objective: The ultrasound-estimated fat-fraction (UEFF) is new biomarker that can facilitate the diagnosis and follow-up of liver steatosis. Different models have been devised to ca lculate UEFF, but these have yet to be compared side-by-side.

Methods

or Background: We retrospectively collected ultrasound parameters from sixty patients with various grades (S0-S3) of metabolic dysfunction- associated steatotic liver disease (MASLD), includi ng attenuation coefficient (AC), backscatter-distribution coefficient (BSC), a nd liver capsule-to-skin distance (CSD). The training set had balanced distr ibution of steatosis grades (S0, S1, S3 - 16 cases each, S2 - 12 cases). Univar iable and multivariable linear and exponential models were trained to predi ct the MRI proton density fat-fraction (PDFF) using repeated cross-validation . The models were tested on fifty cases for which the scanner’s application cal culated an ultrasound fat- fraction (USFF). We compared the R-squared (R2) and the one-way random effect intraclass correlation coefficients (ICC) am ong the models.

Results

or Findings: In low-grade steatosis (≤ S1), the linear model using AC and BSC achieved the best fit (R2= 0.379, p<0.002), and USFF had the best agreement (ICC=0.538, p<0.003) with PDFF. In high-g rade steatosis (≥ S2), the ASC and BSC-based multi-exponential model perfo rmed best (R2= 0.252, p<0.008, ICC=0.422, p<0.012). The fit could be impr oved by including CSD in the model (R2=0.263, p<0.007). The univariable nonl inear AC model resulted in a slightly weaker agreement in both low-grade (I CC=0.236, p<0.019) and high-grade (ICC=0.227, p<0.012) steatosis. A compou nd model using linear regression in low-grade or nonlinear regression in high-grade steatosis outperformed (R2= 0.6, p<0.001, ICC=0.849, p<0.001) other models in predicting the test cases.

Conclusion

The UEFF is a robust method to diagnose liver steat osis across all stages. The compound model substantially improv es the fit and the agreement between UEFF and PDFF.

Limitations

Single center study. Funding for this study: Pál Novák Kaposi was recipient of a research grant from Samsung Medison Ltd. (Contract number: SE4K/20 23/195). Zita Zsombor was recipent of a resarch scholarship from the New National Excellence Program (ÚNKP-23-3-I-SE-23) of the Hunga rian Ministry of Culture and Innovation. Ethics committee - additional information: Semmelweis University’s Regional and Institutional Science and Research Eth ics Committee (Protocol number: SE RKEB 140/2020, 16 July 2020, and SE RKEB 6/2023, 9 February 2023) Author Disclosures: Pál N. Kaposi-Novák: Research/Grant Support: Samsun g Medison Ltd. Aladár David Rónaszéki: Nothing to disclose Pál Maurovich-Horvat: Nothing to disclose Boglárka Zsély: Nothing to disclose Marco Himsel: Nothing to disclose Viktor Bérczi: Nothing to disclose Zita Zsombor: Nothing to disclose Gabriella Györi: Nothing to disclose Dual-Energy CT Liver Fat Quantification as Imaging Biomarker of Mortality and Morbidity in Intensive-Care Patients *J. Erley*, J. Breckow, K. Roedl, A. Duoerkongjiang , G. De Heer, E. Tahir, J. Yamamura, G. Adam, I. Molwitz; Hamburg/DE Purpose or Learning Objective: This study aimed to evaluate the association between liver fat content, assessed using dual-ener gy computed tomography (DECT) material decomposition, in immobilized inten sive care unit (ICU) patients with in-hospital mortality, length of ICU stay, and indicators of morbidity (need for tracheotomy and renal replaceme nt therapy).

Methods

or Background: ICU patients who received a DECT upon ICU admission between November 2019 and December 2022 w ere retrospectively investigated. DECT liver fat fraction (DECT-FF) was determined by material decomposition for fat, liver tissue, and iodine by combining two regions of interest (ROI) in the right and one ROI in the left liver lobe (min. size 3.6 cm2). Cox proportional hazard models were employed, inclu ding DECT-FF, sex, age, body mass index, ICU scoring systems for disease pr ediction, reason for admission, pre-existing malignancies, chronic disea ses, and inflammatory diseases.

Results

or Findings: In total 76 patients were included (33 female, mean age 61±12 years, of which 59% died in the hospital. DEC T-FF at CT1 was 3.3±5.4%. DECT-FF was associated with in-hospital m ortality (hazard ratio 1.09 [95% confidence interval 1.03; 1.15], p=0.004) , with the length of ICU-stay (odds ratio (OR) -4.28 [-6.64; -1.92], p=0.001) and with the need for a tracheotomy (OR: 0.90 [0.80; 0.99], p=0.039). No as sociation was observed between DECT-FF and ICU scoring systems or renal re placement therapy.

Conclusion

A higher liver DECT-FF upon ICU admission was assoc iated with in-hospital mortality and, probably as a consecutiv e effect/bias, with a decreased length of ICU stay and a lower likelihood of tracheotomy in surviving patients. The liver DECT-FF may serve as a predicti ve imaging biomarker of mortality in critically ill patients.

Limitations

The sample size and its heterogeneity. Funding for this study: Not applicable. Ethics committee - additional information: The study has been approved by the ethics committee of the medical association in Hamburg. Author Disclosures: Gerhard Adam: Nothing to disclose Isabel Molwitz: Nothing to disclose Alidan Duoerkongjiang: Nothing to disclose Geraldine De Heer: Nothing to disclose Kevin Roedl: Nothing to disclose Jin Yamamura: Nothing to disclose Jennifer Erley: Nothing to disclose Julia Breckow: Nothing to disclose Enver Tahir: Nothing to disclose Abstract-based Programme 48 Wednesday Accuracy of estimates of liver fat content based on 2point-Dixon in comparison to multi-echo-Dixon sequences in a popul ation-based cohort *M-N. Von Itter*¹, T. Nonnenmacher², T. Norajitra², S. Rospleszcz¹, J. Machann³, F. Bamberg¹, J. Nattenmüller¹; ¹Freibu rg/DE, ²Heidelberg/DE, ³Tübingen/DE Purpose or Learning Objective: The rising prevalence of metabolic dysfunction-associated steatotic liver disease (MAS LD), formerly NAFLD, makes it a leading liver disease and a risk factor for steatohepatitis, cirrhosis, and hepatocellular carcinoma. As it is potentially reversible, early detection is key.

Methods

or Background: A validated nnU-Net liver segmentation model processed 10,636 MRI scans from NAKO participants u sing 2-point- (2p) and multi-echo- (me) Dixon data. Samples with significa nt mask mismatches were excluded. The segmentation masks were used to calcu late mean liver fat content (LFC) from 2p-Dixon data, which was then co mpared to the me-Dixon proton density fat fraction.

Results

or Findings: The overall mean difference between the fat fractio n from me-Dixon measurements and the 2p-Dixon fat est imation is - 1.101±0.012%. 95% of the measured differences are within the range of -3.8% and 1.4%. Mean LFCs from 2p-Dixon data overestimate the fat content compared to me-Dixon, especially for higher fat dep osition. Liver iron content did not lead to a systemic offset in our cohort, th ough only participants with mild to moderate liver iron content were part of ou r sample.

Conclusion

As the 2p-Dixon sequence is often acquired for vari ous clinical questions beyond specific liver imaging, assessment of 2p-Dixon-based LFC can identify individuals with MASLD in possible scr eening programs or MRI studies for other reasons and guide them to the com plementary acquisition of a me-Dixon sequence. This could benefit individuals b y initiating preventive measures and reduce the socio-economic burden of MA SLD caused by its complications.

Limitations

No histological confirmation was performed. Funding for this study: Funding was received from the German Research Foundation (Deutsche Forschungsgemeinschaft, grant number: 428224476). The NAKO is funded by the Federal Ministry of Educa tion and Research (BMBF) [project funding reference numbers: 01ER1301 A/B/C and 01ER1511D], the federal states and the Helmholtz As sociation, with additional financial support by the participating universities and the institutes of the Leibniz association. Ethics committee - additional information: The study received institutional review board approval and written informed consent was obtained from all participants. Author Disclosures: Tobias Nonnenmacher: Nothing to disclose Susanne Rospleszcz: Nothing to disclose Johanna Nattenmüller: Nothing to disclose Jürgen Machann: Nothing to disclose Fabian Bamberg: Nothing to disclose Marc-Nicolas Von Itter: Nothing to disclose Tobias Norajitra: Nothing to disclose Deep learning-based liver volume and fat fraction q uantification from Dixon-MRI: Reference curves from over 66,000 indivi duals and their prognostic value *M. Jung*¹, L. Michel², M. Reisert², S. Jäck², S. R ospleszcz², M. T. Lu¹, F. Bamberg², V. Raghu¹, J. Weiß²; ¹Boston, MA/US, ² Freiburg/DE ([email protected]) Purpose or Learning Objective: Steatotic liver disease (SLD) is a major public health concern with a global prevalence of 3 2.4% and an independent risk factor for cardiometabolic and liver disease. We used a deep learning framework to quantify liver volume and SLD from MRI in a large Western European population to calculate reference curves a nd investigate their prognostic value.

Methods

or Background: We developed a deep-learning model that takes an MRI as input and outputs liver volume (L) and fat f raction (FF, %) using data from the UK Biobank (UKBB) and German National Coho rt (NAKO). Established FF-thresholds were used to define mild( 5%), moderate(15%), and severe(25%) SLD. We computed age-, sex-, and height -normalized reference curves and assessed the prognostic value of liver v olume z-score (z2) and SLD-categories for incident outcomes (diab etes; liver disease; all- cause mortality) in the UKBB (n=35,002). Cox regres sion assessed the association between volume z-score and SLD categori es with outcomes after adjustment for age, sex, BMI, race, and cardiometab olic risk factors (serum glucose, Hb1Ac, lipid panel, prevalent hypertension , history of cancer, alcohol consumption, smoking status).

Results

or Findings: Among 66,664 individuals from the general populatio n (57.7±12.9 years; BMI: 26.2±4.5 kg/m2, 48.3% female), SLD was high with a prevalence of 80.4%. In the UKBB, multivariable-adj usted Cox regression showed that severe steatotic liver disease (SLD) wa s associated with an increased risk of incident diabetes (adjusted hazar d ratio [aHR] 2.66) and liver disease (aHR 6.34) compared with no SLD. A liver vo lume z-score >2 was associated with higher all-cause mortality (aHR 2.2 5) compared with a z-score <1.

Conclusion

SLD and normalized liver volume categories predicte d outcomes beyond traditional risk factors. We will release op en-source reference curves to enhance clinical liver risk assessment and improve comparability in research.

Limitations

Predominantly white population. Funding for this study: This project was conducted with data from the German National Cohort (NAKO) (www.nako.de). The NA KO is funded by the Federal Ministry of Education and Research (BMBF) [ project funding reference numbers: 01ER1301A/B/C, 01ER1511D, and 01ER1801A/B/ C/D], federal states of Germany, and the Helmholtz Association, t he participating universities and the institutes of the Leibniz Asso ciation. This research has been conducted using the UK Biobank Resource under Application Number 80337. We thank all participants who took part in t he NAKO and UKBB study and the staff of these research initiatives. MJ was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Founda tion) - 518480401. VKR was funded by Norn Group Longevity Impetus Gran t, NHLBI K01HL168231, and AHA Career Development Award 93517 6. Ethics committee - additional information: Informed consent was obtained from all participants in the UK Biobank and the Ger man National Cohort study. In addition, we received local IRB approval (IRB of the University of Freiburg: 23-1316-S1-retro and 24-1099-S1-retro). Author Disclosures: Susanne Rospleszcz: Nothing to disclose Marco Reisert: Nothing to disclose Jakob Weiß: Nothing to disclose Lea Michel: Nothing to disclose Matthias Jung: Nothing to disclose Fabian Bamberg: Nothing to disclose Saskia Jäck: Nothing to disclose Vineet Raghu: Nothing to disclose Michael T. Lu: Nothing to disclose Intra-individual quantitative crossover comparision of liver fat measurements between free-breathing radial GRE and conventional cartesian GRE breath-hold methods *T. B. Rodrigues*¹, M. D. Santana¹, V. Hérida¹, N. Almeida¹, T. Castela¹, R. C. Semelka², M. Ramalho¹; ¹Lisbon/PT, ²Chapel Hi ll, NC/US ([email protected]) Purpose or Learning Objective: To evaluate and compare liver fat fraction measurements derived from a free-breathing radial s equence with those obtained from reference cartesian breath-hold techn iques in an intra-individual fashion.

Methods

or Background: The study included 40 subjects (19 males, 21 females; mean age 60.5 ± 13.2) who underwent MRI ex aminations.Three T1- weighted sequences were used: Cartesian 2D-FLASH du al gradient-echo(2D- GRE), Cartesian 3D-GRE with Dixon technique, and fr ee-breathing 3D-GRE with radial data sampling(Radial 3D-GRE). Two indep endent readers measured the mean region of interest (ROI) values o f the liver for in-phase (IP) and out-of-phase (OP) images for each sequence with an equal ROI and similar location. Quantitative liver fat fractions (FF) were calculated using ((SI(IP)−SI(OP))/(2 × SI(IP))×100. The liver FF wer e compared across sequences. The inter-reader agreement was assessed using the intraclass correlation coefficient (ICC).Pearson correlation a nd regression analyses examined relationships among different measurement techniques. P-value of <0.05 was considered significant.

Results

or Findings: The study found strong correlations between liver F F measurements across 2D-GRE, 3D-GRE, and radial 3D-G RE sequences. Pearson correlation coefficients were 0.9788 for 2D -GRE vs.3D-GRE, 0.9506 for 3D-GRE vs.Radial 3D-GRE, and 0.9478 for 2D-GRE vs.Radial 3D-GRE (p<0.0001). Regression analyses confirmed strong ag reement between methods. ICC was 0.9489 (95%CI=0.9279 to 0.9637).

Conclusion

Our findings underscore the promising potential of the new IP/OP free-breathing Radial 3D-GRE sequence as a reliable alternative to traditional breath-hold techniques for liver FF in patients who cannot hold their breath. The excellent correlation between Radial 3D-GRE and the cartesian methods should recommend its use for patients incapable of suspending breathing.

Limitations

Measurement values may be affected by variations in fat distribution within the liver and small sample size .These factors may impact accuracy and reproducibility and introduce selectio n bias. However, identical ROI and matched locations were used in all patients . Funding for this study: No funding was received for this study Abstract-based Programme 49 Wednesday Ethics committee - additional information: Retrospective study Author Disclosures: Vasco Hérida: Nothing to disclose Miguel Ramalho: Nothing to disclose Tiago Castela: Nothing to disclose Nuno Almeida: Nothing to disclose Mariana Domingues Santana: Nothing to disclose Richard C. Semelka: Nothing to disclose Teresa Barata Rodrigues: Nothing to disclose Magnetic resonance elastography of the liver: are t he results reliable? *V. Atamaniuk*, M. Obrzut, L. Hanczyk, M. Cholewa, B. Obrzut; Rzeszów/PL ([email protected]) Purpose or Learning Objective: Magnetic resonance elastography (MRE) has become the gold standard for non-invasive asses sment of liver stiffness, especially in patients with liver fibrosis, replaci ng liver biopsy. While guidelines for MRE protocols and interpretation have been stan dardized by the Quantitative Imaging Biomarkers Alliance (QIBA), th e impact of vibration amplitude and slice thickness on hepatic stiffness measurements remains uncertain. This study aims to evaluate the reliabil ity of MRE under varying vibration amplitudes and slice thicknesses.

Methods

or Background: Twenty volunteers (10 men, 10 women), aged 18– 68, underwent MRE on a 1.5 T whole-body scanner usi ng a 2D GRE sequence. The Resoundant system delivered vibration s at 60 Hz with amplitudes of 25%, 50%, 75%, and 100%. Standard sli ce thickness was 10 mm, and an additional scan was performed at 75% amp litude with a 5 mm slice thickness. Regions of interest (ROIs) were ma nually drawn per QIBA guidelines to ensure consistent location for stiffn ess measurements across tested conditions. Statistical analyses included re peated measures ANOVA and intraclass correlation coefficients (ICC).

Results

or Findings: The effects of both vibration amplitude (p = 0.11) and slice thickness (p = 0.69) on hepatic stiffness wer e not statistically significant. The ICC for different amplitudes was 0.92, and for slice thickness, it was 0.95, indicating excellent agreement across conditions.

Conclusion

MRE provides reliable liver stiffness measurements, with no significant influence from changes in vibration amp litude or slice thickness. Given the excellent agreement across varying imagin g parameters, MRE can be considered a stable and reproducible method for assessing liver stiffness. Further multicentre studies with larger samples and 3D MRE may help confirm these findings and expand clinical applications of this technique.

Limitations

The study's limitations include the small sample si ze and the use of 2D MRE. Funding for this study: No external funding was obtained for this study. Ethics committee - additional information: Approved by the Regional Medical Chamber ethics committee (Resolution No 60/ 2022/B); informed consent was obtained from all participants. Author Disclosures: Marzanna Obrzut: Nothing to disclose Marian Cholewa: Nothing to disclose Lukasz Hanczyk: Nothing to disclose Vitaliy Atamaniuk: Nothing to disclose Bogdan Obrzut: Nothing to disclose Frequency and distribution of steatotic liver disea se in the NAKO study – magnetic resonance imaging of 30,000 participants *M-N. Von Itter*¹, T. Nonnenmacher², E. Grune¹, J. Machann³, J. Weiß¹, J. Nattenmüller¹, T. Norajitra², S. Rospleszcz¹, N. Consortium¹; ¹Freiburg/DE, ²Heidelberg/DE, ³Tübingen/DE Purpose or Learning Objective: Steatotic liver disease (SLD) and its subtype, metabolic dysfunction-associated steatotic liver di sease (MASLD), are risk factors for cardiometabolic disease, liver cirrhosi s, and hepatocellular carcinoma, and represent a major public health burd en. Using magnetic resonance imaging in Germany’s largest population-b ased study (NAKO), we describe the frequency of SLD and MASLD, and evalua te the distribution according to sex, age, BMI, geographic region, and socio-economic status.

Methods

or Background: A validated nnU-Net liver segmentation model processed 29,842 MRI scans from NAKO participants ( 44.1% women) using T1-weighted 6-point Dixon data. The segmentation ma sks were used to calculate mean liver fat content (LFC). SLD was def ined as LFC ≥ 5.56%, and MASLD according to established criteria.

Results

or Findings: Overall frequency of SLD and MASLD was 37.6% and 31.8% in men, and 20.3% and 18.6% in women, respect ively. Frequency increased with increasing BMI and age, with differe nt patterns in men and women. Geographically, the highest frequency of SLD was found in Eastern Germany (40.6%). Frequency was higher in individual s with low socio- economic status, and this difference was more prono unced in women (frequency for high vs. low socio-economic status: 35.7% vs. 47.2% in men, and 17.1% vs. 37.9% in women).

Conclusion

Frequency of SLD and MASLD in Germany is high, with pronounced differences according to sex, age, BMI, geographic region, and socio-economic status. Our findings provide a robus t basis to estimate the public health impact of these liver diseases in Ger many.

Limitations

Ultrasound for diagnosis of SLD was not available. Funding for this study: Funding was received from the German Research Foundation (Deutsche Forschungsgemeinschaft, grant number: 428224476). The NAKO is funded by the Federal Ministry of Educa tion and Research (BMBF) [project funding reference numbers: 01ER1301 A/B/C and 01ER1511D], the federal states and the Helmholtz As sociation, with additional financial support by the participating universities and the institutes of the Leibniz association. Ethics committee - additional information: The study received institutional review board approval and written informed consent was obtained from all participants. Author Disclosures: Tobias Nonnenmacher: Nothing to disclose Susanne Rospleszcz: Nothing to disclose Johanna Nattenmüller: Nothing to disclose Elena Grune: Nothing to disclose Jakob Weiß: Nothing to disclose Jürgen Machann: Nothing to disclose Nako Consortium: Nothing to disclose Marc-Nicolas Von Itter: Nothing to disclose Tobias Norajitra: Nothing to disclose Assessment of using Multimodal Magnetic Resonance I maging (MRI) for Noninvasive Evaluation of type 2 diabetic *M. W. Yang*, W. J. Shao; KunMing/CN ([email protected]) Purpose or Learning Objective: To investigate the capability of intravoxel incoherent motion (IVIM), diffusion kurtosis imagin g (DKI)and diffusion tensor imaging (DTI) to assess the renal function changes of type 2 diabetes.

Methods

or Background: Prospectively included 46 patients diagnosed with T2MD, and divided them into three groups based on e stimated glomerular filtration rate (eGFR) and the presence of diabetic nephropathy: the simple diabetes group (DM), the early diabetic nephropathy group (e-DKD), and the middle-to-late diabetic nephropathy group (m- DKD). At the same time, 33 volunteers（control group,CG） were recruited and underwent MRI examinations to collect images from various sequenc es. The true diffusion coefficient D, pseudo-diffusion coefficient D*, per fusion fraction f, mean diffusion kurtosis MK, mean diffusivity MD, fractio nal anisotropy FA, and apparent diffusion coefficient ADC of the renal cor tex and medulla of the subjects were measured and statistically analyzed.

Results

or Findings: In the e-DKD and m-DKD groups, ACR increased while eGFR decreased. The ACR of the e-DKD group signific antly differed from the DM group and CG (P<0.05). IVIM values, DKI (cortica l and medullary MD), and DTI (cortical and medullary ADC) values from all fo ur groups showed a declining trend with disease progression, while cor tical and medullary MK values from DKI showed an increasing trend. The cor tical MK value of the DM group significantly differed from the CG (P<0.05). Medullary MK value effectively distinguished the CG and e-DKD groups ( AUC=0.881, cutoff=0.593, sensitivity=76.9%, specificity=95.0%).

Conclusion

IVIM, DKI, DTI Sequences can be used to supplement renal dysfunction assessment. Different magnetic resonanc e parameters (IVIM, DKI, and DTI Sequences) identify different renal impairm ent changes in type 2 diabetic patients. Cortical MK value has higher dia gnostic efficiency in the early detection of renal damage in DM patients.

Limitations

The sample size needs to be increased.Further multi center studies are needed to supplement . Funding for this study: The "SKY Imaging Research Fund" by the China International Medical Exchange Foundation. Ethics committee - additional information: Chinese Ethics Review Number:2020143 Author Disclosures: Wei Ju Shao: Nothing to disclose Meng Wei Yang: Nothing to disclose Population-scale MRI body composition analysis: ass ociations between single-slice and volumetric measurements of muscle and adipose tissue *M. Nowak*¹, L. M. Nunez¹, C. Hill¹, S. Marriage¹, R. Salvati¹, M. Pansini², H. B. Thomaides-Brears¹, M. Robson¹; ¹Oxford/UK, ²L ugano/CH Purpose or Learning Objective: Accurate body composition tools are important for assessing adipose and muscle tissue i n both clinical and research settings, including obesity management, sa rcopenia, and weight loss trials. The use of whole-body MRI is limited by cos t and processing demands, while single-slice MRI offers a more efficient alte rnative. This study evaluated the correlations between single-slice and volumetri c assessments of visceral Abstract-based Programme 50 Wednesday adipose tissue (VAT), subcutaneous adipose tissue ( SAT), and muscle tissue, and their associations with cardiometabolic risk fa ctors.

Methods

or Background: We analyzed data from a subset of 67,509 individuals from the UK Biobank with water and fat MRI scans (mean age: 66, 51% male, BMI 26.7). A single axial slice at the L3 vertebra was used to measure VAT, SAT, and skeletal muscle via semi-auto matic segmentations. These were compared with volumetric assessments of VAT, SAT, total lean tissue, and thigh fat-free muscle. Correlation coef ficients were used to assess the relationship between the two methods, and their associations with cardiometabolic risk factors.

Results

or Findings: Single-slice L3 measurements of SAT and VAT demonstrated very strong correlations with SAT (rho =0.94, p<0.001) and VAT volume (rho=0.97, p<0.001), independent of sex, age , BMI, waist circumference, diabetes status, and liver tissue ch aracteristics (SAT: median rho 0.93, VAT: median rho 0.96, all p<0.001). Both body composition assessments showed similar correlations with cardio metabolic risk factors (all p<0.01 for HbA1c, triglycerides, high-density lipop rotein, systolic blood pressure, liver cT1, and liver fat content). Strong correlations were also observed between single-slice skeletal muscle and b oth total lean tissue (r=0.90, p<0.001) and thigh fat-free muscle volume (r=0.91, p<0.001).

Conclusion

Single-slice L3 measurements of VAT, SAT, and muscl e CSA metrics show robust correlations with volumetric as sessments across individuals with diverse cardiometabolic profiles, while exhibiting comparable associations with cardiometabolic risk factors.

Limitations

N/A Funding for this study: Perspectum Ltd. Ethics committee - additional information: In UK Biobank, ethical approval for data collection was received from the North-Wes t Multi-centre Research Ethics Committee and the research was carried out i n accordance with the Declaration of Helsinki of the World Medical Associ ation. Author Disclosures: Luis Miguel Nunez: Nothing to disclose Roberto Salvati: Nothing to disclose Charles Hill: Nothing to disclose Michele Pansini: Nothing to disclose Magdalena Nowak: Nothing to disclose Helena B Thomaides-Brears: Nothing to disclose Matthew Robson: Nothing to disclose Scott Marriage: Nothing to disclose 13:00-14:30 Room G1 Research Presentation Session: Radiographers RPS 414 AI-driven evolution: enhancing image quality, workflow, and professional identity for radiographers Moderators F. Doo; Baltimore, MD/US ([email protected]) S. McFadden; Newtownabbey/UK Author Disclosures: Florence Doo: Equipment Support Recipients: Cloud c redits from Amazon AWS, Microsoft Azure, Google Cloud; Grant Recipient : Funded in part by Association for Academic Radiology (AAR) Clinical E ffectiveness in Radiology Research Academic Fund (CERRAF) in part by GE Healt hcare; Research Grant/Support: Funded in part by the Johns Hopkins Mid-Atlantic Center for Cardiometabolic Health Equity (MACCHE), which is su pported by National Institutes of Minority Health and Health Disparitie s (P50MD017438); Speaker: Honoraria from Eli Lilly R-AI-diographers: a European survey to explore the perceived impact of AI on professional identity, careers, and roles of radiographers N. Stogiannos¹, *G. Walsh*¹, B. K. Ohene-Botwe¹, K. Mchugh², B. Potts¹, J. St John-Matthews¹, M. F. Mcentee³, Y. Kyratsis ⁴, C. Malamateniou¹; ¹London/UK, ²Portsmouth/UK, ³Cork/IE, ⁴Rotterdam/NL ([email protected]) Purpose or Learning Objective: Artificial intelligence is changing radiographer clinical practice and roles. It is the refore vital to understand its impact on the careers, roles and professional ident ity of these professionals.

Methods

or Background: A European-wide, EFRS-endorsed, cross-sectional, mixed methods online survey was designed on qualtri cs. Snowball sampling was used to improve uptake. Survey questions explor ed radiographer perceptions for the short-term and long-term impact of AI implementation on their roles, responsibilities and professional iden tity. The study was translated in 8 languages. Responses were compared between dif ferent demographic groups including gender, age, education and country digital literacy level.

Results

or Findings: 2206 valid responses were received from 37 differen t countries in Europe. 50.4% reported no AI education , and 26.6% were self- taught in AI. Over half (51.1%) thought patient-cen tered care skills will remain the same. 50.9% agreed radiographers will have more time to spend with patients thanks to AI. 57.8% agreed radiographers w ill have to work closer with other MIRT professionals in the future, for efficie nt AI implementation. Men appeared slightly more enthused about the developme nt of technological skills and women about the honing of patient centered care skills, similar to previous studies. Radiographers were overall optimistic abou t the use of AI in healthcare, and optimism was higher in those countr ies with high digital literacy, better education levels and with more AI experience.

Conclusion

Radiographers were overall optimistic about the use of AI in healthcare and strongly believed that AI will advan ce patient-centred care. AI education currently lags for European radiographers , and this should be acutely addressed at the scale and pace required to keep up with current technological developments. Interprofessional colla boration was seen as essential for fostering mutual support among profes sionals.

Limitations

Snowball sampling can lead to selection-bias, but a llows for many recruits. Funding for this study: Funded by the College of Radiographers Industry Partnership Scheme (CoRIPS) [grant number: 2018]. Ethics committee - additional information: Ethics approval was obtained from City St George’s, University of London School of Health and Psychological Sciences Ethics Committee (ETH2223-13 46). Author Disclosures: Gemma Walsh: Nothing to disclose Mark F. Mcentee: Nothing to disclose Yiannis Kyratsis: Nothing to disclose Benard Kwadwo Ohene-Botwe: Nothing to disclose Nikolaos Stogiannos: Nothing to disclose Janice St John-Matthews: Nothing to disclose Ben Potts: Nothing to disclose Christina Malamateniou: Nothing to disclose Kevin Mchugh: Nothing to disclose An investigation into radiographers' perception of quality control auditing of radiographic practice and the potential role of Artificial Intelligence *L. A. Rainford*, M. Mujaydia Alotaibi, J. Mcnulty, J. Potočnik; Dublin/IE ([email protected]) Purpose or Learning Objective: Quality assurance (QA) of radiographic technique is an essential part of radiation protect ion, traditionally performed through Reject Analysis. Digital imaging has increa sed the difficulty in completing radiographic technique auditing and staf f shortages further compromise QA monitoring. This research aimed to se ek radiography opinion on the use of Artificial Intelligence (AI) for QA.

Methods

or Background: An online survey was developed (n=30 questions) to seek information related to QA monitoring of rad iographic technique. Participant demographics, including area of employm ent and country of work, and professional and AI experience, were captured. Current QA auditing details were requested and participant confidence i n these processes. Their opinion was requested on the potential challenges a nd benefits of AI use in QA monitoring. The survey was distributed to affiliate EFRS academic institutions to distribute to their clinical training sites and via Radiography social media.

Results

or Findings: Good representation across all radiography professi onal grades was received from 125 participants (n=22 cou ntries). 19.8% reported QA of radiographic images on at least a weekly basi s, 18.8% stated monthly, whilst 60% reported it occurred far less frequently . 20% of responses stated staff were not individually reviewed. Only 26.8% we re very confident in current QA processes, 48% were somewhat confident and the r emainder not confident or unsure. 80% of participants indicated they perce ived AI as having a role in QA, less than 10% demonstrated concern. Improved qu ality standards and skills were perceived as benefits however considera tion of difficult patients was an identified challenge.

Conclusion

Poor confidence in current QA processes was identif ied and a lack of standardisation of practice. Radiographers identified AI as having the potential to support radiographic technique audits. Benefits and challenges were identified in open comments.

Limitations

Online survey: English language could have limited uptake Funding for this study: Self funded Ethics committee - additional information: University College Dublin, Human Research Ethics Committee – Sciences (HREC-LS ) - LS-LR-24-141- Alotaibi-Rainford. Abstract-based Programme 51 Wednesday Author Disclosures: Meshal Mujaydia Alotaibi: Nothing to disclose Louise A. Rainford: Nothing to disclose Jaka Potočnik: Nothing to disclose Jonathan Mcnulty: Nothing to disclose A comparative study assessing the effectiveness of artificial intelligence and simulation education on reporting radiographer lung cancer detection E. Compton¹, S. Lightfoot¹, R. Shah², S. Ather², P. Taylor¹, *N. H. Woznitza*¹; ¹London/UK, ²Oxford/UK ([email protected]) Purpose or Learning Objective: Chest radiographs (CXRs) are a high- volume test, performed for a broad spectrum of reas ons. Education has been shown to improve CXR reporting accuracy, in particu lar for less experienced reporters. Similarly, artificial intelligence (AI) as a clinical decision support tool provides novice readers with the most benefit. The aim of this study was to compare the impact of education (SIM) with AI in CX R reporting accuracy.

Methods

or Background: A multi-reader, multi-case diagnostic accuracy study was conducted to determine the impact of SIM and AI on reporting radiographer (RR) CXR accuracy. 64 RR consented and completed bank 1 and were randomised stratified by years’ experience (n=32,50% to AI). 43 RRs (24 AI, 19 education) completed both image banks (n =52 CXRs, 26 abnormal).

Results

or Findings: Similar pre and post intervention accuracy was foun d. The AI cohort decreased sensitivity (74%-65%,p=0.01 5) but increased specificity (63%-77%,p<0.0001), the increase SIM se nsitivity (69%- 62%,p=0.115) and decrease in specificity (62%-68%,p =0.217) were not statistically significant. Standalone AI sensitivit y and specificity were 54% and 77% respectively. For the AI arm, when the AI was c orrect specificity improved (67%-86%,p<0.001) with no significant difference in sensitivity (89%- 86%,p=0.31), however when AI was incorrect there wa s a significant decrease in sensitivity (52%-32%,p<0.001) with no difference in specificity (both 57%,p=1). There were four CXRs that only had one (n =3) or three (n=1) pre- intervention correct decisions, suggesting the bank selected comprised of very challenging cases.

Conclusion

In a challenging CXR bank, both education and AI im proved RR performance. As AI tools are adopted for CXR interp retation in clinical practice further work is required to ensure reporters are ed ucation in their use.

Limitations

The enhanced prevalence (50% abnormal) and single p athology (lung cancer) may limit transferability into clinic al practice. Funding for this study: This study was conducted as part of a clinical fellowship supported by NHS England (London). Ethics committee - additional information: Canterbury Christ Church University ETH2223-0246 21st April 2023 Author Disclosures: Paul Taylor: Nothing to disclose Sam Lightfoot: Nothing to disclose Emma Compton: Nothing to disclose Ruchir Shah: Nothing to disclose Sarim Ather: Founder: RAIQC Nicholas Hans Woznitza: Consultant: InHealth Grant Recipient: SBRI Healthcare Employee: NHS London Consultant: SMR Hea lth & Tech Radiographers’ and students’ perspectives on artifi cial intelligence - A cross-sectional online survey *M. R. V. Pedersen*¹, M. W. Kusk², S. Lysdahlgaard² , H. Mork-Knudsen³, C. Malamateniou⁴, J. Jensen⁵; ¹Vejle/DK, ²Esbjerg/DK, ³Bergen/NO, ⁴London/UK, ⁵Odense/DK ([email protected]) Purpose or Learning Objective: The integration of artificial intelligence (AI) into radiography offers potential in enhancing work flow efficiency, image processing, patient positioning, and quality assura nce.. This study aimed to investigate the perspectives and attitudes towards AI in radiography.

Methods

or Background: An online survey including of 29 items was distributed via social media platforms to Nordic st udents and radiographers working in Denmark, Norway, Sweden, Iceland, Greenl and, and the Faroe Islands. The survey included questions on demograph ics, specialization, educational background, place of work, and perspect ives and knowledge on AI. The items were a mix of closed-type and scaled questions, with options for free-text responses when relevant.

Results

or Findings: The survey received 586 responses from all Nordic countries. The mean age was 37.2 years with a stand ard deviation (SD) of ±12.1 years,. A total of 43% (n = 254) of the respondents had not received any AI training in clinical practice, while 13% (n = 76 ) had received AI training during their radiography undergraduate studies. Add itionally, 77.9% (n = 412) expressed interest in pursuing AI education. The ma jority of respondents (82.8%, n = 485) were aware of the potential use of AI, and 39.1% (n = 204) had no reservations about AI.

Conclusion

Overall, radiographers have a positive attitude tow ards AI. However, there has been very limited training or ed ucation provided to radiographers, despite 82.8% reporting plans to imp lement AI in clinical practice. Generally, awareness of AI applications i s high

Limitations

Limitations include language barriers as this surve y was provided in English. Most Nordic radiographers speak, read, and write English very well. Yet, when it come to complex sentences in English t here is a higher risk of skipping items, survey drop out, language misunders tanding or misinterpretation. Funding for this study: No funding Ethics committee - additional information: The study was approved by the Research Ethics Committee at the University of Sout hern Denmark (ID: 22- 58485) Author Disclosures: Helene Mork-Knudsen: Nothing to disclose Martin Weber Kusk: Nothing to disclose Malene Roland Vils Pedersen: Nothing to disclose Janni Jensen: Nothing to disclose Simon Lysdahlgaard: Nothing to disclose Christina Malamateniou: Nothing to disclose An analysis of the user interface preferences of im aging professionals for AI to support clinically relevant decision making *A. Gill*, S. L. Mcfadden, C. Rainey, L. Mclaughlin , J. Mcconnell, C. Hughes, R. Bond; Belfast/UK ([email protected]) Purpose or Learning Objective: This study investigates the cognitive behaviour of imaging staff when interacting with Ex planation User Interfaces (EUI). Data was gathered on user preferences of che st radiograph Artificial Intelligence (AI)-based EUIs.

Methods

or Background: Human and machine interaction involves the EUI that clinicians use to link medical diagnosis or re port. However, there is currently a lack of EUI standardisation within this field (Schalekamp et al, 2022). Building on an international questionnaire u ndertaken at ECR 2024, a multi-methods study was undertaken incorporating ey e-tracking, Think-Aloud and a questionnaire at UKIO 2024. Diagnostic radiog raphers’, radiologists’, trainee radiologists’ and student radiographers’ id entified visual preferences when reviewing four different types of chest radiog raph AI EUIs i.e. 1) salience maps, 2) textual reports, 3) area of interest and 4 )abnormality score EUIs. Participants reviewed the images whilst wearing eye -tracking software and voiced their thought processes i.e. the “Think-Alou d” method. The post study questionnaire asked the participants about their pe rceived level of confidence against the four different interfaces.

Results

or Findings: 24 participants enabled understanding of which components of the chest radiograph EUI are focused on and subsequently preferred. Eye-tracking data relating to fixations and saccades statistically described patterns where maximal attention was dire cted in the interpretation process. Think-Aloud and post-study questionnaire d ata added further insights to participant EUI preferences. The analysis of the eye-tracking study remains ongoing, and completion is aimed for January 2025.

Conclusion

Understanding user preference for chest radiograph AI EUI supports appropriate user engagement with the infor mation provided by the technology. This gives radiographers and radiologis ts the ability to explain this further to patients, positively impacting their und erstanding and subsequent care.

Limitations

Small sample size may have affected the wider gener alisability of findings. Eye-tracking software capabilities Funding for this study: PhD funded by Department for the Economy Ethics committee - additional information: FCNUR-23-084 reference Author Disclosures: Jonathan Mcconnell: Nothing to disclose Laura Mclaughlin: Nothing to disclose Clare Rainey: Nothing to disclose Ciara Hughes: Nothing to disclose Raymond Bond: Nothing to disclose Sonyia Lorraine Mcfadden: Nothing to disclose Avneet Gill: Nothing to disclose Systematic review on advanced image post-processing and workflow optimization in cardiovascular MRI *V. Tambè*, M. Zanardo, C. Torrito, P. Della Cagnol etta, F. Secchi; Milan/IT ([email protected]) Purpose or Learning Objective: Cardiovascular magnetic resonance imaging (CMR) is a critical tool for diagnosing heart disea se, but is hindered by long acquisition times and manual post-processing. This systematic review examines recent advancements in image post-processi ng and workflow optimization, focusing on the integration of artifi cial intelligence (AI). Abstract-based Programme 52 Wednesday

Methods

or Background: A systematic search was conducted using PubMed and EMBASE. Included studies involved the use of AI -based methods to optimize CMR workflows and post-processing. Eligibl e articles were those addressing any of the following: image reconstructi on, segmentation, workflow automation, and clinical integration of AI tools. S tudies without quantitative outcomes related to workflow efficiency or post-pro cessing improvements were excluded.

Results

or Findings: Out of 151 articles screened, 33 studies were inclu ded. Key findings included: automated segmentation repor ted in 15/33 (45%) studies; image reconstruction in 10/33 (30%); workf low automation 8/33 (24%); clinical efficiency in 7/33 (21%); quality control in 5/33 (15%). In the automated segmentation articles, results showed improvements in segmentation speed and accuracy with Dice similarity coefficients exce eding 0.90 in many studies, and reducing manual post-processing time by up to 6 6%. Other studies focused on reducing scan times and enhancing image quality, with AI-based

Methods

reducing scan times by up to 40% while main taining image quality. Articles showed significant improvements in reporti ng times (by up to 30%), while 5 articles presented data on AI-based quality control reducing rescans (by up to 20%).

Conclusion

AI integration into CMR has significantly improved workflow efficiency, reducing acquisition times and enhancin g diagnostic accuracy. Automated segmentation, image reconstruction, and w orkflow automation have accelerated processes, reduced operator dependency, and minimized rescans.

Limitations

Further large-scale validation is needed to fully i mplement AI in CMR across diverse clinical settings. Funding for this study: None Ethics committee - additional information: Not Applicable Author Disclosures: Francesco Secchi: Nothing to disclose Moreno Zanardo: Nothing to disclose Valentina Tambè: Nothing to disclose Paolo Della Cagnoletta: Nothing to disclose Carmelo Torrito: Nothing to disclose Blended intensive program for innovative technologi es and deep learning models (AI) in the radiographer's working environme nt *C. Schneckenleitner*, C. Kamp, C. Vogl, A. Raith, G. Guevara; Vienna/AT ([email protected]) Purpose or Learning Objective: The working environment of radiographers is characterized by permanent technological innovation s. We developed an international intensive program containing theory a nd hands-on training to introduce bachelor and masters’ students to future technologies. This program is designed to prepare students for innovative tech nologies and expand their professional skills in areas such as deep learning models, Computer Assisted Surgery Simulation, 3D Printing, Optical Scanning a nd visualization utilizing mixed reality.

Methods

or Background: During a supervised online phase, students learned how to create patient-specific 3D-printed models, C T-data segmentation for deep learning training with 3D-Slicer, computer-ass isted surgical planning (CAS), how to acquire optical 3D-scans and mixed re ality visualization of the respective 3D-models they created. All this content was worked on by the students in a follow-up international skills lab we ek at the University of Applied Sciences Vienna. Each skills lab block included 6h hours of hands-on training. The students generated deep learning models with th e platform MONAI, created surgical plans with Materialise Mimics, sca nned with optical 3D- scanners from ARTEC and created mixed reality visua lizations.

Results

or Findings: The results were uploaded to an online platform (Moodle) by the students and analyzed based on defi ned criteria. Results, 39 out of 42 students were able to create a segmentati on for deep learning training according to the required criteria. 39 out of 42 successfully performed a CAS plan, 20 created a 3D-printable patient speci fic model and 42 of 42 were able to produce an optical 3D-scan of the face .

Conclusion

The results show that radiography students can prod uce results in adjacent technology areas and expand their compe tence in future technologies.

Limitations

The limited time can generate interest but not a sp ecialization in the profession. Explicit training programs are need ed to deepen radiographers’ expertise in these areas. Funding for this study: No funding. Ethics committee - additional information: The results includes no patient specific information or clinical interventions. Author Disclosures: Christoph Kamp: Nothing to disclose Christian Schneckenleitner: Nothing to disclose Alexander Raith: Nothing to disclose Godoberto Guevara: Nothing to disclose Christoph Vogl: Nothing to disclose Evaluation of ChatGPT as support in image qualitati ve assessment for cardiac sonographers *K. Tissir*¹, G. R. Bonfitto¹, A. Roletto¹, A. Sign oroni²; ¹Milan/IT, ²Brescia/IT ([email protected]) Purpose or Learning Objective: The integration of Large Language Models (LLM) tools like ChatGPT in clinical settings is ch anging how healthcare professionals manage diagnoses and workflow. In car diac clinics, the accurate and timely interpretation of images is crucial for effective diagnosis and monitoring of cardiac conditions. ChatGPT could be particularly beneficial for cardiac sonographers. This study aims to assess whe ther ChatGPT can effectively support cardiac sonographers in the qua litative evaluation of echocardiographic images.

Methods

or Background: A database of 50 anonymized echocardiographic images was retrospectively analyzed, including 2-ch amber, 4-chamber, and apical 3-chamber views. Three evaluators, a junior sonographer, a senior sonographer and ChatGPT-4o conducted the qualitativ e evaluation of the images by identifying scoring them on a 5-point Lik ert scale. The guidelines of the European Association of Echocardiography served as references.

Results

or Findings: Junior sonographer correctly identified views in 84 % of cases (n=42), while ChatGPT correctly identified 58 % of cases (n=29).In comparison to senior sonographer, the junior sonogr apher overestimated 22% of the images (n=11), underestimated 36% (n=18), an d agreed in 42% of the images (n=21). In contrast, ChatGPT overestimated 5 2% of the images (n=26), underestimated 18% (n=9), and agreed in 30% of the images (n=15).

Conclusion

ChatGPT-4o shows limitations in identifying echo ca rdiac views compared to other participants. In addition, ChatGP T is inclined to overestimate image quality. This can be explained b y limited training of the LLM, mainly done with information from guidelines. As other studies in literature showed, more in-depth training could inc rease the performance of ChatGPT. LLM can assist cardiac sonographers in qua litative analysis of images and supporting anomaly evaluation, but conce rns remain about its reliability and bias.

Limitations

The small sample of participants and cases limited the strength of the conclusions of this study. Funding for this study: N/A Ethics committee - additional information: University of Brescia Author Disclosures: Andrea Roletto: Nothing to disclose Karima Tissir: Nothing to disclose Alberto Signoroni: Nothing to disclose Giuseppe Roberto Bonfitto: Nothing to disclose Navigating Artificial Intelligence (AI) Leadership: Radiographers’ Readiness and Challenges in Europe *G. Walsh*¹, Y. Kyratsis², A. Goodall¹, J. St John- Matthews¹, C. Malamateniou¹; ¹London/UK, ²Rotterdam/NL ([email protected]) Purpose or Learning Objective: This study offers unique insights into the preparedness of radiographers to pursue AI leadersh ip roles within healthcare and potential barriers preventing radiographers exc elling in the AI ecosystem.

Methods

or Background: A European-wide, cross-sectional study utilising a mixed methods online survey. Snowball sampling allo wed qualified radiographers, and radiography students, to answer the survey, irrespective of their current role. The survey explored the followi ng areas of interest: a) general radiographer demographics, b) radiographers preparedness and confidence to lead the implementation of AI in heal thcare, c) suggested day-to- day responsibilities of an AI-lead radiographer and d) motivations for considering AI leadership roles.

Results

or Findings: There were 1733 valid responses from 37 European countries. The typical respondent was female (64%), a diagnostic radiographer (59.9%) with >20 years’ experience (31.3%). A lack of education, training and resources led 72.3% of radiographers to feel the pr ofession is not prepared to lead the implementation of AI in healthcare. Those that felt confident to lead in an AI-enabled work environment (50.06%), felt they already have the necessary experience and skills, but also admitted additional resources would be needed. The top two motivators to pursue an AI l eadership role included championing change and the promise of appropriate t raining. Inferential Statistics ongoing as of October 2024.

Conclusion

Radiographers have a unique skill-set making them t he ideal candidates for AI leadership roles within healthcar e. Radiographers do not currently feel confident or prepared to undertake A I leadership roles with education, training and a lack of resources creatin g barriers for this. It is reassuring radiographers feel motivated to undertak e AI leadership roles, however increased training and educational support are needed.

Limitations

Gives a snapshot view of radiographers perceptions. Snowball sampling can lead to selection-bias, but a llows for many recruits. Funding for this study: This research has been funded by the AI special cal l of the College of Radiographers Industry Partnershi p Scheme (CORIPS) of the College of Radiographers (Reference Number 218). Abstract-based Programme 53 Wednesday Ethics committee - additional information: Ethics approval (ETH2223-1346) was granted by City, University of London. Author Disclosures: Gemma Walsh: Nothing to disclose Yiannis Kyratsis Yiannis Kyratsis: Nothing to discl ose Janice St John-Matthews: Nothing to disclose Amanda Goodall: Nothing to disclose Christina Malamateniou: Nothing to disclose MRI deep learning models for assisted diagnosis of knee pathologies and injuries: A systematic review *K. M. Mead*, T. Cross, G. Roger, R. Sabharwal, S. Singh, N. Giannotti; Sydney/AU ([email protected]) Purpose or Learning Objective: Several studies have demonstrated that deep learning (DL) models can be effectively traine d on MRI data to assist clinicians in identifying knee injuries and patholo gies. This systematic review was conducted to explore the current landscape of e xisting DL models developed for detecting knee injuries and pathologi es through magnetic resonance imaging (MRI) and to assess their potenti al clinical applications.

Methods

or Background: Five databases were systematically searched using the following terms ‘Knee AND 3D AND MRI AND Deep L earning’. The Covidence platform was used to screen publications based on title, abstract, and full text. Only original research articles focussing on the automatic detection of knee injuries and pathologies using DL models for MRI were included. The synthesis of results was performed by two independent reviewers.

Results

or Findings: Fifty-four studies were included. The studies focus ed on anterior cruciate ligament injuries (n=19), osteoar thritis (n=9), meniscal injuries (n=13), general abnormal knee appearance (n=10), ti bial plateau fractures (n=1) and synovial fluid detection (n=1). The follo wing convolutional neural network (CNN) infrastructures were used: ResNet, VG G, DenseNet, and DarkNet. The averaged performance outcomes of the D L models demonstrated sensitivity, specificity, AUC-ROC, and accuracy of 87%, 90%, 92%, and 88%, respectively. The DL models for the d etection of a specific injury or pathology outperformed those for general abnormality detection.

Conclusion

This systematic review underscores that fine-tuned DL models for knee pathologies and injuries can effectively suppo rt automatic diagnosis. Further large-scale validation and prospective stud ies are needed to confirm their clinical utility as assistive diagnostic tool s.

Limitations

Inconsistent data reporting across the studies anal ysed resulted in variations in the reporting of DL model performa nce. Sub-group analyses were performed to minimise bias. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is retrospective. Ethics approval was deemed unnecessary by the Research Int egrity and Ethics Committee at the University of Sydney, Australia. Author Disclosures: Keiley Michelle Mead: Nothing to disclose Nicola Giannotti: Nothing to disclose Sahaj Singh: Nothing to disclose Tom Cross: Nothing to disclose Rohan Sabharwal: Nothing to disclose Greg Roger: Nothing to disclose An Innovative AI-Based Interactive Tool for Learnin g Chest X-Ray Anatomy *R. S. T. Ribeiro*, T. Coutaudier, L. Mourot, C. S. D. Reis, L. Raileanu; Lausanne/CH ([email protected]) Purpose or Learning Objective: To enhance the learning process of chest X- ray/(CXR) anatomy for medical imaging/(MI) students by integrating AI segmentation and classification tools into an educa tional web- application/(webapp).

Methods

or Background: The webapp was designed as an interactive platform where students can identify CXR anatomy. T he platform was implemented in Python using Flask to provide the we b-interface, TorchXRayVision to classify and segment key regions (heart, lungs, clavicles, spine, scapula, trachea) with AI. PostgreSQL stored and managedCXR public datasets allowing students’ practice. The user inte rface allows selection and outline regions of interest on the radiographs, tha t are compared to segmentations obtained with AI-algorithms. Feedback is provided through Dice coefficient that assess the accuracy of the user’s segmentation compared to the AI-based reference. The app’s system architectu re is modular.

Results

or Findings: The developed framework successfully integrates AI for fast and accurate segmentation of CXR. Its design a llows users to upload their own radiographs or use others supplied by public da tasets, interact with radiographs by selecting anatomical regions and rec eiving immediate feedback. This functionality aims to support autono mous learning and reduce the need for constant instructor supervision. The m odular architecture ensures scalability, enabling the inclusion of more types o f radiographic images and enhancing its potential for broader applications in medical imaging education.

Conclusion

The web application demonstrates a promising approa ch to improve MI education by providing an interactive an d AI-powered learning tool. Its design is intended to be adaptable/accessible t hrough any web-browser, with potential to expand into areas such as quality assessment. Further development of this framework is planned to test it s impact on the MI students learning process.

Limitations

The app is limited to CXR and relies on AI segmenta tion performance. Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Claudia Sa Dos Reis: Nothing to disclose Ricardo Silva Teresa Ribeiro: Nothing to disclose Lucas Mourot: Nothing to disclose Théo Coutaudier: Nothing to disclose Laura Raileanu: Nothing to disclose 15:00-16:00 Research Stage 1 Research Presentation Session: Breast RPS 502 MRI and contrast-enhanced mammography for treatment planning Moderator M. Dorrius; Groningen/NL ([email protected]) Author Disclosures: Monique Dorrius: Research Grant/Support: KWF PPS Gr ant Radiomic Shape Features for Assessment of Early The rapy Response to Neoadjuvant Chemotherapy of Breast Cancer Patients - Preliminary

Results

of the PREDICTOME-Study *D. A. Resch*, O. Lafcı, P. Clauser, Z. Bago-Horvat h, Y. Tan, G. Langs, T. Helbich; Vienna/AT ([email protected]) Purpose or Learning Objective: To analyze the significance of radiomic shape features derived from dynamic contrast-enhanc ed (DCE) breast MRI in assessing early treatment response in breast cancer patients undergoing neoadjuvant chemotherapy (NAC).

Methods

or Background: We report on the first 29 breast cancer patients of a prospective study, who underwent NAC and received m ultiparametric 18F-FDG PET/MRI imaging at baseline (T0) and after three we eks of NAC (T1). DCE- MRI derived radiomic shape features, including SHAP E_Volume (mL), SHAPE_Volume (vx), SHAPE_Sphericity, SHAPE_Surface( mm²) and SHAPE_Compacity were extracted using LIFEx. The agn ostic shape features were compared to the treatment response assessment by two radiologists. All data were stratified by the pathological complete r esponse status (pCR or non- pCR). Mean change (Δ) of all investigated features were calculated. Pea rson Chi-Square Test, T-Test and Mann-Whitney U Test wer e applied.

Results

or Findings: Twelve out of 29 (41%) patients had pCR and 17/29 (59%) had non-pCR. Radiomic shape features, includi ng ΔSHAPE_Compacity and SHAPE_Volume at T1, were significantly associat ed with pCR (P = 0.015 and P = 0.04, respectively). Radiologist's response assessment (stable disease, partial response, disease progression) was not significantly associated with pathological outcomes (pCR vs. non- pCR) (Pearson Chi- Square: χ2=3.727χ2=3.727, P = 0.155). Similarly, the likelihood rati o test did not show significant results (P = 0.105).

Conclusion

While the radiologist’s assessment did not signific antly correlate with pathological outcomes, radiomic shape features , particularly ΔSHAPE_Compacity and SHAPE_Volume, demonstrated sign ificant associtations with pCR. These findings suggest that DCE derived radiomic shape features might be a promising tool for predic tion of early NAC response in breast cancer patients and may challenge the val idity of RECIST criteria at this stage.

Limitations

This study is limited by the small sample size. We will be able to report on more data in March 2025. Funding for this study: This study is supported by the WWTF (Vienna Science and Technology Fund), grant number LS20-065 . Abstract-based Programme 54 Wednesday Ethics committee - additional information: EK Nr. 510/2009 Author Disclosures: Georg Langs: Nothing to disclose Daphne Ariane Resch: Nothing to disclose Thomas H. Helbich: Nothing to disclose Oğuz Lafcı: Nothing to disclose Zsuzsanna Bago-Horvath: Nothing to disclose Yen Tan: Nothing to disclose Paola Clauser: Nothing to disclose Reactive Cutaneous Capillary Endothelial Proliferat ion in Breast Tissue on MRI during Neoadjuvant Chemoimmunotherapy with C amrelizumab in TNBC *X. Ma*, Q. Xiao, Y. Huang, Y. Gu; Shanghai/CN ([email protected]) Purpose or Learning Objective: To describe the Reactive cutaneous capillary endothelial proliferation (RCCEP) that occurs withi n the breast tissue of triple- negative breast cancer (TNBC) patients undergoing n eoadjuvant chemoimmunotherapy with camrelizumab and to investi gate the potential factors influencing its occurrence.

Methods

or Background: We retrospectively collected 106 cases of TNBC patients from March 2021 to August 2023, including 60 cases who received neoadjuvant chemotherapy (NAC) and 46 cases who und erwent neoadjuvant chemoimmunotherapy (NACI). We analyzed the clinical data, pathological characteristics, MRI at baseline and during each tr eatment cycle for all patients, then identified abnormal lesions after tr eatment and explored their influencing factors.

Results

or Findings: Abnormal enhancement was observed in 30 patients (3 NAC vs. 27 NACI, P<0.001). In the NACI group, cases of abnormal enhancement included 3 cases of ring enhancement, 2 cases of linear enhancement, and 22 cases of mass enhancement, whic h typically appeared after the second cycle and rarely appeared after th e fourth or sixth cycle. These lesions generally decreased in size with cont inued treatment. In the NACI group, the development of new enhancement lesi ons was correlated with younger age (P=0.007), premenopausal status (P=0.01 4), a lack of peritumoral edema on baseline MRI (P=0.007), and the presence o f mass enhancements (P=0.012).

Conclusion

TNBC patients treated with camrelizumab frequently exhibit RCCEP in the breast tissue, primarily presenting as mass enhancements on MRI. These lesions often regress in size even witho ut drug withdrawal, suggesting that recognizing this pattern can preven t unnecessary biopsies and help in adjusting treatment strategies accordingly.

Limitations

First, this study is retrospective and includes a s mall sample size; second, in future studies, it is necessary to exten d the research period to explore the relationship between RCCEP in the breas t and their prognosis. Funding for this study: None Ethics committee - additional information: Fudan University Shanghai Cancer Center Author Disclosures: Yan Huang: Nothing to disclose Yajia Gu: Nothing to disclose Qin Xiao: Nothing to disclose Xiaowen Ma: Nothing to disclose Patient Perspective: Evaluating Imaging Preferences in Women Undergoing Contrast-Enhanced Breast MRI and Contras t-Enhanced Mammography (CEM) N. Caldarelli, *G. Della Pepa*, G. Irmici, E. D'Asc oli, C. De Berardinis, D. Ballerini, A. Bonanomi, C. Depretto, G. P. Scape rrotta; Milan/IT ([email protected]) Purpose or Learning Objective: The study evaluated patient preferences between contrast-enhanced MRI (MRI) and contrast-en hanced mammography (CEM) for breast cancer imaging. MRI has long been the gold standard for breast cancer staging and monitoring neoadjuvant ch emotherapy, while CEM has gained popularity due to its lower cost, faster examination times, and accessibility, offering comparable diagnostic perfo rmance.

Methods

or Background: The aim of this study was to understand patient preferences between these two modalities to support the broader implementation of CEM in clinical practice. The stu dy included 152 patients who underwent both procedures within a six-month in terval between 2018 and 2024. A Likert scale questionnaire was used to asse ss patient preferences focusing on three main aspects: breast positioning (compression for CEM and coil positioning for MRI), sensation during contras t injection, and overall comfort (exam duration, machine noise, and environm ental factors).

Results

or Findings: Results showed that 72.4% of patients preferred CEM , 26.3% preferred MRI, and 1.3% expressed no preferen ce. CEM was significantly more comfortable (p<0.001), with high er median scores than MRI. The main reasons for preferring CEM included faster exam time (28%), lack of claustrophobia (17%), and absence of noise (15%). I n terms of breast positioning, there was a slight preference for MRI (p=0.04). No significant differences were found in the sensation during cont rast injection (p=0.07).

Conclusion

In conclusion, CEM was the preferred option for mos t patients, indicating its potential as an alternative to MRI i n clinical settings. These findings support further exploration of CEM's role in breast cancer imaging.

Limitations

none. We have no limitations. Funding for this study: none. We don't need any funding for this study. Ethics committee - additional information: none. We don't have any Ethics committee. Author Disclosures: Nazarena Caldarelli: Nothing to disclose Gianmarco Della Pepa: Nothing to disclose Elisa D'Ascoli: Nothing to disclose Alice Bonanomi: Nothing to disclose Daniela Ballerini: Nothing to disclose Gianfranco Paride Scaperrotta: Nothing to disclose Claudia De Berardinis: Nothing to disclose Catherine Depretto: Nothing to disclose Giovanni Irmici: Nothing to disclose Non-invasive imaging of the tumor pH in breast canc er with CEST-MRI: A preclinical study *D. Prinz*¹, S. J. Bartsch¹, J. Friske¹, D. Laimer- Gruber¹, T. H. Helbich¹, K. Pinker-Domenig²; ¹Vienna/AT, ²New York, NY/US ([email protected]) Purpose or Learning Objective: Tumor acidosis is a key hallmark of breast cancer (BC). The increased glucose consumption trig gers aerobic glycolysis, leading to the production of lactic acid which resu lts in therapy resistance. Currently, there is no non-invasive tool available to image tumor pH in vivo. We attempted to image the extracellular pH (pHe) with acidoCEST using Iopamidol, and the intracellular pH (pHi) using the CEST-derived AACID (amine and amide concentration-independent detectio n) metric. Non-invasive imaging of tumor pH is of great interest because pH is one of the first biomarkers which changes during treatment.

Methods

or Background: Female athymic nude mice were inoculated with BC cells of HER2+ (SKBR-3, n = 5) and triple-negative (MDA-MB-231, n = 4) molecular subtypes. MRI imaging was performed using a preclinical 9.4T MRI system. CEST images were acquired and ratiometric m easurements were evaluated for the endogenous AACID from baseline im ages and the acidoCEST signal based on post-challenge images.

Results

or Findings: Both acidoCEST and AACID produced reliable and stable signals. The lower pHe and the higher pHi co uld be visualized for HER2+ and triple-negative BC subtypes. Parametric m aps of AACID and acidoCEST revealed differences in the pH gradients between BC subtypes, which correlated with tumor aggressiveness.

Conclusion

We conclude that the AACID-based measurement of pHi sufficiently quantifies the pH gradient between ext racellular and intracellular compartments and may become a promising non-invasiv e contrast-free imaging method to monitor early treatment response in BC.

Limitations

Due to the preclinical nature of this study, the sm all number of mice used presents a limitation. Funding for this study: This work was funded by the Vienna Science and Technology Fund (WWTF), grant no. LS19-018. Ethics committee - additional information: This animal study was approved by Austrian Federal Ministry of Education, Science and Research [66.009/0284-WF/V/3b/2017; 2020-0.363.124; 2022-0.7 26.820] and the Intramural Committee for Animal Experimentation of the Medical University of Vienna. Author Disclosures: Silvester Julian Bartsch: Nothing to disclose Daniela Laimer-Gruber: Nothing to disclose Katja Pinker-Domenig: Nothing to disclose Thomas H. Helbich: Nothing to disclose Daniela Prinz: Nothing to disclose Joachim Friske: Nothing to disclose CEM Background Parenchymal Enhancement: Exploring I ts Clinical and Biological Correlations *C. De Berardinis*¹, C. Depretto¹, G. Della Pepa¹, E. D'Ascoli¹, G. Irmici¹, E. Ancona¹, R. Spiaggia², L. Corradini², G. P. Scap errotta¹; ¹Milan/IT, ²Mussomeli/IT ([email protected]) Purpose or Learning Objective: To assess the correlation between the degree of background parenchymal enhancement (BPE) on contrast-enhanced mammography (CEM) and breast density, menopausal st ate, receptor status, HER 2 expression and proliferation index (Ki-67) of malignant tumors. Abstract-based Programme 55 Wednesday

Methods

or Background: We retrospectively evaluated all patients who underwent CEM at our Institution from January 2023 to April 2024 for pre- operative staging and problem solving. BPE was clas sified as minimal, mild, moderate, or marked, in accordance with the BIRADS lexicon. We assessed the menopausal status and evaluated the receptor st atus and Ki-67 of each malignant lesion. Statistical analysis was performe d using the Spearman’s test to evaluate the correlation between density and BPE . Chi-square test was used to evaluate the correlation between BPE, menop ausal status, receptor status, and Ki-67.

Results

or Findings: A total of 194 patients were included. Spearman's t est analysis demonstrated a weak correlation between BP E and breast density (ρ=0.353, p<0.001, CI:0.219-0.469). The Chi-Square te st revealed a strong association between BPE and menopausal state, with lower BPE levels in postmenopausal patients compared to premenopausal p atients (X²=30.846, p<0.001). There was no statistically significant as sociation between BPE and receptor status (X²=14.494, p=0.270) and HER2+ stat us (X²=1.648, p=0.649). There was suggestive but not statistically signific ant association (X²=6.738, p=0.081) between BPE and Ki67.

Conclusion

A significant correlation was demonstrated between the level of BPE on CEM and menopausal status. Instead, the corr elation with breast density proved to be weak, while no statistically s ignificant correlation was found with tumor receptor status. BPE also appears to have a suggestive correlation with Ki-67% and consequently with tumor aggressiveness.

Limitations

This association needs more data and larger sample size to be confirmed but suggests how BPE might be related to breast cancer risk. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Gianmarco Della Pepa: Nothing to disclose Elisa D'Ascoli: Nothing to disclose Claudia De Berardinis: Nothing to disclose Eleonora Ancona: Nothing to disclose Gianfranco Paride Scaperrotta: Nothing to disclose Catherine Depretto: Nothing to disclose Lisa Corradini: Nothing to disclose Rossana Spiaggia: Nothing to disclose Giovanni Irmici: Nothing to disclose Role of Breast MRI to identify patients with lesion s of uncertain malignant potential (B3) who might avoid surgery: a systemati c review and metanalysis *G. Vatteroni*¹, N. Pötsch², P. Clauser², P. A. Bal tzer²; ¹Milan/IT, ²Vienna/AT ([email protected]) Purpose or Learning Objective: This systematic review and meta-analysis investigates the added value of Contrast Enhanced B reast MRI (CE-MRI) to rule out malignancy in patients with lesions of unc ertain malignant potential (B3) diagnosed at image guided biopsy.

Methods

or Background: A systematic review and meta-analysis were conducted using predefined criteria. Eligible artic les, published in English until August 2024, dealt with CE-MRI in the context of B3 lesions. Two reviewers extracted study characteristics, including true pos itives (TP), false positives (FP), true negatives (TN), and false negatives (FN) . Sensitivity, specificity, negative likelihood ratio, and positive likelihood ratio were calculated using bivariate random effects. Fagan nomograms identifie d the maximum pretest probability at which post-test probabilities of a n egative MRI aligned with the 2% malignancy rate benchmark for downgrading BI-RAD S 4 to BI-RADS 3. I² statistics and meta-regression explored sources of heterogeneity. P-values <0.05 were considered significant.

Results

or Findings: Seven studies investigating 479 patients with 493 h igh risk lesions undergoing CE-MRI were included. The a verage breast cancer prevalence was 17% (88/493). Pooled sensitivity was 91.3% (95%-CI: 82.8%- 95.8%) and pooled specificity was 68.8% (95%-CI 50. 3%-82.8%) using a random effects model. Overall, CE-MRI missed only 6 /493 lesions, all small low-grade Ductal Carcinoma in Situ. Fagan nomograms showed that CE-MRI could rule out malignancy in B3 lesions diagnosed a t image guided biopsy up to a pre-test probability of 13.1%.

Conclusion

CE-MRI in the assessment of B3 lesions could potent ially identify patients who might avoid surgery, saving costs and time as well as reducing patient anxiety and morbidity. Breast cancer can be ruled out up to pre-test probabilities of 13.1%.

Limitations

n/a Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Pascal A.T. Baltzer: Nothing to disclose Nina Pötsch: Nothing to disclose Giulia Vatteroni: Nothing to disclose Paola Clauser: Nothing to disclose Contrast-Enhanced Digital Mammography for the Diagn osis and Determination of Extent of Disease in Invasive Lobu lar Carcinoma: Our Experience and Literature Review *J. Lipin Margaret*, T. Friehmann, S. Tamir, G. Bac har, A. Grubstein; Petach Tikva/IL ([email protected]) Purpose or Learning Objective: In contrast-enhanced imaging for the precise evaluation of Invasive lobular carcinoma (ILC), the aim was to further validate the use of CEM for evaluation of extent of disease in ILC cases, with Digital breast tomosynthesis (DBT) as an adjunct.

Methods

or Background: A retrospective study was conducted in a tertiary medical center. All CEM examinations of ILC patient s between 2017–2023 were re-read by two breast radiologists. The longes t lesion diameter was correlated to pathology findings. For each case, re aders provided commentary on the added value of DBT.

Results

or Findings: Twenty-four CEM examinations were evaluated. The cohort were on average older than expected for ILC patients (74 vs. 63 years) and were unable to undergo MRI due to advanced age and comorbidities. Enhancing lesions, ranging from 4–75 mm, strongly c orrelated to pathology results, with statistical significance, for mass an d non-mass lesions (r = 0.94, p < 0.001 and r = 0.99, p = 0.002, respectively). For most lesions (17/24, 71%), readers remarked that DBT allowed for improved char acterization of lesion margins.

Conclusion

When compared to the pathology findings, ILC was ac curately diagnosed and assessed using CEM. The addition of D BT was reported by the interpreting radiologists as a valuable adjunct for margin analysis.

Limitations

Sample size was small, a general issue among studie s of ILC in CEM (e.g., 30 subjects in the study by Patel et al. in 2018, 31 in Amato et al. in 2019, and 38 in Costantini M et al. in 2022). The d esign was retrospective, and furthermore, since all patients underwent CEM due t o contraindication to MR, introducing selection bias. No patient had extremel y dense breasts or marked enhancement, conclusions regarding the accuracy of CEM in the most challenging breast types cannot be reached. Funding for this study: None. Ethics committee - additional information: Helsinki approval Author Disclosures: Tal Friehmann: Nothing to disclose Johansson Lipin Margaret: Author: Rabin Medical Cen ter, Mammography Unit Ahuva Grubstein: Nothing to disclose Gil Bachar: Nothing to disclose Shlomit Tamir: Nothing to disclose Comparison of additional malignant lesions detectio n in Dense vs. Non- dense breasts with Magnetic Resonance Imaging (MRI) or Contrast- Enhanced Mammography (CEM) performed for loco-regio nal staging *P. Minichetti*, M. Lorenzon, S. Sanità, L. Nardone , L. Cereser, R. Girometti, C. Zuiani; Udine/IT ([email protected]) Purpose or Learning Objective: To compare Additional Malignant Lesions (AML) detection of CEM or MRI (Breast Contrast-Imag ing – BCI) performed for preoperative loco-regional staging of Breast Cancer (BC) in patients with Dense Breast (DB) or Non-DB (N-DB).

Methods

or Background: We retrospectively included 290 patients (median age 62 years) with a biopsy-proven BC who underwent CEM (n=129) or 1.5T- MRI (n=161) at our Institute between January 2022 a nd December 2023. Staging was performed based on EUSOMA criteria or c linical requests. A radiology resident (with >3 years of experience in breast imaging) reviewed all reports written by one-of-four breast radiologists (with 5-25 years of experience). Extracted data included: density accor ding to BI-RADS (C, D classified as DB; A, B as N-DB), background parench ymal enhancement, features of index lesions and AML (not identified b efore BCI, pathologically proven). The BCI detection rate of AML in DB and N- DB groups was assessed and compared using T-test. P-values<0.05 were consi dered statistically significant.

Results

or Findings: In 201 DB patients, preoperative BCI detected 40 AM L (19.9%). In 89 N-DB patients, BCI detected 10 AML ( 8.9%). The difference in the BCI AML detection rate between DB and N-DB grou ps was 11% (p=0.0202). In the DB group, the AML detection rate was found to be more than double that in the N-DB group. Specifically, a mong 71 DB patients, CEM detected 15 AML (21.1%) vs 6 AML in 58 N-DB patient s (10.3%); among 130 DB patients, MRI detected 25 AML (19.2%) vs 4 AML i n 31 N-DB patients (12.9%).

Conclusion

The usefulness of BCI performed for loco-regional s taging is significantly higher in DB than in N-DB patients. T he detection rate of AML is comparable between MRI and CEM in DB and N-DB patie nts.

Limitations

Small cohort, monocentric. Funding for this study: Nothing. Ethics committee - additional information: Institutional Review Board (IRB- DMED) Abstract-based Programme 56 Wednesday Author Disclosures: Chiara Zuiani: Nothing to disclose Silvia Sanità: Nothing to disclose Michele Lorenzon: Nothing to disclose Rossano Girometti: Nothing to disclose Lorenzo Cereser: Nothing to disclose Paola Minichetti: Nothing to disclose Luigi Nardone: Nothing to disclose 15:00-16:00 Research Stage 2 Research Presentation Session: Genitourinary RPS 507 Non-malignant pathology of the female pelvis: insights and imaging approaches Moderator C. Panico; Rome/IT ([email protected]) Diaphragmatic Endometriosis: Correlation with Pelvi c Disease and Symptoms *N. Bogveradze*, A. Santonocito, J. Heine, T. Helbi ch, P. A. Baltzer, R. Wenzl, P. Clauser; Vienna/AT ([email protected]) Purpose or Learning Objective: Diaphragmatic endometriosis (DE) is rare, with limited data on its frequency and management. The deep Pelvic Endometriosis Index (dPEI) and #Enzian classificati on systems improve pelvis magnetic resonance imaging (MRI) accuracy for diagn osing endometriosis by stratifying disease severity. Our study aimed to as sess the prevalence of DE and its correlation with disease severity and assoc iated clinical symptoms.

Methods

or Background: In this IRB-approved retrospective study, consecutive abdominal MRIs performed for endometrio sis (2018-2022) were reviewed by three radiologists in consensus. Positi ve cases of diaphragmatic endometriosis were defined by consensus among reade rs or available histology. Pelvic disease severity was assessed usi ng dPEI and #Enzian scores, classifying endometriosis as mild (score ≤2), moderate (scores 3-4), or severe (scores ≥5). Descriptive statistics were used to analyze the relationship between pelvic disease severity, diaphragmatic invo lvement, and symptoms.

Results

or Findings: DE was detected in 16/108 patients (14.8%). Among 9 2 patients with clinical data, 7/92 (7.6%) showed upp er abdominal symptoms, with DE confirmed in 4/7 (57.1%). No imaging correl ates were identified for three symptomatic patients. Based on dPEI, 2/92 asy mptomatic patients had moderate disease (2.2%), while symptomatic patients , 17/92 had mild (18.5%), 36/92 moderate (39.1%), and 37/92 severe (40.2%) di sease. Among 16 DE patients by dPEI, 2/16 (12.5%) had mild, 7/16 (43.8 %) moderate, and 7/16 (43.8%) severe disease. According to #Enzian, 2/92 asymptomatic patients had severe disease (2.2%), while in symptomatic pat ients, 8/92 had mild (8.7%), 11/92 moderate (12.0%), and 71/92 severe (7 7.2%). By #Enzian, of the 16 DE patients, none had mild, 2/16 (12.5%) had mod erate, and 14/16 (87.5%) had severe disease.

Conclusion

DE detected on MRI was associated with symptoms in 25% of the cases. The presence of diaphragmatic endometrio sis is associated with more severe pelvic disease in both scoring systems.

Limitations

Retrospective Funding for this study: N/A Ethics committee - additional information: Number: 2057/2017 Author Disclosures: Pascal A.T. Baltzer: Nothing to disclose Rene Wenzl: Nothing to disclose Thomas H. Helbich: Nothing to disclose Ambra Santonocito: Nothing to disclose Nino Bogveradze: Nothing to disclose Jana Heine: Nothing to disclose Paola Clauser: Nothing to disclose Adenomyosis and Deep Infiltrating Endometriosis: Th e Role of Adenomyosis Dominant Side and Type in Implant Distr ibution A. Durur Karakaya, *H. Özen Atalay*, V. Samadli, U. Kalkan; Istanbul/TR ([email protected]) Purpose or Learning Objective: The aims of this study are to evaluate the correlation between the types/dominant side of the adenomyosis and the location of deep infiltrating endometriosis (DIE) i mplants, as well as the relationship between the type of adenomyosis and th e presence of endometrioma.

Methods

or Background: We retrospectively evaluated 311 patients with adenomyosis on MRI examinations between January 202 2 and September 2024. The poor quality MRI examination and patients without DIE were excluded from the study, and the final evaluation w as performed with 50 patients with adenomyosis and accompanying DIE. Ade nomyosis was classified into 3 based on the type as focal, diffu se, cystic; and based on the dominant side in the uterus as anterior, posterior, no dominance. DIE implants were also categorized based on the location as ante rior, posterior, or involvement of both compartments. Additionally, the presence and size of endometrioma were evaluated. Statistical significan ce was examined by chi- square tests (p < 0.05).

Results

or Findings: The median patient age was 39 years. A statisticall y significant correlation was found between adenomyos is dominant side and DIE implant location (p = 0.019). Adenomyosis with no d ominant side or diffuse adenomyosis was not significantly associated with e ffecting of both compartments by DIE implants (p = 0.275 and p = 0.1 02, respectively). The correlation between adenomyosis type and endometrio ma presence (p = 0.390) or size >40 mm (p = 0.687) was not statistic ally significant.

Conclusion

To our knowledge, this is the first study evaluatin g the correlation between the dominant side of adenomyosis and DIE im plant location. Previous studies indicated the relationship between the aden omyosis dominant side's and symptom severity, pregnancy loss. Our findings emphasize the need for further investigation into this subject.

Limitations

The limitations are the small sample size and retro spective nature. Funding for this study: Not applicable Ethics committee - additional information: Koc University Biomedical Research Ethics Committee, Istanbul/Turkey Author Disclosures: Afak Durur Karakaya: Nothing to disclose Hande Özen Atalay: Nothing to disclose Vugar Samadli: Nothing to disclose Uzeyir Kalkan: Nothing to disclose MRI for endometriosis: ESUR Consensus for indicatio ns, reporting and classifications *I. Thomassin-Naggara*¹, M. Dolciami², L. Chamie³, A. Guerra⁴, N. Bharwani⁵, S. Freeman⁶, P. Rousset⁷, L. Manganaro²; ¹Paris/FR, ²Rome/IT, ³São Paulo/BR, ⁴Lisbon/PT, ⁵Surbiton/UK, ⁶Cambridge/UK, ⁷Lyon/FR ([email protected]) Purpose or Learning Objective: The ESUR Research Committee appointed two chairs to supervise the development of the upda ted guidelines.

Methods

or Background: These guidelines are recommendations developed by the European Society of Urogenital Radiology (ES UR). A targeted literature search was performed to discover recent evidence co ncerning the imaging of endometriosis. The guidelines were formulated after careful consideration of the available literature by a group of internationa l experts. The panel included 20 experts from 10 different countries, including 1 4 European centers and one non-European institution. The methodology was based on DELPHI process. Each item was classified as follows: “RECOMMENDED” (if agreement ≥ 80%); “OPTIONAL” (if agreement ≥ 70 % but < 80 %); or “NOT RECOMMENDED” (if consensus was not reached, with < 70 % agreement). The survey was composed of three parts: 1) Indications, 2) MR prot ocol and lexicon 3) classification/reporting.

Results

or Findings: Regarding indications, MRI is recommended when TVUS is inconclusive in diagnosing endometriosis or negative, in a symptomatic patient, before surgery or intervention al procedure or after surgical treatment if symptoms persist. Regarding r eporting, MR classification is recommended, especially radiological score (dPEI ). “Patient centered care” is a key dimension of quality care. Good communicat ion with patients, as well as among the healthcare team, has the potential to improve care coordination, enhance safety and outcomes, increase patient satis faction, and reduce healthcare costs. Standardized MR report and drawin g are crucial for improving communication with patients and surgeons.

Conclusion

In conclusion, the ESUR consensus on endometriosis emphasizes the importance of standardized reporting and MR classifications to enhance communication between radiologists and the multidisciplinary team, as well as between radiologists and their patients. This is crucial in managing a disease where optimized communication is essential for providing patient- centered and value-based care Abstract-based Programme 57 Wednesday

Limitations

Consensus paper Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Miriam Dolciami: Nothing to disclose Pascal Rousset: Consultant: ZIWIG Adalgisa Guerra: Nothing to disclose Susan Freeman: Nothing to disclose Isabelle Thomassin-Naggara: Speaker: GE, Siemens, G uerbet, Hologic, Canon, Guebet, Bracco, GSD, Samsung, Fujifilm, Ince pto, ICAD Luciana Chamie: Nothing to disclose Nishat Bharwani: Nothing to disclose Lucia Manganaro: Nothing to disclose MRI for endometriosis: ESUR Consensus for protocol and lexicon *I. Thomassin-Naggara*¹, M. Dolciami², L. Chamie³, A. Guerra⁴, S. Freeman⁵, N. Bharwani⁶, P. Rousset⁷, L. Manganaro²; ¹Paris/FR, ²Rome/IT, ³São Paulo/BR, ⁴Lisbon/PT, ⁵Cambridge/UK, ⁶Surbiton/UK, ⁷Lyon/FR ([email protected]) Purpose or Learning Objective: The ESUR Research Committee appointed two chairs to supervise the development of the upda ted guidelines

Methods

or Background: The panel included 20 experts from 10 different countries

Results

or Findings: Pre imaging fasting, the use of antiperistalsic age nts, a moderately filled bladder and bowel preparation bef ore MRI are highly recommended. Vaginal and rectal opacification shoul d be considered as an option. MR protocol must include multiplanar T2W an d T1W sequence and a sequence that visualizes the kidneys. Superficial e ndometriosis should be described on T1FS as high signal intensity foci on the peritoneal surface. Endometriomas or implants should be described regar ding multiplicity, signal intensity, central or peripheral location and bilat erality. MR evaluation of deep pelvic endometriosis should be performed using a co mpartmental division defining two horizontal and vertical lines. A bladd er nodule should be accurately described according to location measured and the distance to ureteric orifice provided. Uterosacral ligament (US L) of ≤3mm is normal. A USL must be considered as abnormal if a nodule or spicu lation is visible in at least two planes or if a bright T1W spot is detected. A p osterior vaginal wall or pouch of Douglas nodule should be described and measured. The term rectovaginal septum must be accurately used for sub peritoneal n odules. External adenomyosis should be described according to locati on and size. The description of a rectosigmoid nodule includes locat ion, number of nodules, longitudinal extent, distance to the anal verge, an d wall thickening. The lateral compartment includes the anterior distal round liga ment, mediolateral parametrium and posterolateral parametrium. Abdomin al wall nodules, ileocaecal junction and appendiceal nodules, as wel l as sigmoid nodules, must be systematically described.

Conclusion

This lecture will present the 10 ESUR statement reg arding MR Endometriosis protocol and lexicon

Limitations

Consensus Funding for this study: Nond Ethics committee - additional information: Not applicable Author Disclosures: Miriam Dolciami: Nothing to disclose Pascal Rousset: Consultant: ZIWIG Adalgisa Guerra: Nothing to disclose Susan Freeman: Nothing to disclose Isabelle Thomassin-Naggara: Speaker: GE, Siemens, G uerbet, Hologic, Canon, Guebet, Bracco, GSD, Samsung, Fujifilm, Ince pto, ICAD Luciana Chamie: Nothing to disclose Nishat Bharwani: Nothing to disclose Lucia Manganaro: Nothing to disclose MRI features of Accessory Cavitated Uterine Mass (A CUM) and Cystic Adenomyosis *Y. Qi*, Z. Zhao, X. Gao, P. Zhang, H. Xue, Y. He, L. Zhu; Beijing/CN ([email protected]) Purpose or Learning Objective: To assess MRI features in accessory cavitated uterine mass (ACUM) and cystic adenomyosi s by evaluating lesion and uterine morphology.

Methods

or Background: The study included 16 ACUM patients (mean age 25.8 years) and 14 cystic adenomyosis patients (mea n age 35.5 years) with surgically and pathologically confirmed diagnoses. Pre-treatment pelvic MRI scans were performed, assessing lesion location, si ze, shape, cystic cavity, peripheral myometrial thickness, and uterine struct ure.

Results

or Findings: Patients presented with symptoms like dysmenorrhea and abdominal pain. Lesions appeared as layered str uctures with T2 hypointensity peripherally and T1 hyperintensity ce ntrally. ACUM lesions were typically single and located at the fundal lateral wall, while cystic adenomyosis lesions varied in location, 5 adenomyosis lesions w ere irregular. ACUM patients had lesions ranging from 22mm to 45mm, wit h a central cystic cavity diameter of 7mm to 36mm and peripheral myometrial t hickness of 5mm to 10mm. Fifteen ACUM cases showed a concentric ring s tructure and the junctional zone was not significantly thickened. Cy stic adenomyosis lesions ranging from 20mm to 64mm, with a cystic cavity dia meter of 4mm to 60mm and peripheral myometrial thickness of 2mm to 34mm. 5 cases showing uneven peripheral myometrium, only 1 case showing a concentric ring structure. 8 cases showing thickening junctional zo ne and 8 cases showing adenomyosis or fibroid presence, and 7 patients wit h ovarian or tubal endometriosis.

Conclusion

MRI is valuable for diagnosing ACUM and cystic aden omyosis. ACUM patients are younger, with single, concentric lesions at the fundal lateral wall, rarely showing junctional zone thickening or adenomyosis-like changes. Cystic adenomyosis is suggested by lesions on the p osterior wall, irregular shapes, large size, uneven shape, junctional zone t hickening, deep endometriosis, and fibroids.

Limitations

ACUM and cystic adenomyosis are rare diseases, such the number of cases are limited. Funding for this study: National High Level Hospital Clinical Research Funding, 2022-PUMCH-B-069 Ethics committee - additional information: Peking Union Medical College Hospital Ethics committee. Author Disclosures: Xin Gao: Nothing to disclose Zichen Zhao: Nothing to disclose Lan Zhu: Nothing to disclose Huadan Xue: Nothing to disclose Yonglan He: Nothing to disclose Yafei Qi: Nothing to disclose Peng Zhang: Nothing to disclose Comparison of pregnancy outcomes following hysteros alpingography (HSG) with either water soluble contrast media (WSC M) or oil soluble contrast media (OSCM) in routine clinical practice *A. P. Hemingway*, K. M. Wourms, S. Sudderuddin, E. Kashef; London/UK Purpose or Learning Objective: Hysterosalpingography (HSG) represents the mainstay of the imaging evaluation of the paten cy of fallopian tubes in the investigation of subfertility. An apparent associat ion of improved conception following HSG has been recognised since the early 1 900’s. An randomised controlled trial (RCT )published in 2017 establishe d a significant difference between OSCM and WSCM with ongoing pregnancies in 3 9.7% following OSCM and 29.1% following WSCM .The purpose of this study was to determine if these results could be replicated in r outine clinical practice.

Methods

or Background: Imaging and medical records relating to 667 consecutive HSGs performed between January 2020 and December 2021 were reviewed. For both WSCM and OSCM data was anal ysed with respect to age, primary of secondary sub-fertility, normal or abnormal HSGs, number of pregnancies, mode of conception and pregnancy outco me.

Results

or Findings: Records relating to 667 women were reviewed, 76 los t to follow-up. 591 records, 498 following WSCM, 93 f ollowing OSCM were analysed. A total of 354 conceptions were recorded, 291 (58.4%) following WSCM, 63 (67.7%) following OSCM. Spontaneous concep tions accounted for 160 of the WSCM (32.1%) pregnancies and 38 (40.9%) of the OSCM pregnancies. Time to conception was faster in the O SCM group and the pregnancy outcomes were similar in both groups. The re were no observed differences in pregnancy rates following assisted r eproduction in the two groups.

Conclusion

This retrospective study is in accordance with publ ished RCTs and meta analyses that an HSG with WSCM or OSCM is advantageous in increasing the chances of spontaneous conception an d that OSCM has a significant advantage over WSCM.

Limitations

This is a retrospective study. The OSCM group is sm aller than the WSCM group. The study period and analysis is be ing extended to encompass HSGs undertaken in 2022 and 2023 Funding for this study: No external funding. HSGs standard of care. Retrospective data collection Ethics committee - additional information: This retrospective study was approved IRAS Number 254621 Author Disclosures: Katherine Marie Wourms: Nothing to disclose Siham Sudderuddin: Nothing to disclose Anne Patricia Hemingway: Nothing to disclose Elika Kashef: Consultant: Rocket Medical Abstract-based Programme 58 Wednesday Comparing Inter-reader Agreement Between dPEI and # Enzian Classifications: Which Is the Better One? *A. Santonocito*, N. Bogveradze, J. Heine, T. H. He lbich, P. A. Baltzer, R. Wenzl, P. Clauser; Vienna/AT ([email protected]) Purpose or Learning Objective: The deep pelvic endometriosis index (dPEI) and #Enzian classifications have been introduced to facilitate reporting endometriosis patients. While studies show good int er-reader agreement with #Enzian, limited data exist for dPEI. This study ai med to assess the inter- reader agreement of both classifications.

Methods

or Background: In this retrospective, monocentric, IRB-approved study the pelvic MRIs performed in consecutive pati ents for suspected endometriosis from February 2018 to November 2023 w ere evaluated. Two readers (R1, dedicated radiologist, 5 years of expe rience; R2, general radiologist, 2 years of experience) independently a ssessed the presence of lesions using dPEI and #Enzian classifications. The extent of disease was then categorized as mild (score ≤2), moderate (3-4), or severe (≥5). Inter-reader agreement was evaluated by using Cohen’s Kappa.

Results

or Findings: A total of 108 patients (mean age 32.7 ±7.1 years; range 21-50 years old) were included in this study. Using dPEI, R1 classified 22 (20.4%) patients with mild, 43 (39.8%) moderate, an d 43 (39.8%) severe disease; while R2 classified 28 (25.9%) patients wi th mild, 43 (39.8%) moderate, and 37 (34.3%) severe disease. The inter- reader agreement for dPEI was k=0.47 (p<0.001). Using #Enzian, R1 classi fied 10 (9.3%) patients as mild, 12 (11.1%) as moderate, and 86 (79.6%) as severe disease; while R2 classified 21 (19.4%) patients as mild, 34 (31.5%) moderate, and 53 (49.1%) severe disease. The inter-reader agreement for #Enz ian was k=0.31 (p<0.001).

Conclusion

Inter-reader agreement was moderate for dPEI and fa ir for #Enzian. This evidence suggests that dPEI, as a rad iologically based classification, could be easier to use by non-speci alist radiologists. However, a dedicated training is essential to improve inter-re ader agreement in the evaluation of endometriosis.

Limitations

Retrospective study; lacking of assessment of histo logy Funding for this study: None Ethics committee - additional information: Number: 2057/2017 Author Disclosures: Pascal A.T. Baltzer: Nothing to disclose Rene Wenzl: Nothing to disclose Thomas H. Helbich: Nothing to disclose Ambra Santonocito: Nothing to disclose Nino Bogveradze: Nothing to disclose Jana Heine: Nothing to disclose Paola Clauser: Nothing to disclose 15:00-16:00 Research Stage 3 Research Presentation Session: Chest RPS 504 Imaging of diffuse lung diseases: old and new Moderator M. Occhipinti; Florence/IT ([email protected]) Imaging and Clinical Features of Interstitial Lung Abnormalities (ILA) that Predict Progression to Idiopathic Pulmonary Fibrosi s (IPF) *T. Schnitzler*, J. H. Sohn; San Francisco, CA/US ([email protected]) Purpose or Learning Objective: Interstitial lung abnormality (ILA) is often an incidental imaging finding, representing early or m ild fibrosis. While most cases do not progress, some advance to idiopathic pulmona ry fibrosis (IPF), leading to severe outcomes. Accurate risk stratification of ILA on non-contrast chest CT is crucial for guiding follow-up and early treatmen t. This study aims to improve stratification by identifying imaging and clinical features that predict progression from ILA to IPF.

Methods

or Background: This retrospective case-control study included patients from a longitudinal ILD database: a low-ri sk ILA cohort (n = 525) and a high-risk ILA cohort (n = 221). Imaging features an alyzed included subpleural fibrotic reticulation, cranial extent of fibrosis, anterior lung involvement, and emphysema severity. Clinical variables included age and gender. Statistical analyses were conducted using chi-square tests for categorical variables and independent t-tests for continuous variables.

Results

or Findings: The high-risk ILA cohort had significantly higher r ates of subpleural fibrotic changes (78% vs. 36%, p < 0.001 ), cranial extent of fibrosis (61% vs. 14%, p < 0.001), anterior lung involvement (86% vs. 37%, p < 0.001), and severe emphysema (48% vs. 39%, p < 0.001) compa red to the low-risk cohort. The high-risk cohort was also older (mean a ge 72.64 years vs. 70.65 years, p = 0.020), with no significant gender diffe rence.

Conclusion

This study identifies key imaging and clinical pred ictors of ILA progression to IPF, such as subpleural fibrotic cha nges, cranial extent of fibrosis, and older age. These findings could impro ve risk stratification, guiding timely monitoring and interventions to enhance pati ent outcomes. Future research should validate these findings in larger, multi-center cohorts.

Limitations

The main limitation is the retrospective single-cen ter design. Funding for this study: RSNA Research Fellow Grant 2024 Swiss Society for Radiology Research Grant 2023 Bangerter-Rhyner Foundation, Basel, Switzerland Ethics committee - additional information: This study is IRB approved (17-22317). Author Disclosures: Tician Schnitzler: Nothing to disclose Jae Ho Sohn: Nothing to disclose Identifying progressive pulmonary fibrosis on seria l CT: An international multi-observer study *L. Sun*¹, M. A. Mestas Nuñez², J. Jacob¹, S. Piciu cchi³, L. Calandriello⁴, A. Carvalho⁵, R. E. Ledda⁶, M. Chen¹, A. Devaraj¹; ¹London/UK, ²Barcelona/ES, ³Forlì/IT, ⁴Rome/IT, ⁵Porto/PT, ⁶Parma/IT ([email protected]) Purpose or Learning Objective: To evaluate the performance and agreement of thoracic radiologists and interstitial lung dise ase (ILD) physicians in identifying progressive pulmonary fibrosis on seria l CT scans in patients without idiopathic pulmonary fibrosis (IPF).

Methods

or Background: 100 patients with various non-IPF fibrotic lung diseases (median age, 64 years [range, 36 to 85]; m ale, n=40) had serial CTs obtained 6 to 24 months apart, which were reviewed independently by 12 ILD physician and thoracic radiologist readers blinded to clinical data. CTs were reviewed side-by-side and categorised as one of two groups: Stable Disease or Progressive Fibrosis. Groups were compared using contemporary relative change in percentage predicted forced vital capacit y (FVC), per reader and across the cohort, and analysed by Mann-Whitney U t est and mixed-effects modelling. Interobserver agreement was assessed usi ng intraclass correlation coefficient (ICC).

Results

or Findings: Preliminary data are presented. Mean FVC change for all patients was -6.28% (SD, 14.9). For individual readers, there was a significant difference in median FVC decline betwee n corresponding Progressive Fibrosis versus Stable Disease CT group s (range, -6.84% to - 11.44%, p=<0.001–0.015). For the whole reader cohor t, mean FVC decline was significantly greater in Progressive Fibrosis v ersus Stable Disease on CT (-10.70%, 95% CI [-11.89%, -9.50%] versus -1.47%, 9 5% CI [-2.78%, -0.15%]). Interobserver agreement was moderate (ICC = 0.501, 95% CI [0.420, 0.588]).

Conclusion

Among specialist thoracic radiologists and ILD phys icians, visual evaluation of serial CT scans is a valuable method for determining progressive fibrosis in non-IPF fibrotic lung diseases, judged against contemporary FVC decline, though interobserver agreement remains mod erate.

Limitations

Single-centre retrospective study Funding for this study: Nil sought Ethics committee - additional information: Prior IRAS approval for retrospective research within the Royal Brompton Ho spital radiology department Author Disclosures: Sara Piciucchi: Nothing to disclose Roberta Eufrasia Ledda: Nothing to disclose Andre Carvalho: Nothing to disclose Joseph Jacob: Nothing to disclose Mitchell Chen: Nothing to disclose Anand Devaraj: Nothing to disclose Logan Sun: Nothing to disclose Lucio Calandriello: Nothing to disclose Marcos Alejandro Mestas Nuñez: Nothing to disclose Abstract-based Programme 59 Wednesday Diagnostic Delay of Lung Cancer in Interstitial Lun g Disease *T. Schnitzler*, J. H. Sohn; San Francisco, CA/US ([email protected]) Purpose or Learning Objective: Interstitial lung disease (ILD) patients have an increased risk of lung cancer, but detection is challenging due to

Background

fibrosis, leading to diagnostic delays. There is limited research on lung cancer in ILD, particularly regarding diagnost ic delays. This study aims to analyze delayed lung cancer diagnoses in ILD patien ts, including tumor stage at diagnosis, growth rates, treatment regimens, and outcomes.

Methods

or Background: This retrospective study included ILD patients with concomitant lung cancer (pathology proven or >50% r adiologically suspected) from two referral centers. A thoracic radiologist r e-reviewed chest CTs to determine the earliest visible and callable lesion time, when it was first deemed suspicious, and its growth rate. Tumor staging, tre atment regimens, and outcomes were analyzed. Survival curves were genera ted using the Kaplan- Meier method, comparing median survival times betwe en delayed and non- delayed cancer cases with the log-rank test.

Results

or Findings: Seventy-seven cases of concurrent ILD and lung canc er were identified (53 pathology proven, 24 radiologic ally presumed). Delayed diagnoses occurred in 47% (36/77) of cases, with an average delay of 3.42 years. These delayed cases had a mean annual growth rate of 293% and a mean doubling time of 3.3 years. An additional 5% ( 4/61) were diagnosed post- lung transplant. The median survival time was 1269 days for early detection versus 867 days for delayed detection. However, the difference in survival was not statistically significant (p = 0.80).

Conclusion

This study found that 52% of lung cancer cases in I LD had delayed diagnoses, with an average delay of 3.42 ye ars. Despite the delays, there was no significant difference in mortality be tween early and delayed detection cases.

Limitations

The main limitation is the retrospective study desi gn. Funding for this study: Swiss Society for Radiology Research Grant 2023 Bangerter-Rhyner Foundation, Basel, Switzerland Ethics committee - additional information: This study was approved by the local IRB (17-22317) Author Disclosures: Tician Schnitzler: Nothing to disclose Jae Ho Sohn: Nothing to disclose Radiological assessment of bronchial and arterial d imensions and mucus plug presence in 640 bronchiectasis patients: insig hts from the EMBARC registry *Y. Chen*¹, A. Pieters¹, E-R. Andrinopoulou¹, S. Al iberti², M. Loebinger³, P. Ciet¹, J. Chalmers⁴, H. A. W. M. Tiddens¹, On Belalf Of Embarc Study Group⁴; ¹Rotterdam/NL, ²Humanitas Research Hospital, Mila n/IT, ³London/UK, ⁴Dundee/UK ([email protected]) Purpose or Learning Objective: Key features of bronchi in bronchiectasis disease are irreversible widening, wall thickening and mucus plugging. The bronchiectasis registry EMBARC lacks currently obje ctive quantitative metrics for these features. The aim of our study was to ana lyse EMBARC chest CTs using an AI-based algorithm measuring bronchus and artery (BA) dimensions and ratios and counting mucus plugs (MP).

Methods

or Background: 885 CTs from eight EMBARC centres were retrospectively collected for automatic analysis us ing LungQ (Thirona, The Netherlands), which segments the bronchial tree and identifies segmental (G0) and distal (G1,2,3…) generations. For each BA-pair, the following dimensions are computed: diameters of bronchial outer edge (Bo ut), inner edge (Bin), and artery (A), and wall thickness (Bwt) and the follow ing BA-ratios: Bout/A, Bin/A, Bwt/A, and bronchial wall area/outer area (Bwa/Boa) . Cut-offs for mild and severe bronchial widening are Bout/A>1.1 and >1.5, respectively and for thickening (Bwt/A>0.14). The MP analysis automatica lly segments the bronchial tree, detects the total number and volume of MP.

Results

or Findings: 640 CTs were successfully analysed, identifying 141 ,978 BA-pairs from G0 until G29 (222 BA-pairs per CT). B out/A>1.1 or >1.5 were observed in 73% and 39% of all BA-pairs, respective ly. Bwt/A>0.14 was observed in 49% of all BA-pairs. The median(IQR) Bo ut/A, Bin/A, Bwt/A, and Bwa/Boa for G1-6 were 1.34(1.07, 1.72), 1.04(0.81, 1.35), 0.13(0.1, 0.2). MP were found in 83% of CTs, with a median number of 8 plugs and a median volume of 0.44mL per CT.

Conclusion

Our study demonstrates the capability of AI-based a lgorithms to measure BA-dimensions and detect mucus plugs on che st CT scans of bronchiectasis patients. Our findings show widespre ad but heterogeneous bronchial widening and thickening, along with the p resence of mucus plugs, indicative of active infection and/or inflammation.

Limitations

Retrospective study Funding for this study: Supported by the Innovative Medicines Initiative an d The European Federation of Pharmaceutical Industrie s and Associations companies under the European Commission–funded Hori zon 2020 Framework Program and by Inhaled Antibiotic for Bronchiectasi s and Cystic Fibrosis (grant 115721). EMBARC3 is funded by the European Respirat ory Society through the EMBARC3 clinical research collaboration. EMBARC 3 is supported by project partners Armata, AstraZeneca, Boehringer In gelheim, Chiesi, CSL Behring, Grifols, Insmed, Janssen, Lifearc, and Zam bon. J.D.C. is supported by the GlaxoSmithKline/Asthma and Lung UK Chair of Respiratory Research. Ethics committee - additional information: The study received central ethical approval from the Multicentre Research Ethi cs Committee in the UK on Jan 8, 2015 (14/SS/1101) and the study is sponsored by the University of Dundee, Dundee, UK. Author Disclosures: Stefano Aliberti: Grant Recipient: Stefano Aliberti has received grants or contracts fees from INSMED incorporated, CHIESI, Fi sher&Paykel; and received consulting fees from GSK, McGRAW HILL, Zam bon, AstraZeneca, CSL Behring GmbH, Moderna, Chiesi, MSD Italia, Pysi oassist SAS, GlaxoSmithKline Eleni-Rosalina Andrinopoulou: Nothing to disclose James Chalmers: Grant Recipient: James D Chalmers h as received research grants from AstraZeneca, Boehringer Ingelheim, Glax oSmithKline, Gilead Sciences, Grifols, Novartis, Insmed and Trudell; an d received consultancy or speaker fees from Antabio, AstraZeneca, Boehringer Ingelheim, Chiesi, GlaxoSmithKline, Insmed, Janssen, Novartis, Pfizer, Trudell and Zambon. Pierluigi Ciet: Grant Recipient: Pierluigi Ciet has received grants or contracts fees from NOW-Dutch Research Councel and Horizon Pa thfinder; and has received payment or honoraria for lectures, present ations etc from Chiesi and Vertex. Michael Loebinger: Grant Recipient: Michael Loebing er has received consulting fees from Armata, 30T, AstraZeneca, Pari on, Ismed, Chiesi, Zambon, Electromed, Recode and Boehringer Ingelheim ; and received payment or honoraria for lectures, presentations et c from Ismed. Harm A W M Tiddens: Employee: Harm AWM Tiddens has received research grants from Thirona as Chief Medical Officer and Em eritus Professor ErasmusMC Sophia; and has received payment or honor aria for lectures, presentations etc from Vertex. Yuxin Chen: Nothing to disclose On Belalf Of Embarc Study Group: Nothing to disclos e Angelina Pieters: Nothing to disclose Computed Tomography-Derived Quantitative Imaging Bi omarkers enable the prediction of survival and disease severity in patients with Systemic Sclerosis *M. M. Sieren*¹, H. Graßhoff¹, L. Berkel¹, G. Rieme kasten¹, F. Nensa², R. Hosch², J. Barkhausen¹, R. Klöckner¹, F. Wegner¹ ; ¹Lübeck/DE, ²Essen/DE ([email protected]) Purpose or Learning Objective: Systemic Sclerosis (SSc) is a complex connective tissue disorder with variable disease pr ogression and outcome. While chest CT imaging is recommended in all patien ts to evaluate interstitial lung disease, AI-driven body composition analysis ( BCA) can further enhance radiological assessment by providing quantitative i maging biomarkers. This study aims to assess BCA's ability to predict survi val, complications, and disease severity on chest CT.

Methods

or Background: CT scans were obtained from a prospectively maintained cohort of 452 SSc patients, including 12 8 with at least one CT scan and 35 patients with up to three follow-up exams. T he follow-up period averaged 36.5±4.5 months. An AI-based 3D BCA algori thm measured muscle volume, adipose tissue compartments, and bone miner al density. BCA Parameters were evaluated in relation to clinical, laboratory, and functional data on baseline and follow-up scans. Survival pred iction was performed using regression analysis, comparing models based on BCA, BMI, and clinical parameters.

Results

or Findings: The BCA model outperformed BMI and clinical models in predicting survival (BCA AUC=0.74, BMI AUC=0.49, cl inical parameters AUC=0.53). Including longitudinal BCA data further improved the model's AUC to 0.82. Altered BCA parameters were linked to incr eased odds ratios [with 95% confidence interval] for complications like acr al ulcers (1.7 [1.1-1.9]), interstitial lung disease (2.1 [1.4-4.4]), cardiac (2.0 [1.3-3.0]) and gastrointenstinal manifestations (1.6 [1.4-1.9], al l p<0.05).

Conclusion

This study highlights that quantitative body compos ition biomarkers outperform established parameters in pre dicting survival and specific disease manifestations. These findings pro vide a blueprint how radiological assessment can transform from primaril y qualitative assessment to including previously unavailable quantitative data leading to more personalized patient care, potentially improving outcomes for SS c patients.

Limitations

The study's single-center design and small sample s ize may limit generalizability, and variations in CT quality coul d affect AI-based BCA accuracy. Abstract-based Programme 60 Wednesday Funding for this study: None. Ethics committee - additional information: Study/protocol number: AZ 22- 289 Author Disclosures: Lennart Berkel: Nothing to disclose Franz Wegner: Nothing to disclose Roman Klöckner: Nothing to disclose Gabriela Riemekasten: Nothing to disclose Jörg Barkhausen: Nothing to disclose Malte Maria Sieren: Nothing to disclose Hanna Graßhoff: Nothing to disclose René Hosch: Nothing to disclose Felix Nensa: Nothing to disclose Development of imaging protocol and radiomics-based nomogram for assessing lesion reversibility in connective tissue disease-associated interstitial lung disease Y. Zhang, *Y. Wang*, X. Yu, J. Wei, H. Wu; Shanghai /CN ([email protected]) Purpose or Learning Objective: To develop an imaging protocol for assessing lesion reversibility and a radiomics-base d nomogram for predicting lesion reversibility in connective tissue disease-a ssociated interstitial lung disease (CTD-ILD).

Methods

or Background: A retrospective study categorized CTD-ILD patients into training, internal and external validation coh orts. An imaging protocol of serial chest CT scans for assessing lesion reversib ility was developed, classifying patients as completely reversible (CR) and non-CR groups based on CT lesion changes. Lesions were evaluated using morphological CT features and radiomics signatures at the lung-zone level. Visual, radiomics, and combined nomogram models were developed and com pared through receiver operating characteristic (ROC) curve analy sis.

Results

or Findings: Among 153 patients with 575 affected lung zones, a five-feature radiomics signature significantly corr elated with ILD lesion reversibility. The radiomics model showed robust di scrimination, comparable to the visual model in the validation cohorts (interna l: 0.77, 95% CI: [0.68, 0.86] versus 0.87, 95% CI: [0.81, 0.94], p=0.056; externa l: 0.73, 95% CI: [0.66, 0.79] versus 0.78, 95% CI: [0.72, 0.84], p=0.20), and inf erior to the visual model in the training cohort (0.72, 95% CI: [0.66, 0.79] ver sus 0.82, 95% CI: [0.77, 0.87], p=0.02). The combined nomogram model outperf ormed the visual model alone in the training and external validation cohor ts (0.86, 95% CI: [0.81, 0.91], p=0.03; 0.82, 95% CI: [0.77, 0.87]; p=0.048).

Conclusion

An imaging protocol was established for assessing l esion reversibility in CTD-ILD. The radiomics signature p rovided a quantitative approach to predict lesion reversibility. The combi ned nomogram improved the predictive accuracy beyond morphological features a lone.

Limitations

First, clinical information were incomplete and not included in the predictive model. Future research incorporating mor e clinical information is needed. Second, the manual segmentation of lung zon es might introduce bias across various scans. Funding for this study: None Ethics committee - additional information: Ethics approval (No. LY2023- 019-B) was granted by the institutional review boar d (IRB) of Shanghai Jiaotong University, School of Medicine, Renji Hosp ital. The IRB waived informed consent requirement for this retrospective study. Author Disclosures: Yu Wang: Nothing to disclose Jiaxu Wei: Nothing to disclose Ying Zhang: Nothing to disclose Xiao Yu: Nothing to disclose Huawei Wu: Nothing to disclose Pleural Effusion as a Prognostic Indicator in COVID -19: A nationwide Multicenter Analysis *A. M. Bucher*¹, E. Frodl¹, F. G. Meinel², M. M. Si eren³, M. A. Fink⁴, M. S. May⁵, M. S. Kim⁶, T. Vogl¹, A. Surov⁷; ¹Frankfurt/DE, ²Rostock/DE, ³Lübeck/DE, ⁴Heidelberg/DE, ⁵Erlangen/DE, ⁶Essen/DE, ⁷Minden/DE Purpose or Learning Objective: This study evaluates the prognostic significance of pleural effusion (PE) in COVID-19 p atients across 13 German centers, part of the RACOON (Radiological Cooperati ve Network) project. We aimed to assess the relationship between PE and key clinical outcomes, in a large multicentre study.

Methods

or Background: In this retrospective study, 1183 COVID-19 patients (29.3% women, 70.7% men) underwent chest CT to asse ss the presence, volume, and density of PE. We analyzed associations between PE and clinical outcomes including 30-day mortality, ICU admission, and mechanical ventilation. We used univariable and multivariable regression analyses, adjusting for confounders such as the COVID-19 CT s everity score.

Results

or Findings: PE was identified in 31.5% of patients. A significa nt correlation was found between PE and 30-day mortali ty (47.5% in non- survivors vs. 27.3% in survivors, p<0.001). PE pres ence independently predicted mortality with a hazard ratio (HR) of 2.2 2 (95% CI 1.65-2.99, p<0.001). However, PE volume and density were not s ignificantly associated with mortality. ICU admission was necessary in 46.8 % of patients, and 26.7% required mechanical ventilation. PE presence was al so linked to ICU admission and ventilation but not its volume or density.

Conclusion

Pleural effusion is a significant independent predi ctor of 30-day mortality in COVID-19 patients, irrespective of its volume or density. These findings underscore the importance of including PE detection in routine CT assessments to enhance clinical decision-making and patient care.

Limitations

This retrospective study was limited to German tert iary care centers, which may not represent other settings. Funding for this study: Funded by „NUM 2.0“ (FKZ: 01KX2121) Ethics committee - additional information: IRB approval for this retrospective multi centre study was obtained (20-7 19). Author Disclosures: Eric Frodl: Nothing to disclose Alexey Surov: Nothing to disclose Thomas Vogl: Nothing to disclose Moon Sung Kim: Nothing to disclose Matthias Stefan May: Nothing to disclose Matthias Alexander Fink: Nothing to disclose Felix G. Meinel: Nothing to disclose Malte Maria Sieren: Nothing to disclose Andreas Michael Bucher: Nothing to disclose QIP Are Chest Radiographs Being Conducted in Accord ance with the British Thoracic Society Recommendations for Adults Diagnosed with Community Acquired Pneumonia? *M. Mobini*, A. Nehvi, S. Buckingham, L. Mills; Ste venage/UK Purpose or Learning Objective: This audit evaluates whether follow-up chest radiographs are being performed for adults diagnose d with Community- Acquired Pneumonia (CAP) according to British Thora cic Society (BTS) guidelines. A gap was identified when many patients did not receive a follow- up chest X-ray within six weeks, prompting an audit to identify barriers and gaps in care.

Methods

or Background: CAP affects 0.5% to 1% of UK adults annually and carries a mortality risk of 5-14%. Follow-up X-rays are crucial to ensure the resolution of pneumonia and to exclude underlying c onditions such as lung cancer. The first cycle retrospectively reviewed 50 adult patients diagnosed with CAP between November 2023 and March 2024. The second cycle repeated the review from March to July 2024, after implementing several interventions. Data were collected to assess compli ance with follow-up X-ray recommendations, virtual Pneumonia clinic (VPC) ref errals, and Casualty (CAS) alerts.

Results

or Findings: In the first cycle, only 32.6% of patients received follow- up X-rays within six weeks, with 67.4% failing to c omply. Among those discharged from the emergency department, 67.8% did not have a follow-up X- ray. Referrals to the Virtual Pneumonia Clinic (VPC ) were low (23.2%), and only 5% of reports included a CAS alert. Following interventions aimed at raising awareness among doctors, improving document ation, enhancing patient education, and increasing CAS alerts, the s econd cycle showed significant improvement. Compliance with follow-up X-rays increased to 70%, with VPC referrals rising to 77% and CAS alerts rea ching 73%. The compliance rate for follow-up X-rays among patients discharged from emergency care increased from 32.2% to 65%.

Conclusion

A marked improvement in adherence to BTS guidelines after targeted interventions, lead to more follow-up X-ra ys, VPC referrals, and CAS alerts.

Limitations

1) Retrospective Design 2)Small sample size 3) Sing le-Centered Audit Funding for this study: East and North Hertfordshire NHS Trust. Ethics committee - additional information: The study was approved by the trust ethics committee and audit department Author Disclosures: Moein Mobini: Author: second author Aabid Nehvi: Author: 1 Susan Buckingham: Consultant: 2 Author: 2 Lauren Mills: Author: 3 Abstract-based Programme 61 Wednesday 15:00-16:00 Research Stage 4 Research Presentation Session: Cardiac RPS 503 Cardiac CT: plaques and beyond Moderator D. Suchá; Utrecht/NL Author Disclosures: Dominika Suchá: Research Grant/Support: Philips Hea lthcare Research Support received by the Department of Radiology, UM C Utrecht, NL CT coronary calcium scoring to detect obstructive c oronary artery disease in primary care patients with non-typical c hest pain M. Y. Koopman¹, R. Willemsen², B. Kietselaer³, P. M . A. Van Ooijen¹, J-W. Gratama⁴, R. Braam⁴, R. Van Bruggen⁴, P. Van Der Harst⁵, *R. Vliegenthart*¹; ¹Groningen/NL, ²Maastricht/NL, ³Rochester, MN/US, ⁴Apeldoorn/NL, ⁵Utrecht/NL ([email protected]) Purpose or Learning Objective: Computed Tomography coronary calcium scoring (CT-CCS) has higher sensitivity for detecti on of obstructive coronary artery disease (OCAD) than exercise electrography ( x-ECG), but its utility as an initial diagnostic test in primary care remains unclear. This pilot study compares CT-CCS results with x-ECG results in prima ry care and assesses patients’ perspectives.

Methods

or Background: Thirty-eight primary care offices participated in t his study. After cluster randomisation, 19 offices refe rred patients with atypical angina pectoris or non-specific thoracic complaints for CT-CCS and 19 offices used x-ECG as the primary test (standard care). Cli nical data were collected using electronic patient records, and patients’ per spectives on the diagnostic test were assessed through a questionnaire. Outcome measures included CAD diagnosis, initiation of cardiovascular risk manage ment (CVRM), and patient satisfaction.

Results

or Findings: In total, 101 patients were included. In 25 patient s undergoing X-ECG, one (4%) had a positive test resu lt and received CVRM, but no patients were diagnosed with obstructive CAD . CT-CCS was performed in 76 patients. 17 CT-CCS patients (23%) had a posi tive test result (calcium score >100), and 14 (19%) received CVRM. Obstructiv e CAD was diagnosed in four CT-CCS patients (5.3%). Of CT-CCS patients, 31 (43%) perceived the test as ‘very easy’ compared to none of the x-ECG p atients.

Conclusion

CT-CCS is a promising diagnostic tool in primary ca re for the detection of obstructive CAD, offering a more patie nt-friendly experience compared to x-ECG.

Limitations

Small cohort, especially in the x-ECG arm, and low OCAD rate. A few patients received the test result before comple ting the questionnaire. Baseline cardiovascular related risk factors were i nconsistently reported in electronic patient records. Funding for this study: Funding was received from the Dutch Heart Foundation (Hartstichting, grant number: CVON2017-1 4). Ethics committee - additional information: The Medical Ethical Committee of the University Medical Center of Groningen appro ved CONCRETE (number 2018/404). Author Disclosures: Robert Willemsen: Nothing to disclose Pim Van Der Harst: Nothing to disclose Richard Braam: Nothing to disclose Peter M.A. Van Ooijen: Nothing to disclose Moniek Yvonne Koopman: Nothing to disclose Rozemarijn Vliegenthart: Research/Grant Support: Si emens Healthineers Bas Kietselaer: Nothing to disclose Rykel Van Bruggen: Nothing to disclose Jan-Willem Gratama: Nothing to disclose Thin-slice non-contrast CT detects prognostically r elevant calcified plaques missed by conventional calcium scoring *F. Yavuz*, F. Biavati, K. Schulze, S. Tsogias, B. Föllmer, A-M. Stantien, M. Bosserdt, M. Dewey; Berlin/DE Purpose or Learning Objective: To evaluate whether thin-slice non-contrast CT (NCCT) can detect prognostically relevant corona ry plaques missed by conventional 3.0-mm reconstructions.

Methods

or Background: This study included 141 patients from the CAD-Man trial [NCT00844220] (mean age 60.77 ± 11.06 years, 55% female) with available thin-slice NCCT (0.5-mm). The Agatston me thod was used to detect calcified plaques. Sensitivity and specificity for the detection of calcified plaques were calculated using CT angiography (CTA) as the reference standard. Lesion- and patient-level statistics were calculated for plaque volume parameters. Prognostic relevance was assessed by ev aluating plaque progression rates for plaques detected only on thin -slice reconstructions, using median 10-year follow-up data when available.

Results

or Findings: In total 551 calcified plaques were detected. Thin- slice NCCT showed a higher sensitivity (91.83%; 506/551) for detecting coronary calcified plaques compared to 3.0-mm reconstruction (82.76%; 456/551), although standard reconstructions showed an overall per-patient increased mean calcified plaque volume (197.22 mm3 ± 330.05 m m3) compared to thin- slice NCCT (162.65 mm3 ± 284.1 mm3). Conversely, we observed a slightly lower specificity (97.23%; 492/506) for thin-slice NCCT compared to standard reconstructions (99.56%; 454/456). Coronary calcifi ed plaques missed in standard reconstructions were smaller in volume (2. 67 mm3 ± 1.47 mm3) compared to all detected plaques (20.68 mm3 ± 25.56 mm3). Missing calcified plaques on standard reconstructions would have led to the omission of 9 out of 141 patients (6.4%). Additionally, plaques only ide ntified on thin-slice NCCT at baseline were clearly visible at follow-up, with an average 7.9-fold increase in volume.

Conclusion

Coronary calcified plaques detected exclusively on thin-slice NCCT reconstructions showed increased plaque progre ssion rates compared to plaques detected in conventional calcium scoring .

Limitations

This study involved patients from a single-centre, and 10-year follow-up data were not available for all patients with missed plaques. Funding for this study: This study was funded by a grant of the Heisenberg programme. Ethics committee - additional information: The study was approved by ethics committee at Charité (EA1/124/23). Author Disclosures: Kenrick Schulze: Nothing to disclose Ferhat Yavuz: Nothing to disclose Anne-Marieke Stantien: Nothing to disclose Marc Dewey: Board Member: M.D. is European Society of Radiology (ESR) Publications Chair (2022-2025); the opinions expres sed in this presentation are the author’s own and do not represent the view of E SR Grant Recipient: EU (EC-GA 603266 in HEALTH.2013.2.4.2-2) DFG (DE 1361/ 14-1, DE 1361/18-1, BIOQIC GRK 2260/1, Radiomics DE 1361/19-1 [42822292 2] and 20-1 [428223139] in SPP 2177/1), GUIDE-IT (DE 1361/24-1) , Berlin University Alliance (GC_SC_PC 27), G-BA (01NVF23002), Berlin I nstitute of Health (Digital Health Accelerator) Patent Holder: Patent on fractal analysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10,991,109, approved) Author: Cardiac CT (Springer Nature) Research/Grant Support: Siemens, General Electric, Philips, Canon Other: Hands-on cardiac CT courses (www.ct-kurs.de) Instit utional research agreements: Siemens, General Electric, Philips, Can on. Patent on fractal analysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10,991,109, approved) M .D. is European Society of Radiology (ESR) Publications Chair (2022 -2025); the opinions expressed in this presentation are the author’s own and do not represent the view of ESR Sotirios Tsogias: Nothing to disclose Federico Biavati: Nothing to disclose Maria Bosserdt: Nothing to disclose Bernhard Föllmer: Nothing to disclose Association of features derived from segment-level coronary artery calcium scoring with major adverse cardiovascular e vents: A multicentre study *S. Tsogias*, B. Föllmer, M. Mohamed, F. Biavati, K . Schulze, M. Bosserdt, M. Dewey; Berlin/DE ([email protected]) Purpose or Learning Objective: To investigate the association of segment- level coronary artery calcium (CAC) scoring derived features with major adverse cardiovascular events (MACE) compared to ve ssel-based and overall CAC scoring.

Methods

or Background: This subanalysis of the multicentre DISCHARGE trial (NCT02400229) included a total of (N = 1446) patients (mean age 59.9 ± 10.2 years) who had received a calcium scoring CT a nd were followed up over a median timespan of 3.5 years. The definition of M ACE included nonfatal stroke, nonfatal myocardial infarction and cardiova scular death. Associations with MACE were examined for proximal (LM and proxim al segments of the LAD, LCX and RCA) versus non-proximal calcification s and the total number of segments containing calcifications out of 19 (0: No calcification; 1: Limited; 2-9: Moderate, ≥ 10: Extensive). CAC scores were obtained both manu ally and using deep learning-based scoring methods. Analysis was performed using Cox proportional hazards regression adjusting for a ge, sex, body-mass-index, diabetes, dyslipidemia, hypertension, family histor y, smoking status and Agatston categories (< 400; ≥ 400) with hazard ratios (HR) and 95% confidence intervals (CI). Abstract-based Programme 62 Wednesday

Results

or Findings: During follow-up a total of 31 MACE occurred. Proxi mal vessel calcifications were associated with higher r isk for MACE (HR = 3.9, 95% CI [1.02, 14.5], p < .05). A moderate number of cal cified segments [2-9 segments] was also associated with an increased ris k for MACE (HR = 4.2, 95% CI [1.08, 16.1], p < .05).

Conclusion

Proximal vessel calcification as well as moderate s egment calcification were associated with a higher risk fo r MACE.

Limitations

Due to the low number of MACE in this study populat ion 2.1% (31 of 1446) the overall predictive value of the segmen t level CAC scoring may have been underrepresented. Funding for this study: This work was funded by the German Research Foundation through the graduate program BIOQIC (GRK 2260, project-ID: 289347353) and the DISCHARGE project (603266-2, HEA LTH-2012.2.4.-2) funded by the FP7 Program of the European Commissio n. Ethics committee - additional information: The study was approved by The German Federal Office for Radiation Protection and the local or national authorities at each trial site. Author Disclosures: Kenrick Schulze: Nothing to disclose Mahmoud Mohamed: Nothing to disclose Marc Dewey: Grant Recipient: EU (EC-GA 603266 in HE ALTH.2013.2.4.2-2) DFG (DE 1361/14-1, DE 1361/18-1, BIOQIC GRK 2260/1, Radiomics DE 1361/19-1 [428222922] and 20-1 [428223139] in SPP 2 177/1), GUIDE-IT (DE 1361/24-1), Berlin University Alliance (GC_SC_PC 27 ), Berlin Institute of Health (Digital Health Accelerator) Patent Holder: Patent on fractal analysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10,991,109, approved) Other: Hands-on ca rdiac CT courses (www.ct-kurs.de) Other: Institutional research agre ements: Siemens, General Electric, Philips, Canon. Other: M.D. is European S ociety of Radiology (ESR) Publications Chair (2022-2025); the opinions expres sed in this presentation are the author’s own and do not represent the view of E SR. Author: Cardiac CT (Springer Nature) Sotirios Tsogias: Nothing to disclose Federico Biavati: Nothing to disclose Maria Bosserdt: Research/Grant Support: Received fu nding from EU-FP7 Framework Program (DISCHARGE EU FP EC-GA 603266) Bernhard Föllmer: Nothing to disclose Prognostic value of semi-quantitative cCTA scores *E. Bruno*, A. Bettinelli, V. Morrone, A. Colombo, C. Gnasso, F. Pisu, D. Vignale, A. Palmisano, A. Esposito; Milan/IT ([email protected]) Purpose or Learning Objective: Coronary artery disease (CAD) is a global leading cause of morbidity and mortality, with comp lex pathogenesis. Coronary computed tomography angiography (cCTA) is a powerfu l non-invasive tool for diagnosing obstructive CAD. However, most patients have non-obstructive CAD, and risk stratification data are limited. Many cCTA-based risk scores were developed, however with low predictive value a nd reproducibility. This study aims to develop clinical-imaging models to pr edict major adverse cardiac events (MACEs) in patients undergoing cCTA for susp ected CAD.

Methods

or Background: Observational, single-center retrospective study including 4096 out of 10104 patients undergoing cCT A between 2016 and 2020. Patients with cardiovascular comorbidities or terminal cancer were excluded. Demographics, cardiovascular risk-factors , and medical history were collected via phone contact and medical records, to calculate known semiquantitative cCTA scores (CAD-RADS, Leiden risk score, Leaman risk score, SSS, SIS, Calcium score). Patients were comp ared after a minimum 4- year follow-up according to the occurrence of MACEs (cardiovascular death, nonfatal myocardial infarction, all-cause mortality , angina-related hospitalization, late coronary revascularization). Multivariable Cox regression models, adjusted for age and sex, were created usin g significant clinical variables and one cCTA score.

Results

or Findings: Among 1933 patients enrolled (65% men, age:63.5±11. 6 year-old), 353/1933(18%) had MACE. Patients with MA CE had higher rates of hypertension, dyslipidemia, diabetes, and higher cC TA scores(all p<.001). All cCTA scores significantly predicted MACE occurrence in Kaplan-Meier survival analysis(p<.005). Six multivariable models includin g clinical features (diabetes, dyslipidemia, hypertension) and one cCTA score have been developed: in each model cCTA score was the strongest prognostica tor of outcome, with CAD-RADS having the highest HR(2.996, 95%CI 2.374-3 .781, p<.001), followed by CACS(2.103, 95%CI 1.646-2.687, p<.001).

Conclusion

CCTA scores area all predictors of outcome, in part icular CAD- RADS, indicating the highest-grade coronary artery lesion, had the higher Hazard Ratio.

Limitations

No prospective data. Funding for this study: None Ethics committee - additional information: Approved by San Raffaele hospital ethics committee (124/2023) Author Disclosures: Davide Vignale: Nothing to disclose Alberto Colombo: Nothing to disclose Antonio Esposito: Nothing to disclose Elisa Bruno: Nothing to disclose Anna Palmisano: Nothing to disclose Francesco Pisu: Nothing to disclose Chiara Gnasso: Nothing to disclose Vittorio Morrone: Nothing to disclose Andrea Bettinelli: Nothing to disclose Myocardial delayed enhancement with first-generatio n dual-source photon-counting detector CT: an image quality compa rison across available spectral acquisition modes *B. Longere*¹, R. Cusumano¹, C. V. Gkizas¹, A. Rodr iguez Musso¹, F. Dubus¹, C. Croisille², C. Artaud¹, M. Haidar¹, F. A. Pontan a¹; ¹Lille/FR, ²Bordeaux/FR ([email protected]) Purpose or Learning Objective: To compare the image quality of myocardial delayed enhancement (CT-MDE) obtained by two differ ent tube voltages and three distinct cardiac synchronization modes on a f irst-generation dual-source photon-counting detector CT (PCD-CT).

Methods

or Background: Ninety patients (43 women) aged 63 years (54– 73y) referred for cardiac CT with CT-MDE were enrol led. CT-MDE acquisition was performed 5min after injection of 90mL of iodin e contrast medium (400mgI/mL). Tube voltage was set to either 120 or 140kVp. Current was automatically adjusted to a predetermined image qua lity level of 50. CT-MDE was acquired using helicoidal retrospective gating (R120; R140), sequential triggering (S120; S140) or prospective high-pitch g ating (F120; F140). Triggering was set to an RR delay of 300ms. Signal- to-noise ratio (SNR), contrast-to-noise ratio (CNR) and subjective image quality were assessed on virtual monoenergetic images at 65keV (VMI65) and i odine maps.

Results

or Findings: High-pitch acquisitions provided the lowest CT dose index (P<0.001) with no differences in body mass in dex between the 6 groups (P=0.09). No differences were observed in SNR acros s the six acquisition types on VMI65 (P=0.07) and iodine map (P=0.22). F1 20 demonstrated a CNR that was equivalent to or better than that of the o ther acquisitions (VMI65, P=0.01; iodine map, P=0.04). No difference was obse rved in artifact scores (VMI65, P=0.43; iodine maps, P=0.83). Global subjec tive image quality provided by F120-derived VMI65 was better than or e quivalent to that of other series (P=0.04).

Conclusion

High-pitch acquisition at 120kVp provides lower rad iation dose without compromising image quality. This acquisitio n mode should be recommended for the assessment of CT-MDE with PCD-C T.

Limitations

It was a single-center study with a limited sample size. The detectability of iodine enhancement of pathological finding was not assessed as it was considered to be outside the scope of thi s study. Funding for this study: None Ethics committee - additional information: IRB number: CRM-2408-417 Author Disclosures: Benjamin Longere: Nothing to disclose Raphael Cusumano: Nothing to disclose Christos Vasileiou Gkizas: Nothing to disclose François Dubus: Nothing to disclose Mehdi Haidar: Nothing to disclose Cedric Croisille: Employee: Siemens Healthineers Camille Artaud: Nothing to disclose Aimée Rodriguez Musso: Nothing to disclose François Ascagne Pontana: Nothing to disclose Quantification of Extracellular Volume in Acute Myo carditis Using Dual- Source Photon-Counting Detector CT: A Comparative A nalysis with CMR *C. V. Gkizas*¹, J. Limousin¹, W. Ben Mansoura¹, B. Longere¹, A. L. Rodriguez Musso¹, C. Croisille², F. A. Pontan a¹; ¹Lille/FR, ²Bordeaux/FR ([email protected]) Purpose or Learning Objective: The aim of this study was to assess the feasibility and accuracy of myocardial late enhance ment (LE) scanning for extracellular volume (ECV) quantification with dual -source photon-counting detector computed tomography (PCD-CT) in acute myoc arditis.

Methods

or Background: Patients with clinical suspicion of myocarditis who were referred for coronary CT angiography (CCTA) to exclude CAD were included in this retrospective study. The CCTA prot ocol using a first-generation PCD-CT, included a systematic LE acquisition. ECV w as calculated from the iodine ratio of the myocardium and blood pool on th e LE scan. A comprehensive CMR protocol was used as the referenc e method to confirm myocarditis according to the Lake Louise 2018 crite ria. All subjects underwent CCTA and CMR within 24 hours. Abstract-based Programme 63 Wednesday

Results

or Findings: 32 patients were included (mean age 36 years; 13 females). The mean dose length product of the LE sc an was 96± 32 mGy.cm. The mean global ECV between CCT and CMR did not sho w significant difference (29.4% ±4.5 vs 30.0 ±4.1, P=0.69). In patients diagnosed with myocarditis confirmed by CMR (n=25), the mean ECV-C T was notably elevated compared to individuals with normal CCT an d CMR findings (31.6% ±3.6 vs 25.6% ±3.2, P<0.01). ECV-CT value showed a strong positive correlation with LGE mass (r =0.85; p < 0.001).

Conclusion

Calculation of ECV using iodine maps derived from L E cardiac CT images is both feasible and accurate at low radi ation dose. PCD-CT offers a promising non-invasive imaging method in the cont ext of acute myocarditis.

Limitations

Retrospective, single study Funding for this study: None Ethics committee - additional information: All subjects were informed and provided their consent. Author Disclosures: Benjamin Longere: Nothing to disclose Christos Vasileiou Gkizas: Nothing to disclose Cedric Croisille: Nothing to disclose Jean Limousin: Nothing to disclose Aimée Leilen Rodriguez Musso: Nothing to disclose François Ascagne Pontana: Nothing to disclose Wissem Ben Mansoura: Nothing to disclose 16:30-17:30 Research Stage 1 Research Presentation Session: Oncologic Imaging RPS 616 Haematologic malignancies: multimodality imaging Moderator G. Cowell; Glasgow/UK ([email protected]) Prospective assessment of 3T Whole-Body MRI and 18F -FDG PET-CT in diagnosing multiple myeloma and its influence on pa tient care *A. Rossi*¹, D. Bezzi¹, D. Diano¹, A. Prochowski Ia murri¹, A. Cattabriga², E. Antognoni¹, G. Feliciani¹, P. Caroli¹, C. Cerchi one¹; ¹Meldola/IT, ²Bologna/IT ([email protected]) Purpose or Learning Objective: This study aims to compare the diagnostic efficacy of Whole Body-Magnetic Resonance Imaging ( WB-MRI) and 18F- Fluorodeoxyglucose Positron Emission Tomography (PE T-CT) in detecting bone marrow infiltration (BMI) in myeloma patients and assess their impact on patient management.

Methods

or Background: We prospectively enrolled myeloma patients from October 2020 to January 2024. Within a month, patie nts underwent 3T WB- MRI (following MY-RADS criteria) and PET-CT to asse ss BMI, para, and extramedullary disease. Clinical and laboratory dat a were collected. Two haematologists determined treatment plans using Int ernational Myeloma Working Group (IMWG) criteria based on all findings , which were then used to evaluate imaging performance.

Results

or Findings: The cohort included 137 patients (73 male; mean age , 66 years), with 39 having High Risk-Smoldering Mult iple Myeloma (SMM) and 98 with Multiple Myeloma (MM). WB-MRI sensitivity a nd specificity for BMI in MM were 100% and 97%, respectively, while PET-CT sh owed 89% sensitivity and 97% specificity (p=0.02). In SMM, BMI-positive patients had higher paraprotein levels (p=0.01); in MM, they had higher paraprotein (p=0.007) and lower hemoglobin (p=0.002). Clinical management cha nged in 54% of cases based on combined imaging results, with WB-MRI cons istent with management changes in 97% compared to 61% for PET-CT (p < 0.00 1).

Conclusion

WB-MRI and PET-CT play key roles in evaluating myel oma patients. WB-MRI demonstrated superior sensitivity in detecting BMI and had a greater influence on therapeutic decision-making.

Limitations

No limitations were identified Funding for this study: This study was partly funded by the Italian Ministr y of Health for Institutional Research (Ricerca Corrente ) within the research line "Innovative therapies, phase I-III clinical trials, and therapeutic strategy trials based on preclinical models, onco-immunological mec hanisms, and nano vectors. Ethics committee - additional information: the study was approved by C.E.R.O.M comitato etico della Romagna (AccuMRI tri al IRST code 100.15) Author Disclosures: Paola Caroli: Nothing to disclose Davide Bezzi: Nothing to disclose Danila Diano: Nothing to disclose Claudio Cerchione: Nothing to disclose Eleonora Antognoni: Nothing to disclose Arrigo Cattabriga: Nothing to disclose Alice Rossi: Nothing to disclose Andrea Prochowski Iamurri: Nothing to disclose Giacomo Feliciani: Nothing to disclose Prognostic value of maximum tumor spread (Dmax) in lymphoma patients treated with CD19-specific CAR-T cell ther apy *M. Winkelmann*, P. Achhammer, V. Blumenberg, K. Re jeski, G. Sheikh, M. Brendel, J. Ricke, M. Subklewe, W. G. Kunz; Muni ch/DE Purpose or Learning Objective: CD19 specific CAR T-cell therapy (CART) is an effective treatment for relapsed or refractory ( r/r) lymphoma. The maximum distance (Dmax) of lymphoma lesions holds potential as imaging biomarker in lymphoma treated with conventional therapies, but h as not been studied in context of CART. We evaluated Dmax at baseline imag ing as a prognostic tool for assessment of metabolic and overall response, p rogression-free survival (PFS) and overall survival (OS).

Methods

or Background: Consecutive r/r lymphoma patients with (PET/)CT at baseline before CART were included. Dmax was mea sured in cm at BL. Patients were divided into three groups according t o Dmax: low, intermediate and high. The sum of product diameters (SPD) accord ing to Lugano criteria was used to represent tumor burden (TB). Overall re sponse according to Lugano criteria and Deauville score were determined at follow-up imaging.

Results

or Findings: 103 patients were included. Median baseline Dmax wa s 40.0 cm (IQR: 16.4 – 70.3 cm). Median TB was signif icantly higher in the intermediate and high risk group compared to the lo w risk group (p=0.005). Intermediate and high risk group showed significant ly higher Ann Arbor stages (p<0.001). The survival analysis revealed a signifi cantly (p=0.030) shorter PFS in the high-risk group compared to the other patien ts (91 vs 364 days), but no relevant differences in OS (p=0.151). In addition, no significant differences in Deauville score and ORR were detected.

Conclusion

Patients with high Dmax showed a shorter PFS, but n o significant differences in OS. Dmax as an interval-scaled param eter represents a useful alternative to the Ann Arbor classification.

Limitations

Single center study, limited number of subjects. Fe w patients were excluded because of no measurable disease. Som e patients had only CT without PET at FU, with a possible redistribution o f Deauville score among Dmax-based groups. Funding for this study: The work was supported by funding from the research program “Förderung für Forschung und Lehre (FöFoLe) project number 1147” of the Medical Faculty of Ludwig Maximilian Univers ity (LMU) Munich and the Bavarian Cancer Research Center (BZKF) Ethics committee - additional information: All medical records and imaging studies underwent review with approval from the LMU Munich Institutional Review Board (LMU Ethics Committee, project number 19-817). Author Disclosures: Marion Subklewe: Speaker: Amgen; Astra Zeneca; BMS/ Celgene; GSK; Incyte Biosciences; Janssen; Novartis; Pfizer; Seattle Gen etics; Takeda Consultant: Aven Cell; CDR-Life; GSK; Incyte Biosciences; Janss en; Miltenyi Biotec; Molecular Partners Novartis; Pfizer; Takeda Researc h/Grant Support: Amgen; BMS/Celgene; Gilead; Incyte Biosciences; Janssen; M iltenyi Biotec; Morphosys; Novartis; Roche; Seattle Genetics; Taked a Viktoria Blumenberg: Consultant: Kite/Gilead Resear ch/Grant Support: Janssen; BMS/Celgene; Novartis; Takeda; Roche Philipp Achhammer: Nothing to disclose Michael Winkelmann: Nothing to disclose Kai Rejeski: Research/Grant Support: Kite/Gilead; N ovartis Consultant: Kite/Gilead; MBS/Celgene Wolfgang Gerhard Kunz: Advisory Board: Bristol Myer s Squibb; Boehringer Ingelheim; mintMedical; Need, Inc. Gabriel Sheikh: Nothing to disclose Matthias Brendel: Nothing to disclose Jens Ricke: Nothing to disclose Early Whole-Body MRI as a Predictor of long-term Me tabolic Response in Large B-Cell Lymphoma Patients following CAR T-Cell Therapy *C. Neelsen*¹, C. Sachpekidis¹, J. M. E. Jende¹, R. Gnirs¹, F. Kurz², P. Dreger¹, A. Dimitrakopoulou-Strauss¹, H-P. Schlemmer¹; ¹Heid elberg/DE, ²Geneva/CH Purpose or Learning Objective: To evaluate the utility of early whole-body MRI (wbMRI) for the prediction of long-term metabol ic response in patients with large B-cell lymphoma (LBCL) following chimeric ant igen receptor T-cell therapy (CARTT). Abstract-based Programme 64 Wednesday

Methods

or Background: In this prospective, IRB-approved study, we assessed 9 LBCL patients with target lesions identi fied on baseline wbMRI according to adapted Response Evaluation Criteria i n Lymphoma (RECIL), who were fit to undergo wbMRI within two weeks (mea n 11 days) following CARTT in an outpatient setting. Complete remission was defined as the absence of measurable disease and any lesions on di ffusion-weighted imaging. Early wbMRI findings were compared with PE T-CT results at 3-month follow-up, with metabolic responses classified acco rding to the Lugano criteria.

Results

or Findings: At the 3-month PET-CT follow-up 4 patients showed a complete metabolic response (CMR), 2 patients had a partial metabolic response (PMR) and 3 patients demonstrated progress ive disease (PD). Of the 4 patients with CMR, 2 had already achieved complet e remission on the early wbMRI, while the other 2 patients had a minor and p artial response. The 2 patients with PMR exhibited minor responses on earl y wbMRI and finally of the 3 patients with PD, one had a partial and two had m inor responses.

Conclusion

All patients demonstrated some degree of response o n early wbMRI within two weeks after CARTT. Early complete remission appeared to be a predictor of long-term metabolic response. How ever, early wbMRI was inconclusive in patients with initial minor or part ial responses, as these patients may still achieve complete metabolic remission or p rogress over time.

Limitations

The lymphodepleting chemotherapy administered prior to the CAR T-cell infusion complicates the differentiation of the specific effects of chemotherapy versus the CAR T-cell therapy itself. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The study was approved by the institutional review board (S-950/2021) and written informed consent was obtained from all participants. Author Disclosures: Christos Sachpekidis: Nothing to disclose Peter Dreger: Nothing to disclose Regula Gnirs: Nothing to disclose Christian Neelsen: Nothing to disclose Johann Malte Enno Jende: Nothing to disclose Felix Kurz: Nothing to disclose Antonia Dimitrakopoulou-Strauss: Nothing to disclos e Heinz-Peter Schlemmer: Nothing to disclose Diagnostic value Whole-body Magnetic resonance imag ing (WBMRI) short protocols can be useful in Multiple Myeloma p atients *C. R. G. L. O. M. Talei Franzesi*, C. Maino, P. N. Franco, D. Ippolito, R. Corso; Milan/IT ([email protected]) Purpose or Learning Objective: To compare the effectiveness and accuracy of whole-body magnetic resonance imaging (WBMRI) sh ort protocols for the overall assessment of bone marrow involvement in pa tients with multiple myeloma (MM), in comparison with standard whole-bod y MRI protocol

Methods

or Background: Sixty-four patients with biopsy-proven MM, who underwent an WBMRI with full body coverage (from ve rtex to feet) were retrospectively enrolled. WBMRI images were indepen dently evaluated, by two expert radiologists. After identifying the infiltra tion pattern (normal, focal, diffuse and combined), the whole skeleton was divided into six anatomic districts: skull, spine, sternum and ribs, upper limbs, pelvis and proximal two-third of femur, remaining parts of lower limbs, and patients were grouped according to number( 20) and location of the les ions

Results

or Findings: Most of patients showed a focal (59%) and combined (33%) infiltration patterns with lytic lesions pred ominantly distributed in the spine (82%) and pelvis (67%). Locations less freque ntly involved by focal bone lesions were skull and lower limbs (12%, respective ly). Excluding both the anatomic regions mentioned before from the standard MRI protocol, a short MRI protocol with a shorter execution time (saving about 14 minutes) could be obtained, maintaining a good sensitivity (89.9%), s pecificity (66.7%) and diagnostic accuracy (AUROC=0.881; 95%CIs: 0.797-0.9 65)

Conclusion

MRI short protocols could be proposed as an effecti ve and reliable approach to reduce the examination time, p reserving a high diagnostic accuracy and can be more focused on the main involv ed districts

Limitations

None Funding for this study: None Ethics committee - additional information: None Author Disclosures: Cesare Maino: Nothing to disclose Cammillo Roberto Giovanni Leopoldo Oreste Massimili ano Talei Franzesi: Nothing to disclose Rocco Corso: Nothing to disclose Paolo Niccolò Franco: Nothing to disclose Davide Ippolito: Nothing to disclose Dual-Vessel Microcirculation Imaging in Differentia tion of B cell and T cell subtype in intranodal Non-Hodgkin Lymphoma Using Su per-Resolution Ultrasound: An Exploring Study *Y. Dong*; Shanghai/CN ([email protected]) Purpose or Learning Objective: To explore the diagnostic performance of the super-resolution ultrasound (SRUS) imaging in dual- vessel systems, i.e., the microvascular system and the microlymphatic system, for predicting B cell and T cell subtypes in intranodal NHL.

Methods

or Background: Forty-two patients with intranodal NHL were included in this prospective study. All patients un derwent dual-vessel system SRUS imaging via intravenous and intra-lymph node r outes. SRUS parameters such as vessel density, vessel ratio, vessel comple xity level, diameter, distance, tortuosity, and flow velocity were measur ed for the microvascular and microlymphatic circulations. Principal component an alysis (PCA) was utilized to interpret parameters, and a regression model was de veloped to predict NHL subtypes. The areas under the receiver operating ch aracteristic curve (AUC), sensitivity, and specificity were calculated.

Results

or Findings: Among the 42 patients, 35 were diagnosed with B cel l NHL and 7 with T cell NHL. Sixty parameters from du al-vessel SRUS image analysis were obtained for each case. PCA extracted six principal components accounting for 93.1% of the total variance. The reg ression model utilizing these components to distinguish between B-cell and T-cell lymphomas achieved an AUC of 0.927 (95% CI: 0.841-1.000), a sensitivity o f 85.7% (95% CI: 42.1%- 99.6%), and a specificity of 94.3% (95% CI: 80.8%-9 9.3%).

Conclusion

Dual-vessel SRUS imaging, in conjunction with quant itative analysis, could effectively differentiate between B -cell and T-cell NHL, offering a non-invasive diagnostic alternative.

Limitations

First, due to the relatively low prevalence of lymp homa in the general population, the sample size is relatively s mall. Second, the methodology of microlymphatic SRUS imaging is confi ned primarily to intact lymph nodes. In cases of 'bulky mass' lymphomas, wh ere there is extensive disruption of the lymph node architecture, as well as in NHL presenting within the trunk region, this approach might not be approp riate. Funding for this study: No. Ethics committee - additional information: Ruijin Hospital Clinical Research Center Shanghai Jiaotong University, School of Medi cine, Ethic No. 20240116021828990. Author Disclosures: Yijie Dong: Nothing to disclose 16:30-17:30 Research Stage 2 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 605 Artificial intelligence in cardiovascular imaging Moderator T. Leiner; Rochester, MN/US ([email protected]) Author Disclosures: Tim Leiner: Research Grant/Support: Philips Healthc are; Other: Editor-in-Chief, Journal of Cardiovascular Magnetic Resonance (JCMR) CT Deep learning AI quantified fibrosis predicts pr ognosis in Pulmonary Hypertension associated with Chronic Lung Disease *K. Dwivedi*, M. Sharkey, S. Alabed, A. Maiter, C. S. Johns, S. Rajaram, R. Condliffe, D. Kiely, A. J. Swift; Sheffield/UK ([email protected]) Purpose or Learning Objective: Pulmonary Hypertension associated with Chronic Obstructive Pulmonary Disease (PH-COPD) is a heterogenous condition, with a spectrum of predominantly emphyse ma and some overlapping fibrosis. All patients undergo CT, but it is not used for prognostication. The study aim is to investigate th e prognostic value of an AI model that quantifies the percentage of fibrosis on baseline CT, compared to radiological assessment.

Methods

or Background: PH-COPD patients with baseline CT between 2001- 19 were identified from the ASPIRE registry. A vali dated in-house PH specific deep-learning model was run and provided percentage of fibrosis by Abstract-based Programme 65 Wednesday quantifying ground glass change, ground glass with reticulation, and honeycombing. Scans were scored as none/mild/modera te/severe fibrosis by sub-specialist radiologists. Cases with mean pulmon ary arterial pressure ≥ 35 mmHg were classified as severe PH-COPD, and fibrosi s was grouped with a threshold of 3%. Scaled cox regression and Kaplan M eier survival analysis was performed.

Results

or Findings: 157 PH-CLD patients (113 severe PH-COPD) were included. AI quantified fibrosis % was a significan t predictor of mortality (HR 1.46, p<0.001) .There was a significant difference (p=0.001) in survival between patients with more and less than 3% fibrosi s. One and five-year survival was 84% and 35% respectively in those with <3% fibrosis and 63% and 18% respectively in those with ≥3% fibrosis. Radiologist scored mild (HR 2.05, p=0.36) and moderate (HR 2.82, p=0.045) fibro sis was a significant predictor, but not severe fibrosis. In severe PH-CO PD, radiological scoring was not a significant predictor at any level, but AI fi brosis% was a significant predictor (HR 1.37, p<0.001).

Conclusion

CT Deep learning AI model quantified fibrosis is pr ognostic in predicting survival and treatment response in PH-CO PD and provides additional value over radiological assessment in se vere PH-COPD.

Limitations

Single registry analysis, but imaging from 21 hospi tals. Funding for this study: Research conducted during post funded by UK National Institute for Health and Care Research Ethics committee - additional information: Ethical approval was granted by the Institutional Review Board and approved by the National Research Ethics Service (16/YH/0352). Author Disclosures: David Kiely: Consultant: Ferrer, MSD, Janssen, Unit ed Therapeutics, Acceleron Andrew J. Swift: Consultant: Janssen pharamceutical s Grant Recipient: Wellcome Trust, National Institute for Health and C are Research and Janssen pharamceuticals Ahmed Maiter: Nothing to disclose Chris S. Johns: Nothing to disclose Michael Sharkey: Grant Recipient: Wellcome Trust Samer Alabed: Advisory Board: Royal College of Radi ologists AI Working Group Grant Recipient: Wellcome Trust and National Institute for Health and Care Research Smitha Rajaram: Nothing to disclose Robin Condliffe: Consultant: Janssen pharamceutical s and MSD Krit Dwivedi: Advisory Board: Royal College of Radi ologists AI Working Group member Grant Recipient: Wellcome Trust and National Institute for Health and Care Research External Validation of a Deep Learning Cardiac Meta l Artifact Reduction Algorithm (DL-C-MAR) to reduce Metal Artifacts of T ranscatheter Aortic Valves in CT: a retrospective cohort and phantom st udy *I. H. T. Khargi*¹, M. Selles¹, N. Huber², J. Brown e², B. Kietselaer², T. Leiner², M. F. Boomsma¹; ¹Zwolle/NL, ²Rochester, MN/US ([email protected]) Purpose or Learning Objective: To assess the performance of a novel deep learning-based cardiac metal artifact reduction alg orithm (DL-C-MAR) in a retrospective comparison with unedited conventional computed tomography angiograms (CTAs) of transcatheter aortic valve imp lantation (TAVI) valves and phantom experiments.

Methods

or Background: DL-C-MAR was trained using multiple simulated metal implants and artifacts in 1000 CTAs. Performa nce of DL-C-MAR was quantitatively and qualitatively investigated in 50 TAVI patients and compared to unedited conventional CTAs. To quantitatively as sess image quality, noise, contrast-to-noise ratio (CNR), artifact index (AI), and artifact volume were calculated. Diameters of the valve struts were also measured. Images were qualitatively rated on overall image quality, exten t of metal artifacts and valve leaflet definition by two readers on a four-point s cale. Phantom experiments were conducted using four different size steel cyli nders. Diameters of the cylinders were measured by two readers and compared to their conventional counterparts and the ground truth. All images were visually screened for presence of hallucinations.

Results

or Findings: In the CTAs, DL-C-MAR resulted in a higher CNR (9.1±5.8 vs. 7.9±4.8), and lower noise (57.2±33.9 vs. 82.1±54.0), AI (53.0±36.0 vs. 75.7±55.9), and artifact volume (0.02±0.12mL vs. 0.06±0.42mL) compared to unedited conventional CTAs (all p<0.001). The st rut diameter also decreased after DL-C-MAR (1.52±0.26mm vs. 2.05±0.46mm, p=0.005). Initial

Results

from the qualitative analysis suggest incre ased valve leaflet definition and decreased metal artifact severity after DL-C-MA R. In the phantom scans, DL-C-MAR decreased cylinder diameter by 7-67% (p<0. 001), bringing them closer to the ground truth. No hallucinations were observed.

Conclusion

DL-C-MAR increases image quality and reduces metal artifacts in CTAs after TAVI implantation and does not seem to h allucinate on clinical or phantom images.

Limitations

This study did not include impact on clinical decis ion-making outcomes. Funding for this study: No funding was received for this study Ethics committee - additional information: This study was reviewed and approved as exempt with waived informed consent. Re ference no.: RPR - 2024-00000086 Author Disclosures: Mark Selles: Nothing to disclose Tim Leiner: Nothing to disclose Nathan Huber: Employee: Philips Healthcare Bas Kietselaer: Nothing to disclose Indira Hélène Theodora Khargi: Nothing to disclose Jacinta Browne: Nothing to disclose Martijn Franklin Boomsma: Nothing to disclose Diagnostic confidence in coronary stent evaluation using coronary CT angiography. Comparison of Deep Learning Reconstruc tion, Hybrid Iterative Reconstruction and Model Based Iterative Reconstruction *M. Finazzo*¹, M. M. Lagana², F. Graziano³, F. Pint o², C. Duranti¹, F. Finazzo¹; ¹Palermo/IT, ²Milan/IT, ³Monza/IT ([email protected]) Purpose or Learning Objective: The assessment of coronary stents using Coronary CT Angiography (CCTA) can be challenging. Deep Learning Reconstruction (DLR) is an innovative CT image reco nstruction method that reduces noise, enhancing image quality. This study aims to evaluate whether DLR improves diagnostic confidence in coronary sten t evaluation using CCTA images, compared to hybrid iterative reconstruction (HIR) and model-based iterative reconstruction (MBIR).

Methods

or Background: CCTA images of 20 patients with 35 stents were evaluated retrospectively using three reconstructio n methods: HIR, MBIR, and DLR. All examinations were conducted using a 320-ro w whole-heart CT scanner. The diagnostic confidence of the images ob tained with each reconstruction method was evaluated using a Likert score (1=non-diagnostic, 2=poor, 3=acceptable, 4=good, 5=excellent). Stents were divided into proximal and distal according to their location. Stents loca ted in the proximal and intermediate segments of the coronary arteries were considered proximal; stents situated in the distal segments of the main coronary arteries and side branches were considered distal. A cumulative liked mixed model was created in Rstudio version 4.3.1 to examine the differences across reconstruction

Methods

while accounting for the stent position, to explore its potential effect on diagnostic confidence. Post-hoc comparisons were conducted, and the p- values were adjusted using the Tukey method.

Results

or Findings: The reconstruction method had a significant impact, irrespective of stent position. Likert scores were significantly higher for DLR images compared to those reconstructed using HIR an d MBIR (p<0.001), with no significant difference between HIR and MBIR (p=0 .957).

Conclusion

DLR provided the best diagnostic confidence and sig nificantly enhanced the evaluation of coronary stents. As a fu rther development, Super Resolution DLR, a new reconstruction algorithm, cou ld improve spatial resolution, thereby increasing diagnostic confidenc e in coronary stents.

Limitations

Limited number of patients. Qualitative analysis only Funding for this study: No funding Ethics committee - additional information: It's a non-pharmacological retrospective observational study which have been a pproved by the local ethics committee Author Disclosures: Maria Marcella Lagana: Investigator: Canon Medical Systems Investigator Mario Finazzo: Nothing to disclose Francesca Pinto: Investigator: Canon Medical System s Clinical Application Specialist Francesca Graziano: Nothing to disclose Cristiana Duranti: Nothing to disclose Francesca Finazzo: Nothing to disclose Comparing the performance of Large Language Models for automatic CAD-RADS 2.0 classification from cardiac-CT reports *P. Arnold*, M. Russe, E. Kotter, M. T. Hagar; Frei burg/DE ([email protected]) Purpose or Learning Objective: The Coronary Artery Disease-Reporting and Data System (CAD-RADS) 2.0 offers standardized guid elines for interpreting coronary artery disease in cardiac computed tomogra phy (CT). Accurate and consistent CAD-RADS 2.0 scoring is crucial for comp rehensive disease characterization and clinical decision-making. This study investigates the capability of large language models (LLMs) to auton omously generate CAD- RADS 2.0 scores from cardiac CT reports.

Methods

or Background: A dataset of 200 synthetic cardiac CT reports was created to evaluate the performance of several stat e-of-the-art LLMs in generating CAD-RADS 2.0 scores via in-context learn ing. The tested models included GPT-3.5, GPT-4o, Mistral 7b, Mixtral 8x7b, LLama3 8b, LLama3 8b Abstract-based Programme 66 Wednesday with a 64k context length, and LLama3 70b. The gene rated scores from each model were compared to the ground truth, which was provided by an independent committee of two board-certified cardio thoracic radiologists.

Results

or Findings: The GPT-4o model and Llama3 70b achieved the highest accuracy in generating full CAD-RADS 2.0 sc ores including all modifiers, with a performance rate of 93% and 92.5% respectively, followed by Mixtral 8x7b with 78%. In contrast, less advanced L LMs, such as Mistral 7b and GPT-3.5 provided poor performance (16%). Llama3 8b demonstrated intermediate results, with an accuracy of 41.5%.

Conclusion

Advanced LLMs are capable of generating autonomousl y CAD- RADS 2.0 scores for cardiac CT reports with excelle nt accuracy, potentially enhancing both the efficiency and consistency of ca rdiac CT report evaluations. Open-source models not only deliver co mpetitive accuracy but also present the benefit of local hosting, mitigati ng concerns around data privacy.

Limitations

To ensure data privacy and avoid ethical concerns, this study was conducted using synthetically generated cardiac CT reports. Even though these were deemed indistinguishable from real patie nt reports, further research is needed to validate LLM performance in real-world settings. Funding for this study: Hans A. Krebs Medical Scientist Program (Uniklinikum Freiburg) German Research Foundation ( DFG) - SFB 1597 - 499552394 Ethics committee - additional information: None Author Disclosures: Muhammad Taha Hagar: Nothing to disclose Maximilian Russe: Nothing to disclose Philipp Arnold: Nothing to disclose Elmar Kotter: Nothing to disclose Multi-stage deep learning architecture for carotid artery segmentation and stenosis degree evaluation: a comparative study with DSA *Z. Zheng*, X. Cao, W. Liu; Shanghai/CN ([email protected]) Purpose or Learning Objective: HR-MRI provided a non-invasive and radiation-free method for assessing atherosclerosis , with strong advantages for vessel wall visualization. However, efficient segme ntation and stenosis degree evaluation remained a challenging dilemma that is b oth labor- and time- consuming and susceptible to interobserver variabil ity. Thus, a multi-stage deep learning architecture was developed to address above issues.

Methods

or Background: The method contained three modules: artery localization, automatic segmentation, and stenosis degree evaluation modules. The 422 scans were retrospectively collected from t wo tertiary hospitals between 2018 and 2023 with a training-validation se t (372 patients, 545 lesions) and an independent test set (50 patients, 96 lesions). An external validation set (26 patients, 42 lesions) was collec ted prospectively between 2023 and 2024. Subsequently, the artery segmentatio n and stenosis degree evaluation were compared against the ground truth, which was established by consensus among three radiologists and derived from diagnostic results obtained via DSA.

Results

or Findings: The results showed outstanding performance with hig h DSC, IOU, and low RVE, ASSD, and HD95. The concorda nce correlation coefficient (CCC) was 0.985(95% CI: 0.981-0.987), 0 .979(95% CI: 0.963- 0.984), and 0.963(95% CI: 0.944-0.992) for volumes of artery on all datasets. Stenosis degree was evaluated on the NASCET achieve d Acc of 0.8750, 0.8571, AUC of 0.89, 0.80, Sens of 0.8611, 0.9333, and Spec of 0.9167, 0.6667 on the independent test and external validat ion sets, respectively.

Conclusion

The method achieved no less accuracy than manual segmentation by physicians and maintained a high co nsistency with the DSA diagnostic criteria. In addition, by shortening dia gnostic time and minimizing inter-observer variability, it offered an efficient intelligent aid in clinical practice.

Limitations

The method performed in multi-stage may take up a l arge amount of computational resources and modifications to the architecture are required to optimize the inference speed. Funding for this study: This work has received funding from the National Natural Science Foundation of China (82402393, 8210 2132, 8237071280), the Science and Technology Commission of Shanghai Munic ipality (20S31904300, 22TS1400900, 23S31904100, 22ZR1409500) and the Grea ter Bay Area Institute of Precision Medicine (Guangzhou) (KCH231 0094). Ethics committee - additional information: All patients or their guardians gave informed consent to use their anonymized MRI i mages and clinical data for research purposes. Since all data were obtained in the course of daily work, the Ethics Committee waived the need for informed c onsent. Author Disclosures: Zhiji Zheng: Nothing to disclose Wanchen Liu: Nothing to disclose Xin Cao: Nothing to disclose AI-driven joint segmentation of myocardium, scar, a nd microvascular obstruction in bright-blood late gadolinium enhance ment cardiac magnetic resonance imaging *B. Durand*, V. De Villedon De Naide, T. Génisson, M. Stuber, A. Bustin, H. Cochet; Bordeaux/FR ([email protected]) Purpose or Learning Objective: develop and test an AI-driven deep learning model for joint segmentation of healthy myocardium, scar tissue, and microvascular obstruction (MVO) in cardiac MRI usin g bright-blood phase- sensitive inversion recovery (PSIR) imaging.

Methods

or Background: Current methods for scar and MVO quantification in PSIR imaging are manual or semi-automated, time-con suming, and prone to errors and variability. Using a nnUNET architecture , the model was trained on 50 PSIR exams with suspected ischemic heart disease and evaluated on a test set of 20 cases. Data augmentations were applied, a nd manual segmentations by radiologists were used for comparison. To maximi ze performance, a joint segmentation approach was employed, and both magnit ude and phase maps were used together.

Results

or Findings: The AI model demonstrated excellent performance despite only 50 exams in training, in segmenting he althy myocardium (median Dice score 0.96) and good results for scar segmenta tion (median Dice score 0.75). MVO detection was successful in 2 out of 3 c ases. Inference time was under 5 seconds per exam, and no false positives we re identified outside the myocardium.

Conclusion

AI-driven approach showed robust segmentation of my ocardium and scar tissue, with promising results in MVO dete ction. It could streamline clinical workflows for myocardial infarction assess ment by reducing time- counsuming manual segmentations.

Limitations

The model was trained on a small dataset from post- ischemic patients, which limits its generalizability to othe r cardiac conditions such as hypertrophic cardiomyopathy or infiltrative disease s, where scarring patterns differ. We plan to expand the training population t o improve the model's performance in non-ischemic cardiomyopathy. Funding for this study: This research was supported by funding from the French National Research Agency under grant agreeme nt ANR-22-CPJ2- 0009-01, and from the European Research Council (ER C) grant "SMHEART" under the European Union’s Horizon 2020 research an d innovation programme (grant agreement No101076351). Ethics committee - additional information: The study was approved by the Biomedical Research Ethics Committee and all partic ipants provided informed consent for participation. Author Disclosures: Victor De Villedon De Naide: Nothing to disclose Aurelien Bustin: Nothing to disclose Hubert Cochet: Nothing to disclose Thaïs Génisson: Nothing to disclose Baptiste Durand: Nothing to disclose Matthias Stuber: Nothing to disclose 16:30-17:30 Research Stage 3 Research Presentation Session: Physics in Medical Imaging RPS 613 MRI spinning for development and sustainability Moderator T. G. Maris; Iraklion/GR ([email protected]) Revisiting TE selection for T2-weighted spin-echo M RI of the prostate *S. J. Riederer*¹, R. Pabi¹, E. Borisch¹, A. Froemm ing¹, A. Kawashima², N. Takahashi¹; ¹Rochester, MN/US, ²Phoenix, AZ/US ([email protected]) Purpose or Learning Objective: To study if the long echo trains of fast-spin- echo (FSE) prostate T2-weighted imaging (T2-WI) cau se the optimum echo time (TE) for distinguishing normal peripheral zone (PZ) vs. malignancy to deviate from the optimum TE chosen based on standar d T2 decay.

Methods

or Background: All work was done at 3 Tesla. Experiments were done using a standard (NIST) phantom containing ten vials with known T2 relaxation times. Vials 5 (T2=133.3 msec) and 6 (96 .9 msec) were analyzed, Abstract-based Programme 67 Wednesday having values closest to literature-taken T2 measur ements in normal PZ (125 to 150 msec) and PZ malignancy and normal transitio n zone (TZ) (75 to 105 msec). The phantom was imaged using conventional sp in-echo (TE 97, 113, 129, 153 msec; scan time 20 min for each) and a cli nical T2-WI sequence (echo-train-length 21; TE-EFF 104, 114, 135, 145 ms ec; scan time 2:30 for each). 30 consecutive subjects with suspected prost ate cancer were imaged using the clinical T2-WI sequence at both TE-EFF 10 0 and TE-EFF 150.

Results

were compared visually for relative contras t of PZ to TZ and any suspected lesions.

Results

or Findings: Contrast-to-noise ratio (CNR) between Vials 5 and 6 for conventional spin-echo peaked at TE=125 msec, consi stent with theory. However, CNR between the vials for the long-ETL T2- WI sequence was highest at TE-EFF 145 msec, 10% higher vs. 104 msec . In 18 of 30 patient studies the TE-EFF 150 series had superior contrast vs. TE-EFF 100, inferior in 2/30, and equivalent in 10/30.

Conclusion

Fast-spin-echo acquisition in prostate T2-WI artifa ctually prolongs the apparent T2 relaxation, causing the optimum ech o times for distinguishing normal from malignant tissue to be higher than that predicted assuming tabulated T2 values. TE-EFF 150 consistently provid es improved contrast vs. TE-EFF 100 msec.

Limitations

Limited number of subjects Funding for this study: This work was funded by NIH. Ethics committee - additional information: Informed consent was provided by all human subjects. Author Disclosures: Adam Froemming: Nothing to disclose Stephen J. Riederer: Nothing to disclose Eric Borisch: Nothing to disclose Naoki Takahashi: Nothing to disclose Ronard Pabi: Nothing to disclose Akira Kawashima: Nothing to disclose Comparison Between Conventional and Compressed SENS E Sequences on MRI Brain in Paediatric Population *I. S. Shah*, P. C. P. Joshi, V. Jahanvi; Pune/IN ([email protected]) Purpose or Learning Objective: To compare the quality and image acquisition time between conventional and Compresse d SENSE sequences in brain magnetic resonance imaging (MRI) in paediatri c population.

Methods

or Background: Thirty children (below the age of 18 years) undergoing MRI brain were included in this study. I n addition to the routine sequences, one Compressed SENSE sequence was added. 2D - T1, T2, and FLAIR axial sequences were acquired for brain using conventional and Compressed SENSE techniques. One of each sequence w as acquired in 10 patients undergoing an MRI brain study on a 3T MRI using coil 32 channel coil for adults and pediatric 8ch head coil for neonates . Two consultant radiologists (with 35 years and 5 years experience in radiology) independently scored the image quality using the 5-point Likert scale based on resolution, visualization of anatomical regions, grey-white matter differentiati on, sharpness of the image and artefacts. The subjective criteria details for image quality as per the 5-point Likert scale were: non-diagnostic (1), poor (2), mo derate (3), good (4) and excellent (5).

Results

or Findings: The time reduction achieved with 2D T1 at 2 reducti on factor were 60 seconds(24%), with 2D T2 at reductio n factor of 2.2 66 seconds(47.83%) and with 2D FLAIR at reduction fact or 2 66 seconds(40%). Inter-rater agreement for overall diagnostic confid ence was rated higher for Compressed SENSE (k – 0.632) than conventional (k – 0.464). Nonsignificant statistical difference was found regarding image qu ality and image contrast ratio between both techniques.

Conclusion

Compressed SENSE has potential in reducing the imag e acquisition time without compromising the image qua lity and diagnostic confidence. Motion artefacts are also reduced with reduction in time with the use of Compressed SENSE sequence.

Limitations

A small sample size. Funding for this study: None Ethics committee - additional information: Institutional ethics committee approval was obtained Author Disclosures: Priscilla Col Priscilla Joshi: Nothing to disclose Vandana Jahanvi: Nothing to disclose Isha Sandip Shah: Nothing to disclose Comparison and optimization of deep learning enhanc ed 2D ATPw-CEST MRI at 1.5 Tesla and 3 Tesla: A clinically relevant phantom study *L. Wei*¹, A. Volk¹, S. Campana Tremblay², J. Poujo l³, S. Ammari¹, G. Garcia¹, C. Balleyguier¹, N. Lassau¹, F. Bidault¹; ¹Villejui f/FR, ²La Ciotat/FR, ³Buc/FR ([email protected]) Purpose or Learning Objective: The purpose was to compare the CEST effect at 1.5T and 3T on high resolution 2D images provided by a deep- learning algorithm. In addition, the effect of satu ration offset number reduction combined with B0 Mapping on MTRasym values was inve stigated, in order to minimize acquisition time for clinical use.

Methods

or Background: The phantom consisted of 12 tubes filled with BSA at pH 7 for 3 different physiological concentration s, at 37 °C. Acquisitions were performed on 1.5T (GE Artist) and 3T (GE Signa Prem ier) MR scanners. CEST data were acquired with 2D SSFSE using the AIR Reco n Deep Learning option (ARDL) for image reconstruction, CW saturation was used, with 61 offsets. SNR was compared with previously acquired data with out ARDL. B0 correction was performed by using the chemical shifts of Z-spe ctra minima, and by using the 2D B0 Mapping GRE sequence. MTRasym was compare d for different offset numbers ranging from 6 to 61 using B0 Map GR E.

Results

or Findings: 2D CEST SSFSE using ARDL had better SNR compared to the sequence without ARDL at 1.5T and 3 T. For the two B0 correction methods, MTRasym values were similar and increased with protein concentration at 1.5T and 3T. For ΔB00.5ppm, MTRasym were similar for 16 offsets or more.

Conclusion

This study provided a comparison of 2D CEST SSFSE a t 1.5T and 3T on a phantom carried out during the same ima ging session. The findings open up the prospect of high-resolution ti me efficient APTw-CEST clinical MRI at 1.5T.

Limitations

However, the results are preliminary, hence repeata bility studies and proof of concept in patients will be considered . Funding for this study: This material is based upon work supported by the ANRT with a CIFRE fellowship granted to Lecong Wei. Ethics committee - additional information: Not applicable Author Disclosures: Gabriel Garcia: Nothing to disclose François Bidault: Nothing to disclose Corinne Balleyguier: Nothing to disclose Lecong Wei: Employee: Olea Medical, La Ciotat, Fran ce Research/Grant Support: This material is based upon work supported by the ANRT with a CIFRE fellowship granted to Lecong Wei. Samy Ammari: Nothing to disclose Nathalie Lassau: Nothing to disclose Sophie Campana Tremblay: Employee: Olea Medical Julie Poujol: Employee: GE HealthCare Andreas Volk: Nothing to disclose Can we measure extreme brain iron content with Quan titative Susceptibility Mapping? *C. Birkl*¹, M. Panzer¹, C. Kames², A. Rauscher², B . Glodny¹, E. R. R. Gizewski¹, H. Zoller¹; ¹Innsbruck/AT, ²Van couver, BC/CA ([email protected]) Purpose or Learning Objective: Aceruloplasminemia (ACP) is a rare autosomal recessive disorder characterized by progr essive iron accumulation in multiple organs, including the brain, liver, and pancreas. Magnetic Resonance Imaging (MRI) is commonly used to detect iron overload, with Quantitative Susceptibility Mapping (QSM) emerging as a promising method for assessing brain iron levels. Despite its potential, QSM faces challenges such as susceptibility artifacts and a lack of standardi zation. This prospective study aimed to evaluate the performance of different QSM algorithms in measuring brain iron in patients with severe iron overload, c ompared to healthy controls.

Methods

or Background: QSM images were acquired using a 3D multi-echo gradient echo sequence in three patients with ACP a nd three healthy controls. We evaluated six QSM algorithms: (I) Fast Nonlinear Susceptibility Inversion (FANSI), (II) Improved Sparse Linear Equation and L east-Squares (iLSQR), (III) Morphology-Enabled Dipole Inversion (MEDI), ( IV) Streaking Artifact Reduction (STAR) QSM with Rapid Open-source Minimum Spanning Tree (ROMEO) phase unwrapping, (V) STAR QSM with Laplaci an phase unwrapping, and (VI) Multi-Echo Rapid Two-Step (MER TS) QSM. Regional susceptibility values were analyzed in the caudate nucleus, putamen, globus pallidus, and thalamus.

Results

or Findings: We observed significant variability in susceptibili ty values across the different algorithms for patients with ACP. Among the algorithms tested, only one showed consistently ele vated susceptibility values in the globus pallidus of ACP patients compared to healthy controls. Many susceptibility maps showed signal dropouts in brain regions with extreme iron overload. Abstract-based Programme 68 Wednesday

Conclusion

Our findings suggest that only a subset of QSM algo rithms reliably reflect extreme brain iron deposition. Add itionally, the study highlights that performing echo combination prior to phase unw rapping and background field removal may introduce artifacts, resulting in lower-than-expected susceptibility values due to signal dropouts.

Limitations

A limitation is the small sample size. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by the local ethics committee (number 1270/2021) Author Disclosures: Alexander Rauscher: Nothing to disclose Bernhard Glodny: Nothing to disclose Elke Ruth R Gizewski: Nothing to disclose Christoph Birkl: Nothing to disclose Marlene Panzer: Nothing to disclose Heinz Zoller: Nothing to disclose Christian Kames: Nothing to disclose MRI signal intensity comparison of high relaxivity vs standard gadolinium-based contrast agents: Concentration-dep endent effects across different MRI sequences and field strengths *L. Widmer*¹, S. Bhumiwat², F. Porões¹, J. M. M. Fr oehlich³, H. Thoeny¹; ¹Fribourg/CH, ²Phatum Wan/TH, ³Zurich/CH Purpose or Learning Objective: Gadopiclenol (Elucirem™) is a recent high- relaxivity macrocyclic gadolinium-based contrast ag ent (GBCA), with limited data on its detailed concentration dynamics, essent ial for optimizing its use in diverse clinical scenarios. This study compared sig nal intensity (SI) curves from three GBCAs across various MRI sequences, fiel d strengths, coils and concentrations.

Methods

or Background: Signal intensity of gadopiclenol, gadoteric acid an d gadobutrol vials were measured across 18 MRI sequen ces on 1.5T and 3T machines and 8 concentrations ranging from 0 to 25 mmol/L in an experimental in-vitro setting. Relationship between SI and concentrations were compared in SE, FSE, GRE and IR sequences.

Results

or Findings: Concentration had no linear correlation with the SI . At the same concentrations, gadopiclenol produced high er maximal SI than the other two contrast agents in half of the sequences (50%, 9/18). In most sequences (56%, 10/18), gadopiclenol had a left-shi fted curve maximum, reflecting higher SI at lower concentrations. Resul ts of identification of curves patterns by sequence type are still pending.

Conclusion

Signal intensity curve analysis helps optimize imag ing and injection parameters, though in-vivo application re quires considering vessel and tissue distribution. These findings suggest usi ng reduced dose of high- relaxivity agents compared to conventional GBCAs, s upporting sustainable radiology.

Limitations

Experimental Funding for this study: None Ethics committee - additional information: None Author Disclosures: Harriet Thoeny: Nothing to disclose Johannes Malte Maria Froehlich: Nothing to disclose Siwat Bhumiwat: Nothing to disclose Fabio Porões: Nothing to disclose Lucien Widmer: Nothing to disclose Deep Learning-Based Spatial Resolution Improving Al gorithm for MRI: Comparison of Capabilities for Scan Time Reduction and Image Quality Improvement with Conventional Protocol with and wit hout ZIP *D. Takenaka*, H. Nagata, T. Ueda, M. Nomura, T. Yo shikawa, Y. Ozawa, Y. Ohno; Toyoake/JP ([email protected]) Purpose or Learning Objective: Deep learning reconstruction (DLR) and zero fill interpolation (ZIP) technique have been clinic ally applied on routine clinical MRIs. Recently, deep learning-based spatial resolut ion improving algorithm (Precise IQ Engine: PIQE) is developed to transform MR data from low-spatial resolution data to high-spatial resolution data. Th e purpose of this study was to directly compare utilities of PIQE for scan time re duction and image quality improvement of MRIs as compared with DLR with and w ithout ZIP techniques.

Methods

or Background: 28 consecutive patients suspected with 17 brain tumors, 6 spinal diseases and 5 musculoskeletal dis eases were prospectively scanned with conventional MR (224-382×256-512matrix ) and new MR protocols (160-192×192-416matrix). Then, both MR pr otocol data were reconstructed by DLR with and without ZIP technique or PIQE techniques (total five MR data sets). Each standard protocol was dete rmined as conventional MR protocol reconstructed by DLR without ZIP techni que. To compare scan time reduction and image quality improvement among all protocols, mean examination time and signal-to-noise ratios (SNRs) were compared between standard protocol and others by Dunnett's test. To evaluate qualitative image quality improvement, overall image quality, artifac t and diagnostic confidence level were assessed by 5-point scales and compared between standard protocol and others by Steel's multiple comparison test.

Results

or Findings: Mean examination times of new MR protocols were significantly shorter than that of conventional pro tocols (p<0.05), although SNRs had no significant differences. As compared wi th standard protocol, overall image quality and artifact were significant ly improved by conventional protocol reconstructed by DLR with ZIP and new prot ocol with PIQE (p<0.05).

Conclusion

PIQE is equal to or more useful for reduce examinat ion time and image quality improvements as with DLR with and wit hout ZIP technique.

Limitations

LImited study number and no diagnostic performance evaluation Funding for this study: Canon Medical Systems Corporation Ethics committee - additional information: Fujita Health University Hospital Author Disclosures: Yoshiyuki Ozawa: Research/Grant Support: Grant-in-A id for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Smoking Rese arch Foundation Masahiko Nomura: Nothing to disclose Takahiro Ueda: Research/Grant Support: Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Daisuke Takenaka: Nothing to disclose Hiroyuki Nagata: Research/Grant Support: Grants-in- Aid for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Canon Medica l Systems Corporation Takeshi Yoshikawa: Nothing to disclose Yoshiharu Ohno: Research/Grant Support: Canon Medic al Systems Corporation Research/Grant Support: Smoking Researc h Foundation Power Grid Independent Low Field MRI *H-M. Klein*; Burbach/DE ([email protected]) Purpose or Learning Objective: Purpose: Develop a concept for grid independent, power savin g MRI operation using a permanent magnet, solar energy, and a generator sup ported battery system.

Methods

or Background: We installed a 0,4 T MRI system with an open design permanent magnet. Regenerative energy is pro duced with a 29,8 kWp solar array. To achieve grid independency, we insta lled an ´island solution´ using a 22 kWh LiFePO4 battery. For longer periods of power outage, and insufficient solar energy, a specially designed, di rect current (DC), high voltage diesel generator is used. This generator simulates the power profile of a solar array, and is connected to the solar power converte r, feeding the battery.

Results

or Findings: Annual energy uptake of the MRI was 7.022 kWh in 2023. RIS and PACS components consumed 4.959 kWh. H eating and air conditioning consumed 12.500 kWh. Total energy cons umption of the practice was 26.801 kWh. Total energy production was 30.930 kWh. Energy balance was positive with 4.129 kWh. Battery and DC generat or can provide power grid independent operation. Without grid and solar energ y, the practice has an energy consumption rate of max. 1,9 l gasoil/hour.

Conclusion

Grid independent, sustainable MRI operation is poss ible using permanent magnet technology, solar energy productio n, battery storage and a specially designed power generator.

Limitations

Only very few high quality low field MRI with perma nent magnet technology are available in the market. Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Hans-Martin Klein: Nothing to disclose Reduced Energy Consumption with Accelerated MRI Usi ng Deep- Learning Reconstruction: A Phantom Study *Y. Jung*, R. Alizadeh, M. Corwin, L. Hacein-Bey, A . M. Hernandez; Sacramento/US ([email protected]) Purpose or Learning Objective: To quantify image quality, potential cost savings, and greenhouse gas emission reductions in accelerated brain and prostate MRI exams using deep learning reconstructi on (DLR) and phantom imaging.

Methods

or Background: A data logger and current transformer sensor were installed upstream of the power distribution unit o n three 3T MRI platforms (GE, Siemens, United Imaging) to measure power cons umption at 1-second intervals. The ACR phantoms were scanned using the T2 FLAIR sequence from routine brain MRI and the T2-weighted sequence from routine prostate MRI protocols. Phantom scanning was performed using three scan times with different acceleration factors, and images were rec onstructed using conventional inverse Fourier transform (IFT) and DL R at three strength levels: Low, Medium, and High. Signal-to-noise ratio (SNR) and low-contrast Abstract-based Programme 69 Wednesday detectability (LCD) measurements were taken using t he ACR phantom. Total energy consumption was recorded for each acquisitio n.

Results

or Findings: Total energy consumption decreased monotonically wi th reduced scan time across all systems and protocols. SNR and LCD, averaged across all scan times, were generally higher for DL R compared to IFT, and SNR increased with increasing DLR strength. Reducin g the scan time by ~ 5 minutes with medium-strength DLR resulted in a ~65% reduction in energy consumption compared to a non-accelerated acquisiti on, while maintaining comparable SNR and LCD. Extrapolating these savings to all T2 FLAIR and T2-weighted sequences performed annually at our ins titution would result in estimated total savings of 15,944 USD and 74.1 MTCO 2e, equivalent to 16.6 gasoline-powered passenger vehicles driven for one year.

Conclusion

DLR-accelerated MRI exams provide substantial reduc tions in cost and greenhouse gas emissions without compromis ing image quality in phantom imaging experiments.

Limitations

Patient data would be required to assess the actual impact of DLR on image quality and energy savings in clinical practice. Funding for this study: RSNA Emerging Issues Environmental Impact and Sustainability grant Ethics committee - additional information: None Author Disclosures: Ramsey Alizadeh: Nothing to disclose Lotfi Hacein-Bey: Nothing to disclose Youngkyoo Jung: Nothing to disclose Michael Corwin: Nothing to disclose Andrew M Hernandez: Nothing to disclose 16:30-17:30 Research Stage 4 Research Presentation Session: Interventional Radiology RPS 609 Developments in vascular and neurovascular interventions Moderator V. Bérczi; Budapest/HU ([email protected]) Thrombolysis in basilar infarction (TIBI): A novel angiographic scale for evaluating mechanical thrombectomy in basilar arter y occlusion *M. E. Chevasco Hanze*, A. Lopez Rueda, A. Nuñez, E . Ripoll, V. Cuba, S. Aixut, L. Aja, M. A. De Miquel Miquel, O. Chirif e; L'Hospitalet de Llobregat/ES Purpose or Learning Objective: Acute Basilar Artery Occlusion (BAO) has the highest morbidity and mortality in posterior ci rculation strokes. Current reperfusion scores for mechanical thrombectomy (MT) are based on anterior circulation strokes (mTICI score). This study propo ses a basilar artery-specific reperfusion score, based on digital subtraction ang iography (DSA), to assess MT efficacy in acute BAO.

Methods

or Background: A retrospective analysis was conducted on a prospective database of acute BAO patients treated with MT within 24 hours of symptom onset at a stroke center from January 2014 to December 2023. Informed consent was obtained, and institutional re view board approval was granted. Clinical, procedural, and radiological dat a were collected. The Thrombolysis in Basilar Infarction (TIBI) score was developed by grading posterior circulation territories in DSA post-MT, u sing PC-ASPECTS as

Reference

(deducting 1 point for each occipital lob e/cerebellar lobe/thalamus or 2 points for pons/mesencephalon). Successful recana lization was defined as TIBI≥8 and assessed by modified Rankin Scale at 90 days.

Results

or Findings: Ninety-eight patients were included (median age 70, 56 men). Successful recanalization (mTICI2b/3) was ach ieved in 84.7% and TIBI≥8 in 71.4%. Good functional status (mRS≤3) was seen in 50% of patients, with better outcomes in mTICI 2b/3 and TIBI ≥8 groups (p <0.001 and p=0.002). After adjusting for age, NIHSS socre and Glasgow Co ma Sacle, TIBI≥8 was associated with good outcomes at 90 days (OR = 6.18 ; p =0.001), and TIBI≥7 was also linked to good outcomes (OR = 9.45; p <0.0 01).

Conclusion

The TIBI scale is a novel tool for evaluating MT ef ficacy in acute BAO. A TIBI≥8 should be the target for successful MT.

Limitations

unweighted TIBI (<25% were TIBI<7) Funding for this study: No Ethics committee - additional information: Observational study Author Disclosures: Antonio Lopez Rueda: Nothing to disclose Sonia Aixut: Nothing to disclose Victor Cuba: Nothing to disclose Miguel Emilio Chevasco Hanze: Nothing to disclose Enric Ripoll: Nothing to disclose Oscar Chirife: Nothing to disclose Ana Nuñez: Nothing to disclose Maria Angeles De Miquel Miquel: Nothing to disclose Lucía Aja: Nothing to disclose MTICI 2b-stopped or continued after first-pass: int erim results of the Rossetti registry for M1 occlusion *J. I. García García*¹, O. Chirife¹, P. Vega Valdés ², E. Gonzalez³, F. Delgado⁴, G. Dolz¹, A. López-Frías López-Jurado⁵, F. Aparici Robles⁶, A. Lopez Rueda¹; ¹Barcelona/ES, ²Oviedo/ES, ³Barcaldo/ES, ⁴Cordoba/ES, ⁵Madrid/ES, ⁶Valencia/ES ([email protected]) Purpose or Learning Objective: The purpose of this study is to analyze the predictive factors of safety and efficacy of additi onal mechanical thrombectomy in patients with acute ischemic stroke due to M1 oc clusion, who achieve mTICI2B recanalization after the first pass of endo vascular treatment.

Methods

or Background: We retrospectively analyzed patients with acute ischemic stroke due to M1 occlusion from the ROSSET TI registry who achieved mTICI 2b recanalization after the first pa ss of endovascular treatment. Patients were divided into two groups: t hose who stopped the procedure with a mTICI 2b result and those who cont inued treatment for a better angiographic outcome. Among those who contin ued, patients were further split into two subgroups: those with unchan ged results (mTICI 2b) and those with improved results (mTICI 2c/3). Demograph ic, clinical data, procedure details, and outcomes were compared acros s groups.

Results

or Findings: We included 300 patients with acute ischemic stroke M1 occlusion who achieved mTICI2b recanalization score after the first pass of endovascular treatment. 132 patients underwent no f urther passes (group 1), while 168 patients underwent additional passes, wit h 65 of them maintaining a final mTICI 2b score (group 2) and 103 achieving a final mTICI 2c-3 score (group 3). Group 3 exhibited a higher incidence of distal embolism to new territories compared to group 1 (7.8% vs. 0%; p < 0 .001). No significant differences in clinical outcomes were observed betw een the groups and subgroups.

Conclusion

Patients who achieved an improved mTICI score after the first recanalization attempt (mTICI 2c/3) had a higher in cidence of distal embolisms to new territories compared to patients with no fur ther passes. No significant differences in mRS scores at 3 months were observed between groups and subgroups.

Limitations

Inherent limitations of retrospective designs. Funding for this study: None. Ethics committee - additional information: ROSSETTI registry creation was approved by an ethics committee. Author Disclosures: Guillem Dolz: Nothing to disclose Pedro Vega Valdés: Nothing to disclose Eva Gonzalez: Nothing to disclose Fernando Aparici Robles: Nothing to disclose Alfonso López-Frías López-Jurado: Nothing to disclo se Juan Ignacio García García: Nothing to disclose Antonio Lopez Rueda: Nothing to disclose Fernando Delgado: Nothing to disclose Oscar Chirife: Nothing to disclose Comparison of DSA Morphology Parameters In Predicti ng Time To Recanalization of Internal Carotid Artery Saccular Aneurysms Treated With Primary Coiling: Does Neck Angle Matter? *H. Akkaya*, A. I. Soylu, F. Uzunkaya; Samsun/TR ([email protected]) Purpose or Learning Objective: Coil embolization is the most commonly used method of endovascular treatment of narrow-nec k saccular aneurysms. However, recanalization and subsequent aneurysm enl argement and rupture are common in aneurysms embolized only with coils. The aim of this study was to investigate which of the morphology findings dur ing the treatment of internal carotid artery (ICA) aneurysms treated with primary coiling is more successful in predicting the time to recanalization.

Methods

or Background: In this study, DSA images of 51 ICA aneurysms treated with coiling in our center between January 2016 and July 2024 were retrospectively analyzed. Dates of embolization and recanalization times were noted. The segment of the ICA in which the aneurysm s were located, the height of the aneurysm, the diameter of the neck, t he height/diameter ratios, whether there was a parent artery originating from the aneurysm, and the Abstract-based Programme 70 Wednesday angle of the neck of the aneurysm were noted. The r elationship between these parameters and recanalization times was examined.

Results

or Findings: The mean age of the patients was 59.6±13.9 years. 2 4 (47.1 %) patients had recanalization in follow-up e xaminations. The mean duration of recanalization was 12.73±1.7 months. Th e aneurysm neck angle was found to be higher in patients with recanalizat ion (p<0.001). A negative (inverse) moderate correlation was found between th e time of recanalization and aneurysm size/neck diameter ratio: aneurysm nec k angles (r=-0.425; r=- 0.537, respectively).

Conclusion

The angle of the neck of the saccular aneurysm at t he time of treatment is one of the angiographic morphology fin dings that are successful in predicting the time to recanalization.

Limitations

The study has some limitations. First of all, the s tudy was single- centered and the number of patients was small. Anot her limitation is that only aneurysms localized in the internal carotid artery are evaluated. Funding for this study: N/A Ethics committee - additional information: Ondokuz Mayıs University Ethics committee approval was obtained for the study . The requirement for informed consent from the patients was waived due to the ret rospective nature of the study. Author Disclosures: Ayşegül Idil Soylu: Nothing to disclose Fatih Uzunkaya: Nothing to disclose Hüseyin Akkaya: Nothing to disclose Radiation Dose Comparison in Endovascular Clot Retr ieval: General Anaesthetic versus Conscious Sedation Approaches *F. Taylor*, K. Sehgal, D. Carrion, M. Masterson, M . K. Badawy, L-A. Slater; Melbourne/AU ([email protected]) Purpose or Learning Objective: Endovascular clot retrieval (ECR) is a time- sensitive, adjunct treatment for large vessel occlu sion (LVO) strokes. We present the results of this single high-volume stro ke centre, to establish differences in radiation doses between ECR cases pe rformed under general anaesthetic (GA) and conscious sedation (CS).

Methods

or Background: All ECR cases in adult patients between October 2018 and June 2023 were included. Procedure records , patient characteristics, stroke outcome, and radiation dosimetry measures we re collected and analysed retrospectively. Summative data, i.e. medi an and interquartile range for overall dosimetry measures were described overa ll and by subgroup according to anaesthetic strategy. ANOVA was used t o compare radiation dose measures, clot location, stroke severity measu res and endovascular clot retrieval times between anaesthetic strategy.

Results

or Findings: Radiation doses were higher in patients receiving a GA; DAP 9,346 cGy.cm2, (4,996, 17,442) vs. 7,052 cGy.cm 2 (4,529, 10,926),

Reference

air kerma 627 mGy (325, 1,287) vs 507 (30 0, 789), total fluoroscopy time 34 minutes (19, 63) vs 31 (18, 47). This was a ssociated with a greater pre-reperfusion morbidity (NIHSS of 13 (7, 19) vs 1 1 (6, 17), more complex thrombus location (more tandem, less M2, more ICA, more basilar clots) and delayed presentation to hospital (227 minutes (101, 452) vs. 125 (66, 281), p <0.001).

Conclusion

We have demonstrated higher radiation doses under G A versus CS cases and some of the factors that may contribut e to this. These values serve as benchmark ranges for comparable centres pe rforming ECR. Larger multicentre analysis is required to establish more generalisable dose reference levels.

Limitations

Single centre cohort. Affected by the COVID pandemi c; departmental decision to preference GA at several p oints between March 2020-October 2021. TICI grading wasn't core lab adj udicated. Funding for this study: This research did not receive any specific grant fr om funding agencies in the public, commercial, or not- for-profit sectors. Ethics committee - additional information: All procedures performed in studies involving human participants were in accord ance with the ethical standards of the institutional and/or national rese arch committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. For this type of study formal consent is not required. Ethics approval was obtained from the local Human R esearch Ethics Committee (HREC), reference no. QA/91812/MonH-2022- 343316 and the need for informed consent was waived. Author Disclosures: Fergus Taylor: Nothing to disclose Mohamed Khaldoun Badawy: Nothing to disclose Daniel Carrion: Nothing to disclose Kunal Sehgal: Nothing to disclose Lee-Anne Slater: Nothing to disclose Maeve Masterson: Nothing to disclose Long-term outcome results after endovascular abdomi nal aneurysm repair with Zenith stent-graft *J. Reymen*, H. Mufty, A. Laenen, S. Houthoofd, G. Maleux; Leuven/BE Purpose or Learning Objective: To report the long-term outcome of patients presenting with an aortic or aortoiliac aneurysm tr eated with the Zenith Endoprosthesis.

Methods

or Background: A retrospective analysis of the collected data of 2 09 consecutive patients who underwent endovascular ane urysm repair (EVAR) with the Zenith Endoprosthesis (Cook Medical) betwe en January 1998 and December 2009 in an academic, tertiary care centre for aortic disease was performed. Patients’ imaging and clinical follow-up was performed until December 2023 to ensure a follow-up time of at leas t 14 years. The primary end points were overall survival and reintervention -free survival. Secondary end points were endograft-related complications, en doleaks, and reinterventions.

Results

or Findings: Overall survival at 2-, 5-, 10-, and 15-year follow -up was 89.00%, 71.29%, 39.71% and 18.18%, respectively, wi th n=2 (1%) aneurysm- related deaths. Freedom from type I and III endolea k at 5- and 15-years was 92.82% and 86.12%, respectively; type I (n=37 ; 17. 7%) and type III (n=4 ; 2%) endoleaks occurred in the follow-up period until 8 years postoperatively. Reintervention-free survival was 83.08%, 75.62%, 68 .66% and 66.17% at 2-, 5-, 10-, and 15-year follow-up, respectively. Reint erventions occurred meanly in the 0- to 8-year follow-up period.

Conclusion

Endovascular aneurysm repair using the Zenith Endop rosthesis is effective and durable on long-term follow-up, wi th acceptably low endograft- related complications and reinterventions. The numb er of adverse events and reinterventions is minimal after 8 years of follow- up.

Limitations

It is a single-centre retrospective study, suscepti ble to selection bias. The study also has a relatively small patient cohort ( n=209) because not all patients eligible for EVAR were treated with th e Zenith Endoprosthesis. Finally, no comparison with other endograft devices was made. Funding for this study: No funding was received for this study. Ethics committee - additional information: The institutional ethics committee approved this retrospective analysis (MP024667). Author Disclosures: Annouschka Laenen: Nothing to disclose Sabrina Houthoofd: Nothing to disclose Geert Maleux: Nothing to disclose Hozan Mufty: Nothing to disclose Jessie Reymen: Nothing to disclose Assessing abdominal aortic aneurysm growth by using radiomics of different radii of perivascular adipose tissue afte r endovascular repair *R. Lv*, G. Hu, S. Zhang, Z. Zhang, Z. Wang; Beijin g/CN ([email protected]) Purpose or Learning Objective: To investigate the relationship between radiomic features of different radii of perivascula r adipose tissue (PVAT) and abdominal aortic aneurysm (AAA) growth after endova scular aneurysm repair (EVAR).

Methods

or Background: Patients with sub-renal AAA who underwent regular follow-up after EVAR from September 2014 to Septemb er 2024 were retrospectively collected. Two radiologists segment ed the aneurysm and different radii of PVAT (PVAT1-7: 5mm, 7.5mm, 10mm, 12.5mm, 15mm, 17.5mm, 20mm) to evaluate the aneurysm volume chang es during follow-up and calculate radiomic features of different PVAT r egions. Univariable and multivariable logistic regression was performed to construct models to evaluate the growth of AAA based on the radiomic features of PVAT with different radii (5mm-20mm for models 1-7). Calculate the area under the curve (AUC), sensitivity, specificity, and accuracy of the model s, and test the significance of the performance differences among the models.

Results

or Findings: A total of 79 patients (67±7 years, 82% men) were enrolled in this study, 20 of whom had a growing an eurysm. The AUC for Models1-7 are 0.72, 0.77, 0.71, 0.70, 0.69, 0.71, a nd 0.75, respectively. The specificities for Models1-7 are 95%, 95%, 97%, 97%, 93%, 92%, and 97%, respectively. DeLong test and McNemar test: p > 0.0 5 (no statistical significance).

Conclusion

The models constructed using the radiomic features of PVAT with different radii after EVAR showed no significant di fferences in performance for evaluating AAA growth. The models achieved an avera ge specificity of 95%, indicating their effectiveness in minimizing the mi sclassification of non-growing AAA cases as growth cases.

Limitations

The retrospective research highlights the need for prospective and long-term follow-up studies. There are also lim itations in the development of imaging segmentation techniques, which still req uire manual intervention for three-dimensional aneurysm segmentation. Funding for this study: This study has received funding by the National Hig h Level Hospital Clinical Research Funding (2022-PUMC H-B-068). Abstract-based Programme 71 Wednesday Ethics committee - additional information: Institutional review board of Peking Union Medical College Hospital. Author Disclosures: Ge Hu: Nothing to disclose Zhe Zhang: Nothing to disclose Zhiwei Wang: Nothing to disclose Shenbo Zhang: Nothing to disclose Rui Lv: Nothing to disclose Towards Clinical Magnetic Particle Imaging: Safety Measurements of Medical Implants in an Extracorporeally-Perfused Hu man Cadaver Model *F. Wegner*¹, T. Friedrich¹, P. Elfers¹, F. Kleefel dt², D. Peter², P. Gruschwitz², T. Kampf², P. Vogel², V. Hartung²; ¹Lübeck/DE, ²Wür zburg/DE ([email protected]) Purpose or Learning Objective: Magnetic Particle Imaging (MPI) is an emerging, tracer-based, 3D imaging modality on the way to clinical application. It offers high temporal resolution and operates wit hout the use of ionizing radiation, making it particularly advantageous for cardiovascular imaging and real-time interventional monitoring. However, the p otential heating of metallic medical devices within the magnetic fields of MPI s canners is a critical safety concern. This study aimed to assess the thermal beh avior of commercially available medical implants during MPI-scans in a hu man cadaver model.

Methods

or Background: A fiberoptic thermometer probe was introduced into the superficial femoral artery (SFA) of a human cad aver model via a 7 F sheath. A series of commercially available endovasc ular implants (including six stents, five coils, and one vascular plug) were the n positioned sequentially within the SFA. The thermometer probe was retracted sequentially to ensure direct contact with each implant. Additionally, a h ole was drilled in the femur, which contained a gamma nail, and the fiberoptic pr obe was externally inserted to establish contact with the nail. A custom-built human-sized MPI-scanner for interventional purpose was positioned around the ca davers’ thigh, and an MPI- sequence consisting of 40 pulses (4 pulses per seco nd) was applied, with the respective implant centrally located within the sca nner. Throughout the MPI- sequence, the cadavers’ thigh was perfused extracor poreally with a blood- equivalent fluid using an external flow pump.

Results

or Findings: The gamma nail exhibited a temperature increase of 0.04 K during the MPI-sequence, while no detectable heating was observed in any of the endovascular devices tested.

Conclusion

Commonly used medical implants do not heat up signi ficantly in a human-sized MPI-scanner under realistic conditions.

Limitations

Only a limited number of commercial devices was tes ted in this work. Funding for this study: N/A Ethics committee - additional information: Protocol Number 20220413 01 Author Disclosures: Franz Wegner: Nothing to disclose Viktor Hartung: Nothing to disclose Florian Kleefeldt: Nothing to disclose Philipp Gruschwitz: Nothing to disclose Thomas Friedrich: Nothing to disclose Dominik Peter: Nothing to disclose Patrick Elfers: Nothing to disclose Patrick Vogel: Nothing to disclose Thomas Kampf: Nothing to disclose Electroporation with local or systemic bleomycin fo r the treatment of vascular malformations: early results of a prospect ive study *N. Papalexis*, G. Peta, M. Di Carlo, S. Quarchioni , L. Campanacci, M. Carta, M. Miceli, G. Facchini; Bologna/IT ([email protected]) Purpose or Learning Objective: Purpose: To evaluate the safety and efficacy of electrochemotherapy with bleomycin for the treat ment of soft tissue vascular malformations.

Methods

or Background: Materials and Methods: This study analyzes the early results of a prospective study “BESVAM”, desi gned to prospectively evaluate the safety and efficacy of electrochemothe rapy for vascular malformations. 18 patients were enrolled from Febru ary 2023 to July 2024. Bleomycin was injected intralesionally for low-flow vascular malformations or systematically for high-flow vascular malformations . The primary goal was pain control, measured in VAS score at 3,6, and 12 month s follow-up. The secondary goal was the size reduction of the lesion and variations in the QLQ questionnaire.

Results

or Findings: Results: Twelve patients received bleomycin intravenously and six patients intralesional. Basel ine VAS scores averaged 6.9 (SD 2.1), decreasing to 3,1 (SD 3.2) at three month s. Further reduction was observed at 6 and 12 months with scores of 2.1 (SD 2.0) and 1.6 (SD 2.0) respectively. Size was reduced from a mean of 155.1 cm3 (range 56.7 to 515.3 cm3) pre-treatment to a mean of 122.5 cm3 (range 42 .2 to 438.9 cm3) at the 6- month follow-up (p<0.05). (12 months) Twelve patien ts discontinued pain relief therapy. Ten patients experienced skin discoloratio n at the site of insertion of the needles.

Conclusion

Conclusion: The preliminary results are promising, suggesting that electroporation with local or systemic bleomyc in could be a safe and effective tool for the management of vascular malfo rmations.

Limitations

Small sample size, lack of control group Funding for this study: None Ethics committee - additional information: Prospective study approved by the local ethcis committee of Emilia Romagna, Italy . Author Disclosures: Marco Miceli: Nothing to disclose Michela Carta: Nothing to disclose Laura Campanacci: Nothing to disclose Giancarlo Facchini: Nothing to disclose Giuliano Peta: Nothing to disclose Simone Quarchioni: Nothing to disclose Maddalena Di Carlo: Nothing to disclose Nicolas Papalexis: Nothing to disclose 72 Thursday, February 27 Thursday Abstract-based Programme 73 08:00-09:30 Research Stage 1 Research Presentation Session: Abdominal and Gastrointestinal RPS 701 Imaging of the intestines with focus on Crohn's disease Moderator S. A. Taylor; London/UK Author Disclosures: Stuart A. Taylor: Advisory Board: aztrazeneca; Gran t Recipient: takeda; Share Holder: Motilent Improvement of diagnostic performance in low-dose C T enterography: the impact of an artificial intelligence iterative reconstruction algorithm R. Guo¹, W. Zhou², G. Zhang², *T. Wang*², P. Hu¹, Q . Liang¹, P. Rong¹; ¹Changsha/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: To investigate the clinical value of artificial intelligence iterative reconstruction (AIIR) in opt imizing image spatial resolution and diagnostic performance of low-dose CT enterogra phy (CTE) for patients with Crohn's disease (CD), compared with the routin e hybrid iterative reconstruction (HIR).

Methods

or Background: Forty patients with suspected CD were prospectively enrolled to receive low-dose CTE (80k Vp, ref 120mAs) with ilecolonoscopy-guided biopsy as the reference. Imag es were reconstructed using the AIIR and HIR. Diagnosis of CD was made on a per-segment level. The diagnostic confidence was scored with a five-po int scale (1=insufficient, 5=definitely confirmed). Signal-to-noise ratio (SNR ), contrast-to-noise ratio (CNR), and edge rise slope (ERS) of the bowel wall were measured. Diagnostic image quality, including conspicuity of bowel wall enhancement, thickness, luminal narrowing, comb signs, and creep ing fat signs, was evaluated with a five-point scale (1=poor, 5=excell ent).

Results

or Findings: There were 77 bowel segments confirmed with CD, of which 57 were detected on HIR images and 71 on AIIR images. AIIR images showed 124.3% higher SNR, 136.3% higher CNR, 68.4% higher ERS, and more conspicuous diagnostic imaging features (all p <0.001), indicating higher image spatial resolution than HIR. With higher spat ial resolution, AIIR images showed better diagnostic performance for CD detecti on than HIR images (sensitivity: 88.7% vs 71.2%; accuracy: 92.9% vs 87 .9%; specificity: 94.5% vs 94.5%; false-positive-rate: 13.4% vs 16.2%; false-n egative-rate: 4.5% vs 10.9%). The diagnostic confidence was significantly improved by AIIR (4.5±0.6 vs 3.4±0.6, p<0.001).

Conclusion

AIIR improved the image spatial resolution of low-d ose CTE and thus delivered higher diagnostic confidence and bet ter diagnostic performance than HIR. Low-dose CTE with AIIR provides excellent image quality and reliable CD detection, making it a feasible option for follow-up examinations for CD patients.

Limitations

A single-centre study. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study received approval from the Institutional Review Board (R20230019). Author Disclosures: Pengfei Rong: Nothing to disclose Tiantian Wang: Nothing to disclose Wanhui Zhou: Nothing to disclose Guozhi Zhang: Nothing to disclose Rui Guo: Nothing to disclose Qi Liang: Nothing to disclose Pengzhi Hu: Nothing to disclose Comparison of Conventional vs. Abbreviated MR Enter ography in Crohn's Disease: Assessment of Inter-Radiologist Ag reement for Categorizing Disease Activity and Complications *J. R. Rimola Gibert*¹, C. Saavedra¹, N. Capozzi², I. De Kock³, A. R. Radmard⁴, M. Scharitzer⁵, M. C. Masamunt¹, I. Ordás¹, J. Dillman ⁶; ¹Barcelona/ES, ²Bologna/IT, ³Ghent/BE, ⁴Tehran/IR, ⁵Vienna/AT, ⁶Cincinnati, OH/US ([email protected]) Purpose or Learning Objective: Crohn's disease (CD) often requires multiple imaging evaluations. An abbreviated MR enterography (aMRE) protocol, without IV contrast, could improve patient experien ce and reduce costs. This study aims to compare inter-observer agreement for detecting CD activity and complications using conventional MRE protocol (cMRE ) versus an abbreviated protocol (aMRE).

Methods

or Background: Ten radiologists from six countries independently reviewed cMRE and aMRE exams from 80 CD patients, w ith imaging assessments separated by at least one month. The ex ams included both pre- and post-treatment images of CD patients. Interobse rver agreement, Fleiss' Kappa statistics, and Gwet's concordance 1 (AC1), w hich corrects the prevalence dependence in categorizations, were calc ulated at the bowel segment and patient levels for the presence of diff erent findings indicative of active inflammation and complications.

Results

or Findings: Overall (n=80), the concordance for detecting disea se activity was high and comparable between aMRE and c MRE: agreement was 0.83 (0.80-0.86) for aMRE vs. 0.84 (0.82-0.87) for cMRE; Kappa values were 0.41 (0.29-0.53) vs. 0.36 (0.23-0.49); and AC1 valu es were 0.76 (0.64-0.88) vs. 0.79 (0.69-0.89). Agreement for detecting activ e disease was similar across the small bowel and colon, as well as between pre-t reatment (n=51) and post- treatment (n=29) MREs. For detecting strictures, ag reement was 0.76 (0.73- 0.78) for aMRE vs. 0.72 (0.68-0.75) for cMRE; Kappa values were 0.35 (0.24- 0.46) vs. 0.34 (0.24-0.43); and AC1 values were 0.6 1 (0.47-0.74) vs. 0.50 (0.35-0.65). For penetrating complications, agreeme nt was 0.81 (0.79-0.84) for aMRE vs. 0.85 (0.82-0.87) for cMRE; Kappa values we re 0.47 (0.34-0.59) vs. 0.57 (0.45-0.69); and AC1 values were 0.71 (0.56-0. 86) vs. 0.76 (0.63-0.90).

Conclusion

The interobserver agreement for detecting active CD and related complications using an aMRE protocol was comparable to that of the cMRE protocol that supports the adoption of abbreviated MRE protocols.

Limitations

None Funding for this study: None Ethics committee - additional information: Local ethics committee approved the study with the code HCB/2021/0629 Author Disclosures: Maria Carme Masamunt: Nothing to disclose Carolina Saavedra: Nothing to disclose Ingrid Ordás: Nothing to disclose Isabelle De Kock: Nothing to disclose Nunzia Capozzi: Nothing to disclose Jordi Rimola Rimola Gibert: Advisory Board: Janssen Consultant: AstraZeneca, Janssen Alimentiv, Clario, Lument, Ori go Research/Grant Support: Abbvie Martina Scharitzer: Nothing to disclose Amir Reza Radmard: Nothing to disclose Jonathan Dillman: Nothing to disclose Inflammatory Burden in Crohn’s Disease: Insights fr om PET/MR Enterography *N. Bogveradze*, K. Kranz, T. Traub-Weidinger, C. P rimas, A. Macher-Beer, W. Reinisch, T. Mang, M. Hacker, M. Scharitzer; Vie nna/AT ([email protected]) Purpose or Learning Objective: To evaluate the global and regional inflammatory burden in patients with Crohn’s Diseas e (CD) using FDG PET/MR enterography (PET/MRE), in correlation with histopathological findings and relevant biomarkers.

Methods

or Background: Patients with CD undergoing PET/MRE and ileocolonoscopy were included in this retrospective study between 2016-2021. Eight intestinal segments were manually segmented, and the uncorrected total lesion glycolysis values (TLG) were summed to calcu late global CD activity score (GCDAS). GCDAS and highest SUVmax/ patient we re correlated with clinical biomarkers (fecal calprotectin [FC], serum C-reactive protein [CRP], peripheral blood leukocyte counts, Harvey-Bradshaw index [HBI]) to evaluate global inflammation. For assessing regional inflamm ation, SUVmax was correlated with histopathological disease activity. Comparisons were made using Spearman's coefficient and Wilcoxon-W tests.

Results

or Findings: In 41 patients (mean age, 40 years ±14 (SD), 26 men ), the highest segmental SUVmax correlated significant ly with FC (r= 0.443 p < 0.004) and CRP (r = 0.645, p < 0.001). The GCDAS co rrelated significantly with CRP (r = 0.498, p = 0.01) but not with FC (p = 0.49), leukocyte counts (p=0.56) or HBI (p = 0.518). SUVmax values signific antly correlated with grading of active inflammation in corresponding his topathological samples (r = 0.515, p < 0.001) and were higher in segments with severe histopathological inflammation (4.7, 95% CI: 3.8-5.6) compared to seg ments with moderate (3.3, 95% CI: 2.6-4.1) or without active inflammation (2. 0, 95% CI: 1.8-2.0).

Conclusion

SUVmax was found to be a reliable biomarker for ass essing global inflammatory burden and correlated well with histopathological segmental activity. Furthermore, SUVmax enabled gra ding of active inflammation. GCDAS showed good correlation with CR P, but not with FC and may therefore be less useful for quantifying global disease burden.

Limitations

Retrospective study design Funding for this study: N/A Ethics committee - additional information: Ek 1356/2023 Thursday Abstract-based Programme 74 Author Disclosures: Kerstin Kranz: Nothing to disclose Thomas Mang: Nothing to disclose Tatjana Traub-Weidinger: Nothing to disclose Christian Primas: Nothing to disclose Andrea Macher-Beer: Nothing to disclose Martina Scharitzer: Nothing to disclose Nino Bogveradze: Nothing to disclose Marcus Hacker: Nothing to disclose Walter Reinisch: Nothing to disclose Multiparametric and multi-score MRI evaluation of p ediatric Crohn’s disease: characterization of the perianal fistulizi ng phenotype *A. Valenti*, F. Maccioni, L. Busato, L. Bottino, A . Longhi, C. Catalano; Rome/IT ([email protected]) Purpose or Learning Objective: Perianal fistulas are severe complications of pediatric Crohn’s disease (CD), due to a high risk of demolitive surgery. MRI is the gold standard for scoring intestinal and perian al CD. The purpose of this study was to stage severity and activity of both le sions using MRI scores in pediatric CD.

Methods

or Background: A retrospective study was performed on 186 pediatric patients. Inclusion criteria were: proven CD, intestinal and perianal disease, complete MRI. Forty patients were finally included. MR Enterography (MRE) and high-resolution MRI (HRMRI) of the perian al region were used to analyze intestinal and perianal disease. The MEGS s core was applied to assess the severity of intestinal disease, the Park s’ classification and MAGNIFI-CD score to classify perianal disease and q uantify its activity. Correlations between location and activity of intes tinal and perianal lesions were investigated.

Results

or Findings: Simple perianal fistulas (Parks A and B) were found in 82.5% of patients, whereas complex fistulas (Parks C-E or a combination of them) in 17.5 %. Jejunal, ileal and colonic lesions were found in 22,5%, 80% and 35%, variably associated. Considering colorecta l lesions only, 70% were left-sided. Disease of the left colon was associate d with a more extensive disease, > 25 cm in length (p < 0.001), and a more severe activity, MEGS grade 3. Correlations were found between left-sided colonic lesions and severe fistulas, and between MAGNIFI-CD and MEGS scores bo th grade 3 (p<0.05).

Conclusion

The perianal fistulizing phenotype in pediatric pat ients is correlated with the severity of intestinal inflamma tion and with left-sided colonic disease.

Limitations

We used two differente MRI scanners to obtoin our d ata. Funding for this study: No funding. Ethics committee - additional information: We have received the approvation of the ethics commitee of our Hospital Author Disclosures: Alessandro Longhi: Nothing to disclose Alessandra Valenti: Nothing to disclose Lorenza Bottino: Nothing to disclose Carlo Catalano: Nothing to disclose Ludovica Busato: Nothing to disclose Francesca Maccioni: Nothing to disclose Magnetic Resonance Imaging biomarkers in the diagno sis of gastrointestinal acute Graft-versus-Host-Disease *L. Busato*, F. Maccioni, A. Valenti, L. Bottino, A . Iori, U. La Rocca, C. Catalano; Rome/IT ([email protected]) Purpose or Learning Objective: Acute gastrointestinal Graft-versus-Host disease (GI-a GVHD) is one of the most severe compl ications stem cell transplantation, occurring when the transplanted im mune cells attack the host's intestinal tissues. Aim of this study was to evalua te the efficacy of MRI biomarkers in the diagnosis and staging of acute ga strointestinal GI-aGVHD, currently based on clinical and endoscopic criteria only.

Methods

or Background: Thirty-five patients with clinical suspicion of GI- aGVHD were retrospectively and prospectively analys ed, 21/35 retrospectively, 14/35 prospectively, both separately and in conjunc tion. In these patients we investigated 15 MRI biomarkers suggestive of bowel inflammation and GVHD severity. A diagnostic score was tested, based on t he most relevant GVHD biomarkers: small bowel involvement, T2 and post-co ntrast bowel wall stratification, ascites , oedema of the retroperito neal and declivous tissues.

Results

or Findings: GI -GVHD was confirmed by biopsy in 13/35 patients (37%). Analysing the 6 key biomarkers (diagnostic s core), MRI showed 88.2% and sensitivity 100%specificity. In addition to the se siw biomarkers, wall stiffness and mesenteric oedema appeared useful for differentiating GVHD from non-GVHD patients. Most commonly affected segm ents were the proximal, middle and distal ileum (82.3%). Consider ing all the 15 MRI Biomarkers, MRI showed high predictive value on dis ease severity and mortality, superior to the clinical score.

Conclusion

MRI is a noninvasive and accurate tool for the diag nosis of GI- GVHD,which can provide crucial information and impl ement current clinical and endoscopic criteria. Limitations: The small number of patients, due to the low prevalence of this disease.

Limitations

The small number of patients, due to the low freque nce of the disease. Funding for this study: No funding Ethics committee - additional information: Our study has been approved by the ethics committee of our hospital Author Disclosures: Alessandra Valenti: Nothing to disclose Ursula La Rocca: Nothing to disclose Annapaola Iori: Nothing to disclose Lorenza Bottino: Nothing to disclose Carlo Catalano: Nothing to disclose Ludovica Busato: Nothing to disclose Francesca Maccioni: Nothing to disclose Reevaluating MR-Enterography: Value or Overuse? *R. Martín-Márquez*¹, E. Gutiérrez Dorta², D. J. L. Ruiz¹, J. Mesa¹; ¹Córdoba/ES, ²Ourense/ES ([email protected]) Purpose or Learning Objective: To evaluate the profitability of Magnetic Resonance Enterography (MRE) in different clinical contexts, determining if it is both clinically and cost-effective, optimizing reso urces and improving patient care. To establish a protocol for performing MRE in various clinical scenarios.

Methods

or Background: A retrospective study of 615 patients who underwent MRE at Reina Sofía Hospital (Córdoba, Spa in) over one year. Patients were divided into two groups: patients wit h known inflammatory bowel disease (IBD) and those without IBD history. Variab les included the reason for the examination, clinical unit, patient presentatio n, and findings from MRE, colonoscopy, and intestinal biopsy. A descriptive a nalysis was performed, and differences were assessed with chi-square or Studen t's t-tests (p<0.05). Diagnostic indices of MRE were compared to colonosc opy (sensitivity, specificity, positive predictive value (PPV), negat ive predictive value (NPV).

Results

or Findings: A total of 242 (39.3%) patients had a history of IB D, while 373(60.7%) did not. Pathological findings wer e seen in 68.2% of IBD patients versus 23.3% without IBD (p0.05). Among those with elevated fecal calpr otectin, MRE detected more findings in IBD patients (86.4% vs. 18.8%; p<0 .05). Diarrhea occurred in 8.7% of IBD patients and 37.3% of non-IBD patients, with MRE findings in 52.4% and 16.5%, respectively (p<0.05). Diagnostic indices of MRE: IBD group: Sensitivity 70.6%, Specificity 90.9%, PPV 92 .3%, NPV 66.7%. Non-IBD group: S 43.2%, E 89.5%, PPV 70.7%, NPV 73%.

Conclusion

Our study highlights the importance of MRE in IBD p atients and the need to optimize its use in non-IBD patients, e specially with negative colonoscopies.

Limitations

The main limitations of this study are its retrospe ctive design, which may lead to missing information, and the lack of long-term follow-up. Funding for this study: No Ethics committee - additional information: There is no additional information. Author Disclosures: Juan Mesa: Nothing to disclose Eduardo Gutiérrez Dorta: Nothing to disclose Rocío Martín-Márquez: Nothing to disclose Daniel José López Ruiz: Nothing to disclose The diagnostic yield of non-contrast versus contras t-enhanced magnetic resonance enterography (MRE) for small bowel (SB) e valuation in undiagnosed patients: Experience from four centres in the UK *S. Martin*, J. Pancholi, S. Liong; Manchester/UK ([email protected]) Purpose or Learning Objective: This service evaluation compares the diagnostic yield and utilisation trends of non-cont rast versus contrast-enhanced magnetic resonance enterography (MRE) for small bow el (SB) evaluation in a cohort of patients without a diagnosis of inflammat ory bowel disease or clear gastrointestinal symptom aetiology.

Methods

or Background: A retrospective review of all 1,012 MREs performed across four hospitals in Greater Manchester, United Kingdom, between 1 January and 31 December 2023, identified 208 undiag nosed patients. Among these, 92(44.2%) underwent non-contrast and 116(57. 8%) underwent contrast- enhanced MRE. Diagnostic yield was assessed by revi ewing MRE reports and correlating findings with colonoscopic histopatholo gy results (within 6 weeks of MRE) and faecal calprotectin levels (FC, within 12 weeks of MRE). Thursday Abstract-based Programme 75

Results

or Findings: SB abnormalities were observed in a similar minorit y of patients in both the contrast (14/116 [12.1%]) and non-contrast (5/92 [5.4%], p=0.15) groups. Of 45 patients with colonoscopies c ompleted within 6 weeks of MRE, 22 had terminal ileum (TI) biopsies available (non-contrast: n=8; contrast-enhanced: n=14). Compared with TI biopsies , contrast-enhanced MRE has sensitivity of 100% (95%CI 15.8-100%) and s pecificity of 83.3%(95%CI 51.5-97.9%), whereas non-contrast MRE h as sensitivity of 100%(95%CI 2.5-100%) and specificity of 100%(95%CI 59-100%). Contrast- enhanced MRE has positive predictive value (PPV) of 50%(95%CI 22-78%) and accuracy of 85.7(95%CI 57.2-98.2%). Non-contras t MRE has PPV 100%(95%CI 2.5-100%) and accuracy of 100%(95%CI 63. 1-100%). FC was available for 27 patients, but there was no clear r elationship between FC levels and MRE results.

Conclusion

Contrast enhancement did not significantly alter th e diagnostic yield of small bowel pathology in our cohort and is known to require longer acquisition and reporting times, impacting service capacity. Non-contrast MRE has high sensitivity and specificity for diagnosis of SB pathology.

Limitations

Retrospective design. Small number of patients with histopathology and FC available. Funding for this study: None. Ethics committee - additional information: Ethics approval was not required as this was an educational project and retrospectiv e service evaluation. Author Disclosures: Jay Pancholi: Nothing to disclose Sue Liong: Nothing to disclose Sarah Martin: Nothing to disclose Ultrasound is more effective than MRI for monitorin g the response to medical treatment in patients with active ileocolon ic Crohn's disease – a prospective blinded multicenter study J. Brodersen¹, *S. R. Rafaelsen*², M. Agerbæk Jue¹, T. Knudsen¹, J. Keldsen³, M. D. Jensen¹; ¹Esbjerg/DK, ²Vejle/DK, ³Odense/DK ([email protected]) Purpose or Learning Objective: The aim of this study was to evaluate intestinal ultrasound (IUS), magnetic resonance ima ging enterocolonography (MREC), panenteric capsule endoscopy (PCE) and faec al calprotectin (FC) for determining response to medical treatment in patien ts with ileocolonic CD

Methods

or Background: This prospective, blinded, multicentre study included patients with endoscopically active CD. Pa tients were scheduled for IC, MREC, IUS, PCE and FC before and 12 weeks after medical treatment. The vascularity within the affected bowel wall area s was assessed according to the Limberg score. The Simple Ultrasound Score for Crohn’s Disease (SUS- CD) was used for activity assessment. A > 50% reduc tion of the Simple Endoscopic Score for Crohn’s Disease (SES-CD) with IC defined treatment response as gold standard.

Results

or Findings: From 2018 to 2024, 50 patients completed the pre- a nd post-treatment evaluation with IC, and endoscopic r esponse was achieved in 25 (50.0%). PCE was omitted in 12 (24.0%) patients because of stricturing CD. All activity scores decreased in patients achieving endoscopic response: The Simple Ultrasound Score for Crohn’s Disease 2.2 vs. 6.1 (P < 0.001), Magnetic Resonance Index of Activity 29.0 vs. 37.1 (P = 0.05 ), SES-CD with PCE 3.1 vs. 12.8 (P < 0.001) and FC 115.3 vs. 1339.9 mg/kg (P < 0.001). The sensitivity and specificity of IUS, MREC, PCE and FC was 80.0% (95% CI 56.3-94.3) / 77.8% (52.4-93.6), 65.2% (42.7-83.6) / 87.0% (66.4- 97.2), 87.5% (61.7-98.4) / 86.7% (59.5-98.3) and 90.0% (68.3-98.8) / 86.4% (65 .1-97.1), respectively.

Conclusion

IUS, PCE and FC are equally effective for determini ng endoscopic response in patients with active CD. MRE C is insufficient for determining endoscopic response.

Limitations

First, the sample size is limited. Second, IC serve d as gold standard for treatment response, which may favour m odalities assessing mucosal inflammation (PCE and FC). Funding for this study: The study was initiated by the investigators withou t funding from medical imaging companies or the capsu le endoscope manufacturer. Ethics committee - additional information: The study was approved by the Local Ethics Committee of Southern Denmark (S-20170 188). All patients gave informed consent before participation. The study wa s registered: NCT03435016. Author Disclosures: Mie Agerbæk Jue: Nothing to disclose Michael D. Jensen: Nothing to disclose Jacob Brodersen: Nothing to disclose Jens Keldsen: Nothing to disclose Torben Knudsen: Nothing to disclose Sören R. Rafaelsen: Nothing to disclose Gut feels emotion: Psychological distress is associ ated with alterations of magnetic resonance enterography in patients with Crohn's disease *Y. Ke*, R. Zhang, H. Cai, Q. Zeng, S-T. Feng, Z. P eng, X. Li; Guangzhou/CN ([email protected]) Purpose or Learning Objective: Psychological distress may affect bowel disease activity in patients with Crohn's disease ( CD). However, limited studies have investigated its correlation with trans-/peri- intestinal alterations in CD. Therefore, we aimed to investigate the relationship between psychological distress and intestinal abnormalities identified by magnetic resonance enterography (MRE), and to explore their underlying association using blood neurotransmitters.

Methods

or Background: 105 CD patients and 46 healthy controls (HCs) were prospectively recruited. CD patients underwent MRE and provided blood samples for 19 serum neurotransmitters measurement. All participants completed State-Trait Anxiety Inventory (including STAI-Trait and State scores), Beck Depression Inventory (BDI), and Perce ived Stress Scale (PSS) questionnaires to assess psychological distress. Co rrelation analysis, multivariable logistic regression, and causal media tion analyses were employed to investigate the relationship between ps ychological distress and MRE features.

Results

or Findings: Psychological scores of CD patients, including STAI - Trait, PSS, and BDI scores, were significantly high er than HCs (all P<0.001). Among them, STAI-Trait score was significantly corr elated with stricture (r=0.505), mural T2WI hyperintensity (r=0.466), per ianal diseases (r=0.359), and perienteric effusion (r=0.340) (all P<0.05). Mu ltivariable logistic regression analysis indicated that STAI-Trait score significan tly influenced the odds of perienteric effusion (OR: 1.124; 95% CI: 1.007-1.25 5; P=0.036). In causal mediation analysis, a direct effect of STAI-Trait s core on perienteric effusion (P=0.04) was observed; STAI-Trait score and tryptop han had a combined effect on perienteric effusion (P=0.06), approachin g statistical significance. Negative correlation between tryptophan level and p erienteric effusion (r=- 0.220, P<0.05) was also found.

Conclusion

Psychological state is associated with MRE-detectab le intestinal morphological changes, and neurotransmitters may se rve as mediators in establishing this connection.

Limitations

This was a single-center study with small sample si ze. To enhance reliability and validity, future investigat ions should consider conducting multicenter studies with larger sample s izes. Funding for this study: This study was financially supported by National Natural Science Foundation of China (82070680, 8227 0693, 82271958, 82471948, and 82072002). Ethics committee - additional information: The study was approved by the institutional ethics review board of our hospital ( No. [2021]215-2). Author Disclosures: Shi-Ting Feng: Nothing to disclose Zhenpeng Peng: Nothing to disclose Yaoqi Ke: Nothing to disclose Qiaoling Zeng: Nothing to disclose Ruonan Zhang: Nothing to disclose Xuehua Li: Nothing to disclose Huasong Cai: Nothing to disclose MRI neurophenotype reflecting brain-gut interaction s to predict intestinal disease progression in patients with Crohn’s diseas e *R. Zhang*, X. Shen, Y. Wang, J. Lin, L. Huang, W. He, S-T. Feng, X. Li; Guangzhou/CN ([email protected]) Purpose or Learning Objective: There is considerable recent interest in the role of brain-gut axis in the pathogenesis and mani festations of Crohn’s disease (CD). We developed a multimodal neuroimagin g-based model to characterize the neurophenotype of CD patients and predict intestinal disease progression, using multi-omics data to demonstrate its validity.

Methods

or Background: This prospective study enrolled 109 CD patients who underwent baseline tests (including multimodal neuroimaging, psychological scales, MR enterography, ileocolonosc opy) and fecal/blood samples collection within one week. The neurophenot ype of patients with different intestinal inflammation levels was charac terized using a radiomics model, developed from 13 out of 13,870 neuroimaging features. This neurophenotype in predicting disease progression du ring follow-up was evaluated using Kaplan-Meier curves and Cox regress ion analysis. Multi-omics data (including fecal microbiome, fecal/blood metab olomics, intestinal/blood- brain-barrier permeability, and blood neurotransmit ter) were used to elucidate how this neurophenotype reflecting brain-gut intera ctions.

Results

or Findings: The model enabled accurate characterization of neurophenotypes in patients with different intestin al inflammation levels in training and test cohorts (AUC=0.824-0.842, both P0.05). Multi-omics analysis revealed that elevated intestinal inflamma tion was correlated with Thursday Abstract-based Programme 76 increased intestinal permeability and specific gut microbiota (e.g., Enterococcus) and metabolites (e.g., caproic acid), which collectively contributed to high-risk neurophenotype (all P<0.05 ). High-risk neurophenotype subsequently associated with intestinal disease pro gression by establishing correlations with six blood neurotransmitters (e.g. , tryptophan) (all P<0.05).

Conclusion

The neurophenotype varies among CD patients with di fferent intestinal inflammation levels and can predict inte stinal disease progression. Multi-omics data offer biological evidence to suppo rt its validity.

Limitations

This was a single-centre study, and the potential m echanisms underlying the brain-gut axis in our study have yet to be validated. Funding for this study: This study was financially supported by National Natural Science Foundation of China (82070680, 8227 0693, 82271958, 82072002, 82170537, and 82222010), Guangdong Basic and Applied Basic Research Foundation (2023B1515020070 and 2023A15150 11097), 2023 SKY Imaging Research Fund of the Chinese International Medical Foundation (Z- 2014-07-2301), and National Key R&D Program of Chin a (2023YFC2507300). Ethics committee - additional information: The study was approved by the institutional ethics review board of The First Affi liated Hospital of Sun Yat-sen University (No. [2021]215-2) Author Disclosures: Shi-Ting Feng: Nothing to disclose Li Huang: Nothing to disclose Weitao He: Nothing to disclose Ruonan Zhang: Nothing to disclose Xuehua Li: Nothing to disclose Yangdi Wang: Nothing to disclose Xiaodi Shen: Nothing to disclose Jinjiang Lin: Nothing to disclose First Findings from the BIPOCUS Study: Progress and Educational Impact on Ultrasound Training in the Practical Year *E. Höhne*¹, V. Schäfer², S. Petzinna², A. Wittek², J. Gotta¹, P. Reschke¹, F. Recker²; ¹Frankfurt/DE, ²Bonn/DE Purpose or Learning Objective: Point-of-care ultrasound (POCUS) is increasingly important in clinical settings, leadin g to a growing demand for comprehensive ultrasound training in medical educat ion. This study marks the University of Bonn's first attempt to integrate ult rasound courses and handheld devices into the regular curriculum for final-year medical students and assess their utilization.

Methods

or Background: Forty students in their practical year received a handheld ultrasound device for four months and were invited to participate in eight optional ultrasound courses, where they acqui red and rated images using a developed rating system. At the end of the tertia l, they could complete a voluntary survey on equipment usage.

Results

or Findings: Participation in the optional ultrasound courses wa s positive, with the Introduction and FAST module att racting the most participants (29). Lung images received the highest average rating (18.82 out of possible 23 points, SD ± 4.30), while aorta and vena cava images scored lowest (16.62, SD ± 1.55). The overall mean score f or all images was 17.47 (SD ± 2.74). Only 21 students responded to the survey, and 67% used the device independently four times or fewer during the tertial.

Conclusion

The study aimed to enhance improving students' ultr asound skills, but device usage was unexpectedly low, with most students using it only once a month or less. This raises concerns about re source justification, prompting future initiatives to focus on technical improvements, better login data access, and closer monitoring of usage and pro gress to emphasize practical ultrasound training in medical education.

Limitations

This study's limitations include a small sample siz e, single- institution focus, incomplete usage data, and low r esponse rates, which hinder the generalizability and reliability of the finding s regarding ultrasound device integration into student routines. Funding for this study: None Ethics committee - additional information: The local ethics committee of the Uuniversity of Bonn approved the study (253/23-EP) . Author Disclosures: Agnes Wittek: Nothing to disclose Valentin Schäfer: Nothing to disclose Florian Recker: Nothing to disclose Philipp Reschke: Nothing to disclose Jennifer Gotta: Nothing to disclose Elena Höhne: Nothing to disclose Simon Petzinna: Nothing to disclose 08:00-09:30 Research Stage 2 Research Presentation Session: Cardiac RPS 703 New techniques and clinical applications of CMR Moderator K.-F. Kreitner; Mainz/DE ([email protected]) Flip angle mapping and “without-gadolinium” enhance ment: A new approach to acute myocardial infarction in magnetic resonance imaging *G. Lucchi*, G. D. Aquaro, A. Marcucci, M. Lombardo , L. Faggioni, R. Lencioni, D. Cioni, E. Neri; Pisa/IT Purpose or Learning Objective: To evaluate Flip Angle Mapping (FAM) as a non-contrast alternative to Late Gadolinium Enhance ment (LGE) for assessing ischemic core in acute myocardial infarction using balanced Steady-State Free Precession (bSSFP) MRI sequences.

Methods

or Background: This study included 11 patients suspected of Myocardial Infarction with Non-Obstructive Coronary Arteries (MINOCA) and eight healthy controls. Two bSSFP datasets were acq uired with flip angles of 15° and 60°, respectively. Signal intensity differe nces were mapped and compared to LGE images. Pathological areas were man ually segmented on both FAM and LGE images. A statistical analysis was conducted to evaluate the correlation and concordance of the two methods in estimating the pathological area.

Results

or Findings: The diagnosis of MINOCA was confirmed in five patients, while four others were diagnosed with myo carditis and two with Takotsubo syndrome. Comparison between FAM and LGE images showed a complete overlap of pathological areas. Linear regr ession analysis revealed a strong positive correlation between the extent of L GE and FAM abnormalities (r=0.99; p<0.001). Bland-Altman analysis confirmed good agreement between the two methods (mean difference: -0.3%; 95% limits of agreement: -4.3 to 3.6%).

Conclusion

These findings suggest that FAM could potentially r eplace LGE for acute myocardial damage assessment, dispensing with contrast agents. If validated in larger studies, FAM could represent a significant advancement in non-invasive cardiac imaging.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is a methodological proof of principle. Author Disclosures: Giacomo Lucchi: Nothing to disclose Emanuele Neri: Nothing to disclose Alessandro Marcucci: Nothing to disclose Giovanni Donato Aquaro: Nothing to disclose Riccardo Lencioni: Nothing to disclose Lorenzo Faggioni: Nothing to disclose Marilena Lombardo: Nothing to disclose Dania Cioni: Nothing to disclose One-shot black-blood late gadolinium enhancement im aging for rapid, motion-free, and diagnostically accurate scar imagi ng *V. De Villedon De Naide*¹, K. Narceau¹, B. Durand¹ , T. Küstner², M. Villegas-Martinez¹, P. Jais¹, M. Stuber³, H. Coc het¹, A. Bustin¹; ¹Bordeaux/FR, ²Tübingen/DE, ³Lausanne/CH ([email protected]) Purpose or Learning Objective: Multi-shot black-blood LGE imaging is increasingly being used to assess myocardial scars and overcome poor scar- blood contrast, often observed with conventional br ight-blood LGE imaging. However, this method is time-consuming, requires mu ltiple breath-holds, and is prone to residual motion artifacts. Here, we introd uce a one-shot black-blood LGE sequence combined with image denoising to provi de rapid, motion-free, and diagnostically accurate scar imaging.

Methods

or Background: The 2D black-blood ECG-triggered LGE sequence acquires multiple single-shot short-axis images per slice using a non-selective 180° inversion pulse, followed by a T1-rho preparat ion. A dummy heartbeat is added between shots, for magnetization-recovery. Si ngle-shot images are averaged to enhance quality. The proposed one-shot sequence eliminates dummy heartbeats and employs a patch-based low-rank denoising algorithm (PROST) to achieve image quality comparable to mult i-shot techniques. 19 Thursday Abstract-based Programme 77 patients with ischemic heart disease underwent 1.5T CMR (Siemens Area) using reference PSIR and five-shot black-blood LGE imaging 12min after gadolinium injection. One-shot images were retrospe ctively selected from multi-shot datasets and were PROST-denoised. A blin ded radiologist graded diagnostic confidence, documented eventual residual motion artefact and extracted scar volume and signal intensities (blood , scar, remote myocardium) using Circle CVI42 for the three datasets.

Results

or Findings: Acquisition times were in average 4min shorter for black- blood one-shot PROST compared to reference sequence s. No statistically significant differences were observed between black -blood multi-shot and one- shot PROST in signal intensities or in scar detecti on, while scar volume agreement was excellent. Diagnostic confidence was rated good or excellent in 95% of black-blood multi-shot and 89% of one-shot P ROST scans. No residual motion artefacts were found in black-blood one-shot PROST datasets.

Conclusion

Black-blood one-shot PROST provides rapid, motion-f ree, and diagnostically accurate scar imaging, offering a mo re efficient and patient- friendly solution.

Limitations

Prospecting testing is now warranted. Funding for this study: This research was supported by funding from the French National Research Agency under grant agreeme nt ANR-22-CPJ2- 0009-01, and from the European Research Council (ER C) grant "SMHEART" under the European Union’s Horizon 2020 research an d innovation programme (grant agreement No101076351). Ethics committee - additional information: The study was approved by the Biomedical Research Ethics Committee and all partic ipants provided informed consent for participation. Author Disclosures: Victor De Villedon De Naide: Nothing to disclose Aurelien Bustin: Nothing to disclose Hubert Cochet: Nothing to disclose Kalvin Narceau: Nothing to disclose Manuel Villegas-Martinez: Nothing to disclose Pierre Jais: Nothing to disclose Baptiste Durand: Nothing to disclose Matthias Stuber: Nothing to disclose Thomas Küstner: Nothing to disclose Incremental Value of Multiparametric Cardiac MRI fo r Non-invasive Identification of Significant Acute Cardiac Allogra ft Rejection: a Prospective and Biopsy-proven Study *P. Zhou*, Z. Dong, S. Zhao; Beijing/CN ([email protected]) Purpose or Learning Objective: Using endomyocardial biopsy as the

Reference

standard, this study aimed to 1) evaluate the association between cardiac MRI (CMR) multiparameters and significant a cute cardiac allograft rejection (SR), and 2) assess the incremental value of CMR multiparameters over conventional serum examinations for identifyin g SR in heart transplantation (HTx) recipients.

Methods

or Background: HTx recipients with endomyocardial biopsy and healthy controls were prospectively recruited for C MR assessment. CMR feature tracking (CMR-FT) was performed to evaluate the left ventricular (LV) global strain in all three directions. The last ser um examinations including N- terminal pro brain natriuretic peptide (NT-proBNP) before anti-rejection therapy were recorded. Participants were divided into 3 gro ups: control, SR (acute cellular rejection grade≥2R and/or antibody-mediated rejection [AMR] grade≥pAMR1), and NSR (non-SR).

Results

or Findings: Finally, thirty controls (43.3±13.6 years, 26 male) and 51 HTx recipients comprising 23 SRs (48.6±12.6 years, 24 male) and 28 NSRs (42.7±14.9 years, 16 male) were enrolled for analysis. Compared with NSRs, SRs showed elevated NT-proBNP (7797.0±7527.6pg/ml v s 3334.6±5935.3pg/ml, p<.001), worse LV global longit udinal strain (GLS) (- 9.7±3.1% vs -13.1±2.9%, p<.001), and increased native T1 (1384±80.1ms vs 1321±69.9ms, p<.001) and T2 values (50.9±2.7ms vs 45.7±4.3ms, p<.001). In multivariable analysis, LVGLS (OR=0.76, 95%CI, 0.59 to 0.98, p=.03) and T2 value (OR=1.35, 95%CI, 1.10 to 1.65, p=.01) were in dependently associated with SR after NT-proBNP adjustment. Furthermore, th e likelihood ratio test showed LVGLS (p=.002) and T2 value (p<.001) had inc remental value over NT-proBNP for identifying SR.

Conclusion

LV GLS and T2 value were independently associated w ith SR, providing incremental value for non-invasive identi fication of significant rejection in HTx recipients.

Limitations

Although a relatively small participant sample, thi s is a prospective and biopsy-proven study with comprehens ive cardiac examinations, including T1 and T2 mapping of CMR. Funding for this study: This study is supported by the National Key R&D Program of China (Nos. 2021YFF0501400 and 2021YFF05 01404) and the Key Project of National Natural Science Foundation of C hina (No. 81930044). Ethics committee - additional information: The ethics committee from Fuwai Hospital. Author Disclosures: Pengyu Zhou: Nothing to disclose Zhixiang Dong: Nothing to disclose Shihua Zhao: Nothing to disclose Multiparametric cardiac MRI for the detection of ch imeric antigen receptor T-cell therapy associated myocardial chang es *D. Kravchenko*¹, L. Bischoff¹, A. Isaak¹, T. Holde rried¹, T. S. Emrich², A. Varga-Szemes², N. Mesropyan¹, D. Kütting¹, J. A. Luetkens¹; ¹Bonn/DE, ²Charleston, SC/US ([email protected]) Purpose or Learning Objective: New chimeric antigen receptor (CAR)-T cell therapy has demonstrated advantages over traditiona l cancer therapies for treatment of highly refractory or relapsing hematol ogical malignancies. Unfortunately, there is a paucity of data regarding cardiotoxic cardiac MRI (CMR) findings of therapy associated cytokine relea se syndrome (CRS).

Methods

or Background: Consecutive patients were enrolled for CAR-T cell therapy and received a standard 1.5 T CMR examinati on consisting of functional cines, parametric mapping, late gadolini um enhancement (LGE), and featuring tracking strain, before therapy (base line), during acute cytokine release syndrome (CRS; as determined by treating on cologist and lab parameters), and at 6-month follow-up (mean 184±20 days). Data was compared using RM-ANOVA with Tukey’s posthoc test.

Results

or Findings: 29 patients were available for analysis (mean age 6 0±15 years, 23 males [79%]). The most common malignancy was diffuse large B-cell lymphoma (13 [45%]). CRS was observed at a median t ime of one day (IQR 1- 2 days) after CAR T-cell therapy and reached a medi an degree of 1 (IQR 1-2). One patient passed away due to non-cardiac related CRS. One patient developed therapy associated heart failure. No new instances of LGE were observed in any cases. There were no differences fr om baseline to CRS or to follow-up scans for left ventricular ejection fract ion (61±5 vs 60±6 vs 59±7%, p=0.39), T1 relaxation times (969±18 vs 988±26 vs 972±22 ms, p=0.11), T2 relaxation times (53.3±2.4 vs 54.0±3.0 vs 52.9±1.8 ms, p=0.36), global longitudinal strain (-16.3±2.2 vs -15.6±2.4 vs -14.9±2.8, p=0.18), global circumferential strain (-12.5±2.9 vs -13.0±2.7 vs -12.2±1.8, p=0.51), or global radial strain (28.3±12.5 vs 29.2±8.1 vs 30.5±6.0, p=0.67).

Conclusion

CAR T-cell therapy related low degree CRS does not produce significant myocardial changes on multiparametric C MR from baseline to acute CRS or follow-up.

Limitations

Small study size. Funding for this study: None Ethics committee - additional information: Ethikkommission der Medizinischen Fakultät Bonn Geb. 74, 4. OG Venusberg-Campus 1 53127 Bonn Author Disclosures: Julian Alexander Luetkens: Nothing to disclose Narine Mesropyan: Nothing to disclose Alexander Isaak: Nothing to disclose Leon Bischoff: Nothing to disclose Dmitrij Kravchenko: Speaker: Philips Daniel Kütting: Nothing to disclose Tilman Stephan Emrich: Grant Recipient: Siemens Tobias Holderried: Nothing to disclose Akos Varga-Szemes: Grant Recipient: Siemens Consult ant: Elucid Cardiovascular magnetic resonance–derived upper ven tricular septal scar can predict the prognosis of left bundle branc h area pacing *Y. Fan*, X. Zhu; Nan Jing/CN ([email protected]) Purpose or Learning Objective: As a novel technique, left bundle branch pacing (LBBP) can achieve excellent resynchronizati on in patients with left bundle branch block (LBBB). The study is to use car diovascular magnetic resonance (CMR) to evaluate myocardial scars of dif ferent segments to predict the prognosis of patients with LBBB.

Methods

or Background: Consecutive patients with LBBB, left ventricular ejection(LVEF)≤35% and who underwent CMR examination and successfu l LBBP were retrospectively enrolled. The myocardial scar of different segments is assessed by CMR. LVEF response was defined as a 15% increase in LVEF assessed by echocardiography at 6 months.

Results

or Findings: Among 68 patients who were included, and 52 patient s patients showed a favorable LVEF response . The res ponders had lower global, septal scar burden by CMR (P<0.001). The sc ar burden of AHA 8 is independently associated with the prognosis of LBBP (AUC: 0.877 [95% CI: Thursday Abstract-based Programme 78 0.782, 0.971]) and the linear equation was that ΔLVEF= -0.4161(scar burden) + 23.229 (r=-0.60,P＜0.001), indicating that each 1% increased in scar b urden, LVEF decreased by 0.4161%. Moreover, the patterns o f scar of AHA8 and the scar morphology are independent of the improvement of LVEF (P＞0.05).

Conclusion

The scar burden of AHA 8, as a common implantation area, can predict LVEF improvement. And we recommend to pay m ore attention to the extent of mycardial LGE rather than the patterns an d morphology.

Limitations

The limitations are that firstly, it is a retrospec tive single-center investigation and it may lead to an inclusion bias in that some patients who did not undergo CMR before surgery were not included in the examination. Secondly, the study only investigated patients with low LVEF, and further confirmation is needed on the relationship between the prognosis of patients with high LVEF. Funding for this study: Project supported by the Young Scientists Fund of t he National Natural Science Foundation of China (No.82 302163 ) Ethics committee - additional information: Given that the study was retrospective, the committee has been waived. Author Disclosures: Xiaomei Zhu: Nothing to disclose Yin Fan: Nothing to disclose Wideband myocardial T2 mapping with implantable car diac device: A preliminary evaluation in healthy volunteers at 1 .5 T *P. Gut*¹, D. Kim², H. Cochet³, F. Sacher³, P. Jais ³, M. Stuber¹, A. Bustin⁴; ¹Lausanne/CH, ²Northwestern/US, ³Pessac/FR, ⁴Bordeaux/FR ([email protected]) Purpose or Learning Objective: Myocardial T2 mapping allows assessment of myocardial inflammation and edema, but is impact ed by artefacts related to implantable cardioverter defibrillators (ICDs), lea ding to image artifacts and inacurate T2 values. This study aimed to integrate a wideband T2 preparation into a T2 mapping sequence and evaluate its perform ance against conventional T2 mapping in healthy subjects with an d without ICDs.

Methods

or Background: Three short-axis slices covering the heart at the basal, mid-ventricular, and apical levels were acqu ired in eight healthy volunteers (2 females, age: 26±6y) at 1.5T (MAGNETOM Aera, Siemens) during end-expiration in mid-diastole using both co nventional and wideband T2 mapping, with and without an ICD taped below the le ft clavicle (~10 cm from the heart). The T2 preparation module (duration=0, 27, 55ms) included two adiabatic hyperbolic secant refocusing pulses of 1. 6 kHz (conventional) and 5.0 kHz (wideband). Common parameters included: res olution=1.4mmx1.4mm, slice thickness=8mm, FA=15°, GRAPPA x2, partial Fou rier phase 6/8, TE/TR=2.09/3.95ms, readout bandwidth=1221Hz/pixel, FOV=360mmx287mm, and gradient recalled-echo (GRE) readout. T2 maps w ere reconstructed using a 2-parameter (M0 and T2) fitting model. Myocardial T2 values were manually extracted in 16 heart segments. Statistical analyse s were performed using repeated measures ANOVA and Bonferroni correction.

Results

or Findings: Without ICD, T2 values were not significantly diffe rent between conventional (mean: mean: 43.5, SD: 2.21) a nd wideband sequences (mean: 44.0, SD: 2.15) (P=0.111). With ICD, convent ional T2 values significantly decreased (mean: 36.4, SD: 5.91) (P<0 .01), especially in apical anterior, apical inferior, mid-ventricular anterior , and basal anterior segments. Wideband T2 values remained unchanged (mean: 42.7, SD: 1.91) (P=0.377).

Conclusion

Wideband T2 mapping effectively reduces ICD-related artifacts, providing more accurate myocardial T2 values than c onventional T2 mapping.

Limitations

The study was conducted solely on healthy individua ls. Validation in clinical populations is warrented. Funding for this study: This research was supported by funding from the French National Research Agency under grant agreeme nts Equipex MUSIC ANR-11-EQPX-0030, ANR-22-CPJ2-0009-01, ANR-21-CE17- 0034-01, and Programme d’Investissements d’Avenir ANR-10-IAHU04- LIRYC, and from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement 10 1076351). Ethics committee - additional information: The study was approved by the Biomedical Research Ethics Committee and all partic ipants provided informed consent for participation. Author Disclosures: Pauline Gut: Nothing to disclose Aurelien Bustin: Nothing to disclose Hubert Cochet: Nothing to disclose Daniel Kim: Nothing to disclose Pierre Jais: Nothing to disclose Frederic Sacher: Nothing to disclose Matthias Stuber: Nothing to disclose Left ventricular remodelling index to predict ventr icular tachyarrhythmia in nonischemic dilated cardiomyopathy with ejection fraction <35% *X. Jia*, S. Zhao; Beijing/CN ([email protected]) Purpose or Learning Objective: Based on current guidelines, only a few dilated cardiomyopathy (DCM) patients with left ven tricular ejection fraction (LVEF) <35% receive appropriate implantable cardiov erter-defibrillator therapy, leading to increased medical costs and patient comp lications. We explored the predictive value of LV remodeling index (LVRI) for ventricular tachyarrhythmia (VTA) in nonischemic DCM with LVEF <35%.

Methods

or Background: In this retrospective single-center study, consecutive nonischemic DCM patients with LVEF <35% (n=271) who underwent cardiac magnetic resonance (CMR) imaging were followed up for VTA events, including sustained ventricular tachyca rdia, ventricular fibrillation/flutter, sudden cardiac death (SCD), a nd aborted SCD. The newly derived LVRI was defined as the cubic root of the L V end-diastolic volume divided by the maximal LV wall thickness. Competing risk regression analysis and Kaplan-Meier analysis were used to evaluate the association of LVRI with VTA.

Results

or Findings: During a median follow-up of 71 months (interquarti le range: 17–134 months), 35 (12.9%, mean age 46.7 yea rs, 27 males) participants reached VTA events. The presence of la te gadolinium enhancement (LGE) (62.9% vs. 60.2%, p=0.761) and LV EF (23.3±6 vs. 21.9±10.3, p=0. 197) were not significantly differe nt between the patients with and without VTA events. Kaplan-Meier curve analysis showed that participants with LVRI ≥7.5 were more likely to experience VTA (p<0.0001). In the multiple competing risk analysis, when heart transplantation and heart failure-related death were counted as competing risks, LV mass inde x (hazard ratio [HR], 0.983; 95% confidence interval [CI]: 0.968-0.999; p =0.033) and LVRI ≥7.5 (HR, 2.496; 95% CI: 1.213-5.138; p=0.013) were observed as the independent predictors of VTA after adjusting for age, sex and left bundle branch block.

Conclusion

In the cohort of patients with nonischemic DCM with LVEF <35%, CMR-assessed LVRI ≥7.5 was an independent predictor of VTA events.

Limitations

Not applicable. Funding for this study: Funding was received from the National Key R&D Program of China (No. 2021YFF0501400, 2021YFF050140 4); Key Project of National Natural Science Foundation of China (No. 8 1930044). Ethics committee - additional information: The study received institutional review board approval and written informed consent was obtained from all participants. Author Disclosures: Shihua Zhao: Nothing to disclose Xi Jia: Nothing to disclose The Effect of Obesity on Cardiac Structure and Func tion: A Magnetic Resonance Study of the Hamburg City Health Cohort *J. Erley*, D. G. Aydemir, K. Muellerleile, E. Cavu s, G. Adam, M. Meyer, E. Tahir; Hamburg/DE Purpose or Learning Objective: To analyze the effect of waist-to-hip ratio (WHR) and body mass index (BMI) on cardiac structur e and function using magnetic resonance imaging (CMR).

Methods

or Background: The Hamburg City Health Study (HCHS) is a population-based cohort study. Individuals between 45-74 years of age underwent 3T CMR. Subjects with cardiac diseases (e .g., coronary artery disease, myocardial infarction), and previous cardi ac interventions were excluded. Linear regression models were conducted, adjusted for age and sex.

Results

or Findings: 1671 subjects (41% female, mean age 64±8 years) wer e analyzed. Median WHR was 0.95 [interquartile range: 0.88; 1.01] and median BMI was 26.2kg/m² [23.8; 29.2]. Concerning BMI cut- off values, 44% of subjects were overweight (BMI 25-29.9kg/m²) and 20% obese (BMI ≥ 30kg/m²). According to the WHR, 81% of subjects were obese (W HR ≥0.85 in females and 0.90 in males). An increase in WHR was associat ed with a 5.2% [0.1-10.2] higher left ventricular (LV) ejection fraction (p=0 .044), and a 43.6g [27.6; 59.5] higher LV end-diastolic mass (EDM) (p<0.001), but l ower LV and right ventricular (RV) end-diastolic volumes (EDV) (LV: - 37.3ml [-57.4; -17.3], p<0.001; RV: -34.0ml [-56.0; -12.1], p=0.002) and e nd-systolic volumes (LV: - 18.7ml [-28.5; -8.8], p<0.001; RV: -16.0ml [-28.8;- 3.1], p=0.015), leading to lower stroke volumes (SV) (LV: -18.5ml [-32.2;-4.7] , p=0.008; RV: -18.8ml [- 33.6;-4.1], p=0.013). An increase in BMI was associ ated with a 1.9g [0.2; 2.2] higher LVEDM (p<0.001), higher EDV (LV: 0.5ml [0.2; 0.9], p=0.002; RV: 0.4ml [0.1; 0.7], p=0.047) and a 0.4ml [0.2; 0.7] higher LVSV (p<0.001).

Conclusion

An increase in WHR is associated with a higher left ventricular mass and lower volumes as a sign of concentric remo deling, while an increase in BMI is associated with ventricular dilatation.

Limitations

Analyses are not adjusted for other cardiovascular risk factors. Thursday Abstract-based Programme 79 Funding for this study: The HCHS is supported by the Innovative medicine initiative (IMI) under Grant No. 116074, by the Fon dation Leducq under Grant Number 16 CVD 03, by the euCanSHare Grant Agreement No. 825903- euCanSHare H2020 and the DFG under project Grant TH 1106/5-1; AA93/2-1. The DIFE provides the licence for the Food Frequenc y and Physical activity. Technical equipment is provided by SIEMENS accordin g to a contract for 12 years as well as by the Schiller AG on a loan basis for 6 years and by Topcon on a loan basis from 2017 until 2022. The Hamburg C ity Health Study is additionally supported by an unrestricted Grant (20 17–2022) by Bayer. Project- related analyses are supported by Amgen, Astra Zene ca, BASF, Deutsche Gesetzliche Unfallversicherung (DGUV), DKFZ, DZHK, Novartis, Seefried Stiftung and Unilever. The study is further support ed by donations from the “Förderverein zur Förderung der HCHS e.V.”, TEPE (2 014) and Boston Scientific (2016). A current list of the supporters is online available on www.uke.de/hchs. Ethics committee - additional information: The study was approved by the local ethics committee of the medical association i n Hamburg. Author Disclosures: Gerhard Adam: Nothing to disclose Ersin Cavus: Nothing to disclose Mathias Meyer: Nothing to disclose Jennifer Erley: Nothing to disclose Kai Muellerleile: Nothing to disclose Destina Gizem Aydemir: Nothing to disclose Enver Tahir: Nothing to disclose Lipomatous hypertrophy of the atrial septum in card iac magnetic resonance *A. Fortunati*, D. Fazzini, S. Papa, M. Alì, F. Dar vizeh, F. Secchi; Milan/IT ([email protected]) Purpose or Learning Objective: Lipomatous hypertrophy of the atrial septum is a benign anomaly, consisting in a fatty infiltra tion of the interatrial septum with sparing of the fossa ovalis and typically asso ciated with elderly. This retrospective study aim to define the prevalence of LHAS through cardiac magnetic resonance and its correlation with age or functional biventricular parameters.

Methods

or Background: A retrospective analysis of 621 patients who underwent CMR from March 2020 to March 2022 was per formed, with the following inclusion criteria: the presence of a 4-c hamber sequence and the presence of volume analysis. All images were review ed by a reader to evaluate the presence of LHAS. The atrial septum thickness w as measured, and the functional biventricular parameters retrieved from the clinical report. The statistical analysis was conducted using the Spearm an’s correlation test.

Results

or Findings: Among the 619 patients included in the study, 241 patients were found with LHAS and 150 showed lipoma tous deposition of the upper half of the atrial septum with a mean thickne ss of the atrial septum of 3 mm. A significant negative correlation was found be tween the degree of LHAS and left ventricle end-diastolic volume (r = -0.21, p<0.001) and systolic volume (r = -0.20, p<0.001). A significant negative correl ation was also recognized between the degree of LHAS and right ventricle end- diastolic volume (r = -0.25, p<0.001) and systolic volume (r = -0.18, p<0.001).

Conclusion

LHAS reached a 39% prevalence. Findings confirm a s ignificant negative correlation between LHAS and biventricular end-diastolic volumes according to a consecutive reduction of atrial and ventricular volumes to offset the volumetric increase of atrial septum and a prog ressive increase of LHAS disease prevalence with age.

Limitations

The study has some limitations due to its monocentr ic nature and the relatively small sample size. Funding for this study: None Ethics committee - additional information: The local Ethics Committee approved this retrospective study. Author Disclosures: Francesco Secchi: Nothing to disclose Sergio Papa: Nothing to disclose Deborah Fazzini: Nothing to disclose Marco Alì: Nothing to disclose Alice Fortunati: Nothing to disclose Fatemeh Darvizeh: Nothing to disclose Risk Stratification of Sudden Cardiac Death in Non- Ischemic Cardiomyopathy: Towards Arrhythmogenic Substrate As sessment in Cardiac MRI D. Zhou, M. Lu, *Y. Wang*; Beijing/CN Purpose or Learning Objective: Magnetic resonance imaging (MRI)-derived arrhythmogenic substrate is indicative of sudden ca rdiac death (SCD) in patients with non-ischemic cardiomyopathy (NICM). A key issue that needs to be addressed is what extent of T1 mapping metric co ntributes to the prognosis for SCD over late gadolinium enhancement (LGE).

Methods

or Background: A total of 837 NICM patients who underwent T1 mapping MRI were consecutively enrolled in this stu dy. The primary endpoint is a composite of SCD-related events, including SCD , appropriate implantable cardioverter-defibrillator shock and resuscitated c ardiac arrest.

Results

or Findings: Over a median follow-up of 58.3 months, 78 patients reached the primary endpoint, and 198 patients reac hed the secondary endpoint. In the adjusted analysis, LGE ≥ 7.2%(HR: 4.748, p < 0.001), extracellular volume (ECV) fraction ≥ 31.8% (HR: 2.913, p = 0.001), and native T1 z-score ≥ 2.1 (HR: 1.686, p = 0.035) were associated with SC D-related events. Patients with LGE (-) and ECV ≥ 31.8% were at higher risk of experiencing SCD events compared to those with ECV < 31.8% and LGE between 0-7.2% or mid-wall/focal LGE. Patients stra tified by LGE ≥ 7.2% exhibited a high risk of experiencing SCD-related e vents with an annual event rate of 4.65%, regardless of ECV. Patients with LVE F > 35%, LGE < 7.2%, and ECV < 31.8 exhibited an actual low risk of SCD with an annual event rate of 0.2%.

Conclusion

LGE ≥ 7.2% was strongly associated with high SCD risk, s uperior to LGE distribution and pattern. ECV serves a cruci al role in differentiating patients at low to moderate risk, particularly thos e with negative LGE or focal/mid-wall LGE.

Limitations

This is a retrospective study. Funding for this study: High-level research projects of the National Health Commission (2022-GSP-QZ-5) Ethics committee - additional information: Fuwai Hospital Author Disclosures: Minjie Lu: Nothing to disclose Yining Wang: Nothing to disclose Di Zhou: Nothing to disclose One-click joint bright- and black-blood late gadoli nium enhancement and T2 mapping for advanced myocardial imaging in the a cute STEMI population *V. De Villedon De Naide*¹, E. Gerbaud¹, B. Durand¹ , M. Villegas-Martinez¹, A. I. Schmid², P. Jais¹, M. Stuber³, H. Cochet¹, A. Bustin¹; ¹Bordeaux/FR, ²Vienn/AT, ³Lausanne/CH ([email protected]) Purpose or Learning Objective: CMR imaging enables post-infarction risk- stratification by identifying prognostic markers, s uch as infarct size (IS), the presence of microvascular obstruction (MVO), ejecti on fraction, area-at-risk (AAR) and myocardial salvage (MS). However, collect ing these markers requires the use of several MRI sequences. Here, we propose a unified 'one- click' sequence jointly collecting scar, MVO, and M S information for seamless planning, fast acquisition, and enhanced image quan tification and analysis in the acute STEMI population.

Methods

or Background: The proposed 2D whole-heart SPOT-MAPPING acquisition is a single-shot breath-held sequence g athering black- and bright- blood LGE images, averaged for optimal measurement of IS and MVO. For the bright-blood shots, a T2 preparation module with in creasing duration is used to generate a T2 map for myocardial tissue quantificat ion (MS and AAR). Seven patients with acute STEMI underwent CMR (1.5T Sieme ns). Pre-contrast T2 maps and post-contrast PSIR, SPOT and SPOT-MAPPING images were collected in a random order 12 after injection of g adolinium. Left ventricular wall and scar contours were drawn by a radiologist using Circle CVI42. Prognostic markers were extracted according to lite rature, along with T2 values in remote and injured myocardium. Acquisition times were recorded.

Results

or Findings: Acquisition times for PSIR, SPOT, T2 mapping and proposed SPOT-MAPPING were 10, 10, 13 and 10 heartb eats per slice, respectively. No significant differences were found between SPOT-MAPPING and PSIR for IS and between SPOT-MAPPING and T2 map ping for AAR, and T2 values. By imaging both IS and AAR in a co-regis tered fashion, SPOT- MAPPING enabled the measurement of the MS and the M VO.

Conclusion

SPOT-MAPPING enables easy planning, fast acquisitio n and enhanced image quantification and analysis for pati ents with acute STEMI.

Limitations

Further validation in a larger cohort is warranted, as SPOT- MAPPING clinical application is still in its early stages. Thursday Abstract-based Programme 80 Funding for this study: This research was supported by funding from the French National Research Agency under grant agreeme nt ANR-22-CPJ2- 0009-01, and from the European Research Council (ER C) grant "SMHEART" under the European Union’s Horizon 2020 research an d innovation programme (grant agreement No101076351). Ethics committee - additional information: The study was approved by the Biomedical Research Ethics Committee and all partic ipants provided informed consent for participation. Author Disclosures: Victor De Villedon De Naide: Nothing to disclose Aurelien Bustin: Nothing to disclose Hubert Cochet: Nothing to disclose Albrecht Ingo Schmid: Nothing to disclose Manuel Villegas-Martinez: Nothing to disclose Pierre Jais: Nothing to disclose Baptiste Durand: Nothing to disclose Edouard Gerbaud: Nothing to disclose Matthias Stuber: Nothing to disclose 08:00-09:30 Research Stage 3 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 705 Artificial intelligence in breast imaging Moderator J. Teuwen; Nijmegen/NL ([email protected]) Optimal utilization of an AI diagnostic software in a mammography screening program in Switzerland *M. Blum*, A. Geissler, D. Ehlig, J. Vogel, J. Sube lack, R. Morant; St.Gallen/CH ([email protected]) Purpose or Learning Objective: The goal of this study is to evaluate Profound AI® (pAI) in the screening process of the organized mammography screening program (MSP) “donna”. We aim to identify the optimal utilization of pAI in the MSP regarding its effectiveness (sensiti vity and specificity) and its influence on required resources.

Methods

or Background: In this retrospective study, we analyse all mammographies from one screening round, i.e., the y ears of 2022 and 2023, of the MSP “donna” in the Swiss canton of St.Gallen (approximately 27,600 mammographies) using pAI by iCAD, which will assign each mammography a case and predictive risk score. We use optimization models, such as the receiver operating characteristics curve, to find t he optimal threshold for case

Discussion

in a consensus conference. We simulate m ultiple AI implementation scenarios within the MSP, including AI as a substit ute for one of the two radiologists and AI as a preselection tool to ident ify mammographies for double reading.

Results

or Findings: First results of this study are expected in early 2 025 with anticipation to determine an optimal threshold when a mammography should be further discussed in a consensus conference. Thi s threshold is expected to increase the effectiveness by increasing the breast cancer detection rate. In the simulated scenarios, we expect that the workloa d of radiologists can be reduced significantly, thus increasing the efficien cy of the MSP, without loss of effectiveness.

Conclusion

Our study will contribute to identifying the optima l implementation of AI in the screening process of an MSP, optimize its effectiveness, i.e., increasing the cancer detection rate, and its effic iency, as well as initiate a

Discussion

about the future of organized screening.

Limitations

This study’s limitation lies in its retrospective d esign and the initial omission of interval carcinomas. Funding for this study: This study is partly funded by the Cancer League of Eastern Switzerland. Ethics committee - additional information: This study has been approved by the Ethics Committee of Eastern Switzerland (EKOS) under the project ID 2024-01310. Author Disclosures: Jonas Subelack: Nothing to disclose Alexander Geissler: Nothing to disclose David Ehlig: Nothing to disclose Marcel Blum: Nothing to disclose Justus Vogel: Nothing to disclose Rudolf Morant: Nothing to disclose Artificial intelligence mammography interpretation systems are affected more by mammographic image quality issues than radi ologists are *S. D. Verboom*, J. M. D. S. Boita, M. Broeders, I. Sechopoulos; Nijmegen/NL ([email protected]) Purpose or Learning Objective: To determine how common image quality issues in mammograms affect the performance of arti ficial intelligence (AI)- based mammography interpretation systems compared t o expert breast radiologists.

Methods

or Background: Five common image quality issues were simulated on 80 digital screening mammograms (40:20:20, cance r:benign:normal). Each issue was simulated at two levels: the lowest quali ty that was acceptable to radiologists, and a realistic quality that was not acceptable. Thirteen expert breast radiologists from five countries and two com mercial AI systems assessed all mammograms and scored the mammograms w ith a probability of malignancy (PoM) and a recall decision. The AI reca ll decision was obtained by matching the specificity on standard quality ima ges to that of the radiologists. The area under the receiver operating characteristics curve (AUC) and recall decisions of radiologists and AI for the two lower quality levels were compared to those for the standard quality images.

Results

or Findings: The radiologists’ original mean AUC of 0.76 (95%CI 0.68-0.84) was not affected by the lower image qual ity (p=0.77, 0.46). The AUCs of AI system A were 0.72 (0.60-0.83) on the or iginal quality, 0.68 (0.55- 0.80) (p=0.47) for the lower-acceptable quality, an d 0.61 (0.49-0.74) (p=0.06) for the unacceptable quality. For system B, the AUC decreased from 0.95 (0.90-1.0) to 0.91 (0.84-0.96) (p=0.25) and to 0.87 (0.78-0.95) (p=0.02), respectively. Radiologists gave the same recall dec ision in 83% and 82% of the cases for each quality level. Meanwhile, system A gave the same recall decision in 75% (p=0.06) and 68% (p=0.001) of the c ases and system B in 80% (p=0.47) and 78% (p=0.27) of the cases.

Conclusion

Image quality can affect AI performance and recall decision more than radiologists’, even when radiologists’ perform ance is not affected.

Limitations

Retrospective study with limited sample size. Funding for this study: aiREAD financed by the Dutch Research Council (NWO), Dutch Cancer Society (KWF), Health Holland ( HH). Ethics committee - additional information: Approval of the etics committe was not applicable due to the retrospective nature of this study with anonymized data that was previously approved for re trospective use. Author Disclosures: Mireille Broeders: Speaker: Hologic and Siemens Hea lthcare Research/Grant Support: Hologic, Screenpoint Medical, Sectra Benel ux, Volpara Healthcare, Lunit, and iCAD Sarah Delaja Verboom: Nothing to disclose Ioannis Sechopoulos: Research/Grant Support: Siemen s Healthcare, Canon Medical Systems, ScreenPoint Medical, Sectra Benelu x, Volpara Healthcare, Lunit Advisory Board: Koning Corp. Speaker: Canon, Siemens Healthcare Joana Maria Dos Santos Boita: Employee: After compl eting this project employee at Canon Medical Systems Europe Evaluation of a Digital Breast Tomosynthesis Cancer Detection AI Algorithm Using the Personal Performance in Mammogr aphic Screening Scheme (PERFORMS) G. Partridge¹, P. Phillips², J. James¹, N. Sharma³, K. Satchithananda⁴, R. Butler⁵, J. Lewin⁵, M. Michell⁴, *Y. Chen*¹; ¹Nottingham/UK, ²Lancaster/UK, ³Leeds/UK, ⁴London/UK, ⁵New Haven, CT/US ([email protected]) Purpose or Learning Objective: To compare the performance of a Digital Breast Tomosynthesis (DBT) Artificial Intelligence (AI) model as a standalone reader to that of a large cohort of breast imaging readers, using the Personal Performance in Mammographic Screening (PERFORMS) sc heme. The performance of a subset of readers, assisted by the DBT AI during image interpretation, will also be reported.

Methods

or Background: 75 challenging combined DBT and Synthetic 2D mammography (S2D) screening cases were collated int o a PERFORMS test- set. Test-set images were analysed by a prototype s erver allowing batch- processing of a commercial AI model (Hologic Genius AI Detection [GAID] v2.0). The set was also distributed to 156 readers from 8 UK National Health Service (NHS) hospitals that use DBT in screening a s part of the PROSPECTS trial, and to 6 readers from 1 US institution that employs DBT in routine screening. The AI performance will be benchmarked a gainst the performance of this reader cohort. The US readers will addition ally re-review the test-set Thursday Abstract-based Programme 81 with AI-markup available for decision support, foll owing a 6-8 week washout period. Performance with and without AI-support wil l be investigated and compared to the AI as a standalone reader.

Results

or Findings: The AI model achieved an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.935, and a sensitivity of 89.5% and specificity of 85.7% at the optimal threshold (=33) . Human readers are currently undertaking the case review, but their da ta will be reported at the conference.

Conclusion

This international, Multiple Reader Multiple Case ( MRMC) study enables the comparison of a very large cohort of br east imaging readers to a DBT AI model, as well as investigating the affect o f reading DBT with AI- support.

Limitations

The test-set is enriched with malignant cases which may influence human reader decisions. Funding for this study: Funding was acquired from Hologic Inc. Ethics committee - additional information: This study is classed as a clinical audit for quality assurance for improvement of the breast screening programme. Ethics Reference No: 88-1223. Author Disclosures: Nisha Sharma: Nothing to disclose Reni Butler: Nothing to disclose Keshthra Satchithananda: Nothing to disclose Michael Michell: Nothing to disclose George Partridge: Nothing to disclose Peter Phillips: Nothing to disclose John Lewin: Nothing to disclose Jonathan James: Nothing to disclose Yan Chen: Nothing to disclose Evaluation of an AI System for Cancer Detection in Abbreviated Breast MRI *K. Eppenhof*¹, A. Rodriguez Ruiz¹, W. B. Veldhuis² , C. Van Gils², A. M. Rosanò³, R. Yang⁴, D. E. Lehrer⁵, L. Çelik⁶, R. Mann¹; ¹Nijmegen/NL, ²Utrecht/NL, ³Sion/CH, ⁴East Brunswick, NJ/US, ⁵Buenos Aires/AR, ⁶Istanbul/TR ([email protected]) Purpose or Learning Objective: To investigate the performance of an AI system for breast cancer detection in abbreviated D CE-MRI.

Methods

or Background: A combination of high-risk screening and diagnostic DCE-MRI exams from five hospital groups and a publi c data set (Duke-Breast- Cancer-MRI) were acquired. Each MRI exam was proces sed by an AI system, which takes as input the pre-contrast and a single post-contrast T1 image (abbreviated breast MRI), detects suspicious region s, and outputs a malignancy score per breast between 1 and 10. Addit ionally, the AI system was evaluated on an enriched screening dataset from the DENSE trial.

Results

or Findings: Area under the Receiver Operating Characteristic cu rve (AUROC) was computed for classifying exam malignanc y for exams from four hospital groups located in Argentina (41 of 780 exa ms containing biopsy- proven cancer, AUROC 0.891 (95% CI=0.828-0.944)), S witzerland (98/3499, 0.863(0.824-0.896)), Turkey (33/164, 0.955(0.898-0. 998)), and the US (153/1096, 0.904(0.877-0.929)). The consistency in AUROCs indicates robustness across populations, protocols, and scann ers. Because Duke- Breast-Cancer-MRI exams all contain cancer, a breas t-level analysis was done where breasts without cancer were used as the negat ive class (904/1808 breasts containing cancer). The AUROC (0.965(0.957- 0.972)) is similar to an earlier published AI that used two post contrast im ages. For exams that had a BIRADS assessment, the agreement between the AI (sc ore >= 9) and the radiologist interpretation (BIRADS 1 or 2 vs. 4 or 5) was found to be moderate (Cohen kappa=0.502(0.449-0.555)). The performance o n screening-only data was measured in exams from the fifth hospital locat ed in the Netherlands (66/2920 exams containing cancer, AUROC 0.812(0.753 , 0.868)), and exams from the DENSE trial (83/517, AUROC 0.803(0.747-0.8 56)).

Conclusion

A first evaluation of an AI system for abbreviated DCE-MRI shows potential for decision support in detecting b reast cancer.

Limitations

The study has a retrospective design. Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Carla Van Gils: Nothing to disclose Alejandro Rodriguez Ruiz: Employee: ScreenPoint Med ical Koen Eppenhof: Employee: ScreenPoint Medical Daniel Eduardo Lehrer: Nothing to disclose Wouter B. Veldhuis: Nothing to disclose Levent Çelik: Nothing to disclose Roger Yang: Nothing to disclose Anna Maria Rosanò: Nothing to disclose Ritse Mann: Advisory Board: ScreenPoint Medical Validating the impact of real-world live use of AI as an additional reader in breast cancer screening (BCS) *A. Ng*¹, E. Ambrozay², E. Szabó², B. Glocker¹, P. Kecskemethy¹; ¹London/UK, ²Budapest/HU ([email protected]) Purpose or Learning Objective: To validate that the measured impact of deploying AI as an additional reader is a result of the AI intervention and not purely the result of additional reading.

Methods

or Background: Live-use of an AI-system as an additional reader (XR) to flag cases for additional review that it su ggested to recall but the standard double reading (DR) decision was “no recal l”, has been demonstrated to result in a 0.8/1000 increase in cancer detectio n rate (CDR) and a 0.1% increase in positive predictive value when 6.0% of cases are additionally reviewed, compared to DR. To validate that the incr eased effectiveness in early cancer detection of the XR AI-workflow is not purely from additional reading, the maximum CDR increase opportunity due t o third-human-reading a random 6.0% of cases was simulated, calculated as i nterval cancer rate (ICR) times the portion of cases to be third-human-read ( 6.0%), times the portion of human-detectable interval cancers (ICs) i.e. visibl e on priors. A range of ICR of 0.84-2.11/1000 was used (DOI:10.1038/s41523-017-001 4-x). Studies have measured that 22% of ICs are human-detectable (DOI: 10.1007/s00330-020- 07130-y), however, a wider range of 22-100% was use d. Third-human-reading was assumed to have an unrealistic 100% sensitivity among human-detectable ICs.

Results

or Findings: For the lower and upper end of assumptions, respectively, the maximum CDR increase opportunity calculated for third- human-reading is 0.01 and 0.13/1000, which is 98.9% and 87.2% less than the XR AI-workflow, suggesting that the increased CDR i mpact of XR is 6-70 times more effective than third-human-reading without AI.

Conclusion

Simple simulations show that CDR improvements from third- human-reading a random set of cases would be margin al, validating that the substantial CDR increase demonstrated by XR is a di rect effect from using AI to flag cases for additional review.

Limitations

Single AI assessed Funding for this study: Kheiron Medical Technologies Ethics committee - additional information: Not required Author Disclosures: Annie Ng: Nothing to disclose Ben Glocker: Employee: Kheiron Medical Technologies Ltd Eva Ambrozay: Nothing to disclose Peter Kecskemethy: CEO: Kheiron Medical Technologie s Ltd Endre Szabó: Nothing to disclose Adding artificial intelligence (AI) case scoring in a breast screening programme to optimize reading workflow and workload : a retrospective study *A. Nitrosi*, R. Vacondio, L. Verzellesi, M. Creola , M. Bertolini, P. Giorgi Rossi, V. Iotti, P. Pattacini, C. Campari; Reggio Emilia/I T ([email protected]) Purpose or Learning Objective: The objective of this study was to retrospectively evaluate a strategy to optimize rea ding workflow and readers’ workload based on the iCAD Case Malignancy Score (C MS).

Methods

or Background: We analyzed 122,216 2D mammography screening reading times (RT) corresponding to 61,108 exams in cluding 244 proven tumours, consequentially acquired in Reggio Emilia Breast Screening Program (BSP) starting from January 2023 to June 2024 and e laborated by iCAD Inc. ProFound AI 2D system. ICAD Case Malignancy Scores (CMS) represents the relative confidence that a case is malignant on a s cale of 0% to 100%. A pool of radiologists performs blinded double reading plu s arbitration framed in work- shift. Packs are assigned to a reader respecting a numerical criterion of maximum readings per work-shift. We analyzed the co rrelations (Spearman) between the RT of individual readers (normalized on the personal median) with the CMS and the breast density (D). The analysis wa s repeated considering only the women recalled / not recalled / true posit ive (TP).

Results

or Findings: A positive correlation was demonstrated between CMS and RT (R = 0.76) and slightly between D and RT (R = 0.52), overall and in recalled and non-recalled women separately. Using C MS, packs could be optimized based on individual reader characteristic s to maximize the number of exams for each reader’s pack with constant recall r ate (and TP): first simulations show up to 14% increase in the number o f exams read over 4 hours effective reading period.

Conclusion

This scenario would not undermine the reading scree ning workflow while ensuring resource optimization nor i ntroduce any cognitive bias influencing the readers since each session would ha ve similar expected recall and detection rate.

Limitations

Cases refer only to Reggio Emilia BSP, limiting thi s study. Funding for this study: This study was partially supported by the Italian Ministry of Health - Ricerca Corrente Thursday Abstract-based Programme 82 Ethics committee - additional information: Compliance with Ethical Standards Institutional Review Board approval was n ot required because it is a Clinical Audit about a technical development. This study was conducted in accordance with the routine quality assurance proce dures established by the Local Health Authority for its screening programmes . The Reggio Emilia Cancer Registry, which routinely collects the scree ning history of each case of breast cancer, has been approved by the Provincial Ethics Committee. Author Disclosures: Laura Verzellesi: Nothing to disclose Valentina Iotti: Speaker: Invited speaker Cinzia Campari: Nothing to disclose Marco Bertolini: Nothing to disclose Pierpaolo Pattacini: Speaker: Invited speaker Andrea Nitrosi: Speaker: Invited speaker Rita Vacondio: Speaker: Invited speaker Paolo Giorgi Rossi: Nothing to disclose Martina Creola: Nothing to disclose Assessment of an AI-system in indicating breast lat erality for screen- detected and interval cancers in breast screening i n a large-scale retrospective study A. Ng¹, C. Oberije¹, G. Fox¹, R. Currie², A. Redman ³, A. Leaver³, W. Teh⁴, B. Glocker¹, *P. Kecskemethy*¹; ¹London/UK, ²Exeter /UK, ³Gateshead/UK, ⁴Harrow/UK ([email protected]) Purpose or Learning Objective: Assess the utility of an AI-system in indicating breast laterality in breast screening.

Methods

or Background: Employing a commercially available AI-system as an independent reader, utilising its case-wise reca ll suggestions, within double reading has previously been shown to maintain/impro ve screening performance, while providing substantial workload s avings. This has been demonstrated in a large-scale retrospective clinica l study (306,839 cases from 236,739 participants between 2017-2021), involving three Hologic sites across the UK’s major genetic clusters (South-East/West/No rth), including more diverse ethnicities in London. To further assess th e AI-system’s utility in supporting follow-up investigations for recalls, it s breast laterality recommendation for screen-detected cancers (SDCs) a nd interval cancers (ICs) were compared to pathology information.

Results

or Findings: The study included 2592 SDCs and 379 ICs. The AI- system correctly recalled 2304 SDCs (88.9% sensitiv ity) and 152 ICs (40.1% IC detection rate). Among the correctly recalled SD Cs, the AI-system: A) indicated pathology-agreeing laterality in 84.5% (8 3.5% unilateral/1.1% bilateral), B) recalled unilateral cases as bilater al in 13.9%, C) recalled one side in bilateral cancer cases in 0.4%, and D) indi cated the opposite side not assessed in 1.2%. The respective results for ICs we re: A) 58.8% (58.1% unilateral/0.7% bilateral), B) 18.4%, C) 2.2%, and D) 20.6%. For category D, it is unknown if an early abnormality could be present as the AI-indicated side was not assessed by biopsy nor by additional diagno stic imaging. The AI- system provides screening utility in scenarios A-C, which comprises 98.8% for SDCs, 79.4% for ICs, 88.1% for ICs diagnosed within 1 year, and 97.1% for false negative ICs (FNICs).

Conclusion

The AI-system’s laterality detection demonstrated u tility in almost all SDCs/FNICs, and the large majority of ICs, show ing it can support the clinical workflow with laterality information for f ollow-up assessments.

Limitations

Single AI assessed Funding for this study: NIHR AI in Health and Care Award Ethics committee - additional information: UK HRA REC reference: 21/HRA/4830 Author Disclosures: Cary Oberije: Employee: Kheiron Medical Technologie s Ltd Annie Ng: Employee: Kheiron Medical Technologies Lt d William Teh: Nothing to disclose Ben Glocker: Employee: Kheiron Medical Technologies Ltd Alice Leaver: Nothing to disclose Georgia Fox: Employee: Kheiron Medical Technologies Ltd Alan Redman: Nothing to disclose Peter Kecskemethy: CEO: Kheiron Medical Technologie s Ltd Rachael Currie: Nothing to disclose Repurposed AI-Based Mammography Interpretation in D iverse Clinical Scenarios *H. Ngo*¹, J. Neubauer¹, A. L. Palacios Acedo², M. Windfuhr-Blum¹, E. Kotter¹, F. Bamberg¹, J. Weiß¹; ¹Freiburg/DE, ²Marseille/FR ([email protected]) Purpose or Learning Objective: This study evaluates the diagnostic performance of an artificial intelligence (AI) tool originally developed for screening mammography, now repurposed for use in va rious clinical scenarios, including diagnostic mammograms in 1) asymptomatic women, 2) symptomatic women and 3) patients with a personal history of br east cancer (PHBC).

Methods

or Background: A total of 601 women with were retrospectively included and categorized into three subgroups: diag nostic mammograms of 1) asymptomatic women (n = 423), 2) symptomatic women (palpable abnormality, suspicious sonography, n=66) and 3) patients with P HBC (n =112). The AI-tool provided continuous scores (1 to 100) for potential malignancy, with histopathological confirmation and/or follow-up ≥2 years as reference standard.

Results

or Findings: The AI-tool showed high performance across all thre e cohorts, with areas under the curve (AUC) for diagn ostic mammograms of 1) asymptomatic women: 0.75 (95% CI: 0.51-0.98), 2) sy mptomatic women: 0.92 (95% CI: 0.81-1.0), and 3) patients with PHBC: 0.71 (95% CI: 0.52-0.90). Excluding women with extremely dense breasts (ACR D ) increased the AUC for diagnostic mammograms of 1) asymptomatic women to 0.79 (95% CI: 0.41- 1.0), 2) symptomatic women: 0.92 (95% CI: 0.81-1.0) , and 3) patients with PHBC: 0.73 (95% CI: 0.51-0.95). Using a threshold o f the highest 10% AI- scores to binarize the continuous AI-output resulte d in sensitivity 0.92 and specificity 0.50 for subgroup 1); 0.96 and 0.77 for 2) and 0.81 and 0.67 for 3), respectively.

Conclusion

Repurposed AI-tools can enhance malignancy detectio n across diverse patient groups, especially in less dense br easts. Optimizing thresholds for specified populations, such as asymptomatic and symptomatic cohorts, may further improve AI's diagnostic effectiveness.

Limitations

Varying breast densities, particularly extremely de nse breast, can pose detection challenges, and the sample size of 6 01 may influence the generalizability of the findings. Funding for this study: Unrestricted research grant from Lunit. Ethics committee - additional information: Approved by local IRB. Author Disclosures: Helen Ngo: Research/Grant Support: Lunit Jakob Neubauer: Research/Grant Support: Lunit Jakob Weiß: Research/Grant Support: Lunit Ana Luisa Palacios Acedo: Employee: Lunit Europe Fabian Bamberg: Research/Grant Support: Lunit Marisa Windfuhr-Blum: Research/Grant Support: Lunit Elmar Kotter: Research/Grant Support: Lunit Patient perceptions towards the use of artificial i ntelligence (AI) in breast cancer imaging D. Velazquez-Pimentel, S. Khan, T. Falco, S. Hickma n, S. Dani, *T. Suaris*; London/UK ([email protected]) Purpose or Learning Objective: The aim of this study is to evaluate patient perceptions towards the use of artificial intellige nce (AI) in breast cancer imaging

Methods

or Background: Women presenting to a single breast cancer unit in East London were invited to participate in a prospe ctive survey. Baseline knowledge and attitude towards technology in daily living and attitude towards the use of AI in mammography screening was measured using a 4-point Likert scale. Demographic data including age, ethnicity, e ducation was collected.

Results

or Findings: 944 responses were analysed. Of these, 90% (n=853/944) expressed a preference for combined com puter-physician reading with more women expressing confidence in the accura cy of combined computer-physician (93%, n=882/944) reading over co mputer reading alone (54% n=513/944). Self-reported understanding of tec hnology was associated with a higher level of concern. In patients with li mited understanding 46% expressed concern with regards to the accuracy of c omputer read mammograms compared to 38% in patients with expert understanding. Level of concern was not significantly associated with ag e, ethnicity or education level (p > 0.05). Regardless of level of concern, t he majority of respondents expressed a positive opinion on the impact computer read mammograms can have on improving both efficiency (85%, n=798/944) and pick up rate (84%, n=797/944).

Conclusion

Despite confidence in the ability of AI to improve efficiency and pick up rate there is a strong preference expressed by patients towards combined computer-physician read mammograms. This s tudy demonstrates that this remains true regardless of age, ethnicity or level of education. Level of concern is associated with self-reported understand ing of technology; targeted patient education programs may support implementati on of AI workflow in breast screening programs.

Limitations

Survey responses are subject to bias. Funding for this study: No funding was received for this study. Ethics committee - additional information: Patient Survey - local research lead confirmed no formal ethics application necessa ry Author Disclosures: Tamara Suaris: Nothing to disclose Thamiris Falco: Nothing to disclose Sarah Khan: Nothing to disclose Diana Velazquez-Pimentel: Nothing to disclose Thursday Abstract-based Programme 83 Sarah Hickman: Other: -SEH have research collaborat ions with Vara, Screen- Point, Lunit, Google, Volpara, iCAD, Therapixel, Cu reMetrix, Sunnybrook Research Institute, and Massachusetts Institute of Technology. -SEH is a Radiology AI: Trainee Editorial Board member. Shefali Dani: Nothing to disclose ADMEDVOICE – The Pathway to Polish Language Automat ic Structured Reporting in Breast Ultrasound using Voice Recognit ion and Large Language Models *M. Bobowicz*, D. Szplit, A. Dąbkowska, J. Bogdan, K. Gwozdziewicz, J. Omernik, B. Graff, A. Czyżewski, K. Narkiewicz; Gdansk/PL ([email protected]) Purpose or Learning Objective: Breast ultrasound (BUS) equipped with the ACR BI-RADS lexicon is a well-described diagnostic procedure with mandatory fields and a relatively closed vocabulary. This stu dy aims to generate BUS- structured reports automatically using voice recogn ition and topic modelling in Polish.

Methods

or Background: A dataset of 6269 BUS radiology reports from the University Clinical Center’s Hospital Information S ystem covering 2013-2023 was obtained. The reports were created by more than ten experienced breast radiologists and multiple residents. They covered v arious clinical scenarios, including diagnosis, treatment, and follow-up tests in breast cancer, benign disease, mutation carriers, and studies without pat hology.

Results

or Findings: From 6269 reports, 48721 text fragments were obtain ed, representing specific parts of the BUS report used as training data. We identified specific ‘topics’ relating to ‘ontologie s’ in these fragments. Topics represented parts of the radiologist’s report that could be structured into subsections: 1) reference letter information, 2) ti ssue composition, 3) pathology descriptors (masses and calcifications), 4) associa ted features, 5) axillary and intramammary lymph node descriptors, 6) other speci al cases, 7) conclusions, 8) recommendations, 9) final remarks. For automatic text recognition BERTOPIC was explored. As a next step, we invited 2 5 specialist radiologists, residents, medical students and other HCPs to recor d 3328 separate sentences for voice recognition algorithms training .

Conclusion

The presented research, which involved topic modell ing, is a first step towards creating Polish language automatic str uctured BUS reporting using voice recognition and LLMs. The resulting dat abase with voice samples at three quality levels will be released soon. It w ill allow AI training to reduce the radiology reporting burden with more natural vo ice commands being transferred to structured reports.

Limitations

The single-centre design, restriction to the Polish language, and lack of external validation. Funding for this study: Funding for the ADMEDVOICE Project was provided by the Polish National Centre for Research and Deve lopment; Infostrateg IV action; grant number: INFOSTRATEG-IV/003/2022. Ethics committee - additional information: The study was approved by the Bioethics Committee for Scientific Research of Medi cal University of Gdansk. Author Disclosures: Andrzej Czyżewski: Nothing to disclose Beata Graff: Nothing to disclose Maciej Bobowicz: Nothing to disclose Krzysztof Narkiewicz: Nothing to disclose Katarzyna Gwozdziewicz: Nothing to disclose Anna Dąbkowska: Nothing to disclose Dariusz Szplit: Nothing to disclose Julia Bogdan: Nothing to disclose Justyna Omernik: Nothing to disclose Contrastive Learning in Breast MRI: MLIP as the Bas e Foundation Model *N. Rasoolzadeh*¹, T. Zhang², R. Mann¹; ¹Nijmegen/N L, ²Amsterdam/NL ([email protected]) Purpose or Learning Objective: To explore the potential of utilizing a contrastive language image pretraining approach for 3D breast MRI images.

Methods

or Background: A dataset of 15005 pairs of dynamic contrast- enhanced (DCE) and subtraction MRI images with corr esponding radiological reports from the Netherlands Cancer Institute were used for training a model to find the most similar image-text pairs by contrasti ng positive pairs (similar) against negative pairs (dissimilar) samples. Full M RI images and complete Dutch reports were utilized. The image and text emb eddings were obtained using a 3D ResNet50 architecture and RadioLOGIC as the image and text encoders, respectively. Two inference scenarios wer e tested: image retrieval by text queries and BI-RADS prediction. The area un der the curve (AUC) was used to evaluate the model's performance. The devel oped Multimodal Breast MRI Language-Image Pretrained (MLIP) model was firs t used for the zero-shot BI-RADS prediction task and was later fine-tuned.

Results

or Findings: The preliminary results show an AUC of 0.717 (95% C I: 0.604, 0.824) for BI-RADS 4/5 abnormal MRI images r etrieval, 0.640 (95% CI: 0.538, 0.740) for dense breast retrieval, and 0.601 (95% CI: 0.505, 0.698) for low background parenchymal enhancement (BPE) retrie val. In the second inference, the performance of MLIP was compared to that of a fine-tuned model. The fine-tuned model demonstrated improved a ccuracy, with a reduction in the number of originally benign cases misclassified as malignant.

Conclusion

In this study, a multi-modal breast MRI pretrained model was developed. The preliminary results suggest MLIP can be adjusted to perform diagnostic tasks and radiology report generations, holding the potential to serve as a foundation model for breast MRI analysis .

Limitations

The model needs to be validated on larger datasets and across more downstream tasks. Funding for this study: Funding was provided by the ODELIA project (from the European Union’s Horizon Europe research and in novation programme under grant agreement, No 101057091) Ethics committee - additional information: This study did not require formal ethics committee approval, as it exclusively used f ully anonymized MRI images and reports. No identifiable personal data was coll ected or used in the analysis. All MRI data was anonymized prior to acce ss, ensuring that no individual participants can be identified from the data Author Disclosures: Tianyu Zhang: Nothing to disclose Nika Rasoolzadeh: Nothing to disclose Ritse Mann: Nothing to disclose Generating virtual T2w-fat-saturated breast MRI acq uisition using neural- networks *A. Liebert*¹, D. Hadler¹, C. M. Ehring¹, H. Schrei ter¹, F. B. Laun¹, M. Uder¹, E. Wenkel², S. Ohlmeyer¹, S. Bickelhaupt¹; ¹Erlange n/DE, ²Munich/DE ([email protected]) Purpose or Learning Objective: Multi-parametric breast MRI protocols typically include T2-weighted fat-saturated(T2w-FS) sequences, which are used for tissue characterization. However, their ac quisition can significantly increase scan time. This study aims to evaluate, wh ether a 2D-U-Net neural- network can generate virtual T2w-FS images(VirtuT2) from other acquisitions of a routine multiparametric breast MRI protocol.

Methods

or Background: This IRB-approved, retrospective study included n=914 breast MRI examinations performed between Jan uary 2017 and June 2020 at University Hospital Erlangen. The dataset w as divided into training(n=665), validation(n=74), and test(n=175) sets. 2D-U-Net was trained on T1w, DWI, and DCE sequences to generate VirtuT2 . Quantitative metrics and a qualitative multi-reader assessment by two ra diologists were used to evaluate the VirtuT2 images. For qualitative readin gs radiologist were asked to identify, whether an image is original T2w-FS or Vi rtuT2 image, evaluate the diagnostic image quality(DIQ) and wheter they can i dentify presence of edema around the mass-lesions.

Results

or Findings: VirtuT2 images demonstrated high structural similarity(SSIM=0.87) and peak signal-to-noise rati o(PSNR=24.90) compared to original T2w-FS images. High level of the freque ncy error norm(HFEN=0.87) indicates strong blurring presence in the VirtuT2 i mages, which was also confirmed in qualitative reading. Radiologists corr ectly identified VirtuT2 images with 92.3% and 94.2% accuracy, respectively. No significant difference in DIQ was noted for one reader(p=0.21), while the other reported significantly lower DIQ for VirtuT2(p<=0.001). Moderate inter-rea der agreement was observed for edema detection on T2w-FS images( ƙ=0.43), decreasing to fair on VirtuT2 images(ƙ=0.36).

Conclusion

Neural-networks can technically generate VirtuT2 im ages with high similarity to real T2w-FS images, using T1w, D WI and DCE acquisitions, however blurring remains a limitation. Future inves tigations with different architectures and using larger datasets are needed to improve clinical applicability.

Limitations

Limited dataset from a single site was used. Qualit ative reading was performed on just n=52 cases. Funding for this study: This project is funded by the Bavarian State Minist ry of Science and the Arts in the framework of the bid t Graduate Center for Postdocs. L.B. is funded by the DFG Grant No: 51868 9644 Ethics committee - additional information: The study protocol was approved by the ethics committee of the Friedrich-Alexander Universität Erlangen- Nürnberg. The ethics comitee waived the need for in formed consent. Author Disclosures: Sebastian Bickelhaupt: Research/Grant Support: Siem ens Healthineers Evelyn Wenkel: Nothing to disclose Dominique Hadler: Nothing to disclose Frederik Bernd Laun: Nothing to disclose Michael Uder: Research/Grant Support: Siemens Healt hineers Hannes Schreiter: Nothing to disclose Sabine Ohlmeyer: Nothing to disclose Andrzej Liebert: Nothing to disclose Chris Matthias Ehring: Nothing to disclose Thursday Abstract-based Programme 84 08:00-09:30 Research Stage 4 Research Presentation Session: Neuro RPS 711 Stroke and neurovascular imaging: inside and outside of the blood vessels Moderator E. Papadaki; Iraklion/GR ([email protected]) Advancing Neurovascular Imaging with Ultra-High-Res olution Photon- Counting Detector CT: Optimization of Reconstructio n Kernel and Quantum Iterative Reconstruction *A. Toth*, Y. (. Cho, E. Wilson, J. Crow, E. Bass, J. Joyce, M. G. Matheus, S. Tipnis, M. V. Spampinato; Charleston, SC/US ([email protected]) Purpose or Learning Objective: Our goal was to identify the optimal combination of dedicated neurovascular reconstructi on kernels and quantum iterative reconstruction (QIR) levels for ultra-hig h-resolution (UHR) photon- counting detector (PCD)-CT angiography of the head and neck.

Methods

or Background: 18 patients with intracranial aneurysms were prospectively included in this study. CT angiograms were obtained in UHR mode using a clinical dual-source PCD-CT scanner. I mages were reconstructed with a slice thickness of 0.2 mm, uti lizing six strength levels of a dedicated neurovascular kernel (Hv48/Hv56/Hv64/Hv72 /Hv80/Hv89) and four levels of QIR (1-4). We assessed image noise, signa l-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and vessel sharpness for all reconstructions.

Results

or Findings: With higher kernel sharpness and lower QIR, image noise continuously increased. The best performing r econstructions in terms of CNR were Hv48 and Hv72 in combination with QIR-4. V essel sharpness improved with higher kernel levels, reaching a plat eau with the Hv64 and Hv72 reconstructions, as observed in the small intracran ial arteries (maximum ΔHU values of 260.59 and 255.11, respectively). Based o n the results of the quantitative analysis, the kernels identified as th e top performers and selected for further evaluation in the qualitative analysis were Hv56, Hv64, and Hv72.

Conclusion

Vessel sharpness increased with higher kernels leve ls, reaching a plateau at Hv64 and Hv72. Overall, Hv56, Hv64 and Hv72 were recognized as the best performing kernels based on the quantit ative results. In the following steps, qualitative image quality evaluati on will be conducted by three readers using a 5-point Likert scale. This evaluati on will focus on the best- performing kernels in combination with the availabl e QIR levels to assess overall image quality and diagnostic performance.

Limitations

The limitations of the study are the relatively sma ll patient cohort and the fact that quantitative measurements were pe rformed by a single observer. Funding for this study: Funding was provided by Siemens Healthineers (research grant). Ethics committee - additional information: The study was approved by the Institutional Review Board of the Medical Universit y of South Carolina (Pro00123327)). Author Disclosures: Eric Bass: Nothing to disclose Yongjoo (Jennifer) Cho: Nothing to disclose John Crow: Nothing to disclose Evan Wilson: Nothing to disclose Adrienn Toth: Nothing to disclose Maria G. Matheus: Nothing to disclose Jennifer Joyce: Nothing to disclose Sameer Tipnis: Nothing to disclose M. Vittoria Spampinato: Research/Grant Support: Sie mens Healthineers Performance of Dual-Layer Spectral Detector Non-con trast Computed Tomography in Identifying Early Ischemic Changes in Acute Ischemic Stroke Patients *Y. Wang*, H. Zhu, J. Wen, S. Ma, S. Yang; Beijing/ CN ([email protected]) Purpose or Learning Objective: The study aimed to evaluate the effectiveness of non-contrast dual-layer spectral c omputed tomography (DLCT) for detecting early ischemic changes in patients wi th acute ischemic stroke (AIS).

Methods

or Background: NCCT is a common imaging technique for suspected AIS patients. This study involved 27 AIS patients who underwent both DLCT and MRI within 12 hours of symptom onset. A retrospective analysis was conducted on the imaging data, focusin g on quantitative measurements from regions identified as acute infar ction on diffusion-weighted MRI. Various parameters, including conventional CT values, virtual monoenergetic (monoE) CT values, and electron densi ty relative to water (EDW) were compared between ischemic and normal bra in. Statistical analyses, including the Mann-Whitney U test and ROC curve analysis, were performed to assess the diagnostic performance of t hese parameters. Z test was performed to compare the ROC curves of differen t parameters.

Results

or Findings: This study analyzed 59 lesions in 27 patients to ev aluate the diagnostic performance of various spectral para meters. The study found that EDW had the highest area under the curve (AUC) of 0.957, with an sensitivity of 95%, specificity of 85%. MonoE at 10 0 keV achieved an AUC of 0.955, with high sensitivity (95%) and negative pre dictive value (94%). In contrast, monoE at 40 keV showed the lowest perform ance, with an AUC of 0.701 and sensitivity of 66%. Conventional CT image s had an AUC of 0.887, comparable to monoE at 70 keV (0.910). Statisticall y significant differences were noted between the AUCs of EDW/100 keV and conv entional CT, while no significant difference was found between EDW and 10 0 keV.

Conclusion

Our findings indicated that EDW and monoE CT images obtained from DLCT can improve the detection of AIS compared to conventional non- contrast CT imaging.

Limitations

The sample size was small. Funding for this study: None Ethics committee - additional information: This retrospective study was approved by the local Ethics Committee of Civil Avi ation General Hospital (2024-L-K-122). Author Disclosures: Jing Wen: Nothing to disclose Haifeng Zhu: Nothing to disclose Yujie Wang: Nothing to disclose Shanrui Ma: Nothing to disclose Shan Yang: Nothing to disclose DTI-derived Perivascular Space Diffusion Index coul d mirror the polarization of AQP4 following cerebral ischemia *X. Hao*, J. Tian, Z. Yao; Shanghai/CN ([email protected]) Purpose or Learning Objective: The purpose of this study was to investigate the dynamic changes of the perivascular space diffu sion index (ALPS) and its correlation with aquaporin 4 (AQP4) polarization fo llowing cerebral ischemia in rats, using advanced diffusion tensor imaging (DTI) technique.

Methods

or Background: Rats were divided into a normal group (n=5) and an ischemic group (n=25). The ischemic group underwent transient middle cerebral artery occlusion (tMCAO) and was further s ubdivided into five subgroups (n=5 each) based on the time post-ischemi a (1, 3, 7, 14, and 28 days). Rats underwent MRI scans, including DTI, T2- weighted imaging (T2WI), and susceptibility-weighted imaging (SWI). Subseque ntly, immunofluorescence staining for AQP4 and glial fibrillary acidic prote in (GFAP) was performed. The ALPS index was analyzed based on T2W, SWI, and frac tional anisotropy (FA) obtained by DTI post-processing. And regions of int erest were selected on the ipsilateral periventricular area, the ipsilateral c orpus callosum/cingulate area and their mirror areas of the contralateral side. T he AQP4 polarization was analyzed by GFAP/AQP4 in the non-glial scar area ar ound the infarction in cortex and striatum.

Results

or Findings: ALPS indexes were markedly reduced in ischemic rats , particularly on the affected side, with a notable d rop on day 1, then rising at days 14 and 28. AQP4 polarization mirrored this tre nd, falling initially, then rising significantly by days 14 and 28. The ALPS in dex closely aligns with AQP4 index fluctuations.

Conclusion

The DTI-based ALPS index mirrors changes in AQP4 po larization after stroke, dropping sharply in the hyperacute ph ase and recovering in the early chronic phase, which could be a useful biomar ker for glymphatic pathway function following stroke.

Limitations

This experiment requires high consistency of the le sions. Funding for this study: National Natural Science Foundation of China (No. 81801660) and the grant of National Natural Science Foundation of China (No. 82272061) Ethics committee - additional information: Institutional Animal Care and Use Committee of Fudan University Author Disclosures: Xiaozhu Hao: Nothing to disclose Zhenwei Yao: Nothing to disclose Jiaqi Tian: Nothing to disclose Thursday Abstract-based Programme 85 A novel model to quantify blood transit time in cer ebral arteries using ASL-based 4D magnetic resonance angiography with ex ample clinical application in moyamoya disease A. Bhogal¹, S. Uniken Venema¹, *P. T. Deckers*¹, K. Van De Ven², M. Versluis², K. Braun¹, B. Van Der Zwan¹, J. Siero ¹; ¹Utrecht/NL, ²Best/NL ([email protected]) Purpose or Learning Objective: Angiography is critical for visualizing cerebral blood flow in intracranial steno-occlusive diseases. Current 4D magnetic resonance angiography (MRA) techniques pri marily focus on macrovascular structures, yet few have quantified h emodynamic timing. This study introduces a novel model to estimate macrovas cular arterial transit time (mATT) derived from arterial spin labeling (ASL)-ba sed 4D-MRA. We provide examples of our method that visualize mATT differen ces throughout the brain of patients with intracranial steno-occlusive disea se (moyamoya), as well as changes in mATT resulting from the cerebrovascular reactivity (CVR) response to acetazolamide (ACZ).

Methods

or Background: The study population consisted of twelve patients with intracranial steno-occlusive disease, with a c linical indication to undergo hemodynamic imaging with an ACZ challenge to measur e CVR. CVR is measured using multi-PLD ASL-MRI, acquired pre- and post-ACZ, and includes a four-dimensional dynamic MRA sequence us ing an ASL-scheme. The scan indications varied, but mostly involved ce rebrovascular hemodynamic evaluation before or after neurosurgica l intervention.

Results

or Findings: We provide examples of our method that visualize mA TT differences throughout the brain of patients with i ntracranial steno-occlusive disease (moyamoya), as well as changes in mATT in r esponse to an ACZ injection. Furthermore, we present a method that pr ojects sparse arterial signals into a 3D native brain-region atlas space a nd correlates regional mATT with other hemodynamic parameters of interest, such as tissue transit time and CVR.

Conclusion

Our approach offers a non-invasive, quantitative as sessment of macrovascular dynamics, which enhances the understa nding of large-vessel and tissue-level hemodynamics and augment monitorin g of treatment outcomes in steno-occlusive disease patients. This can directly be used in stroke trial stratifications and peri-procedural tr eatment monitoring. Furthermore, it sets the stage for more in-depth in vestigations of the macrovascular contribution to brain hemodynamics.

Limitations

Pilot data in a small subgroup of patients (n=12). Funding for this study: This work was supported by the W.M. De Hoop Foundation, the Janivo Foundation and Friends of UM C Utrecht & Wilhelmina Children’s Hospital, and an NWO VIDI grant awarded to A.A.B. (VI.Vidi.223.085). JCWS is supported by the Brain C enter Young Talent Fellowship 2019 of the University Medical Center Ut recht, The Netherlands. Ethics committee - additional information: The Medical Ethics Review Committee of the University Medical Centre Utrecht declared that the Medical Research Involving Human Subjects Act (WMO) did not apply to the present research since all study measures were part of rout ine clinical practice. All patients or their legal representative (i.e., paren t or guardian) provided written informed consent to use their data. Healthy subject s were acquired under a sequence development ethical protocol, which was ap proved by the Medical Ethics Review Committee of the University Medical C entre Utrecht. Informed consent was given by each healthy subject. Author Disclosures: Kees Braun: Nothing to disclose Bart Van Der Zwan: Nothing to disclose Alex Bhogal: Nothing to disclose Simone Uniken Venema: Nothing to disclose Maarten Versluis: Other: Works for Philips Healthca re and provided the advanced ASL patch used in this work. No financial support was provided. Jeroen Siero: Nothing to disclose Pieter Thomas Deckers: Nothing to disclose Kim Van De Ven: Other: Works for Philips Healthcare and provided the advanced ASL patch used in this work. No financial support was provided. The use and pitfalls of hemodynamic MRI using multi delay arterial spin labelling for intracranial steno-occlusive dis ease in clinical practice: a single-center experience *S. Uniken Venema*, P. Deckers, J. W. Dankbaar, B. Van Der Worp, J. Hendrikse, B. Van Der Zwan, K. Braun, A. Bhogal, J. Siero; Utrecht/NL ([email protected]) Purpose or Learning Objective: The primary objective is to describe a clinically feasible advanced neuroimaging protocol developed at an academic medical center that uses multi-delay arterial spin labeling (ASL) and blood oxygen level dependent (BOLD)-MRI with acetazolamid e. This protocol is designed to assess cerebrovascular reactivity (CVR) in patients with intracranial steno-occlusive disease (e.g. moyamoya ), while avoiding the

Limitations

of PET scans.

Methods

or Background: Image acquisition on 3-Tesla MRI involves acetazolamide-augmented multi-delay ASL and dynamic BOLD, in addition to structural sequences. Image processing is done usin g customized MATLAB- based toolboxes. ASL-CVR is calculated by subtracti ng pre-acetazolamide cerebral blood flow (CBF) from post-acetazolamide C BF and additional hemodynamic maps, such as arterial transit time (AT T), are generated. Imaging interpretation includes assessment of scan quality and success of the hemodynamic challenge.

Results

or Findings: Since 2018, approximately 100 patients were scanned using this protocol. Multi-delay ASL enables a more accurate assessment of CBF and CVR compared to single-delay ASL in patient s with prolonged ATT owing to their stenosis, and enables quantifying AT T simultaneously – a useful marker in itself. While CVR assessment is primarily done using ASL-derived CVR maps, BOLD-CVR provided useful complementary in formation in some cases. A typical patient with an intracranial steno sis experiencing ischemic symptoms will demonstrate lower baseline CBF, lower CVR and prolonged ATT in the affected hemisphere.

Conclusion

Our multi-delay ASL-based protocol demonstrates cli nical feasibility and utility, allowing detailed cerebral hemodynamic evaluations of individual patients that is useful for clinical dec ision-making. This work serves as a practical guide for clinicians and MRI experts seeking to implement these advanced imaging methods in their institutions.

Limitations

Potential pitfalls in imaging acquisition and inter pretation, including motion artefacts, inadequate labeling, th e effects of anesthesia on CVR and the uncertainties of acetazolamide-augmente d BOLD, must be carefully considered. Funding for this study: This work was supported by the W.M. De Hoop Foundation, the Janivo Foundation and Friends of UM C Utrecht & Wilhelmina Children’s Hospital. Ethics committee - additional information: NedMec (study number 21-406) Author Disclosures: Kees Braun: Nothing to disclose Jan Willem Dankbaar: Nothing to disclose Bart Van Der Zwan: Nothing to disclose Alex Bhogal: Nothing to disclose Pieter Deckers: Nothing to disclose Bart Van Der Worp: Board Member: Past president at the European Stroke Organization Advisory Board: Liva Nova and Bayer Gr ant Recipient: Dutch Heart Foundation, Stryker, and the European Union Simone Uniken Venema: Nothing to disclose Jeroen Siero: Nothing to disclose Jeroen Hendrikse: Nothing to disclose Predictive Value of Venous Outflow in SAH (PreViOS) *H. Briody*, J. Henry, R. Bruen, P. Mchugh, P. Roha n, M. Javadpour, P. Nicholson; Dublin/IE ([email protected]) Purpose or Learning Objective: Favorable cortical venous outflow (VO) is linked to better outcomes in acute ischemic stroke. It’s role in aneurysmal subarachnoid hemorrhage (aSAH) remains unclear. Thi s study investigates the association between VO profiles and functional outc omes in aSAH.

Methods

or Background: Patients with aSAH referred to a tertiary neurosurgery center between 2016 and 2023 were incl uded if presentation computed tomographic angiography (CTA) demonstrated satisfactory venous system opacification. VO was assessed using the cor tical vein opacification score (COVES). The primary outcome was poor functio nal outcome (Glasgow Outcome Scale [GOS] 1-3) at 90 days. Associations b etween COVES and outcomes were assessed using univariable and multiv ariable (adjusted for World Federation of Neurosurgical Societies [WFNS] grade) binomial regression.

Results

or Findings: Of 675 patients with aSAH and available CTA, 204 (30%) met inclusion criteria. The median age was 54 .2 years (range 12-85). 182 (89%) had favorable VO. No significant associat ion was found between unfavorable VO and poor 90-day functional outcome ( RR 0.78, 95% CI 0.38- 1.58, p=0.48), even after adjusting for WFNS grade (RR 1.04, 95% CI 0.38- 2.83, p=0.94). Unfavorable VO was not associated wi th poor outcome at discharge, need for cerebrospinal fluid diversion, or shunt dependence.

Conclusion

The study failed to demonstrate a link between veno us outflow (as measured by COVES) and outcomes in aSAH. This i s an important negative finding. It suggests that, unlike in ische mic stroke, venous outflow might not be a major determinant of outcome in aSAH . Prospective studies are needed to definitively assess the role of VO in aSA H.

Limitations

This retrospective, single-center study may be subj ect to selection bias due to the inclusion criteria requir ing adequate jugular bulb opacification on CTA. The impact of delayed cerebra l ischemia, a major determinant of outcome in aSAH, was not specificall y assessed. Funding for this study: None Ethics committee - additional information: Institutional review board Thursday Abstract-based Programme 86 Author Disclosures: Patrick Nicholson: Nothing to disclose Mohsen Javadpour: Nothing to disclose Paul Mchugh: Nothing to disclose Richard Bruen: Nothing to disclose Hayley Briody: Nothing to disclose Pat Rohan: Nothing to disclose Jack Henry: Nothing to disclose Enhanced Detection of Cerebral Lesions in Cerebral Amyloid Angiopathy Using 7T MRI: Insights into Cognitive Correlation a nd Clinical Implications *D. Botta*, A. Cusin, L. Sveikata, K-O. Loevblad, F . T. Kurz; Geneva/CH Purpose or Learning Objective: The aim of this study is to evaluate the detection of cerebral microbleeds (CMBs) and cortic al microinfarcts using 7 Tesla (7T) MRI in patients with cerebral amyloid an giopathy (CAA) and to correlate these findings with cognitive performance as measured by the Montreal Cognitive Assessment (MoCA).

Methods

or Background: 19 patients with probable or possible CAA were scanned using both 7T and 3T MRI. Imaging protocols at 7T included susceptibility-weighted imaging (SWI) for CMB detec tion with an in-plane resolution of 0.15x0.15mm, and 3D FLAIR and T1 MP2R AGE sequences for cortical microinfarcts detection. Cognitive functio n was assessed using the MoCA score. Statistical analyses were conducted to assess correlations between lesion burden and MoCA scores.

Results

or Findings: 7T MRI detected 379 CMBs compared to 179 at 3T (average: 18.37±34.18 CMBs at 7T vs 9.53±15.03 at 3T). Additionally, 7T identified 27 cortical microinfarcts across 5 patie nts, while none were detected at 3T. MoCA scores ranged from 9 to 30 with a mean of 23.0±4.82. Weak correlations were found between the number of CMBs and MoCA scores (7T: r =-0.28; 3T: r =-0.24).

Conclusion

7T MRI at high-resolution is superior to 3T for det ecting CMBs and cortical microinfarcts in CAA patients. However , the weak correlation between lesion burden and cognitive decline suggest s that other factors may also contribute to cognitive impairment in these pa tients.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Written informed consent was obtained from all participants, the study was appro ved by the institutional review board of Geneva University Hospitals. Author Disclosures: Professor Karl-Olof Loevblad: Nothing to disclose Lukas Sveikata: Nothing to disclose Alexandre Cusin: Nothing to disclose Felix T Kurz: Nothing to disclose Daniele Botta: Nothing to disclose Diagnostic Performance of Low-Dose Cerebral CTA Ima ges Using Artificial Intelligence Iterative Reconstruction fo r Differentiating intracranial Aneurysms and Infundibula H. Chen¹, S. Xu², G. Zhang², *J. Wang*¹, X. Yin¹; ¹ Baoding/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: Intracranial aneurysms and infundibula with similar morphology and anatomical location are diff icult to distinguish using low-dose cerebral CT angiography (CTA). This study evaluated the diagnostic performance of cerebral low-dose CTA with artificia l intelligence iterative reconstruction (AIIR) for differentiating intracran ial aneurysms and infundibula.

Methods

or Background: Sixty-four patients (38 male, mean age 62.2 ± 12.5 years) with suspected intracranial aneurysms were p rospectively enrolled. Each patient underwent routine-dose (RD) and low-do se (LD) cerebral CTA. The RD protocol used 100kVp, ref. 180mAs, and hybri d iterative reconstruction (HIR), whereas the LD protocol used 100kVp, ref. 30 mAs, and AIIR. Two radiologists, blinded to scan/reconstruction parame ters, independently detected aneurysms and infundibula on low-dose imag es. The diagnostic reports of RD CTA served as references. Diagnostic performance in differentiating aneurysms and infundibula was asses sed using receiver operating characteristic (ROC) analysis, calculatin g sensitivity, specificity, positive predictive value (PPV), negative predictiv e value (NPV), accuracy, and area under the curve (AUC) with 95% confidence inte rval (CI).

Results

or Findings: A total of 64 lesions were identified, including 44 aneurysms and 20 infundibula. Reader 1 detected 62 out of 64 lesions (96.9%) on low-dose images, while Reader 2 detected 61 out of 64 (95.3%). Two infundibula with sizes of 1.5 mm and 2.4 mm were mi ssed by both readers, whereas one aneurysm of 1.8 mm was missed by Reader 2. In differentiating aneurysms and infundibula, the sensitivity, specifi city, PPV, NPV, diagnostic accuracy, and AUC for Reader 1 were 97.72%, 100%, 1 00%, 94.74%, 98.39%, and 0.989 (95% CI: 0.963–1.015), while for Reader 2 they were 97.67%, 100%, 100%, 94.74%, 98.36%, and 0.988 (95% CI: 0.96 2–1.015).

Conclusion

The AIIR shows the potential in reducing the radiat ion dose of the cerebral CTA when diagnosing intracranial aneurysms and infundibula.

Limitations

Not applicable. Funding for this study: the Key Research and Development Program of Hebei Province (grant number 202330604010017) Ethics committee - additional information: This study was approved by the local institutional review board. Author Disclosures: Guozhi Zhang: Employee: at United Imaging Healthcar e Shijie Xu: Nothing to disclose Jianing Wang: Nothing to disclose Haoyan Chen: Nothing to disclose Xiaoping Yin: Nothing to disclose Temporal muscle trophicity as a prognostic factor f or functional recovery in non-traumatic intracerebral hemorrhage *S. Nataf*¹, O. Curtinot², T. Maghfour², G. Bouloui s², A. Aignatoaie¹, C. Ozsancak¹, M. Pasi², C. Cohen¹; ¹Orleans/FR, ²To urs/FR ([email protected]) Purpose or Learning Objective: Intracerebral hemorrhage (ICH) is associated with poor outcome. Identifying patients with higher risk of disability is a key feature of optimal care. Recently, tempora l muscle thickness (TMT) has been shown to predict ability after ischemic st roke. We explored the relationship between temporal muscle trophicity and the functional ability of non-traumatic ICH patients.

Methods

or Background: Patients with acute non-traumatic ICH (2021-2022) from two university centers were retrospectively in cluded. Imaging ICH parameters were retrieved (e.g. volume, location). Temporal trophicity was assessed through thickness (TMT), area (TMA) and de nsity (TMD) on baseline brain computed tomography (CT). Good prognosis at 3 -6 months was defined as a modified Rankin Scale (mRS)<3. Association bet ween TMT, TMA and TMD and 3-6months mRS was evaluated using student T -test.

Results

or Findings: Among 453 ICH patients, 216 with both CT and 3- 6months mRS were analyzed (49% women, 70±15.9y). Me an hemorrhage volume was 47.3±46mm³, mainly lobar (46%) and locat ed in basal ganglia (41%). Mean TMT was 7.32±2.70mm in the good prognos is group, vs 6.37±2.13mm in the poor prognosis group (p=0.005). Mean TMA was 408±203mm² in the good prognosis group, vs 343±161mm² in the poor prognosis group (p=0.01). Prognosis was not signifi catively associated with TMD (p=0.051). 3-6months mRS was associated with he morrhage volume (p=0.01).

Conclusion

To our knowledge, this is the first study to analyz e and demonstrate that temporal muscle trophicity serves as a prognostic factor in intracerebral hemorrhage (ICH). Temporal muscle thi ckness (TMT) is a simple and practical indicator for assessing overall healt h and guiding patients toward targeted rehabilitation.

Limitations

Since many patients underwent only MRI, further res earch is needed to evaluate the association between temporal muscle trophicity and prognosis using this imaging modality. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was approved by a Comité de protection des personnes (CPP). Author Disclosures: Olivier Curtinot: Nothing to disclose Marco Pasi: Nothing to disclose Simon Nataf: Nothing to disclose Grégoire Boulouis: Nothing to disclose Andreea Aignatoaie: Nothing to disclose Canan Ozsancak: Nothing to disclose Clara Cohen: Nothing to disclose Tasnym Maghfour: Nothing to disclose 3D variable flip angle turbo spin echo black-blood MRI for diagnosing cerebral venous thrombosis: a systematic review and meta-analysis *A. Akhavi Milani*; Tabriz/IR ([email protected]) Purpose or Learning Objective: To evaluate the performance of 3D variable flip angle turbo spin echo black-blood MRI (BB-MRI) in diagnosing cerebral venous thrombosis (CVT). The secondary objectives i ncluded, comparing BB- MRI with conventional MRI, MRV, MPRAGE, and SWI in diagnosing CVT, and assessing the utility of BB-MRI in estimating throm bus age.

Methods

or Background: This study was registered in PROSPERO [ID: blinded]. The PubMed/MEDLINE, Web of Science, Scopu s, and Embase databases were systematically searched and studies were selected based on predefined eligibility criteria. The risk of bias w as assessed using the QUADAS-2 tool. Meta-analysis was performed to calcu late pooled sensitivity, specificity, and AUC. Thursday Abstract-based Programme 87

Results

or Findings: Nine studies were included in the review. Two of th em insufficiently reported the quantitative data; ther efore, seven studies involving 176 CVT patients and 217 controls, encompassing 610 thrombosed and 3,279 normal cerebral venous segments were included in th e meta-analysis. The studies demonstrated a high bias risk in the patien t selection and reference standard domains. The pooled sensitivity and specif icity of BB-MRI were 96% [95% CI: 92%–98%] and 96% [95% CI: 93%–98%] on a pa tient-based level, and 92% [95% CI: 87%–95%] and 98% [95% CI: 92%–99%] on a venous segment-based level, respectively. The AUC was 0.98 for patient-based data and 0.96 for venous segment-based data. For seconda ry objectives, a narrative summary indicated that BB-MRI outperforms conventional MRI, MRV, and MPRAGE. It also outperforms SWI in assessing ce rebral cortical veins. Moreover, BB-MRI can prove beneficial in thrombus a ge estimation.

Conclusion

BB-MRI demonstrates significant potential in diagno sing CVT. Further comparative studies are required to specify its role in clinical decision- making for CVT.

Limitations

The small number of the retrieved studies. Funding for this study: No funding was received for this study. Ethics committee - additional information: This was a systematic review study for which ethics committee approval was waive d. Author Disclosures: Ali Akhavi Milani: Nothing to disclose The probability of cerebral amyloid angiopathy acco rding to the Simplified Edinburgh CT criteria in a large, unsele cted lobar intracerebral hemorrhage population *A. Hillal*¹, T. Ullberg², J. Wassélius¹; ¹Lund/SE, ²Malmö/SE ([email protected]) Purpose or Learning Objective: Early identification of the underlying cause of intracerebral hemorrhage (ICH) is important for tre atment and prognosis. This study aims to investigate the association of hemato ma volume and other clinical parameters on the distribution of Cerebral Amyloid Angiopathy (CAA) probability according to the simplified Edinburgh C T criteria in a large, unselected intracerebral hemorrhage (ICH) populatio n.

Methods

or Background: Patients with spontaneous ICH residing in Skane county registered with clinical data in the Swedish Stroke Register 2016–2020 were included. Radiological parameters were evaluat ed using baseline non- contrast CT (NCCT) for categorization according to the simplified Edinburgh CT criteria by the presence of subarachnoid hemorrh age (SAH) and fingerlike- projections (FLP). Multivariable logistic regressio n analysis was used to determine factors associated with an increased (int ermediate/high) CAA probability.

Results

or Findings: Of 666 patients with lobar ICH, 190 (29%) had high CAA probability, 92 (14%) intermediate, and 384 (58%) l ow CAA probability. Patients with increased CAA probability more often presented with decreased level of consciousness, larger hematoma volumes, an d had higher 90-day mortality. Female sex, age, and increasing baseline hematoma volume (Odds Ratio up to 30) were associated with increased odds of having an increased CAA probability.

Conclusion

We identified a strong association between baseline hematoma volume and an increased probability of CAA in lobar ICH patients on NCCT, indicating that large hematoma volumes per se may c ontribute to the occurrence of FLP and SAH, and act as a confounder for the Simplified Edinburgh CT Criteria. Validation against MRI is wa rranted.

Limitations

The lack of MRI studies to allow for the correlatio n between CT imaging characteristics and the gold imaging standa rd MRI Boston criteria Funding for this study: ALF grants to Teresa Ullberg and Johan Wasselius, the Crafoord Foundation to JW, VINNOVA to Johan Was selius, and by SUS Stiftelser & Fonder to Johan Wasselius. Ethics committee - additional information: This study was approved, and individual informed consent was waived by the Swedi sh Ethical Review Authority (reference number 2020-06800). Author Disclosures: Teresa Ullberg: Nothing to disclose Johan Wassélius: Nothing to disclose Amir Hillal: Nothing to disclose Evaluation of a FLAIR Hyperintensity Algorithm for the prediction of DWI- FLAIR Mismatch in Acute Ischemic Stroke *C. M. Offersen*, J. Johansen, A. H. Brandt, T. C. Truelsen, A. Pai, S. Darkner, M. B. Bachmann Nielsen, J. F. Carlsen; Copenhagen/D K Purpose or Learning Objective: Moderate inter-rater variability of Diffusion- Weighted Imaging (DWI) – Fluid-Attenuated Inversion Recovery (FLAIR) mismatch in wake-up stroke raises concerns about th e potential exclusion of eligible patients for treatment with thrombolysis. A novel FLAIR algorithm has shown promising potential to perform this mismatch assessment but has only been evaluated on a small dataset. In the present s tudy, we aimed to evaluate an updated version of the FLAIR algorithm for predi cting the DWI-FLAIR mismatch in a large cohort of wake-up stroke patien ts.

Methods

or Background: We conducted a single-centre, retrospective study. A consecutive cohort of patients suspected of wake- up stroke, who underwent MRI between 2019 and 2021 was included. Two radiolo gists and one resident, blinded to clinical data, manually assessed DWI-FLA IR mismatch according to the current clinically used binary categorisation. Cohens Kappa was calculated for the inter-rater agreement. The FLAIR algorithm depends on a DWI segmentation. We used a commercial DWI segmentation model and then tested the ability of the FLAIR algorithm on the id entified ischemic lesions to predict manual mismatch. This was analysed with log istic regression test.

Results

or Findings: The DWI model identified 495 patients with possible ischemic lesions. Manual radiological assessments f ound 365 of those patients to have actual ischemic lesions. Inter-rater agreem ent for binary DWI-FLAIR mismatch assessment was moderate (κ = 0.461 ± 0.028SD). We found a high accuracy (82.3 ± 3.3SD). Low sensitivity (59.1 ± 9.1SD). Specificity (91.5 ± 4.2SD) and AUC (0.845 ± 0.03SD) were high in the au tomatic mismatch assessment.

Conclusion

The FLAIR algorithm predicted DWI-FLAIR mismatch st atus with a high AUC, which suggests the algorithm could prov ide a more standardized decision on mismatch, and reduce the inter-rater va riability through an

Objective

assessment to assist the radiologist.

Limitations

This was a single-centre, retrospective study. Funding for this study: Innovation Fund Denmark Ethics committee - additional information: Danish National Center for Ethics Author Disclosures: Thomas Clement Truelsen: Nothing to disclose Cecilie Mørck Offersen: Grant Recipient: Received f unding from IFD for part of the work focused on automated assessment in wake up stroke. This project was in collaboration with Cerebriu A/S. Michael B Bachmann Nielsen: Nothing to disclose Akshay Pai: Founder: Founder and CTO at Cerebriu A/ S, a company that makes automated diagnostic software for radiology. Jacob Johansen: Other: Earlier employment as an ind ustrial PhD student at Cerebriu A/S Andreas Hjelm Brandt: Nothing to disclose Sune Darkner: Nothing to disclose Jonathan Frederik Carlsen: Nothing to disclose 10:00-11:00 Research Stage 1 Research Presentation Session: Head and Neck RPS 808 Imaging the skull base and face Moderator A. Bernaerts; Antwerp/BE ([email protected]) Dual-layer spectral detector CT for differentiating middle ear cholesteatoma and chronic suppurative otitis media *S. Zhou*¹, L. Mei¹, H. Liu¹, X. M. Liu², J. Li¹; ¹ Changsha/CN, ²Guangzhou/CN ([email protected]) Purpose or Learning Objective: To compare the diagnostic performance of dual-layer spectral detector CT (DLCT) and high-res olution CT (HRCT) in differentiating middle ear cholesteatoma and chroni c suppurative otitis media.

Methods

or Background: This prospective, institutional review board- approved study included sixty-six patients who were preliminary diagnosed as cholesteatoma or otitis media, and received DLCT sc anning before surgery. Thirty-three patients were finally diagnosed choles teatoma based on intraoperative or pathological findings. Two blinde d readers (Reader 1: one radiologist; Reader 2: one otologist) provided diag noses and diagnostic confidence scores using a five-point scale, based o n HRCT images and DLCT multi-parameter images, including virtual mono-ener getic image at 40keV (VMI 40keV) and effective atomic number (Zeff). Diagnost ic accuracy of HRCT and DLCT maps were compared using McNemar’s test. Inter observer agreement was evaluated by Kappa statistic.

Results

or Findings: HRCT and DLCT identified a total of 27/33 and 31/33 cholesteatomas by Reader 1, 19/33 and 30/33 cholest eatomas by Reader 2 respectively. The sensitivity, specificity, PPV, NP V, accuracy of HRCT and DECT by Reader 1 were 81.8, 75.8, 77.1, 68.2, 78.8% and 93.9, 69.7, 75.6, 92.0, 87.5%, by Reader 2 were 57.6, 90.9, 86.3, 68. 2, 74.2% and 90.9, 75.8, Thursday Abstract-based Programme 88 78.9, 89.3, 83.3%, respectively. Compared to HRCT, the diagnostic sensitivity of DLCT increased for both readers, with a statisti cally significantly improvement in Reader 2 (p＜0.05). Diagnostic confidence scores of DLCT versus HRCT by Reader 1 and Reader 2 both improved significantly (4.85±0.36 vs 3.77±1.30, 4.44±0.73 vs 3.45±0.98, respectively, all p＜0.05). Interobserver reproducibility was higher for diagno ses made with DLCT maps (k =0.717) than for that made with HRCT images (k = 0.495).

Conclusion

Spectral CT improved diagnostic performance and int erobserver reproducibility of determination of cholesteatoma v ersus otitis media.

Limitations

None Funding for this study: None Ethics committee - additional information: lRB of Xiangya Hospital Author Disclosures: Shuangyuan Zhou: Nothing to disclose Xiao Min Liu: Nothing to disclose Juan Li: Nothing to disclose Lingyun Mei: Nothing to disclose Heng Liu: Nothing to disclose Redefining Sinonasal Cancer response assessment to induction chemotherapy with tumor volumetry, results from two prospective multicentric trials: SINTART-1 and SINTART-2 P. Rondi¹, *E. Massoni*¹, A. Borghesi¹, P. Bossi², C. Resteghini², D. Farina¹, M. Ravanelli¹; ¹Brescia/IT, ²Milan/IT ([email protected]) Purpose or Learning Objective: Aim of this study is to identify the best response criteria in patients with sinonasal cancer undergoing induction chemotherapy (IC).

Methods

or Background: Patients enrolled in SINTART-1 and SINTART-2 were included in this study. Unidimensional diamete rs (antero-posterior, AP; latero-lateral, LL; cranio-caudal, CC), maximum axi al area (Amax) and volume (V) were performed on MRI by two radiologists. RECI ST 1.1 assessment was included as a categorical variable. Variables were evaluated at baseline, after 1st chemotherapy cycle and at best response. Intero bserver repeatability was analyzed. Stepwise univariable and multivariable Co x proportional-hazards regression models were used to correlate variables with Disease-Free Survival (DFS) and Overall Survival (OS).

Results

or Findings: 60 patients were included in this study. Interobser ver correlation at baseline and after 1st IC cycle was excellent for V (0.916 and 0.928 respectively), CC diameter (0.96 and 0,863) a nd AP diameter (0.846 and 0.796); good for LL diameter and moderate for Amax. RECIST 1.1 criteria after 1st IC cycle and at best response were not correlat ed with OS and DFS. Volume variation after 1st IC cycle and at best res ponse was the variable most correlated with OS (p<0.0001and p=0.002) and DFS (p <0.0001and p=0.005). At the multivariable analysis V variation after 1st IC cycle and at best response resulted to be the only variable statistically asso ciated with OS (p<0.001 and p=0.0019 respectively) and DFS (p=0.0004and p=0.004 respectively).

Conclusion

Volume variation should be preferred to RECIST 1.1 and other measurements as objective radiological response in sinonasal cancer.

Limitations

The main limitation of this study is that despite c onsidering two prospective studies the size of the cohort is small and this could reduce the statistical power. Funding for this study: This study has received no funding. Ethics committee - additional information: The ethics committee of participating centers have approved this study Author Disclosures: Elena Massoni: Nothing to disclose Andrea Borghesi: Nothing to disclose Paolo Bossi: Nothing to disclose Davide Farina: Nothing to disclose Paolo Rondi: Nothing to disclose Marco Ravanelli: Nothing to disclose Carlo Resteghini: Nothing to disclose Diagnostic accuracy of MRI for orbital and intracra nial invasion of sinonasal malignancies: a systematic review and met a-analysis *U. B. Abdullaeva*¹, B. Pape², J. Hirvonen³; ¹Tashk ent/UZ, ²Turku/FI, ³Tampere/FI ([email protected]) Purpose or Learning Objective: To review the diagnostic accuracy of MRI in detecting orbital and intracranial invasion of sino nasal malignancies using histopathological or surgical evidence as the refer ence standard.

Methods

or Background: The systematic review protocol was pre-registered in the Prospective Register of Systematic Reviews ( PROSPERO) under registration number CRD42024492090. A systematic se arch of the studies in English was conducted in PubMed and Embase, limited to articles published since 1990. We included studies that used preoperat ive MRI to detect intracranial and orbital invasion of sinonasal mali gnancies, using histological or surgical confirmation as the reference standard, an d reported patient numbers in each class required for assessing diagnostic acc uracy. The outcome measures were sensitivity, specificity, positive pr edictive value (PPV), and negative predictive value (NPV). Heterogeneity was assessed with the Higgins inconsistency test (I2).

Results

or Findings: Seven original articles with 546 subjects were incl uded in the review, six of these in the meta-analysis. P ooled overall accuracy for orbital invasion was higher at 0.88 (95% CI, 0.75-0 .94) than for intracranial invasion - 0.80 (95% CI, 0.76-0.83). Meta-analytic estimates and their 95% confidence intervals were as follows for intracrani al/orbital invasion: sensitivity 0.77 (0.69-0.83)/ 0.71 (0.40-0.90); specificity 0.7 9 (0.74-0.83)/0.91 (0.78-0.97); PPV 0.76 (0.64-0.85)/0.78 (0.61-0.88); and NPV 0.82 (0.72-0.89)/0.90 (0.63- 0.98). Significant heterogeneity was observed in th e Higgins inconsistency test (I2) for orbital invasion (84%, 83%, and 93% for se nsitivity, specificity, and NPV, respectively).

Conclusion

MRI yielded moderate to high diagnostic accuracy fo r intracranial and orbital invasion, but there are limitations lea ding to false diagnoses. Loss of the hypointense zone on MRI predicts dural invas ion. Infiltration of the extraconal fat beyond the periorbita is an MRI feat ure of orbital invasion.

Limitations

Limitations include a small number of predominantly retrospective studies, some with a small subset of patients. Funding for this study: Funding was provided by the Sigrid Jusélius Foundation, grant number 240053. Ethics committee - additional information: Since this is a systematic review, Institutional Review Board approval was not necessa ry. Author Disclosures: Bernd Pape: Nothing to disclose Jussi Hirvonen: Nothing to disclose Umida Bafoevna Abdullaeva: Nothing to disclose Advanced MRI Techniques for Evaluation of Sinonasal Masses: Exploring the Additive Utility *M. Saini*, S. Manchanda, A. S. Bhalla, D. Kandasam y; New Delhi/IN Purpose or Learning Objective: Sinonasal area is affected by a wide spectrum of benign and malignant tumours presenting with nonspecific symptoms and differentiation solely based on conven tional magnetic resonance imaging has only limited specificity To evaluate the role of newer imaging techniques li ke intravoxel incoherent motion (IVIM), diffusion kurtosis imaging (DKI) and dynamic contrast enhanced MRI (DCE-MRI) in differentiating benign and maligna nt sinonasal masses

Methods

or Background: A prospective study was performed on 30 patients with sinonasal masses (18 malignant and 12 benign) who underwent routine MRI, DWI, IVIM and DCE MRI. Apparent diffusion coef ficient (ADC) from diffusion weighted imaging, true diffusion coeffici ent (Dt), Pseudodiffusion coefficient (Dx), perfusion fraction (f) from IVIM, apparent kurtosis coefficient (Kapp) and apparent diffusion coefficient (Dapp) fr om DKI, semiquantitative and quantitative perfusion parameters from DCEMRI w ere measured and compared between two groups.

Results

or Findings: The mean ADC, Dt and Dapp values were significantly lower in malignant sinonasal lesions than in benign sinonasal lesions with p values of 0.000, 0.015 and 0.030 respectively. The mean Kapp value was higher in malignant lesions than in benign lesions (p value of 0.001). There was no significant difference Dx, f and in semiquan titative and quantitative perfusion parameters.

Conclusion

The mean ADC derived from the DWI, Dt derived from IVIM, and Dapp & Kapp derived from the DKI can be used as a n on-invasive method to differentiate benign and malignant sinonasal masses . Among these, ADC is the best parameter to differentiate, however there is no incremental role of DKI and IVIM over conventional DWI. The perfusion param eters showed no significant difference

Limitations

The sample size was small and a heterogeneous group of pathologies were included in the final analysis. In addition, the b values were set arbitrarily in the IVIM-DKI sequence. Funding for this study: No funding was provided for this study Ethics committee - additional information: Study was preapproved by the Institute Review Board (IRB) [Ref No: IECPG-487/25. 08.2021]. Author Disclosures: Manish Saini: Nothing to disclose Ashu Seith Bhalla: Nothing to disclose Smita Manchanda: Nothing to disclose Devasenathipathy Kandasamy: Nothing to disclose Thursday Abstract-based Programme 89 Presence of bone exposure to the aerodigestive trac t as an important imaging feature in patients with skull base osteomy elitis following prior irradiation for nasopharyngeal carcinoma *H. S. Leung*, K. K. F. Tsoi, Q-Y. H. Ai, A. D. Kin g; Hong Kong/HK Purpose or Learning Objective: Osteoradionecrosis is one the long-term complications associated with radiotherapy for naso pharyngeal carcinoma (NPC), and may result in the significant complicati ons particularly skull base osteomyelitis (SBOM). A limited number of clinical case series have reported on SBOM and their predictors, imaging features and associations with other radiation-induced complications remain poorly under stood. This study is to evaluate the imaging factors associated with SBOM i n patients with previous irradiation for NPC.

Methods

or Background: This is a retrospective matched case-control study, of patients with clinically proven SBOM and compute r tomography (CT) scan performed at diagnosis selected as cases, while con trols were selected within NPC post-RT patients and without SBOM, matched by i nitial staging and time from initial treatment. CT studies were reviewed fo r the presence of bone exposure to aerodigestive tract, bony sclerosis, bo ne loss and dehiscence and abscess formation; baseline demographics and clinic al outcomes were analyzed by logistic regression and survival analys es.

Results

or Findings: A total of 31 SBOM cases and 31 controls were analysed. Presence of bone exposure to aerodigestiv e tract is the only independent factor associated with SBOM (p<0.001 by McNemar’s test), while the degree of bone loss also shows borderline signi ficance (p=0.052 by Wilcoxon sign rank test). Bone exposure to upper ae rodigestive tract remains significant upon regression controlling for bone lo ss and staging of initial NPC. SBOM patients had worse survival with a higher inci dence of other RT-related complications including carotid occlusion, blowout or RT-induced malignancy.

Conclusion

The presence of bone exposure to upper aerodigestiv e tract is an independent factor associated with SBOM, which coul d be helpful in early identification and treatment to avoid complications of SBOM which adversely affects survival.

Limitations

Recall bias from retrospective study. Funding for this study: Nil Ethics committee - additional information: This study has been approved by Joint CUHK-NTEC Clinical Research Ethics Committee; The Chinese University of Hong Kong (Reference number: CREC 202 4.306) Author Disclosures: Kelvin Kam Fai Tsoi: Nothing to disclose Ann Dorothy King: Nothing to disclose Ho Sang Leung: Nothing to disclose Qi-Yong Hemis Ai: Nothing to disclose Can HU analysis by used for jaw lesions differentia tion? *C. Nadler*, Y. Pakanaev-Levi, H. Rushinek, N. Yavn ai, Y. Zadik, I. Zeevi; Jerusalem/IL ([email protected]) Purpose or Learning Objective: Bone lesions demonstrated on Computed Tomography (CT) images may be differentiated by sev eral methods including Hounsfield Unit (HU) analysis. However, the routine use of this method in pre- operative assessment of jaw lesions remains underex plored. We aimed to distinguish using HU analysis 3 types of jaw lesion s.

Methods

or Background: We retrospectively included pre-operative Multidetector CT (MDCT) scans of patients with unil ocular hypodense jaw lesions, with histologically proven, non-inflame ei ther odontogenic keratocyst OKC, central giant cell granuloma CGCG or unicystic ameloblastoma UA. Demographic data and anonymized DICOM files were re trieved. Two observers, blinded to the lesions’ diagnosis, measu red three HU values for each lesion, in axial slices, on Philips IntelliSpa ce Portal software. Statistical analysis included intra and inter-observer reliabil ity and validity evaluations as well as comparisons of mean HU values between the d ifferent lesions and between the same lesion in different jaws

Results

or Findings: Our cohort included 30 jaw lesions (17 OKCs, 8 CGCG s and 5 UAs). Mean HU values for OKCs, CGCGs and UAs were 27.99±13.8, 70.68±46.3 and 31.38±7.4, respectively. Statistically significant difference was found between mean HU values of OKC, CGCG and UA (P =0.035). Following additional pooled analysis, mean HU values of CGCG was statistically higher than OKC and UA (<0.001). No statistically differen ce was found between HU values of different lesions in different jaws.

Conclusion

Pre-operative differentiation using HU analysis may be used to diagnose CGCG from OKC and UA. Future multi-center studies with additional types of lesions are needed to substantiate our res ults

Limitations

Our limitation included: small sample size, as is a result of stringent inclusion criteria and the transition of referrals from MDCT to CBCT and the fact that all cases were from a single medi cal center. Funding for this study: None Ethics committee - additional information: The study protocol was approved by the Institutional Review Board Author Disclosures: Yehuda Zadik: Nothing to disclose Yehuda Pakanaev-Levi: Nothing to disclose Itai Zeevi: Nothing to disclose Nirit Yavnai: Nothing to disclose Chen Nadler: Nothing to disclose Heli Rushinek: Nothing to disclose Multiparametric magnetic resonance imaging in deter mining disease activity of thyroid-associated ophthalmopathy: Adde d value from dynamic contrast-enhanced and diffusion-weighted im aging *X-Y. Pu*, H. Hu, J. Zhou, L. Jinling, X-Q. Xu, F-Y . Wu; Nanjing/CN ([email protected]) Purpose or Learning Objective: To evaluate the performance of dynamic contrast-enhanced MRI (DCE-MRI) and diffusion-weigh ted imaging (DWI) in determining disease activity of thyroid-associated ophthalmopathy (TAO) and to establish their additional value for staging TAO compared to conventional T2-weighted imaging (T2WI).

Methods

or Background: Seventy-two patients with TAO (48 active, 96 eyes; 24 inactive, 48 eyes) who underwent DCE, DWI and T2 WI with fat suppression were prospectively enrolled. Simplified histogram p arameters (mean, max, min) of DCE-MRI derived parameters (Ktrans, Kep, Ve), ap parent diffusion coefficient (ADC) and signal intensity ratio (SIR) at extraocular muscles were calculated for each orbit and compared between the active and inactive groups. Multivariate analyses were used to identify independent predictors. Receiver operating characteristic curves analyses a nd DeLong tests were performed to evaluate and compare the performances of the identified significant imaging parameters and their combinatio ns.

Results

or Findings: Active TAO patients showed significantly higher mea n and maximum Ve, higher minimum, mean and maximum AD C, higher minimum, mean and maximum SIR than inactive patient s (P < 0.05). The mean Ve, mean ADC and mean SIR were found to be independ ent predictors for active TAO (all P < 0.05). Combination of mean Ve, mean ADC and mean SIR outperformed mean SIR alone in staging TAO (AUC, 0. 839 vs 0.769, P = 0.016).

Conclusion

DCE-MRI and DWI could help to determine the disease activity of TAO. DCE-MRI-derived Ve and DWI-derived ADC values could provide added value to conventional T2WI-derived SIR in staging T AO

Limitations

First, the sample size was relatively small. More r esearch should be done to increase the number of datasets to verif y our findings. Second, only multiparametric MRI-derived simplified histogram me trics were analysed. Future studies integrating radiomics and machine le arning have the potential to further improve staging performance. Funding for this study: Jiangsu Province Hospital (the First Affiliated Hos pital with Nanjing Medical University) Clinical Capacity Enhancement Project (JSPH-MC-2021-8 to Xiao-Quan Xu) , Jiangsu Province Capability Improvement Project through Science, Technology and Education (JSDW202243 to Fei-Yun Wu) and National Natural Sci ence Foundation of China (NSFC) (81801659 to Hao Hu), Ethics committee - additional information: This study was approved by the Institutional Review Board of the First Affiliated Hospital of Nanjing Medical University (IRB No.2021-SRFA-024) Author Disclosures: Fei-Yun Wu: Nothing to disclose Jiang Zhou: Nothing to disclose Lu Jinling: Nothing to disclose Xiao-Quan Xu: Nothing to disclose Hao Hu: Nothing to disclose Xiong-Ying Pu: Nothing to disclose Thursday Abstract-based Programme 90 10:00-11:00 Research Stage 2 Research Presentation Session: Musculoskeletal RPS 810 Imaging in metabolic and inflammatory arthropathies Moderator T. Diekhoff; Berlin/DE ([email protected]) Author Disclosures: Torsten Diekhoff: Advisory Board: Eli Lilly, AbbVie , UCB; Speaker: Novartis, MSD, UCB, Janssen, Eli Lilly, Canon MS, Berlinflame , Bracco Evaluation of contrast-enhanced ultrasound for rheu matoid arthritis activity in patients who do not respond to second-l ine biologic therapy compared with superb microvascular imaging: first r esults *S. Lavalle*¹, A. Montana², Y. Dal Bosco², F. Aiell o³, R. Foti², G. Privitera², R. Foti², P. Romeo²; ¹Milan/IT, ²Catania/IT, ³Enna/ IT ([email protected]) Purpose or Learning Objective: Detection of synovitis is essential for assessing rheumatoid arthritis (RA) activity and ch anging the therapy. This study aim to evaluate the level of agreement and co rrelation between DAS 28 (Disease activity score 28) and contrast-enhanced u ltrasound (CEUS) and Superb microvascular imaging (SMI) in the classific ation of disease severity index in patients with RA who did not respond to se cond-line biologic therapy.

Methods

or Background: SMI and CEUS were applied to 25 patients with active RA not respond to second-line biologic thera py. We evaluate the radiocarpal joint of both wrists. Differences in po sitive synovial vascularity (SV) and its semi-quantitative scale were observed, and the correlations of SMI and CEUS results with DAS-28.

Results

or Findings: The results indicate that CEUS method shows high- moderate agreement with DAS 28 clinical method (Kap pa = 0.406), 95% CI (0.1916, 0.5854), p = 0.00584, while SMI has weaker agreement (Kappa = 0.121) 95% CI (0.0098, 0.2466), p = 0.0846 (Kappa = 0.121). The correlation between CEUS and SMI is very strong (ρ = 0.828), CI 95% (0.6911, 0.9044), suggesting that the two radiological methods tend t o produce very similar classifications, although CEUS is more in line with the gold standard. The Wilcoxon signed-rank test showed significant differ ences between DAS 28 and each of the two radiological methods, with CEUS app earing closer to the clinical method

Conclusion

Use of CEUS to detect vessels in the synovium and v isualization of local SV is the method that most correlates with disease severity in relation to DAS 28 in patients with synovial arthritis who d o not respond to second-line biologic therapy compared with SMI.

Limitations

Small sample size and the need for larger multicent er studies to confirm our findings. Funding for this study: No funding Ethics committee - additional information: Not applicable Author Disclosures: Placido Romeo: Nothing to disclose Salvatore Lavalle: Nothing to disclose Roberta Foti: Nothing to disclose Fabio Aiello: Nothing to disclose Rosario Foti: Nothing to disclose Angelo Montana: Nothing to disclose Giambattista Privitera: Nothing to disclose Ylenia Dal Bosco: Nothing to disclose Role of imaging in inflammatory hand arthritis with diagnostic ambiguity: how complementary MRI findings in clinically establ ished DIP arthritis may facilitate the specific diagnosis *Y. Yaraşir*, G. Ayan, H. Avci, L. Kılıç, Ü. Aydingöz, U. Ka lyoncu, A. E. Yildiz; Ankara/TR ([email protected]) Purpose or Learning Objective: Ascertaining whether synovium or synovioentheseal complex (SEC) is predominantly inv olved and pattern of any degeneration would help radiologists in distinguish ing inflammatory hand arthritis (IHA). We aimed to characterize the role of MRI in reaching specific diagnosis in IHA.

Methods

or Background: Patients aged ≥18 years with suspicious IHA in at least one joint (but not treated other than with NS AIDs) were consecutively enrolled in this prospective study. 3T-MRI with a f ine-tuned protocol was utilized, whereby differential diagnoses were made according to the predominant involvement of synovium or SEC, and/or specific degenerative findings. Physical examination, laboratory and imag ing findings, treatment response, and already-established rheumatological c lassification criteria were used to reach final diagnosis.

Results

or Findings: Of 80 patients initially enrolled, 57 (42 females; mean age, 54 [range, 28-79]) constituted the final group with eventual clinical diagnoses of 11 psoriatic arthritis (PsA), 14 rheum atoid arthritis (RA), 11 erosive osteoarthritis and/or calcium pyrophosphate dihydrate deposition disease, 21 arthritis with distal interphalangeal j oint involvement (ADIPI) not otherwise classified into any group. MRI revealed n o difference between PsA and ADIPI groups, except for nail-bed enthesitis. C omparison between PsA and RA disclosed that enthesitis (p=0.033) and peri articular soft tissue edema (p=0.042) were more frequent in PsA. When ADIPI and PsA groups were combined, enthesitis and periarticular soft tissue edema were more common than in other groups (p<0.001). Those with enthesit is were 24 times more likely to be in the PsA+ADIPI group than those without ent hesitis (95% CI: 2.6–63.3). Accurate classification rate of the model was 83.7% , and area under the curve (AUC) value was 0.81.

Conclusion

SEC inflammation and periarticular edema on MRI are strong predictors of PsA, especially in patients with DIP arthritis who don’t meet rheumatological classification criteria.

Limitations

Small sample size Funding for this study: Funding was provided by Hacettepe University Scientific Research Projects Coordination Unit Ethics committee - additional information: Our study was approved by Hacettepe University Clinical Studies Ethics Commit tee (2021/23-20) Author Disclosures: Umut Kalyoncu: Nothing to disclose Gizem Ayan: Nothing to disclose Levent Kılıç: Nothing to disclose Adalet Elcin Yildiz: Nothing to disclose Yasin Yaraşır: Nothing to disclose Üstün Aydingöz: Nothing to disclose Hanife Avci: Nothing to disclose Hemosiderin Quantification in Hemophilic Arthropath y of the Knee using Quantitative Magnetic Resonance Imaging *S. Sedaghat*¹, P. Leutz-Schmidt¹, J. Park², E. Fu² , H. Jang²; ¹Heidelberg/DE, ²Davis/US ([email protected]) Purpose or Learning Objective: This study aims to establish quantitative magnetic resonance imaging (qMRI) as a precise, non invasive tool for evaluating hemosiderin deposition in hemophilic art hropathy (HA) of the knee.

Methods

or Background: This prospective study included nine ex-vivo knee synovial tissues from HA patients and the same tiss ues from healthy controls. All tissues underwent standardized qMRI protocols u sing quantitative susceptibility mapping (QSM), based on ultrashort e cho time MRI, which was optimized to detect and quantify hemosiderin deposi ts. Also, standard MRI sequences were employed. The HA tissues were proces sed histologically using Perl’s Prussian Blue (PPB) staining to identi fy iron contents. Several regions of interest were drawn in each tissue. Usin g specialized algorithms, voxel-wise magnetic susceptibility was calculated t o assess iron deposition within the knee tissues objectively.

Results

or Findings: qMRI demonstrated high sensitivity in detecting and quantifying hemosiderin deposition, whereas convent ional imaging showed no abnormalities. The estimated susceptibility values (ESVs) showed significant differences between HA and control samples. HA tiss ues presented a mean ESV of 0.48 ± 1.08 ppm and control tissues of 0.13 ± 0.12 ppm (p<0.05). A significant linear correlation was found between th e iron level quantified by histology and the ESV estimated by QSM (R = 0.908, p < 0.01). There was a significant difference in the susceptibility in hig h load (HL) tissues compared to low load (LL) tissues (ESV = 5.57 ± 1.23 ppm for HL vs. 0.57 ± 0.85 ppm for LL, p<0.001).

Conclusion

This study establishes qMRI, particularly QSM, as a noninvasive and highly sensitive technique for quantifying hemo siderin in HA of the knee. By providing an objective measure of hemosiderin de position, qMRI offers potential as a tool for early diagnosis and disease monitoring in patients with hemophilic arthropathy.

Limitations

Main limitation: ex-vivo study design. Funding for this study: This study was funded by the National Institutes of Health (NIH R01AR078877) and the Deutsche Forschung sgemeinschaft (DFG SE 3272/1-1) Ethics committee - additional information: The Institutional Review Board (IRB) of the University of California San Diego app roved the study. Thursday Abstract-based Programme 91 Author Disclosures: Jinil Park: Nothing to disclose Sam Sedaghat: Nothing to disclose Hyungseok Jang: Nothing to disclose Eddie Fu: Nothing to disclose Patricia Leutz-Schmidt: Nothing to disclose Diabetes-related foot disease: the added value of z te M. Di Diego, D. Perla, A. Infante, A. M. Costantini , M. L. Angeli, C. Gullì, *G. Ferrara*; Rome/IT ([email protected]) Purpose or Learning Objective: The primary endpoint is to evaluate the additional diagnostic information obtained from ZTE sequences added to the standard MRI protocol (particularly compared to T1- weighted images) for improved assessment of bone structures in diabetes- related foot disease. Secondary endpoints include assessing accuracy comp ared to CT in the evaluation of bone alterations and investigating in terobserver agreement between three musculoskeletal radiologists with dif ferent expertise.

Methods

or Background: This retrospective single-center study analyzed 32 MRIs with ZTE sequences from 31 patients (22 males, 10 females; age range: 49-87 years) from March 2024 to September 2024. Inc lusion criteria included patients >18 years old with a confirmed diagnosis o f diabetes mellitus. In 11 cases, comparison between ZTE sequences and CT bone imaging was possible.

Results

or Findings: ZTE sequences compared to standard MRI protocol (particularly to T1-weighted images) were superior in identifying soft tissue air (43% vs 31%), bone pneumatosis (19% vs 9%), bone er osions (77% vs 67%), bone exposure (20% vs 15%), bone sclerosis (49% vs 41%), periosteal reaction (19% vs 17%), and bone fragments (39% vs 2 5%). Compared to CT, ZTE demonstrated high sensitivity (75-100%) and spe cificity (92-100%) for all musculoskeletal alterations analyzed. Interobserver agreement between musculoskeletal radiologists was excellent (k-range 0.82).

Conclusion

ZTE sequences provided additional musculoskeletal i nformation compared to T1-weighted MRI sequences, particularly for the morphological evaluation of bones affected by diabetes-related fo ot disease. This is crucial when hypointensity on T1-weighted images reduces th e ability to visualize bone structures and their alterations. Moreover, th e high sensitivity and specificity values compared to CT suggest that ZTE is a valid alternative. The interobserver agreement for the qualitative evaluat ion of ZTE sequences was excellent, indicating the ease of interpretation.

Limitations

Small sample size. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Amato Infante: Nothing to disclose Consolato Gullì: Nothing to disclose Maria Luigia Angeli: Nothing to disclose Daniele Perla: Nothing to disclose Giuseppe Ferrara: Nothing to disclose Mario Di Diego: Nothing to disclose Alessandro Maria Costantini: Nothing to disclose The assessment of rheumatoid arthritis and other ar thropathies through power Doppler and superb microvascular imaging: is there any difference? *P. Del Nido Recio*, J. D. Aquerreta, A. Paternain Nuin, M. R. López De La Torre Carretero, M. Jiménez Vázqu ez, C. Mbongo, C. Urtasun Iriarte, D. A. Zambrano, M. B. Barrio Pi queras; Pamplona/ES ([email protected]) Purpose or Learning Objective: This study aims to analyze whether the upgrade Doppler activity, when comparing SMI and PD , is significantly different in Rheumatoid Arthritis (RA), compared to other art hropathies.

Methods

or Background: Between May 2023 and April 2024, we prospectively analyzed a cohort of 57 joints of 21 different patients. Most of them were previously diagnosed with RA and other ar thropathies, such as Osteoarthritis, Psoriatic Arthritis, Gout, etc. PD and SMI imaging were obtained in all joints and the individual grades for Doppler Activity were registered for each joint with active synovitis, raging from 0 to 3, according to the EULAR - OMERACT US Score. Mann-Whitney U test was applied t o calculate means in independent samples. Two-tailed p-values of <0.05 w ere considered statistically significant.

Results

or Findings: 21 joints of 8 patients with RA and 36 joints of 13 patients with other arthropathies were studied. Met acarpophalangeal joints were the most frequently analyzed in the RA group ( 11), and interphalangeal joints were the most frequently analyzed in the oth er group (18). When comparing Doppler activity, the mean PD was signifi cantly higher in the RA group compared to the other group (1.14 vs. 0.75, p = 0.043). When upgrading the Doppler activity with SMI, we did not find stat istically significant differences (2.52 vs 2.69, p = 1.04). Nevertheless, when compar ing the mean of the difference between SMI and PD, it was significantly lower in the RA group (1.38 vs. 1.94, p = 0.016).

Conclusion

Our results indicate that a higher upgrade between PD and SMI exams is more likely seen in arthropathies such as oligoarthritis, Psoriatic Arthritis or Gout, rather than in RA.

Limitations

The number of patients. Interobserver variability. The degree of arthritis and type of treatment. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study received institutional review board approval and written informed consent was obtained from all participants Author Disclosures: Pablo Del Nido Recio: Nothing to disclose Carmen Mbongo: Nothing to disclose Cesar Urtasun Iriarte: Nothing to disclose Jesús Dámaso Aquerreta: Nothing to disclose Miguel Barrio Barrio Piqueras: Nothing to disclose Manuel Rafael López De La Torre Carretero: Nothing to disclose Daniel Alfonso Zambrano: Nothing to disclose Alberto Paternain Nuin: Nothing to disclose Marcos Jiménez Vázquez: Nothing to disclose Prevalence of diffuse idiopathic skeletal hyperosto sis (DISH) according to recently established imaging criteria and coexis tent pelvic enthesophytes *V. Yaman*, A. E. Yıldız, B. Fırlatan, H. Avcı, O. Karadağ, U. Kalyoncu, Ü. Aydingöz; Ankara/TR ([email protected]) Purpose or Learning Objective: Recent (2019) criteria that supplement classic DISH criteria were developed to diagnose th is condition at an earlier stage. The aims of the present study were to invest igate the prevalence of the early-stage (ES-DISH) as well as classic (late-stag e) (LS-DISH) disease and to determine pelvic enthesophyte status in patients wi th ES-DISH and LS-DISH.

Methods

or Background: 636 consecutive patients aged ≥18 years who underwent thorax CT during October 2023 in a tertia ry medical center were retrospectively evaluated. CTs were scored accordin g to 2019 DISH criteria by two independent observers. Pelvic (including parasa croiliac) enthesophytes in patients who also had a simultaneous abdomen CT wer e examined by a musculoskeletal radiologist blinded to patients’ DI SH status as well as clinical information, yielding a personal “pelvic enthesophy te load score” (PELS). Malignancy and metabolic syndrome data from patient s were also analyzed.

Results

or Findings: Prevalences of ES-DISH and LS-DISH were 16.7% and 15.4%, respectively. Intra- and interobserver agree ment were “almost perfect” (ICC = 0.88; 95% CI, 0.77–0.99) and “substantial” ( ICC = 0.73; 95% CI, 0.68– 0.78), respectively. Mean ages (and age range) of “ no DISH”, ES-DISH, and LS-DISH subgroups were 55.1 (18–86), 65.5 (46–90), and 69.2 (48–96), respectively. DISH prevalence was similar in patien ts without and with cancer. When age was factored in, frequency of metabolic sy ndrome in DISH patients did not differ significantly compared to those with out DISH. Mean±SD PELS were 6.2±3.5, 8.9±3.4 and 10.1±3.2 in “no DISH”, ES-DISH and LS-DISH groups. Logistic regression analysis yielded an ove rall accuracy of 68% for PELS in predicting DISH status.

Conclusion

This study demonstrated that prevalence of DISH dou bled when ES-DISH criteria were used. Pelvic enthesophytes ar e more profusely seen in ES- or LS-DISH than in patients without DISH.

Limitations

Retrospective study Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was approved by Hacettepe University Clinical Studies Ethics Commit tee (SBA 24/166). Author Disclosures: Umut Kalyoncu: Nothing to disclose Hanife Avcı: Nothing to disclose Büşra Fırlatan: Nothing to disclose Üstün Aydingöz: Nothing to disclose Omer Karadağ: Nothing to disclose Vedat Yaman: Nothing to disclose Adalet Elçin Yıldız: Nothing to disclose Thursday Abstract-based Programme 92 10:00-11:00 Research Stage 3 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 805 Artificial intelligence in neuroimaging Moderator M. M. Serra; Avignon/FR ([email protected]) Author Disclosures: Maria Mercedes Serra: Employee: BC Platforms; Resea rch Grant/Support: Sanofi; Share Holder: BC Platforms Impact of data quality variations caused by dose an d image reconstruction on AI assessment of intracranial ane urysms L. Gölz, A. Laudani, U. Genske, M. Scheel, G. Bohne r, H-C. Bauknecht, S. Mutze, *P. Jahnke*; Berlin/DE ([email protected]) Purpose or Learning Objective: To assess the performance of a commercial AI algorithm in detecting intracranial aneurysms wh en scan data quality variations occur due to changes in dose and image r econstruction.

Methods

or Background: Consistency testing of AI performance was performed using a realistic head CT phantom designe d for AI evaluation. The phantom simulated a patient with three intracranial aneurysms located in the anterior communicating artery (ACoA), middle cerebr al artery (MCA), and basilar artery (BA). The phantom was repeatedly exa mined at 21 dose levels (0.47 to 20.09 mGy) using iterative reconstruction and filtered back projection. Aneurysm labeling by an FDA-approved and CE-marked AI solution was analyzed. In addition, five neuroradiologists rated aneurysm visiblity in all examinations.

Results

or Findings: AI detection rates varied by aneurysm type, with detection rates of 74.6% for the ACoA, 92.9% for th e MCA, and 2.4% for the BA aneurysm across all examinations. The AI respons e was inconsistent at doses below 8 mGy with iterative reconstruction and at doses below 7 mGy and above 14 mGy with filtered back projection. In contrast, readers consistently reported 100% visibility for all aneur ysms at doses above 2 mGy regardless of image reconstruction.

Conclusion

AI approved for managing intracranial aneurysms sho ws performance issues due to variations in data qualit y and requires different data quality standards than neuroradiologists.

Limitations

This prospective study was limited to a single AI a pplication, a single scanner system, and three intracranial aneur ysms. Funding for this study: This work has not received any funding. Ethics committee - additional information: Ethics committee of the Charité Author Disclosures: Hans-Christian Bauknecht: Nothing to disclose Paul Jahnke: Shareholder: PhantomX GmbH Employee: P hantomX GmbH Ulrich Genske: Nothing to disclose Georg Bohner: Nothing to disclose Leonie Gölz: Nothing to disclose Angelo Laudani: Nothing to disclose Michael Scheel: Shareholder: PhantomX GmbH Sven Mutze: Nothing to disclose Improving diagnostic precision: a deep learning sys tem for differentiating multiple sclerosis from small vessel disease using standard non- enhanced brain MRI scans K. Firouznia, *M. Arab Ahmadi*, H. Hashemi, M. Boro omand-Saboor, R. Ghavami Modegh, M. Akhlaghpasand, H. Dashti, M. Gity, M. Mohammadzadeh; Tehran/IR ([email protected]) Purpose or Learning Objective: The diagnosis of Multiple Sclerosis (MS) primarily depends on clinical evaluation, bolstered by magnetic resonance imaging (MRI) interpreted by skilled radiologists.H owever, the typical imaging characteristics of MS can resemble those of other c entral nervous system disorders.One such condition is Cerebral SVD, which can complicate the radiologist's ability to make a diagnosis.This diff erential diagnosis can be particularly challenging in the early stages of the disease.The objective of this study is to create and assess a Computer-Aided Diag nosis (CAD) system utilizing brain MRI images to differentiate between MS and SVD.

Methods

or Background: Brain MRI scans were obtained from a 3 Tesla scanner for patients diagnosed with MS during acute attacks and silent phases, alongside individuals diagnosed with SVD ba sed on cardiovascular risk factors. MRI sequences included FLAIR,T1, and T2. An expert neuroradiologist identified white matter lesions, w hich were segmented using artificial intelligence software. The dataset was d ivided into 80% for training, 10% for validation, and 10% for testing. A neurorad iologist then evaluated the AI results against established clinical and imaging criteria.

Results

or Findings: The study included 80 MS patients with 265 lesions compared to 67 SVD patients with 218 lesions. The A I tool achieved a sensitivity of 78.57% and specificity of 93.33% (P- value < 0.05). It also demonstrated a positive predictive value (PPV) of 9 1.67%, a negative predictive value (NPV) of 82.35%, balanced accuracy of 85.95%, and an area under the curve (AUC) of 78.71.

Conclusion

The findings suggest that artificial intelligence c an effectively differentiate MRI images of MS from those of SVD us ing routine sequences.Implementing AI in distinguishing between MS and SVD lesions could enhance diagnostic accuracy and improve patie nt management in clinical practice.

Limitations

Sample volume and one center study listed as some o f the limitations. Funding for this study: None Ethics committee - additional information: This study was approved by an institutional ethics committee. Author Disclosures: Masoumeh Gity: Nothing to disclose Kavous Firouznia: Nothing to disclose Mohammadhosein Akhlaghpasand: Nothing to disclose Hassan Hashemi: Nothing to disclose Melika Boroomand-Saboor: Nothing to disclose Maryam Mohammadzadeh: Nothing to disclose Rassa Ghavami Modegh: Nothing to disclose Hamed Dashti: Nothing to disclose Mehran Arab Ahmadi: Nothing to disclose Previously proposed radiomics features for ruptured intracranial aneurysm classification: Overview, auto-segmentatio n, and external validation *D. Zhu*, Y. Yang; Wenzhou/CN ([email protected]) Purpose or Learning Objective: To automatically segment and extract radiomics features of intracranial aneurysms (IAs), validate existing radiomics predictors for ruptured IAs, and construct machine learning (ML) and deep learning (DL) models for classifying ruptured IAs.

Methods

or Background: In this retrospective study, we used data from the MIRACLE Cohort, registered with the Chinese Clinica l Trial Registry (ChiCTR2400084601). IAs were segmented automaticall y using the DGIS method. We systematically reviewed studies reportin g radiomics predictors for ruptured IAs and externally validated those predict ors. We developed five ML and DL models for classifying ruptured IAs, employi ng the SHapley Additive exPlanations (SHAP) method to enhance model interpr etability.

Results

or Findings: The study included 632 patients with 668 aneurysms, divided into training (n=593) and external testing (n=75) datasets. The DGIS

Method

achieved great segmentation accuracy with Di ce coefficients of 0.98 and 0.75 in the source and target domains, respecti vely. When comparing radiomics features derived from manual and automati c segmentations, the original_shape_VoxelVolume, MeshVolume, and Surface Area showed the highest stability (with all ICC of >0.9). Upon exte rnal validation of radiomics predictors from 12 studies, the AUCs ranged from 0. 59 to 0.71 in the training dataset and 0.48 to 0.65 in the external testing da taset. The original_shape_ Elongation feature emerged as the most frequently u tilized predictor. The Gradient Boosting and DRE models performed well in classifying ruptured IAs, with AUCs reaching 0.995 and 0.95 in the training d ataset, and 0.85 and 0.80 in the external testing dataset, respectively.

Conclusion

This study presents a comprehensive workflow for au tomatic IAs rupture risk analysis and an overview of existing r adiomics predictors. After external validation, certain original shape feature s demonstrated significant stability, utility, and predictive power. The ML an d DL models offer a promising tool for risk stratification of IAs.

Limitations

Not applicable. Funding for this study: This study was supported by the Wenzhou Major Program of Science and Technology Innovation (Grant No. ZY2020012) and Key Laboratory of Novel Nuclide Technologies on Pre cision Diagnosis and Treatment & clinical Transformation of Wenzhou City (Grant No. 2023HZSY0012). Ethics committee - additional information: Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University Author Disclosures: Yunjun Yang: Nothing to disclose Dongqin Zhu: Nothing to disclose Thursday Abstract-based Programme 93 Diagnostic Performance of Neural Network Algorithms in Skull Fractures Detection in CT Scans: A Systematic Review and Meta -Analysis *R. Hajibeygi*¹, G. Sharifi¹, M. Fathi¹, A. Bahrami ², R. Eshraghi², I. Dixe De Oliveira Santo³, A. Mirjafari⁴, J. Chan⁴, L. Tu³; ¹Tehran/IR, ²Kashan/IR, ³New Haven, CT/US, ⁴Los Angeles, CA/US ([email protected]) Purpose or Learning Objective: The potential intricacy of skull fractures as well as the complexity of underlying anatomy poses diagnostic hurdles for radiologists evaluating CT scans. The necessity for automated diagnostic tools has been brought to light by the shortage of radiol ogists and the growing demand for rapid and accurate fracture diagnosis. C onvolutional Neural Networks (CNNs) are a potential new class of medica l imaging technologies that use deep learning (DL) to improve diagnosis ac curacy. The objective of this systematic review and meta-analysis is to asse ss how well CNN models diagnose skull fractures on CT images.

Methods

or Background: PubMed, Scopus, and Web of Science were searched for studies before February 2024 that used CNN models to detect skull fractures on CT scans. Meta-analyses were con ducted for area under the receiver operating characteristic curve (AUC), sens itivity, specificity, and accuracy. Egger's and Begg's tests were used to ass ess publication bias.

Results

or Findings: Meta-analysis was performed for 11 studies with 207 98 patients. Pooled average AUC for implementing pre-t raining for transfer learning in CNN models within their training model’ s architecture was 0.96 ± 0.02. The pooled averages of the studies' sensitivi ty and specificity were 1.0 and 0.93, respectively. The accuracy was obtained 0 .92 ± 0.04. Studies showed heterogeneity, which was explained by differ ences in model topologies, training models, and validation techniq ues. There was no significant publication bias detected.

Conclusion

CNN models perform well in identifying skull fractu res on CT scans. The results suggest that CNNs have the poten tial to improve diagnostic accuracy in the imaging of acute skull trauma. To f urther enhance these models' practical applicability, future studies cou ld concentrate on the utility of DL models in prospective clinical trials.

Limitations

One of the limitations is lack of homogeneity in CT image quality across studies. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Reza Eshraghi: Nothing to disclose Mobina Fathi: Nothing to disclose Janine Chan: Nothing to disclose Guive Sharifi: Nothing to disclose Ashkan Bahrami: Nothing to disclose Irene Dixe De Oliveira Santo: Nothing to disclose Arshia Mirjafari: Nothing to disclose Ramtin Hajibeygi: Nothing to disclose Long Tu: Nothing to disclose AI as a second reader in post-traumatic head CT at Oktoberfest 2024: A prospective performance monitoring study *M. B. Steinberger*, M. Bock, A. S. Duque, B. F. Ho ppe, J. P. Rudolph, Y. Dikhtyar, P. Reidler, W. Flatz, D. Hinzmann, V. Bogner-Flatz, J. Ricke, C. C. Cyran; Munich/DE ([email protected]) Purpose or Learning Objective: To prospectively assess the impact of an AI algorithm on radiologists’ diagnostic confidence in detecting intracranial haemorrhage (ICH) in post-traumatic head CT at Okto berfest 2024.

Methods

or Background: A mobile CT scanner (Somatom go.Top, Siemens Healthineers) was operated on-site for triaging pat ients with mild to moderate traumatic head injuries. This prospective study inc luded n=219 patients who underwent head CT. Instant AI analysis was provided via auto-routing to a fully PACS-integrated, GDPR-compliant clinical AI platfor m. Initially, one of 15 board-certified radiologists, alternating in shifts , read the head CT unassisted, rating ICH likelihood on a 5-point Likert scale (-2 , “very low”; 2, “very high”). After submitting this evaluation, algorithm results were made available for reassessment of ICH likelihood. Performance monitor ing of the AI tool was implemented in PACS (Visage Imaging) via Fast Healt hcare Interoperability Resources (FHIR) pop-up forms.

Results

or Findings: AI support was utilised in 66% (146/222 scans) of t he readings, varying significantly between readers (43 %-100%). At a probability threshold of 0.1, the AI tool correctly identified 6 out of 7 ICH, 139 true negatives, no false positives (sens=0.857, spec=1.0 00, acc=0.993, ppv=1.000, npv=0.993). AI assistance increased radiologists' c onfidence in ruling out ICH in 19 cases (-1 to -2) and confirming ICH in two ca ses (1 to 2). In two borderline cases, AI aided in excluding ICH (0 to - 1). Overall, diagnostic confidence was significantly higher with AI support (p<0.001).

Conclusion

AI assistance significantly improved diagnostic con fidence of radiologists reading trauma head CT at Oktoberfest 2024, serving as a virtual second reader in this emergency setting. PACS-integ rated FHIR forms set a framework for seamless monitoring of AI performance and its impact on diagnostic workflow.

Limitations

In 40 cases, no AI analysis was performed due to in correct specifications or failed auto-routing. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee notification can be found under the number UID 24-0813. Written informed consent was obtained from all participants and the study was re gistered in the German Clinical Trial Register (DRKS00034969). Author Disclosures: Jan Philipp Rudolph: Nothing to disclose Boj Friedrich Hoppe: Nothing to disclose Yevgeniy Dikhtyar: Nothing to disclose Clemens C. Cyran: Nothing to disclose Wilhelm Flatz: Nothing to disclose Dominik Hinzmann: Nothing to disclose Matthias Bock: Nothing to disclose Paul Reidler: Nothing to disclose Maria Barbara Steinberger: Nothing to disclose Veronika Bogner-Flatz: Nothing to disclose Anna Sophie Duque: Nothing to disclose Jens Ricke: Nothing to disclose Impact of Defacing Procedures on Brain Age Gap Esti mation *V. L. Ivan*¹, J. Caspers¹, M. Vach¹, D. M. Hedderi ch², D. Weiß¹, C. Rubbert¹; ¹Düsseldorf/DE, ²Munich/DE Purpose or Learning Objective: Removal of facial features from MRI brain scans (“Defacing”) is mandatory from data privacy p erspective. We investigated the impact of defacing on Brain Age Ga p Estimation (BrainAGE), an imaging biomarker used in various research areas such as atypical aging.

Methods

or Background: A total of 364 Alzheimer’s disease (AD) patients and 717 cognitively normal (CN) participants were a nalyzed including unaccelerated (AD:n=290; CN:n=386) and accelerated 3DT1 imaging (AD:n=203;; CN:n=500). BrainAGE was computed after defacing using either afni_refacer, fsl_deface, mri_deface, mri_reface, P yDeface, or spm_deface and without defacing. For BrainAGE, gray matter fea tures were extracted using CAT12 for SPM12. BrainAGE was calculated as predict ed age minus chronological age. A subset of participants (AD:n=7 4, CN:n=84) had within- session repeat imaging available and were processed without defacing, serving as a benchmark for BrainAGE differences. Me an absolute error (MAE), and mean squared error (MSE) were calculated. Outli ers due to defacing were identified using Grubbs’s tests.

Results

or Findings: Benchmark analysis found MAE of 1.15 and MSE of 2.25 for BrainAGE differences between initial and r epeat scans without defacing in CN, and an MAE of 1.43 and MSE of 3.29 for AD. Among defacing methods, PyDeface exhibited the best performance wi th an overall MAE of 0.33 and MSE of 0.27, showing a mean BrainAGE diffe rence of 0.08±0.52. PyDeface also had the fewest outliers (n=99) based on the benchmark criteria. Grubbs’s test identified 23 outliers after PyDeface , with 11 found after mri_reface and 20 after spm_deface.

Conclusion

Defacing can be employed for data privacy without s ignificantly affecting the reliability of BrainAGE as an imaging biomarker. PyDeface is recommended.

Limitations

BrainAGE may be affected by defacing, however, in m ost approaches this influence is lesser than the variab ility observed in BrainAGE in repeat non-defaced imaging. Funding for this study: No Ethics committee - additional information: No Author Disclosures: Marius Vach: Nothing to disclose Julian Caspers: Nothing to disclose Christian Rubbert: Nothing to disclose Daniel Weiß: Nothing to disclose Vivien Lorena Ivan: Nothing to disclose Dennis M Hedderich: Nothing to disclose Identification of depression subtypes in Parkinson' s disease patients via structural MRI whole-brain radiomics: an unsupervis ed machine learning study *Z. Shu*; Zhejiang, Hangzhou/CN ([email protected]) Purpose or Learning Objective: Unsupervised machine learning methods based on whole-brain radiomic analysis were used to identify subtypes of depression that occur during the progression of Par kinson's disease (PD).

Methods

or Background: Data from 272 PD patients in the PPMI database were used, among which 81 experienced depression in Parkinson's Disease (DPD) during a 5-year follow-up period. Quantitativ e radiomic features were extracted from the whole-brain magnetic resonance s tructural images of each Thursday Abstract-based Programme 94 patient, and principal component analysis (PCA) was used for feature dimensionality reduction. All of the cases were cla ssified into different subtypes by unsupervised cluster analysis (UCA). The high-ri sk subtypes were selected through comparative analysis. The high-risk subtype data were divided into training subgroups and testing subgroups at a 7:3 r atio. On the basis of the clinical characteristics of the training subgroups, multiple logistic regression analysis was performed to confirm the risk factors for DPD subtypes. The DPD subtypes were subsequently identified on the basis of the risk factors. A prediction model was constructed via decision trees , and the diagnostic accuracy of the model was evaluated via receiver op erating characteristic (ROC) curves.

Results

or Findings: Logistic regression analysis based on high-risk sub type groups revealed that rem, updrs1_score, updrs2_scor e, and ptau were independent predictors of DPD. The prediction model based on high-risk subgroups had AUC values of 0.853 and 0.81 in the t raining and testing subgroups, sensitivities of 0.765 and 0.786, and sp ecificities of 0.771 and 0.815, respectively. The AUC, sensitivity, and spec ificity in the non-high-risk subgroup were 0.859, 0.654, and 0.852, respectively .

Conclusion

An UCA based on MRI structural imaging features can identify high-risk subtypes of DPD, and the constructed mode l can also predict the progression of DPD well.

Limitations

This study was designed as a retrospective analysis . Funding for this study: The work was supported by the Natural Science Foundation of Zhejiang Province of China (LGF22H090 021) Ethics committee - additional information: The case data used in this study came from the Parkinson's Progression Markers Initi ative (PPMI) (http://www.PPMI-info.org) database, and data colle ction was approved by institutional review board; For ethical review info rmation on the data, please refer to the website. Author Disclosures: Zhenyu Shu: Nothing to disclose A Machine-Learning Model Based on US Radiomics to C lassify Benign and Malignant Thyroid Nodules A. Guerrisi¹, *V. Dolcetti*¹, L. Miseo¹, A. Valenti ¹, F. Elia¹, G. Del Gaudio¹, F. Raponi², E. David², V. Cantisani¹; ¹Rome/IT, ²Ca tania/IT ([email protected]) Purpose or Learning Objective: The aim of this work was to develop a machine learning model based on thyroid ultrasound images in order to classify nodules into benign and malignant classes. Ultrasound and fine needle biopsy are the most reliable diagnostic methods to date, but they have some limitations. Radiomics and machine learning could b e useful to improve diagnosis while reducing invasive procedures.

Methods

or Background: Ultrasound images from 142 subjects were collected: 40 patients belonged to "malignant" and 102 to "benign" class, according to histological diagnosis (fine-needle as piration). Those images were used to train, cross-validate and internal test thr ee different machine learning models, using the “Trace for Research” software. A robust radiomic approach was applied, and the models (random forests, SVM an d k-NN classifiers) were evaluated. Finally, the best model was externally t ested on an additional cohort of 21 patients.

Results

or Findings: The best model (ensemble of random forest) showed ROC-AUC (%) of 85 (majority vote), 83.7** (mean) [8 0.2-87.2], accuracy (%) of 83, 81.2** [77.1-85.2], sensitivity (%) of 70, 67.5 ** [64.3-70.7], specificity (%) of 88, 86.5** [82-91], PPV (%) of 70, 66.5** [57.9-75. 1], and NPV (%) of 88, 87.1** [85.5-88.8] (*p<0.05, **p<0.005) in the inte rnal test cohort. This model was then externally tested, achieving an Accuracy o f 90.5%, a sensitivity of 100%, a specificity of 86.7%, a PPV of 75% and an N PV of 100%.

Conclusion

The best model could successfully identify all the malignant nodes and the consistent majority of benign in exte rnal testing cohort. Further investigations could be conducted by testing the mo del with images of nodules from different centers.

Limitations

Additional external tests should be performed, with images from different ultrasound machines and different healthc are centers to increase variability of target population. Funding for this study: This research was supported by Italian Ministry of Health Ethics committee - additional information: This study was performed in line with the principles 417 of the Declaration of Helsi nki. Approval was granted by the Ethics Committee of IRCCS 418 IFO-Fondazione GB Bietti (Date: 23/01/2023, N: 1820/23) Author Disclosures: Antonino Guerrisi: Nothing to disclose Giovanni Del Gaudio: Nothing to disclose Alessandro Valenti: Nothing to disclose Fulvia Elia: Nothing to disclose Vincenzo Dolcetti: Nothing to disclose Flavia Raponi: Nothing to disclose Ludovica Miseo: Nothing to disclose Vito Cantisani: Nothing to disclose Emanuele David: Nothing to disclose 10:00-11:00 Research Stage 4 Research Presentation Session: Abdominal and Gastrointestinal RPS 801 Advances in liver imaging Moderator J.-I. Choi; Seoul/KR ([email protected]) Author Disclosures: Joon-Il Choi: Grant Recipient: Guerbet Korea, Sieme ns Healthineers, Samsong Medison, Bracco Korea; Speaker: Bayer Heath care Multiparametric spectral imaging for characterizati on of small hypoattenuating liver lesions *N. Abou Zeid*, C. Nelles, Z. Gurbanova, N. Große H okamp, T. Persigehl, S. Lennartz; Cologne/DE Purpose or Learning Objective: To investigate the diagnostic utility of spectral reconstructions for determining cystic nat ure of small, hypoattenuating liver lesions.

Methods

or Background: Patients with portal venous phase dual-layer dual- energy CT (dlDECT) who were diagnosed with hypoatte nuating liver lesions smaller than one centimeter that were verified as c ysts in corresponding MRI examinations were retrospectively included. ROI-bas ed measurements were conducted by two raters in conventional images (CI) , virtual unenhanced images (VUE) and iodine images. CT-based determinat ion of cystic nature of the lesions was conducted using a HU attenuation th reshold of less than 20 HU or an iodine concentration threshold of less tha n 0.5 mg/dl, the latter of which has been reported as the scanner-specific low er limit of iodine detection. Accuracy for determining cystic nature was compared between CI, VUE and iodine images.

Results

or Findings: 77 patients with 287 small liver cysts were include d. Mean attenuation in CI for small liver cysts was 20 .8 ± 24.4 HU, and 11.1 ± 16.8 HU in VUE images. Mean iodine concentration wa s 0.46 ± 0.57 mg/dl. Using the 20 HU threshold in CI resulted in an accu racy of 60.3 % (173/287), whereas the corresponding accuracy using the same t hreshold in VUE images was 76.7 % (220/287). Accuracy solely based on the iodine threshold was 57.5 % (165/287). Combining the VUE and iodine threshold resulted in an accuracy of 92.7 % (266/287) for determining cystic nature o f the lesions in dlDECT.

Conclusion

Combining quantitative VUE and iodine measurements using established thresholds facilitated correctly diagno sing 92.7% of small hypoattenuating liver lesions as cysts, compared to an accuracy of 60.3 % when using HU measurements in conventional images. This approach may help reducing correlative imaging and thereby accel erating staging of cancer patients.

Limitations

Retrospective, mono-centric study Funding for this study: None to report Ethics committee - additional information: IRB waiver due to retrospective nature of the study. Author Disclosures: Simon Lennartz: Speaker: Amboss GmbH Author: Amboss GmbH Nils Große Hokamp: Research/Grant Support: Philips Speaker: Philips Speaker: Amboss GmbH Christian Nelles: Nothing to disclose Nour Abou Zeid: Nothing to disclose Zuleykha Gurbanova: Nothing to disclose Thorsten Persigehl: Nothing to disclose Thursday Abstract-based Programme 95 Could MRI-radiomics predict Liver Metastasis Recurr ence and Overall Survival after surgery or ablation treatment? *T. Russo*, A. Belardo, A. Della Corte, D. Santange lo, F. Calabrese, M. M. Vincenzi, M. Mori, C. Fiorino, F. De Cobelli; Milan/IT ([email protected]) Purpose or Learning Objective: To investigate the potentials of MRI radiomics to predict recurrence (R), hepatic recurr ence (IR), and overall survival (OS) in a cohort of patients with colorect al liver metastases (CLM) who underwent microwave ablation (MWA) performed alone or in combination with surgical resection.

Methods

or Background: 121 CLM patients with pre-operative Gadoxetic acid-MRI treated at our Institute between October 2 015 and December 2022 were analyzed. One observer manually segmented the largest CLM on T2 scans. The abdominal aorta at the level of second l umbar vertebrae was used for the z-score normalization of the lesion. Cox mu ltivariate analysis was run to establish a few-features radiomic model (RAD-T2), t o predict recurrences and death. A bootstrap-based methodology for robust fea ture selection, including redundancy filtering, was optimized to select the b est combination of two, three, four features. Correction of the models for optimism was then performed by internal bootstrap-based validation.

Results

or Findings: For R, IR and OS the median follow-ups were respectively 12, 13 and 23 months; the number of ev ents were 80, 68 and 34. After corrections for optimism, the resulting best RAD-T2 models were based on the combination of 2-3 features; they were able to predict R with C- index=0.65 (p=0.0002), IR (C-index=0.64, p=0.0029) and OS (C-index=0.71, p=0.0046). As an example, based on the best cut-off value of the RAD-T2 index, OS at 2 year was 58% and 88% when the cohort was stratified accordingly.

Conclusion

T2-MRI-based radiomic evaluation of CLMs is feasibl e and potentially useful for outcome prediction.

Limitations

The limited number of patients and the retrospectiv e nature of the study. Funding for this study: The limited number of patients and the retrospectiv e nature of the study. Ethics committee - additional information: All procedures were carried out in accordance with the Declaration of Helsinki (196 4) and its later amendments. Author Disclosures: Monica Maria Vincenzi: Nothing to disclose Angelo Della Corte: Nothing to disclose Claudio Fiorino: Nothing to disclose Domenico Santangelo: Nothing to disclose Alfonso Belardo: Nothing to disclose Martina Mori: Nothing to disclose Tommaso Russo: Nothing to disclose Francesca Calabrese: Nothing to disclose Francesco De Cobelli: Nothing to disclose Intraindividual Comparison of Half-dose Gadopicleno l and Standard Dose of Gadobenate Dimeglumine for Abdominal MRI *A. Del Gaudio*¹, K. Kalisz¹, D. Kruse¹, F. Ria¹, D . De Santis², L. Lofino¹, D. Marin¹; ¹Durham, NC/US, ²Rome/IT ([email protected]) Purpose or Learning Objective: To intraindividually compare image quality and lesion conspicuity of abdominal MRI using gadop iclenol at 0.05 mmol/kg and gadobenate dimeglumine (Gd-BOPTA) at 0.1 mmol/k g.

Methods

or Background: From September 2023 to March 2024, consecutive patients who had undergone two clinically indicated abdominal MRIs within 12 months using gadopiclenol and Gd-BOPTA on the same scanner were retrospectively enrolled. One independent radiologi st manually measured the signal intensity of abdominal organs, arterial and venous vessels, and abdominal lesions (liver, pancreas, and kidneys) on unenhanced, late arterial, venous, and equilibrium phases. SNR, CNR, and magni tude of contrast enhancement (ΔE) were calculated for all organs and vessels on ea ch contrast-enhanced phase. Percentage enhancement (%E ) was calculated for all lesions on contrast-enhanced phases. Subjective image quality was assessed using a 5-point Likert scale, including: o rgans' and vessels' enhancement, liver-to-vessels contrast, and overall image quality. Lesion characteristics were also evaluated. Statistical an alysis employed paired t- and Wilcoxon tests.

Results

or Findings: One hundred subjects (64 years ± 14; 55 men) and 21 abdominal lesions were included. Compared to Gd-BOP TA, gadopiclenol yielded significantly higher CNR and SNR for pancre as, porta, and kidney in the late arterial phase (p ≤ .040). No significant differences in CNR and SNR were observed between gadopiclenol and Gd-BOPTA acr oss all organs in the portal venous and equilibrium phases. Gadopiclenol showed significantly higher pancreatic ΔE in all contras-enhanced phases (p ≤ .049) compared to Gd-BOPTA. The %E of abdominal lesions was comparabl e between gadopiclenol and Gd-BOPTA for all contrast-enhanced phases (p ≥ .100). No significant differences were observed in readers’ p erception of image quality and lesions’ characteristics.

Conclusion

Gadopiclenol at 0.05 mmol/kg yields similar image q uality and improved pancreatic enhancement compared to Gd-BOPT A at 0.1 mmol/kg.

Limitations

Retrospective study design Funding for this study: None Ethics committee - additional information: Written informed consent was waived and Institutional Review Board approval was obtained Author Disclosures: Danielle Kruse: Nothing to disclose Ludovica Lofino: Nothing to disclose Domenico De Santis: Nothing to disclose Daniele Marin: Nothing to disclose Antonella Del Gaudio: Other: Bracco research fellow ship Francesco Ria: Nothing to disclose Kevin Kalisz: Nothing to disclose Preclinical profile of a new macrocyclic MRI liver agent *J. Lohrke*, T. Brumby, S. Herbert, T. Frenzel, G. Jost, M. Berger, H. Pietsch; Berlin/DE ([email protected]) Purpose or Learning Objective: The established liver-specific gadolinium- based magnetic resonance imaging (MRI) contrast age nts, gadoxetate disodium and gadobenate dimeglumine are based on li near, DTPA like ligands. In the present study a new early preclinical macroc yclic liver-specific candidate will be presented.

Methods

or Background: The MRI efficiency (r1-relaxivity) of the candidate BAY 3393081 was determined at 37°C 1.41 T in human plasma. The kinetic inertness of the complex stability was investigated using an established zinc transmetallation assay. The in vitro liver cell upt ake was assessed in rat hepatocytes and human transfected organic anion tra nsporter protein 1B1 or 1B3 embryonic kidney cells. Pharmacokinetic paramet ers were evaluated in rodent (rats) and non-rodent (dogs) species by anal yzing the gadolinium (Gd) concentrations in plasma over time. The in vivo liv er elimination was examined in bile-duct-cannulated rats and the bile was analy zed using inductively coupled plasma mass spectroscopy. Contrast-enhanced liver MRI was performed in mice, pigs and a VX2 tumor model in ra bbits.

Results

or Findings: The relaxivity of BAY3393081 was determined with 8.7±0.3 L/(mmol·s) in human plasma at 1.41 T. The k inetic inertness of the complex stability was comparable to marketed macroc yclic GBCAs. The in vitro cell uptake results revealed that BAY3393081 is spe cifically taken up by rat and human OATPs. BAY 3393081 showed a high plasma c learance in rat and dog. In rats ~80% of the injected dose were elimina ted in an unchanged form via the bile. Strong signal enhancement of liver pa renchyma was demonstrated in rats, rabbits and pigs.

Conclusion

The preclinical candidate BAY3393081 showed the hig h kinetic inertness of macrocyclic GBCAs and a strong liver p arenchyma enhancement in the MRI.

Limitations

Limited transferability of preclinical data to huma n liver elimination due to significant interspecies variability in hepa tobiliary transporter expression. Funding for this study: Funding for this preclinical study was provided by Bayer AG. Ethics committee - additional information: All animal experiments were approved by LaGeSo. Author Disclosures: Simon Herbert: Employee: Bayer AG Markus Berger: Employee: Bayer AG Gregor Jost: Employee: Bayer AG Thomas Frenzel: Employee: Bayer AG Jessica Lohrke: Employee: Bayer AG Thomas Brumby: Employee: Bayer AG Hubertus Pietsch: Employee: Bayer AG Non-invasive diagnosis of chronic liver disease and portal hypertension using intravoxel incoherent motion imaging and magn etic resonance elastography *D. Catucci*, S. U. Von Däniken, V. Obmann, A. Berz igotti, L. Ebner, J. T. Heverhagen, A. Christe, P. Vermathen, A. T. H uber; Bern/CH Purpose or Learning Objective: This study aimed to analyse the performance of intravoxel incoherent motion (IVIM) imaging parameters (tissue-diffusivity D, perfusion-fraction PF and ps eudo-diffusion-coefficient D*) and liver stiffness (LS) measured by magnetic reson ance elastography (MRE) to screen for chronic liver disease (CLD) and clini cally significant portal hypertension (CSPH) on liver MRI examinations.

Methods

or Background: This prospective study included 103 patients without CLD (noCLD-group, n=103) and 82 patients wi th biopsy-proven CLD who underwent liver MRI examinations including MRE and IVIM imaging Thursday Abstract-based Programme 96 between 03/2016 and 11/2023. Patients with CLD were subdivided based on their liver fibrosis degree: early CLD (F0-F1; eCLD -group, n=21), intermediate CLD (F2; iCLD-group, n=19), advanced CLD (F3-F4; aC LD-group, n=20) and aCLD with CSPH according to the BAVENO VII consensu s (aCLDPH-group; n=22). IVIM imaging parameters (D, PF and D*) of th e liver as well as LS were measured in all patients. Statistical analysis incl uded the Kruskal-Wallis test for group comparison and receiver operating characteris tic (ROC) curve analysis with multiple logistic regression analysis for grou p differentiation.

Results

or Findings: D, PF, D* and LS differed significantly between all groups (p3.2 kPa (sensitivit y 67%/specificity 96%) respectively >3.8 kPa (sensitivity 95%/specificity 88%), both with p<0.001. For CLD-detection, a combination of D 3.2 kPa increased the AUC from 0.89 with LS to 0.95. For CS PH-detection, a combination of D* 3.8 k Pa increased the specificity (90%) with a slightly lower sensitivity (91%, p<0.001).

Conclusion

IVIM imaging parameters (D, PF and D*) as well as L S measured by MRE allow non-invasive screening for CLD and CSP H.

Limitations

This was a single center study and should be extern ally validated. Funding for this study: This study received funding by the Swiss National Science Foundation (SNF), grant number 188591. Ethics committee - additional information: This study was approved by the cantonal ethics committee of Bern (Kantonale Ethikk ommission Bern). Author Disclosures: Sandro Urs Von Däniken: Nothing to disclose Johannes T. Heverhagen: Nothing to disclose Damiano Catucci: Nothing to disclose Verena Obmann: Nothing to disclose Peter Vermathen: Nothing to disclose Lukas Ebner: Nothing to disclose Andreas Christe: Nothing to disclose Adrian Thomas Huber: Nothing to disclose Annalisa Berzigotti: Nothing to disclose Evaluation of Artificial Intelligence supported Thi rd Harmonic B-mode in gallbladder ultrasound *P. Spiesecke*, T. Fischer; Berlin/DE Purpose or Learning Objective: Superharmonic imaging is a useful B-mode ultrasound technology increasing spatial resolution . Currently, there is a novel technology available which produces B-mode images b y a combination of different harmonics up to third harmonic using an A rtificial Intelligence (AI) driven algorithm. The aim of the present study is t he first evaluation of this technology.

Methods

or Background: For this prospective study, overall 52 healthy test persons and patients were included. Standard and no vel B-mode images of the gallbladder were captured in each subject – in each case by recording several combinations of additional parameters such as Dynam ic Range and Speckle Reduction. For this purpose, a premium ultrasound s ystem was used (Canon Aplio i800 including third-harmonic imaging). The i mages were manually segmented manually and were subjected to computer-a ided analysis to analyze artifacts and edge sharpness. Additionally, Radiologists in different stages of training and subspecialisation rated the images by means of different parameters on a Likert scale.

Results

or Findings: N = 26 data sets each with and without gallbladder pathology were included. In gallbladder B-mode ultr asound, ratings of the Third Harmonic Imaging-derived images showed a significan t reduction of artifacts in the gallbladder lumen as well as a higher sharpness of interfaces. Subjective analysis revealed higher image quality in Third Har monic Imaging compared to standard B-mode.

Conclusion

Our results suggest, that the AI-driven Third Harmo nic Imaging can be a useful tool to increase the sharpness of i nterfaces and reduce the artifacts in gallbladder B-mode ultrasound.

Limitations

The present study is a single center study which fi rst evaluates this novel AI-driven Third Harmonics Imaging B-mode ultrasound technology. Further studies are necessary to evaluate this tech nology more detailed. Funding for this study: None. Ethics committee - additional information: Local ethics committee. Author Disclosures: Paul Spiesecke: Nothing to disclose Thomas Fischer: Nothing to disclose 12:30-13:30 Research Stage 1 Research Presentation Session: Physics in Medical Imaging RPS 913 Striving for lower radiation dose and better image quality Moderator D. Kostova-Lefterova; Sofia/BG ([email protected])

Results

from a decade (2012-2021) of periodical pat ient dose surveys for CT in Belgium *A. S. L. Dedulle*, T. Vanaudenhove, K. Van Slambro uck, A. Fremout; Brussels/BE ([email protected]) Purpose or Learning Objective: In Belgium, diagnostic reference levels (DRLs) are established based on periodical patient dose surveys carried out by the regulatory body. This study evaluates trends in doses from CT scans, using data from these surveys.

Methods

or Background: From 2012 to 2021, 10 periodical patient dose surveys were conducted in Belgium, covering CT equi pment nationwide, as participation in the surveys is mandatory. Anonymou s patient dose data were collected for 10 types of CT examinations (abdomen, chest angiography, coronary angiography, colon, cervical spine, lumbar spine, skull, sinus, thorax, thorax-abdomen). For each type of examination, dose data (CTDIvol, DLP) were registered for minimal 30 adult patients per C T device. The typical dose value was calculated (median) for each type of exam ination and each CT device. DRLs for each type of examination along wit h other statistical parameters were derived from the distribution of th ese values.

Results

or Findings: The participation rate exceeded 85% across all 10 surveys. Between 2012 and 2021, the 75th percentile of the typical DLP-values for complete examinations showed a decrease between 22% and 63%, depending on the type of examination. Additionally, the data spread narrowed between 8% and 65%, and the 95th percentiles decrea sed between 16% and 62%. The DRL for CTDIvol per acquisition reduced be tween 31% and 71%, while the 95th percentile of the typical CTDIvol-va lues decreased between 38% and 71%. For most examination types, the larges t decrease in DRL was obtained during the first five periodical surveys.

Conclusion

Over the 10-year period, patient doses from CT scan s in Belgium substantially decreased. This is reflected in both lower DRLs and a reduced spread in dose data. This decrease is likely the re sult of improved protocols and the introduction of advanced CT technology.

Limitations

No limitations were identified. Funding for this study: No funding was received. Ethics committee - additional information: Not applicable. Author Disclosures: An Saskia Luc Dedulle: Nothing to disclose Thibault Vanaudenhove: Nothing to disclose Katrien Van Slambrouck: Nothing to disclose An Fremout: Nothing to disclose Accounting for imaging dose in Hodgkin’s lymphoma p atients undergoing PBS proton therapy and photon VMAT: a SI NFONIA study *M. Azizi*¹, M. Romero-Expósito², I. Múñoz², A. Gka vonatsiou², O. Norrlid², C. Goldkuhl³, D. Molin², I. Toma-Dasu¹, A. Dasu²; ¹ Stockholm/SE, ²Uppsala/SE, ³Gothenburg/SE ([email protected]) Purpose or Learning Objective: This project aimed to fill a knowledge gap on the magnitude of secondary doses including the out- of-field and the imaging doses contributing to the risks from photon and pro ton radiotherapy.

Methods

or Background: A framework was developed for determining and integrating the imaging and therapy doses for indiv idual determination of total organ doses. VirtualDose software [1] was used for dose determinations from individual CT scans, while Monte Carlo simulations were used for CBCT dose determinations. Synthetic whole body CTs from the i ndividual planning CTs were generated using IS2aR-software [2]. Neutron do ses in proton radiotherapy were calculated using MCNP. Out-of-fie ld doses in photon therapy patients were determined with Periphocal3D [3]. Thursday Abstract-based Programme 97

Results

or Findings: To our knowledge, this was the first systematic assessment of total dose administered to patients t hroughout the course of the radiotherapy, encompassing clinically relevant freq uency of use of the imaging procedures. The numbers of kV-CBCT ranged from 3 to 17 and 3 to 11 CTs, respectively in photon and proton plans. Imaging do ses contribute 60-570 mSv for photon and 6-200 mSv for proton treatments over the entire treatment course for organs close to the target. Distant orga ns like the stomach, bladder, and liver showed a 13.5% increase in imaging dose r elative to the photon treatment dose, while PBS indicates a 400% increase (though with lower absolute doses), indicating its greater relative im pact.

Conclusion

Radiation burden in high precision radiotherapy dep ending on the imaging protocols will have to be taken into accoun t in epidemiological studies on the incidence of second cancers in future patien t cohorts. References: [1] A. Ding et al., Phys Med Biol; 2015. [2] I. S. Muñoz-H ernández et al., Phys. Medica, 2023. [3] B. Sánchez-Nieto, et al., Front. Oncol., 2022.

Limitations

No limitation was identified. Funding for this study: This project has received funding from Euratom’s research and innovation programme 2019-20 under gra nt agreement no. 945196. Ethics committee - additional information: The study is retrospective. Author Disclosures: Mona Azizi: Nothing to disclose Alexandru Dasu: Nothing to disclose Maite Romero-Expósito: Nothing to disclose Ola Norrlid: Nothing to disclose Angeliki Gkavonatsiou: Nothing to disclose Iuliana Toma-Dasu: Nothing to disclose Christina Goldkuhl: Nothing to disclose Daniel Molin: Nothing to disclose Isidora Múñoz: Nothing to disclose Fetal radiation dose from iodine-125 seeds in pregn ant breast cancer patients *J. Pluim*¹, J. Van De Kamer², E. Heeling², I. Ploe g², D. Hulsen¹; ¹'s-Hertogenbosch/NL, ²Amsterdam/NL Purpose or Learning Objective: The treatment of breast cancer during pregnancy (PrBC) requires careful consideration of consequences for both maternal and fetal health. In non-pregnant patients , the use of radioactive iodine-125 (125I)-seeds is standard practice for lo calising non-palpable breast tumors before breast-conserving surgery. However, t he use of 125I-seeds in pregnant patients has been avoided due to concerns about fetal radiation exposure.

Methods

or Background: This study developed a mathematical model to estimate the fetal absorbed dose based on several f actors: the radioactivity of the 125I-seed, the duration of implantation, and th e distance between the 125I- seed and fetus as a function of maternal anatomy, g estational age, and fetal development. Three scenarios, representing a range of maternal and fetal anatomy, were evaluated, including a worst-case sce nario from a radiation safety perspective.

Results

or Findings: The results show that the fetal absorbed dose varie s across the three scenarios, with ranges of 0–1.6 mG y, 0.0–1.0 mGy, and 0.0– 0.4 mGy, depending on when the 125I-seed was implan ted and when it was removed. These dose ranges are similar to conventio nal diagnostic x-ray scans. The maximum calculated absorbed dose (1.6 mG y) is unlikely to be reached in practice and is well below the 100 mGy t hreshold associated with possible fetal malformations. The associated cancer risk increase (0.016%) is minimal.

Conclusion

The use of 125I-seeds as localisation method of bre ast tumors in pregnant patients results in low fetal radiation do ses and should not be avoided due to dose concerns.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: Not applicable Author Disclosures: Eva Heeling: Nothing to disclose Iris Ploeg: Nothing to disclose Dennis Hulsen: Nothing to disclose Jip Pluim: Nothing to disclose Jeroen Van De Kamer: Nothing to disclose Implementing novel optimization strategies in x-ray interventional cardiology imaging for paediatric examinations with a simulation framework R. Massera, *N. W. Marshall*, H. Bosmans; Leuven/BE ([email protected]) Purpose or Learning Objective: To apply novel optimization strategies in the search for optimal x-ray technique factors in paedi atric interventional cardiology examinations.

Methods

or Background: A simulation framework previously developed for adult interventional radiology examinations was ada pted to use paediatric phantoms. The optimization framework implemented th e Monte Carlo (MC) code PENELOPE(2018)/penEasy(2020) for dose and imag e quality (IQ) calculations, combined with a ray-tracing routine t o calculate the attenuation through the patient and table. A figure of merit (F OM) defined as SDNRw²(u)/Dose was used. SDNRw(u) is a signal-diffe rence-to-noise ratio weighted for the impact of geometric blurring from the focal spot and from object motion. This was evaluated for the task of d etecting a 0.36 mm diameter iron guidewire. Dosimetric quantities comprised inc ident air kerma (AK) at the

Reference

point, used to approximate skin dose, and the effective dose (Deff), used to estimate stochastic risk. To calculate Deff and SDNRw(u), ICRP female paediatric phantoms of 1- and 5-year-old wer e used. The tube voltage, spectral copper filtration and x-ray focus that yie lded the highest FOM value for a particular dose quantity were found, taking into consideration x-ray tube loading limitations.

Results

or Findings: For the 1-year-old and 5-year-old phantoms, optimal FOM values were achieved at respectively 65kV/0.7 m m Cu/micro focus and 64kV/0.5 mm Cu/small focus, when AK was the cost fu nction. Using effective dose as the cost function gave optimal factors of 5 9kV/0.5 mm Cu/micro focus and 60kV/0.3 mm Cu, for the 1-year-old and 5-year-o ld cases, respectively.

Conclusion

The framework was successfully adapted to work with paediatric phantoms. Optimization based on effective dose sele cted lower tube voltages and copper spectral filtration thicknesses compared to a typical optimization using incident air kerma.

Limitations

A limited number of phantoms were used in the simul ations. Funding for this study: This study is the result of a research agreement wi th Siemens Healthineers. Ethics committee - additional information: Na Author Disclosures: Nicholas William Marshall: Nothing to disclose Rodrigo Massera: Nothing to disclose Hilde Bosmans: Other: Research agreement with Sieme ns Healthineers Image Quality in lung cancer screening LDCT: compar ing the NELSON trial to current conventional and photon-counting t horacic CT *K. Torfs*¹, D. Petrov¹, L. D'Hondt², M. Lefere³, K . Bacher², A. Snoeckx⁴, W. De Wever¹, H. Bosmans¹; ¹Leuven/BE, ²Gent/BE, ³B onheiden/BE, ⁴Zandhoven/BE ([email protected]) Purpose or Learning Objective: Current guidelines for lung-cancer-screening (LCS) with low-dose chest CT (LDCT) are focused on dose, without specifying image quality (IQ) targets. This study compares noi se and resolution in patient scans between the NELSON LCS trial, an ultra-low-do se (ULDCT) LCS study and current clinical standard-dose (SDCT) and LDCT on both energy- integrating (EIDCT) and photon-counting CT (PCCT).

Methods

or Background: IQ was measured in 54 patient scans (24-26cm water-equivalent-diameter, sharp, 1mm slice-thickne ss reconstructions) of 6 protocols: LDCT-NELSON (Siemens Sensation 16), SDCT -EIDCT, LDCT- EIDCT and ULDCT-EIDCT (Siemens SOMATOM Force) and L DCT-PCCT and SDCT-PCCT (Siemens Naeotom Alpha). Noise was comput ed per scan by averaging global-noise-levels (GNL) for soft tissue (0-170HU) from 50 equidistant slices. Resolution was quantified using AUC of the digital modulation-transfer-function (MTF) measured from th e patient skin-air- interface. Protocol averages were presented as: [kV p|reconstruction kernel|CTDIvol(mGy)|GNL-soft(HU)|MTF-AUC(mm-1)]. To assess standardized-condition-protocols, patient-specific influence of pixel-size and dose was removed by predicting GNL at 1.6mGy CTDIvo l and measuring presampled-MTF.

Results

or Findings: The results can be summarized as follows: LDCT- NELSON [120|B50|1.6±0.2mGy|155±8HU|0.52±0.06mm-1] ULDCT-EIDCT [Sn100|Br64-IR3|0.16mGy|151±7HU|0.49±0.04mm-1] SDCT-EIDCT [120|Br54|5.8±1.6mGy|70±4HU|0.49±0.06mm-1] SDCT-PCCT [120|Bl56|4.8±0.6mGy|127±9HU|0.86±0.05mm-1] LDCT-PCCT [Sn100|Bl56- IR1|1.08±0.15mGy|129±3HU|0.71±0.06mm-1] Compared to LDCT-NELSON scans, noise was significantly lower (p<0.001) in S DCT-EIDCT, SDCT-PCCT and LDCT-PCCT and the AUC-MTF significantly sharper (p<0.001) in SDCT- PCCT and LDCT-PCCT. ULDCT had similar noise and res olution properties as LDCT-NELSON, at a mean dose of only 0.16mGy versus 1.6mGy. However, for standardized-conditions, LDCT-PCCT, SDCT-EIDCT and ULDCT-EIDCT protocols were inherently less noisy (p<0.01) than NELSON, with SDCT- PCCT, LDCT-EIDCT and LDCT-PCCT being significantly sharper (p<0.001).

Conclusion

We have proposed a method to compare IQ of successf ul historical LCS scans to current state-of-the-art ca ndidates with a dose – image quality evaluation from patient CT scans. Taking th e NELSON setting as minimal reference, there are several candidate (ult ra)LDCT protocols, with the LDCT on PCCT outperforming.

Limitations

Limited no. cases Funding for this study: This work was performed with a grant from Kom op Tegen Kanker (G0B1922N), a Flemish NGO active in th e fight against cancer Thursday Abstract-based Programme 98 Ethics committee - additional information: Study approved under internal

Reference

number S68527 Author Disclosures: Mathieu Lefere: Nothing to disclose Klaus Bacher: Nothing to disclose Kwinten Torfs: Nothing to disclose Louise D'Hondt: Nothing to disclose Annemiek Snoeckx: Nothing to disclose Hilde Bosmans: Nothing to disclose Dimitar Petrov: Nothing to disclose Walter De Wever: Nothing to disclose Combining rapid kVp-switching and photon-counting d etectors for high- resolution spectral CT imaging at ultra-low doses *O. Sandvold*¹, R. Proksa¹, A. Perkins², P. Noël¹; ¹Philadelphia, PA/US, ²Cleveland, OH/US Purpose or Learning Objective: This work presents a CT acquisition paradigm utilizing sparse spectral imaging to deliv er both high spatial resolution and spectral imaging, specifically desig ned for pediatric imaging.

Methods

or Background: Combining spectral imaging with high spatial resolution at ultra-low doses is challenging with c urrent technology. In our method, most of the scan is captured using single l ow tube voltage with the detector operating in non-spectral, high-resolution mode by combining x-ray photons across energy bins (excluding electronic no ise). During sparse intervals, the system switches to rapid kVp mode, l everaging the detector's spectral capabilities. Data is continuously acquire d and combined to generate both high-resolution and spectral images. A Monte C arlo simulation demonstrated this pediatric imaging protocol, using 70 kVp with intermittent 110 kVp pulses for spectral data. The detector pixe l size was set to 0.5x0.5 mm², with an additional sampling protocol investiga ted using 1x1 mm² pixels. The simulated phantom represented a 150 mm pediatri c patient. Spectral SNR in monoenergetic images was estimated using the Cra mér-Rao Lower Bound, and the area under the monoenergetic curve (AUMC) w as calculated as the total SNR over 35–120 keV.

Results

or Findings: The sparse spectral protocol improved AUMC spectral SNR by 220% compared to a constant 100 kVp photon-c ounting scan using the same dose level and pixel size. Binning pixels to measure 1x1 mm², the sparse spectral performance was 475% the 100 kVp re ference scan AUMC. At 62% of the 100 kVp dose, the sparse protocol AUMC w as 170% greater than

Reference

AUMC.

Conclusion

Pediatric spectral CT faces three main challenges: achieving high spatial resolution, obtaining low-noise spectral da ta, and minimizing radiation dose. Our proposed acquisition method combines mult iple technologies to address these challenges. Future clinical translati on promises improved pediatric care with minimal radiation exposure.

Limitations

None Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Olivia Sandvold: Nothing to disclose Peter Noël: Nothing to disclose Roland Proksa: Nothing to disclose Amy Perkins: Employee: Philips Healthcare A Machine Learning-based method for predicting norm alized glandular dose coefficients and associated uncertainty in dig ital mammography and digital breast tomosynthesis *A. Sarno*¹, R. Massera², G. Paternò³, P. Cardarell i³, N. W. Marshall², H. Bosmans², K. Bliznakova⁴; ¹Milan/IT, ²Leuven/BE, ³Ferrara/IT, ⁴Varna/BG ([email protected]) Purpose or Learning Objective: To investigate the use of a machine learning algorithm and patient-derived digital breast phanto ms for predicting normalized glandular dose (DgN) coefficients and factors that influence the DgN uncertainty in digital mammography (DM) and digital breast tomosynthesis (DBT).

Methods

or Background: Monte Carlo dosimetry calculations were performed for a set of 126 anatomically realistic digital bre ast phantoms to establish the ground truth DgN. The DgN was then predicted using a linear regression with an Automatic Relevance Determination Regression alg orithm from 5 anatomical breast features: compressed breast thick ness, glandular fraction, total glandular volume, center of mass and standard deviation of the glandular tissue distribution in the cranio-caudal direction. An algorithm for data imputation was explored to account for the cases wh ere the latter two features are not available. The regression algorithm was val idated using 5-fold Cross Validation.

Results

or Findings: With the use of all 5 selected anatomical features, average difference between predicted DgN and the gr ound truth was 1%, with 50% of cases differing from the ground truth by les s than 3%; estimated uncertainty on the DgN values was 9%. Uncertainty o n DgN coefficients increased to 17% when the features related to the g landular distribution were excluded; however, this had only a minor impact on the prediction accuracy. The data imputation algorithm reduced the uncertain ty on the predicted values, but could not match the prediction performance obta ined by using all the available anatomical features.

Conclusion

The proposed methodology predicts the normalized gl andular dose in DM and DBT with an error of 1%, on average, and with an estimated uncertainty of only 9%. 50% of the predicted DgN co efficients differed by less than 3% from the ground truth.

Limitations

Limited to single DM/DBT geometry Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Nicholas William Marshall: Nothing to disclose Gianfranco Paternò: Nothing to disclose Kristina Bliznakova: Nothing to disclose Rodrigo Massera: Nothing to disclose Hilde Bosmans: Nothing to disclose Antonio Sarno: Nothing to disclose Paolo Cardarelli: Nothing to disclose 12:30-13:30 Research Stage 2 Research Presentation Session: Paediatric RPS 912 Insights into foetal imaging Moderator M. Rebollo Polo; Barcelona/ES ([email protected]) Validation of fetal brain 3D slice-to-volume regist ration (SVR) in detecting the cause of antenatal ventriculomegaly confirmed b y neonatal scan *W. H. E. Hamed*, G. Kendall, L. Dyet, L. Srinivasa n, D. Peebles, A. David, M. Sokolska, K. P. Baruteau; London/UK ([email protected]) Purpose or Learning Objective: Validates 3DSVR for detecting anatomical and structural pathologies in fetal MRI and assesse s quality improvement in a cohort of antenatal ventriculomegaly confirmed by n eonatal MRI.

Methods

or Background: Detecting subtle anatomical abnormalities in fetal brain MRI is challenging due to motion artefacts an d the limited spatial resolution of 2D slices. Recently, slice-to-volume reconstruction (SVR) software has been developed to realign multiple 2D stacks into a high- resolution 3D volume (3DSVR), enabling better visua lisation through multiplanar reconstruction. However, clinical valid ation of 3D-SVR is limited due to lack of ground truth data. A retrospective c ross-sectional study was conducted on pregnancies with ventriculomegaly. Inc lusion criteria included fetal and neonatal MRI performed with standard prot ocols and 3DSVR. The median gestational age at fetal MRI was 28weeks (ra nge 21-33w), and at neonatal MRI, 1week (range 1d-4w). Ventriculomegaly causes were assessed on fetal 2DT2w-HASTE and 3DSVR and confirmed with 2 DT2w-TSE on neonatal scans. Eleven brain structures were scored on a 3-point visibility scale, and image quality was rated based on signal- to-noise ratio (SNR) and motion artefacts. Statistical analysis was performe d using the Wilcoxon signed- rank test.

Results

or Findings: Of 20 subjects, eight had aqueduct stenosis identif ied on neonatal MRI. This was confirmed in 3/8 on 2D and 8 /8 on 3DSVR. Fetal 3DSVR improved visibility scores in six of eleven s tructures, with significant differences in PLIC (0.65vs1.85, p<0.001), Sylvian aqueduct (1.05vs1.9, p<0.001), olfactory bulbs (0.9vs1.7, p<0.01), and g rey-white matter contrast (0.9vs1.9, p<0.01). SNR improved in 35% of scans, a nd motion artefacts were reduced in 25%.

Conclusion

3DSVR provides improved diagnoses of aqueduct steno sis, as evidenced by comparison with neonatal ground-truth scans. This is achieved by improving the visibility and overall quality of fetal brain MRI. Future work will validate 3DSVR in other pathologies.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The study is retrospective. Thursday Abstract-based Programme 99 Author Disclosures: Magdalena Sokolska: Nothing to disclose Donald Peebles: Nothing to disclose Giles Kendall: Nothing to disclose Kelly Pegoretti Baruteau: Nothing to disclose Leigh Dyet: Nothing to disclose Weaam Hamed Elsayed Hamed: Nothing to disclose Latha Srinivasan: Nothing to disclose Anna David: Nothing to disclose Quantification of pathological fetal brain developm ent through segmentation: a novel generative AI approach for sy nthetic pathological data generation *M. Kaandorp*¹, H. Asma-Ull², H. G. Kim², D. Agbele se¹, K. Payette¹, A. Jakab¹; ¹Zurich/CH, ²Seoul/KR ([email protected]) Purpose or Learning Objective: Fetal MRI is increasingly used for the quantification of the developing brain through segm entation of various anatomical structures. One of the challenges is tha t current segmentation algorithms perform poorly in cases with ventriculom egaly due to limited high- quality annotations. Additionally, privacy concerns often restrict data sharing. We aimed to overcome these challenges through the g eneration of realistic synthetic pathological MRIs from manipulated health y label images using generative AI.

Methods

or Background: We trained a stable diffusion model for 3D brain MRI synthesis (Med-DDPM) using 727 fetal and preter m neonatal MRI-label image pairs. We generated 47 synthetic ventriculome galy label images (S47- ventriculomegaly) from 33 healthy fetal MRIs (R33-h ealthy) by dilating ventricular labels. Combining label images from R33 -healthy and S47- ventriculomegaly, we generated 80 synthetic MRIs (S 80-ventriculomegaly) using Med-DDPM and visually assessed their quality. To evaluate segmentation performance, we trained two nnUNet mod els on S80- ventriculomegaly or R33-healthy MRI-label pairs and measured performance using Dice score on 40 test cases from FeTA2021 Cha llenge and 26 spina bifida severe ventriculomegaly cases (SPINABIFIDA). Performance was also assessed for 80 healthy and pathological MRI-label pairs (R80) and their synthetic equivalent (S80).

Results

or Findings: Med-DDPM generated diverse, high-quality synthetic fetal MRIs. In segmentation tasks, S80-ventriculome galy outperformed R33- healthy, achieving higher Dice scores in FeTA2021 ( 0.773 vs. 0.751) and SPINABIFIDA (0.759 vs. 0.722), especially for ventr icles (0.780 vs. 0.672). S80 also surpassed R80 in SPINABIFIDA (0.822 vs. 0.815) , with comparable performance in FeTA2021 (0.766 vs. 0.773).

Conclusion

This study demonstrates that generating realistic p athological fetal MRIs by manipulating labels from normally dev eloping subjects can enhance data augmentation and data anonymization in prenatal imaging. This is an important step towards addressing privacy con cerns while improving segmentation performance.

Limitations

Our method could be expanded to include more pathol ogies, increasing clinical applicability. Funding for this study: Funding was provided by SNF grant: IZKSZ3_218590 Ethics committee - additional information: The study has been approved by the ethics committee of Zurich Children's Hospital, decision number: 2022- 01157) Author Disclosures: Hyun Gi Kim: Nothing to disclose Andras Jakab: Nothing to disclose Hosna Asma-Ull: Nothing to disclose Kelly Payette: Nothing to disclose Damola Agbelese: Nothing to disclose Misha Kaandorp: Nothing to disclose Feasability of fetal cardiac MRI in the prenatal ev aluation of congenital heart defects in comparison to US *G. Biechele*¹, B. Stos², D. Laux², S. Stöcklein¹, D. Grevent², L. J. Salomon²; ¹Munich/DE, ²Paris/FR ([email protected]) Purpose or Learning Objective: Congenital heart defects (CHD) are common severe birth defects, resulting in a clinical need for precise prenatal assessment. Magnetic Resonance Imaging (MRI) of the fetal heart recently became feasible with the advent of novel gating tec hniques , but its feasibility has not yet been systematically evaluated in pregna ncies with fetal CHD.

Methods

or Background: This study yet evaluated 42 singleton pregnancies, consisting of 25 fetuses (60%) with diagnosis of CH D and 17 healthy control fetuses (40%). Diagnosis of CHD was refined by a ca rdiopediatrician. All MRI images were reviewed by a single blinded operator. Image quality of cardiac sequences was rated from 0 (not usable) to 5 (high image quality in multiple orientations). Cardiac anatomy was evaluated and co rrelation between MRI findings and US diagnosis was assessed.

Results

or Findings: Fetal cardiac MRI was feasible in 40 patients (imag e quality score >0) due to extensive fetal movement i n n=2 cases. Among, in n=13 (32.5%), n=21 (52.5%) and in n=6 (15.0%) cases , image quality was rated high (image quality score 5), medium (image q uality score 3-4) and poor (image quality score 1-2), respectively. Among the 25 fetuses with CHD, fetal cardiac MRI was capable to fully confirm US-diagnos is in n=19 cases (76.0%) and to partly confirm US-diagnosis in n=6 (24.0%) c ases, mostly due to incomplete recording of structures of interest. The re was a substantial agreement between MRI and US (κ-value 0.70 with an accuracy of 0.85).

Conclusion

Our preliminary study suggests a similar diagnostic performance of MRI to US.

Limitations

With improvements in sequences, gating techniques a nd operator experience, MRI examination might soon become an es sential part of the prenatal management of CHD, not only to identify as sociated abnormalities but also to reinforce the assessment of the heart itsel f. Funding for this study: DFG Walter-Benjamin-Stipendium (BI 2563/1-1, G.B.) Ethics committee - additional information: Local ethic commitee number NCT 04142606 Author Disclosures: Laurent J. Salomon: Nothing to disclose Sophia Stöcklein: Nothing to disclose David Grevent: Nothing to disclose Bertrand Stos: Nothing to disclose Daniela Laux: Nothing to disclose Gloria Biechele: Equipment Support Recipient: Compa ny Northh Medical for the lend of a cardiac gating device during the peri od of a fellowship abroad. Thoracic findings at early gestation fetal post-mor tem micro-CT *I. C. Simcock*¹, A. Lamouroux², S. C. Shelmerdine¹ , C. Hutchinson¹, N. Sebire¹, O. Arthurs¹; ¹London/UK, ²Nimes/FR ([email protected]) Purpose or Learning Objective: To identify the range and frequency of thoracic diagnoses at less-invasive autopsy, follow ing a post-mortem micro-CT investigation for early gestation pregnancy loss.

Methods

or Background: Micro-CT provides high-resolution imaging for early gestation fetuses (<300g), typically following misc arriage or termination of pregnancy, allowing parents a less invasive autopsy option. We retrospectively analysed micro-CT diagnoses made on an unselected p opulation of over 1190 early gestation fetuses between 2017 and 2024 at ou r tertiary referral institution.

Results

or Findings: Thoracic abnormalities on micro-CT were identified in (147/1190; 12.3%), comprising 148 cardiac and 79 no n-cardiac individual abnormalities. The commonest cardiac abnormalities were septal defects (50/147; 34.0%), aortic abnormalities e.g. coarctat ion (23/147; 15.6%), congenital heart disease particularly tetralogy of Fallot (16/147; 10.9%) and ventricular abnormalities e.g. hypoplastic left hea rt (11/147; 7.5%). Most common non-cardiac chest abnormalities were abnorma l fluid accumulation (41/147; 27.9%), which included pleural effusion (3 4/147; 23.1%), hydrops (4/147; 2.7%), and pericardial effusion (3/147; 2%) . A thoracic wall defect was observed in (18/147; 12.7%) of cases.

Conclusion

A range of thoracic abnormalities were made by micr o-CT in our cohort, commonly from cardiac causes. This informat ion is useful for parents regarding the likelihood of congenital abnormalitie s in subsequent pregnancies and provides an alternative to conventional invasiv e autopsy.

Limitations

Single centre data from a large specialist centre. Not all parents consented to invasive autopsy, so some diagnoses co uld not be histologically confirmed. Funding for this study: ICS was funded by a National Institute for Health Research (NIHR) Clinical Doctoral Research Fellowsh ip (ICA-CDRF-2017-03- 53), Development and Skills Enhancement Award (NIHR 302390) and a Research for Patient Benefit Award (NIHR206174), OJ A was funded by a NIHR Career Development Fellowship (NIHR-CDF-2017-1 0-037) and SCS is supported by a NIHR Advanced Fellowship Award (NIHR -301322), and the work is funded by the Great Ormond Street Hospital Children’s Charity. AL was funded by two mobility funding from Nimes and Montp ellier university hospital All research at Great Ormond Street Hospital NHS Fo undation Trust and UCL Great Ormond Street Institute of Child Health is ma de possible by the NIHR Great Ormond Street Hospital Biomedical Research Ce ntre. The views expressed are those of the author(s) and not necess arily those of the NHS, the NIHR or the Department of Health & Social Care. Thursday Abstract-based Programme 100 Ethics committee - additional information: We obtained ethical approval from the UK NHS Health Research Authority (HRA) Eth ics Committee (IRAS ID: 131395). Author Disclosures: Ian C. Simcock: Nothing to disclose Neil Sebire: Nothing to disclose Audrey Lamouroux: Nothing to disclose Ciaran Hutchinson: Nothing to disclose Susan Cheng Shelmerdine: Nothing to disclose Owen Arthurs: Nothing to disclose Optimizing Outcomes in PPROM: Ultrasound-Based Feta l Lung Maturity Analysis *A. Verma*, A. Malik; New Delhi/IN ([email protected]) Purpose or Learning Objective: To evaluate fetal lung maturity using multiparametric ultrasound in pregnant women with p reterm premature rupture of membranes (PPROM), and to determine the accuracy of ultrasound parameters in predicting neonatal respiratory distr ess in this population. This study highlights the role of multiparametric ultras ound in predicting neonatal respiratory distress.

Methods

or Background: We conducted an 18 month observational prospective cohort study on 81 women with singleton pregnancies under 37 weeks gestation, complicated by PPROM with fetuses between the 10th and 90th weight percentiles. Fetal biometric measuremen ts included biparietal diameter (BPD), head circumference (HC), abdominal circumference (AC), and femur length (FL). We also assessed distal femoral and proximal tibial epiphyseal ossification centers, placental maturity (Grannum system), and fetal main pulmonary artery (MPA) Doppler flow measuring resistive index (RI), pulsatility index (PI), acceleration time, and ejec tion time. These parameters were correlated with neonatal respiratory distress to predict fetal lung maturity in PPROM cases.

Results

or Findings: BPD demonstrated highest diagnostic accuracy (81.48%) for predicting neonatal respiratory distre ss syndrome (RDS). The appearance of proximal tibial epiphyses showed rema rkable sensitivity (92.31%) in predicting RDS. Among the Doppler param eters, fetal MPA resistive index (RI) showed the highest diagnostic accuracy (88.89%). Notably, combined fetal main pulmonary artery indices yielde d the highest diagnostic accuracy (95.06%), followed by combined fetal biome try (91.36%).

Conclusion

Fetal lung maturity, a key determinant for neonatal respiratory distress can be assessed by grayscale ultrasound an d Doppler parameters. Multiparametric ultrasonographic assessment is a pr omising tool in prediction of neonatal RDS in women with PPROM.

Limitations

The study's generalizability is constrained by its limited sample size. Additionally, assessment of the fetal main pu lmonary artery via ultrasonography is subject to inter-observer variab ility and may be affected by inherent artifacts in ultrasound and Doppler imagin g techniques. Funding for this study: The cost of this study was covered by VMMC and Safdarjung Hospital, which provided essential resou rces and infrastructure to ensure its completion. Ethics committee - additional information: This study was approved by Institutional Ethics Committee, VMMC and Safdarjung Hospital, New Delhi (06/2022/CC-265) Author Disclosures: Amita Malik: Nothing to disclose Animesh Verma: Nothing to disclose Prenatal Prediction of Fetal Lung Maturity Using 3D Lung Volume, Lung- to-Liver Intensity Ratio Tissue Histogram and Pulmo nary Artery Doppler Indices A. Omar, A. Mohamed Tharwat, M. Aboelnasr, H. Abo-A li Hamza, *S. A. Hassanein*, W. Gaber Eldamaty; Shebin El Kom /EG ([email protected]) Purpose or Learning Objective: Fetal lung maturity assessment is the most critical factor for identifying the optimal deliver y time. A non-invasive sonographic technique is necessary to evaluate feta l lung development. we aimed to predict maturity of fetal lung utilizing 3 D lung volume ultrasound, lung to liver intensity ratio, and pulmonary artery dopp ler indices measurement

Methods

or Background: A prospective observational study was conducted on 200 pregnant females with gestational age of 32 to 40 weeks age underwent 3D ultrasound (3DUS) for determination th e fetal lung volume (FLV) and fetal lung-to-liver intensity ratio (FLLIR) (tissue histogram) with doppler examination of the main pulmonary artery (MPA) for the following parameters; acceleration-time to ejection-time ratio (At/Et), p ulsatility index (PI) and resistive index (RI) during a period of one week fr om delivery and comparing the results to the neonatal outcome.

Results

or Findings: Of 200 fetuses investigated; 113 cases (56.5%) were found to have respiratory distress syndrome. The MP A RI and PI were significantly greater in fetuses with respiratory d istress syndrome comparing with those without (2.6± 0.3 and 0.9±0.05 vs. 1.9±0.3 and 0.8±0.2, respectively with p-value < 0.001 for both). MPA At/Et was signi ficantly lesser for fetuses with RDS than fetuses without RDS (0.2±0.1 vs. 0.3±0.1 respectively, p-value less than 0.001). FLLIR was significantly lesser in RDS +ve group comparing with RDS -ve group (0.9±0.2 versus 1.3±0.3 respectively, p-value < 0.001) and FLV was significantly smaller in fetuses with respi ratory distress syndrome comparing with those with no (31.5±2.5 vs. 38.1±2.8; p-value < 0.001).

Conclusion

The utilization of main pulmonary artery Doppler in dices, together with mean fetal lung volume and FLLIR for assessing fetal lung is a quick non- invasive accurate technique for estimation of neona tal lung maturity and respiratory distress syndrome

Limitations

A uni-center study Funding for this study: No funding was present. Ethics committee - additional information: Menoufia faculty of medicine research ethics committee under code no: (4/2022OBS G35 Author Disclosures: Amal Omar: Nothing to disclose Mohamed Aboelnasr: Nothing to disclose Ahmed Mohamed Tharwat: Nothing to disclose Shaimaa Abdelhamid Hassanein: Nothing to disclose Wael Gaber Eldamaty: Nothing to disclose Haytham Abo-Ali Hamza: Nothing to disclose Intrauterine blood transfusion causes dose- and tim e-dependent signal alterations in the liver and the spleen on fetal ma gnetic resonance imaging M. Schwarz, V. Schmidbauer, *N. M. Nowak*, P. Kiena st, D. Bettelheim, J. Binder, T. Reiberger, D. Prayer, G. Kasprian; Vi enna/AT Purpose or Learning Objective: This study aimed to investigate the effects of IUTs on MRI findings in the fetal liver and spleen.

Methods

or Background: Intrauterine transfusions (IUTs) are a life-saving treatment for fetal anemia. However, with each tran sfusion iron bypasses uptake regulation through the placenta and accumula tes in fetal organs. Unlike other imaging modalities, fetal magnetic resonance imaging (MRI) is capable of non-invasively assessing fetal liver disease and/or organ iron overload. This study aimed to investigate the effects of IUTs on M RI findings in the fetal liver and spleen. For this retrospective study, we includ ed eight fetuses undergoing IUT and prenatal MRI from 2014 to 2023. The fetuses were gestational age- matched with a cohort that received fetal MRI for o ther indications, but no IUTs. Signal intensity (SI) and volumetric analyses of the liver and the spleen were performed.

Results

or Findings: Fetuses receiving transfusions had significantly la rger volumes of both liver (p=0.003) and spleen (p=0.029 ). T1 SI inversely correlated with the number of IUTs (Pearson’s r=-0. 43, p = 0.099). This effect regressed over time (r=0.69, p=0.057). T2 SI did no t correlate significantly with transfusion frequency but showed a strong positive correlation with the number of days between IUT and MRI (r=0.91, p=0.002). For splenic SI measures, similar effects were observed regarding T1 SI reduc tion per received transfusion (r=-0.36, p=0.167) and recovery of T2 S I after IUT (r=0.88, p=0.004).

Conclusion

This is the first study to report the effects of IU Ts on MRI data of fetal livers and spleens. We observed considerable dose- and time-dependent SI alterations of the liver and spleen following IU T. Furthermore, fetal hepatosplenomegaly can be expected following IUT.

Limitations

Two fetuses have had parvovirus B19 infection, whic h may cause hepato-/splenomegaly, but may also mandate the need for IUT. Funding for this study: Medical University of Vienna Ethics committee - additional information: In concordance with the principles of the Declaration of Helsinki, approved by the ethics committee of the Medical University of Vienna. Author Disclosures: Victor Schmidbauer: Nothing to disclose Nikolaus Michael Nowak: Nothing to disclose Julia Binder: Nothing to disclose Michael Schwarz: Nothing to disclose Dieter Bettelheim: Nothing to disclose Daniela Prayer: Nothing to disclose Thomas Reiberger: Nothing to disclose Patric Kienast: Nothing to disclose Gregor Kasprian: Nothing to disclose Thursday Abstract-based Programme 101 Improving image quality and decreasing SAR with hig h dielectric constant pad in 3.0T fetal MRI *Z. Zhu*, C. Yan, Z. Lin, B. Zhang; Nanjing/CN ([email protected]) Purpose or Learning Objective: To assess the potential of high dielectric constant (HDC) pad in increasing image quality and decreasing specific absorption rate (SAR) in 3.0T fetal MRI.

Methods

or Background: This prospective single-center observational study included 168 pregnant participants taking 3.0T feta l MRI scanning with and without HDC pad between 1 May 2021 and 31 November 2023. Quantitative Image-quality analysis included signal-to-noise rat io (SNR) and contrast-to- noise ratio (CNR). Qualitative analysis was perform ed by three radiologists with four-point scale to evaluate overall image quality, dielectric artifact and diagnostic confidence. Whole-body total SAR was als o compared. Correlation between image quality variable changes and particip ant clinical characteristics was evaluated using spearman correlation.

Results

or Findings: 128 participants (mean gestational age 30.25±3.53 weeks, range 22-37 weeks) undertook balanced steady state free precession (bSSFP) sequence and 40 participants (mean gestatio nal age 30.38±3.50 weeks, range 23-37 weeks) undertook single-shot fas t spin-echo (SSFSE) sequence. With HDC pad, SNR and CNR was significant ly higher (41.45% increase in SNR, 54.05% increase in CNR on bSSFP, p <0.001; 258.76% increase in SNR, 459.55% increase in CNR on SSFSE, p<0.001). Overall qualitative image quality, dielectric artifact and diagnostic confidence improved significantly (p<0.001). Adding HDC pad significant ly reduced Whole-body total SAR (32.60% on bSSFP, p0.05).

Conclusion

In a clinical setting, adding HDC pad can increase overall quantitative and qualitative image quality while re ducing dielectric artifact and SAR.

Limitations

Due to technical limitations, we could not conduct further analysis to compare the radiofrequency power deposited on ut erus, whole fetal body and fetal brain separately. We only conducted exper iments on magnetic resonance machines produced by United Imaging and d idn’t include machines from other manufactures. Funding for this study: National Science and Technology Innovation 2030 (2022ZD0211800) Ethics committee - additional information: Ethic committe of Nanjing Drum Tower Hospital Author Disclosures: Bing Zhang: Nothing to disclose Zhengyang Zhu: Nothing to disclose Zengping Lin: Nothing to disclose Chenchen Yan: Nothing to disclose 12:30-13:30 Research Stage 3 Research Presentation Session: Hybrid, Molecular and Translational Imaging RPS 906 Exploring the frontiers in hybrid and molecular imaging Moderator M. Naik; London/UK ([email protected]) Low-dose Fibronectin-targeted Gd-based contrast age nt Enables Early and Accurate Assessment of Chemotherapy Response in Pancreatic Cancer *W. Zhang*, X. Liang, Y. Du, J. Tian, N. Hong; Beij ing/CN Purpose or Learning Objective: Albumin-bound paclitaxel and gemcitabine (AG) chemotherapy is a mainstay in treatment of pan creatic ductal adenocarcinoma (PDAC), unfortunately not all patien ts respond to this treatment. Clinical imaging techniques cannot preci sely evaluate and predict the response to AG therapies over several weeks. A strong fibrotic reaction is a hallmark of drug-resistance while depletion of fibr osis is a positive response to AG. Extradomain-B fibronectin (EDB-FN) is an import ant element of fibrosis in PDAC. Here, we prepared EDB-FN targeted Gd-based co ntrast agent (EDB- Gd) to perform molecular MRI for early, noninvasive and quantitative assessment of treatment response in PDAC.

Methods

or Background: BxPC-DR was pre-treated with AG to establish acquired drug resistance. Subcutaneous and orthotop ic models with BxPC-DR or BxPC were established. Mice were intravenously i njected with EDB-Gd or Gd-DOTA. The ratio of T1 value reduction (T1d%) wer e compared quantitatively. For chemotherapy monitoring, MRI wa s performed before and after AG treatments. Histological analyses were use d for validation.

Results

or Findings: Molecular MRI with EDB-FN could specifically detect and quantify fibrogenesis in PDAC xenografts at a l ow dose 0.05mmol/kg, which is half of clinical dosage of Gd. The optimal imaging time point was 30min after injection of EDB-Gd. In addition, the t argeted probe generated more robust contrast-enhanced and longer retention compared to traditional Gd-DOTA. For chemotherapy montoring, T1d% were sign ificantly increased 2.5-fold in drug-resistance xenografts group in fib rotic tumor areas compared to AG-sensitive group (p < 0.05). Comparing the T1d % before and 5 days after AG predicted treatment response.

Conclusion

This study indicates EDB-FN-targeted molecular MRI possesses clinical applications in accurate assessment and pr ediction of AG chemotherapy.

Limitations

Correlations between the observed fibrotic changes on MRI with pathologic markers of treatment response and even s urvival and prognosis are needed in the future. Funding for this study: This study was funded by the Beijing Natural Scienc e Foundation (Grant No. 7244524, 7212207) and Nationa l Natural Science Foundation of China (Grant Nos. 62027901, 82272111, 92159303, 82071896, 81871422, 81871514, and 81227901). Ethics committee - additional information: All experimental studies were approved by the Ethics Committee of the Peking Univ ersity People's Hospital (2024PHE048). Author Disclosures: Jie Tian: Nothing to disclose Yang Du: Nothing to disclose Nan Hong: Nothing to disclose Wenjia Zhang: Nothing to disclose Xiaolong Liang: Nothing to disclose Investigating the Relationship of Endothelin Recept or Expression and Tumor Hypoxia by Optoacoustic Tomography *A. Helfen*, M. Mallik, M. Stölting, E. Hoffmann, C . Höltke; Münster/DE ([email protected]) Purpose or Learning Objective: A significant prognostic factor of tumor malignancy is the formation of a hypoxic supportive tumor microenvironment (TME). The endothelin (ET)M signaling network is li nked to tumor hypoxia through stabilized hypoxia-inducible factor 1 in a feedback loop. Here, we examined the interrelation of both cellular signali ng systems by multispectral optoacoustic tomography (MSOT).

Methods

or Background: Murine syngeneic 4T1 breast tumors were examined in vivo using MSOT depicting deoxygenated (Hb) and oxygenated (HbO2) hemoglobin content to detect hypoxic regions over one week. An exogenous fluorescent endothelin-A receptor (ETAR) probe served for evaluating ETAR expression status. Therapeutic inte rventions (anti-angiogenic and macrophage depletion) were evaluated concerning Hb/HbO2 ratio and ETAR expression changes. Treatment response to Beva cizumab, Clodronate and Sorafenib was detected over one week.

Results

or Findings: MSOT was capable of delineating and quantifying hypoxia within tumor lesions. 4T1 tumors were highl y hypoxic compared to healthy tissue, represented by oxygen saturation (s O2) of 0.33 vs. 0.79, respectively. Baseline data of the ETAR probe showe d an initial rise in signal intensity from day 0 to day 8, corresponding to tum or growth. Therapeutic interventions showed that the ETAR signal intensity could be significantly reversed, while all applied therapies did not lead to significant tumor growth reduction. However, Sorafenib and Bevacizumab led t o a significant increase in sO2 values. MSOT data were supported by subseque nt immunohistochemistry.

Conclusion

Tumor hypoxia within syngeneic murine breast cancer can be visualized non-invasively by MSOT evaluating hemogl obin (oxygenated and deoxygenated ratios) as well as ETAR expression. Fu rthermore, MSOT was able to depict therapeutic effects already in the e arly course of treatment representing a potential imaging biomarker.

Limitations

Limited penetration depth of MSOT is significantly improved compared to optical imaging, but with regard to tra nslation still mainly suitable for superficial tissues. Funding for this study: Financial support from the German Research Foundation (DFG, SF656 A04), from the medical facul ty of the University of Münster (IMF: I-HÖ111709) and from the Joachim Herz Stiftung is gratefully acknowledged. Ethics committee - additional information: All animal experiments described in this study were approved by the responsible auth orities (“Landesamt für Natur, Umwelt und Verbraucherschutz NRW”, Germany, Protocol No. 84- 02.04.2017.A011). Thursday Abstract-based Programme 102 Author Disclosures: Miriam Stölting: Nothing to disclose Carsten Höltke: Nothing to disclose Anne Helfen: Nothing to disclose Emily Hoffmann: Nothing to disclose Moushami Mallik: Nothing to disclose Theranostic Innovation: SPCCT-Triggered Photodynami c X-ray Therapy (XPDT) in a Murine Breast Cancer Model *P. Akl*, A. Carret, A. Houmeau, A. Gautheron, I. G oddard, J-B. Langlois, B. Montcel, F. Lerouge, P. C. Douek; Lyon/FR ([email protected]) Purpose or Learning Objective: Photodynamic therapy (PDT) uses light to activate photosensitizers (PS) but is limited by sh allow penetration. X-ray PDT (XPDT) overcomes this by using X-rays for deeper ac tivation. Spectral Photon- Counting Computed Tomography (SPCCT) enhances imagi ng with K-Edge capabilities. This study aims to optimize XPDT in v itro and in vivo using SPCCT and gadolinium nanoparticles (GdNp) as contra st and therapeutic agents.

Methods

or Background: GdNp doped with terbium were synthesized, functionalized for biocompatibily with different co atings (PEG and Silica) and the photosensitizer, rose bengal (RB) and irradiate d at different concentrations (0.02-1 M) in vitro using SPCCT with different X-ra y dose parameters (80-140 KVp and 10-300 mAs). Luminescence emission was reco rded using optical fiber immersed in the GdNp solution. 63 athymic nud e mice with MDA-MB-231 cell-derived breast cancer xenograft models were ir radiated using SPCCT(120kVp and 300mAs) 60 axial acquisitions/ 1 s ec, meandose 30mGy, 1800mGy in total ,24 hours after intra tumoral inje ction. Safety, biodistribution, and therapeutic effects were assessed through mice examination every 3 days for tumor volume to assess tumor growth delay .Biod istribution was monitored using SPCCT imaging.

Results

or Findings: Fluorescence emission increased with X-ray dose and concentration of GdNp. SPCCT imaging revealed speci fic distribution of GdNp, with luminescence emission proportional to the admi nistered dose in vitro and in vivo. The silica + RB nanoparticle group showed a slight improvement in efficacy compared to controls, with delayed tumor g rowth. Biodistribution studies indicated relatively different patterns of intratumoral and peritumoral localization, and heterogenous distribution.

Conclusion

This technology shows potential for customizing and enhancing the efficacy of XPDT, modulate luminescence intensi ty by adjusting X-ray dose parameters and nanoparticle concentration. Silica + RB showed a slight improvement in efficacy compared to controls.

Limitations

Though further studies are required to confirm sign ificance. Funding for this study: EU Horizon 2020 grant agreement 899549 Ethics committee - additional information: ethical comittee IRB APAFIS#44558. Author Disclosures: Isabelle Goddard: Nothing to disclose Jean-Baptiste Langlois: Nothing to disclose Angele Houmeau: Nothing to disclose Frederic Lerouge: Nothing to disclose Philippe Charles Douek: Nothing to disclose Bruno Montcel: Nothing to disclose Alison Carret: Nothing to disclose Arthur Gautheron: Nothing to disclose Pia Akl: Nothing to disclose [18F]FDG-PET/CT imaging biomarkers for time point-m atched response characterization of experimental melanomas to anti- PD-L1/anti-CTLA-4 immunotherapy *M. J. Antons*, S. Kloiber-Langhorst, H. Hirner-Epp eneder, F. Herr, S. Ziegler, M. Brendel, J. Ricke, M. Heimer, C. C. Cyran; Munic h/DE ([email protected]) Purpose or Learning Objective: Three-time point [18F]FDG-PET/CT imaging allows for in vivo monitoring of a combined anti-PD -L1/anti-CTLA-4 immunotherapy in a murine melanoma model validated by time point-matched multiparametric immunohistochemical reference stand ard

Methods

or Background: Melanoma cells (B16-F10) were injected subcutaneously into the abdominal flank of C57BL/6 mice (n=40). Following a baseline scan after day 7, the therapy group receiv ed 5 injections (i.p.) of anti- PD-L1 and anti-CTLA-4 antibodies on days 7, 9, 11, 13 and 15. The control group received sham treatment. Follow-up scans were performed on day 13 and 19. Tumor allografts were harvested for time po int-matched immunohistochemistry (CD8, Ki-67, TUNEL) to validat e PET/CT parameters (MTV, SUVmax) as imaging biomarkers of early therap y response.

Results

or Findings: At follow-up 1 (FU-1), the therapy group exhibited significantly lower MTV compared to the control gro up (p=0.0037). By follow-up 2 (FU-2), both MTV and SUVmax were significantly lo wer in the therapy group versus the control group (MTV: p=0.0078; SUVmax: p= 0.00034). Ex vivo analysis revealed significant anti-tumor effects in the therapy group, with a significantly higher apoptosis rate at FU-1 (p= 0.0 12) and FU-2 (p= 0.001). Moreover, the therapy group demonstrated a signific ant increase in CD8- positive T-cells at FU-2 (p=0.0027), while tumor ce ll proliferation was significantly lower at both follow-up time points ( FU-1: p=0.012; FU-2: p=0.012).

Conclusion

Multi-time point [18F]FDG-PET/CT allowed for the ea rly non- invasive monitoring of a combined immunotherapy wit h anti-PD-L1/anti-CTLA-4 in experimental melanomas, validated by multiparame tric immunohistochemistry. The significantly lower tumor glucose metabolism was paralleled by significant pro-immunogenic, pro-apop totic and anti-proliferative effects of the combined immunotherapy.

Limitations

The allograft model of melanoma may have limited tr anslational relevance to human tumor pathophysiology. Secondly, the observation period was relatively short, and no clinical endpoints suc h as overall survival of the animals were determined. Funding for this study: None Ethics committee - additional information: All animal experiments were performed in accordance with the guidelines for the use of living animals in scientific studies and the animal study was officia lly approved (ROB-55.2- 2532.Vet_02-19-32). Author Disclosures: Clemens C. Cyran: Nothing to disclose Maurice Heimer: Nothing to disclose Felix Herr: Nothing to disclose Sibylle Ziegler: Nothing to disclose Melissa J. Antons: Nothing to disclose Heidrun Hirner-Eppeneder: Nothing to disclose Sandra Kloiber-Langhorst: Nothing to disclose Matthias Brendel: Nothing to disclose Jens Ricke: Nothing to disclose Identification of intratumoral clusters in breast c ancer xenograft tumors by simultaneous multiparametric [18F]FMISO-PET/MRI *S. J. Bartsch*¹, J. Friske¹, M. Hacker¹, D. Laimer -Gruber¹, D. Prinz¹, T. Wanek¹, T. H. Helbich¹, K. Pinker-Domenig²; ¹Vie nna/AT, ²New York, NY/US Purpose or Learning Objective: Hypoxia is a driver of breast cancer (BC) progression, inducing more aggressive phenotypes an d intratumoral neovascularization. The quantification of hypoxia a nd neovascularization with simultaneous multiparametric [18F]FMISO-PET/MRI wou ld benefit the characterization of the hypoxic tumor microenvironm ent in BCs. We aim to combine simultaneous multiparametric [18F]FMISO-PET /MRI biomarkers and use them for the identification of intratumoral clu sters for a holistic assessment of hypoxia and neovascularization in BCs.

Methods

or Background: Female athymic nude mice (n = 32) were inoculated with luminal A, HER2+ or triple negative BC cells. PET/MRI was performed on a Bruker 94/30USR system, combined wit h a Bruker PET-insert. Hypoxia was evaluated using [18F]FMISO-PET, and by hyperoxic blood oxygen level dependent (BOLD) MRI. Neovascularizati on was assessed via dynamic contrast enhanced MRI, and non-contrast-enh anced intravoxel incoherent motion MRI. Intratumoral clusters were i dentified based on a hierarchical cluster analysis using R (version 4.2. 3).

Results

or Findings: The cluster analysis of [18F]FMISO-PET/MRI revealed distinct clusters in all BC subtypes. Clusters corr esponding to hypoxia showed elevated SUVbw values and the lowest ktrans and ve, indicating dense tissue and limited vessel permeability, along with little change in R2* following hyperoxic BOLD-MRI. The least hypoxic cluster had t he lowest SUVbw values and the highest ktrans and ve of all clusters and h igh change in R2*. Hypoxic clusters were more prevalent in HER2+ and triple ne gative BCs than in luminal A BCs, while less hypoxic clusters were most common in luminal A BCs.

Conclusion

Simultaneous multiparametric [18F]FMISO-PET/MRI pro vides a holistic perspective on hypoxia and neovascularizat ion in three BC molecular subtypes. These insights enable a non-invasive char acterization of BC, and may be used for assessing treatment response.

Limitations

The use of intratumoral cluster identification for the assessment of treatment response has to be confirmed in upcomi ng studies. Funding for this study: This work was funded by the Vienna Science and Technology Fund (WWTF), project number LS19-018. Ethics committee - additional information: This animal study was approved by Austrian Federal Ministry of Education, Science and Research [66.009/0284-WF/V/3b/2017; 2020-0.363.124; 2022-0.7 26.820] and the Intramural Committee for Animal Experimentation of the Medical University of Vienna. Thursday Abstract-based Programme 103 Author Disclosures: Silvester Julian Bartsch: Nothing to disclose Daniela Laimer-Gruber: Nothing to disclose Katja Pinker-Domenig: Nothing to disclose Thomas H. Helbich: Nothing to disclose Daniela Prinz: Nothing to disclose Marcus Hacker: Nothing to disclose Joachim Friske: Nothing to disclose Thomas Wanek: Nothing to disclose Comparative Radiomics and Feature Consistency: In V ivo CT vs. Ex Vivo Micro-CT in Classifying Lung Cancer Subtypes *L. Brizzi*, L. Preda, C. Bortolotto, S. Megalizzi, D. Malerba, F. Checchin; Pavia/IT ([email protected]) Purpose or Learning Objective: The study aimed to compare radiomic features between conventional in vivo CT and micro- CT of ex vivo lung tumor blocks following lobectomy. This analysis included 60 patients with lung cancer, comprising 30 adenocarcinomas (ADK) and 30 squamous cell carcinomas (SCC). The goal was to assess the variat ion and significance of features to improve predictive accuracy in distingu ishing between tumor types.

Methods

or Background: The dataset comprised 107 radiomic features, compliant with IBSI standards, extracted using Pyra diomics software. Statistical analyses, including a percentage variat ion calculation and t-tests, were performed to evaluate the correlation and cons istency of features between in vivo CT and ex vivo micro-CT. The analys is focused on features with potential utility in discriminating between AD K and SCC tumors.

Results

or Findings: Out of the 107 radiomic features, 82 showed less th an 10% variation between the two imaging modalities, w ith 46 features exhibiting a variation of less than 1%. Shape features and app roximately 90% of GLCM features demonstrated strong consistency. T-tests r evealed that 21 radiomic features had a p-value < 0.05, indicating statistic al significance. These features included various shape characteristics and first-or der statistics that are crucial for tumor classification.

Conclusion

The findings indicate a strong correlation between the radiomic features extracted from in vivo CT and micro-CT, pa rticularly in shape and GLCM features. The identified significant features offer promising potential for improving predictive models for lung cancer classif ication between ADK and SCC.

Limitations

The study was limited to a specific subset of lung tumors (ADK and SCC), and the generalizability to other lung ca ncer subtypes remains uncertain. Additionally, technical variations in CT acquisition parameters might affect feature extraction consistency. Further vali dation with larger datasets is necessary. Funding for this study: Founding are provided by Research Foundings on AI of the IRCCS Policlinico San Matteo Ethics committee - additional information: Protocol 25657/2024 Author Disclosures: Leonardo Brizzi: Nothing to disclose Silvia Megalizzi: Nothing to disclose Davide Malerba: Nothing to disclose Filippo Checchin: Nothing to disclose Lorenzo Preda: Nothing to disclose Chandra Bortolotto: Nothing to disclose 12:30-13:30 Research Stage 4 Research Presentation Session: Chest RPS 904 Imaging of pulmonary embolism and pulmonary hypertension Moderator N. J. Screaton; Cambridge/UK ([email protected]) Radiomics parameters of epicardial adipose tissue p redict mortality in acute pulmonary embolism *H-J. Meyer*¹, S. Zimmermann¹, J. Borggrefe², A. Su rov²; ¹Leipzig/DE, ²Minden/DE Purpose or Learning Objective: Accurate prediction of short-term mortality in acute pulmonary embolism (APE) is very important. T he aim of the present study was to analyze the prognostic role of radiomi cs values of epicardial adipose tissue (EAT) in APE.

Methods

or Background: Overall, 508 patients were included into the study, 209 female (42.1%), mean age, 64.7 ± 14.8 years. 4.6%and 12.4% died (7- and 30-day mortality, respectively). For external v alidation, a cohort of 186 patients was further analysed. 20.2% and 27.7% died (7- and 30-day mortality, respectively). CTPA was performed at admission for every patient before any previous treatment on multi-slice CT scanners. A tr ained radiologist, blinded to patient outcomes, semiautomatically segmented the E AT on a dedicated workstation using ImageJ software. Extraction of ra diomic features was applied using the pyradiomics library. Patients were random ly assigned to a training and a validation cohort with a ratio of 7:3. We cha racterized two models (30- day and 7-day mortality).

Results

or Findings: We fitted the characterized models to a validation cohort (n = 169) in order to test accuracy of our models. We observed an AUC of 0.776 (CI 0.671-0.881) and an AUC of 0.724 (CI 0.62 8-0.820) for the prediction of 30-day mortality and 7-day mortality, respective ly. The overall percentage of correct prediction in this regard was 88% and 79% i n the validation cohorts. Lastly, the AUC in an independent external validati on cohort was 0.721 (CI 0.633-0.808) and 0.750 (CI 0.657-0.842), respective ly.

Conclusion

Radiomics parameters of EAT are strongly associated with mortality in patients with APE.

Limitations

It is a retrospective study, which should be evalua ted in a prospective multi center analysis. Funding for this study: None Ethics committee - additional information: Nr. 145/21, Ethics Committee, Otto-von-Guericke University of Magdeburg, Magdebur g, Germany) Author Disclosures: Alexey Surov: Nothing to disclose Silke Zimmermann: Nothing to disclose Hans-Jonas Meyer: Nothing to disclose Jan Borggrefe: Nothing to disclose Feasibility Study on the Use of 6ml Iodine Contrast Agent in Pulmonary Artery CT *H. Shang*, Y. Gao, D. D. Tian, K. Li, X. Zhang, P. Cao; Xi'an/CN ([email protected]) Purpose or Learning Objective: To assess the viability of using a low dose of iodine contrast medium (CM) and slow injection rate in single-energy 40keV CT imaging for pulmonary artery angiography (CTA).

Methods

or Background: Seventy patients, clinically suspected of pulmonary embolism and treated between January and September 2024, were randomly assigned to either an experimental group (35 patien ts) or a control group (35 patients). The experimental group underwent imaging at 100kV with 6ml CM at an injection rate of 2.5ml/s, triggered at a thresh old of 60HU. The control group was imaged at 120kV with 35-40ml CM at an injection rate of 3.5ml/s, triggered at 80HU. All scans targeted the pulmonary artery tr unk and utilized the non- ionic, water-soluble iodine contrast agent iomeprol (400mg/ml). Both groups were scanned under free breathing conditions. Image quality was subjectively rated on a 5-point scale, and objectively assessed based on pulmonary vascular enhancement, CT values, signal-to-noise ra tio (SNR), contrast-to- noise ratio (CNR), and radiation dose metrics inclu ding dose-length product (DLP) and volumetric CT dose index (CTDIvol). Thursday Abstract-based Programme 104

Results

or Findings: The CT values of the main pulmonary artery showed n o significant difference between groups (P>0.05). How ever, the experimental group demonstrated significantly higher CT values i n the right upper, middle, interlobar, and lower lobe arteries, as well as the left upper and lower lobe arteries (P 0.05). The experimental group also had significantly lower CTDIvol and DLP values compared to the control group (P0.05).

Conclusion

CTA of the pulmonary artery at 40keV with 6ml CM an d an injection rate of 2.5ml/s can provide diagnostic im age quality.

Limitations

Not applicable Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Pan Cao: Nothing to disclose Dong Dong Tian: Nothing to disclose Xiaoyue Zhang: Nothing to disclose Kai Li: Nothing to disclose Haitao Shang: Nothing to disclose Yanjun Gao: Nothing to disclose Double-low protocol CTPA for Pulmonary Embolism det ection: Enabling artificial intelligence with deep learning-reconstr uction based Images *L. Shen*, J. Lu, C. Zhou, Z. Bi, X. Ye, M. Zeng, W . Mingliang; Shanghai/CN ([email protected]) Purpose or Learning Objective: To assess the effectiveness of artificial intelligence software (AI) in detecting pulmonary e mbolism (PE) using low-dose CT pulmonary angiography (CTPA) enhanced by deep le arning reconstruction (DLR) and contrast-enhancement boost (CE-boost) tec hnique.

Methods

or Background: This prospective two-center study included 180 patients who underwent CTPA for suspected PE. Patie nts were randomly divided into two groups: the routine CTPA group wit h 50 mL contrast medium (CM) was reconstructed using HIR, and the low-dose CTPA group with 25 mL CM was reconstructed using DLR. The CE-boost was ad ditionally performed to generate DLR-boost images in the low-dose group. Si gnal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of pulmonary arte ries were quantitatively assessed. For qualitative image quality assessment, two experienced radiologists independently rated CT images (5, best ; 1, worst). A subset of 46 randomly selected patients in each group (1:1 ratio ) were evaluated by the AI software (Discover PE, uAI) for the presence of PE. Reference standard was established by expert consensus. The diagnostic acc uracy (sensitivity and specificity) of the AI interpretations were compare d between methods by bootstrapping.

Results

or Findings: DLR-boost images produced lower noise, higher SNR and CNR, and superior subjective image quality comp ared to HIR images in the routine group (p < 0.05). For detecting PE, com pared to HIR images, DLR- boost images showed comparable sensitivity (97.67% vs. 93.02%, p 0. 05). The effective dose of the double-low group and the routine group was 1.19 ± 0.45 mSv and 2.69 ± 0.49 mSv, respectively.

Conclusion

DLR-boost significantly enhances CTPA image quality at reduced radiation and contrast doses. AI software achieves diagnostic performance comparable to traditional reconstruction methods, s upporting its use in clinical practice.

Limitations

N/A Funding for this study: This study has received funding by Shanghai Anticancer Association EYAS PROJECT (Grant NO. SACA CY22C15). Ethics committee - additional information: This study was approved by the Ethics Committee of Shanghai Geriatric Medical Cent er(B2024-009). Author Disclosures: Xiaodan Ye: Nothing to disclose Leilei Shen: Nothing to disclose Zhenghong Bi: Nothing to disclose Wang Mingliang: Nothing to disclose Chun Zhou: Nothing to disclose Mengsu Zeng: Nothing to disclose Jinjuan Lu: Nothing to disclose AI-driven pulmonary vascular analysis with computed tomography in patients with chronic thromboembolic pulmonary dise ase *A. Cisarri*, A. Valentini, E. M. Bassi, I. Fiorina , A. D'Onorio De Meo, G. Rodolico, M. Zacchino, K. Ellena, L. Preda; Pavi a/IT ([email protected]) Purpose or Learning Objective: Chronic thromboembolic pulmonary disease (CTEPD) is a rare condition in which patients may o r may not develop pulmonary hypertension. Although both groups are tr eated in the same way, through pulmonary thromboendarterectomy (PEA), the pathogenic mechanisms by which some patients develop pulmonary hypertension while others do not have yet to be clarified. The study a ims to investigate vascular changes using AI-driven quantification of pre- and post-operative CT scans in patients undergoing PEA.

Methods

or Background: The study analyzed 35 patients with chronic thromboembolic disease, divided into three groups b ased on preoperative mean pulmonary arterial pressure (mPAP), from 2017- 2022. Pre- and post- operative CT scans of 22 of these patients were ana lyzed with AI-software to quantify vascular morphology, including vessel numb er, diameter, and blood volume at various pleural depths. Hemodynamic param eters such as mPAP and pulmonary vascular resistance (PVR) were also a ssessed.

Results

or Findings: No significant differences in vascular quantificati on parameters were observed between patient groups pre -operatively. Post- treatment analysis showed a reduction in the number of small vessels (p=0.0065) and blood volume of small vessels (BV5) at 24 mm depth (p=0.036), with an increase in mean vessel diameter (p=0.0005). A significant correlation was found between BV5 reduction and PVR improvement (p=0.01).

Conclusion

CT quantification revealed significant post-operati ve vascular changes in CTEPD patients, especially in the medium -caliber vasculature. The findings suggest that PVR reduction is primarily li nked to the reperfusion of larger vessels, with limited improvement in smaller vessel volumes. Future studies should explore these parameters as potentia l biomarkers for diagnosis and prognosis in CTEPD.

Limitations

The small sample size and limited postoperative fol low-up restrict the study’s ability to generalize findings. Further prospective studies with larger cohorts are needed. Funding for this study: No funding Ethics committee - additional information: Approved Author Disclosures: Giuseppe Rodolico: Nothing to disclose Emilio Maria Bassi: Nothing to disclose Ilaria Fiorina: Nothing to disclose Alessandro D'Onorio De Meo: Nothing to disclose Andrea Cisarri: Nothing to disclose Michela Zacchino: Nothing to disclose Lorenzo Preda: Nothing to disclose Katia Ellena: Nothing to disclose Adele Valentini: Nothing to disclose Automated Quantified CT analysis of Morphological D ifferences in Chronic Thromboembolic Pulmonary Disease and chroni c thromboembolic pulmonary hypertension *W. Xu*, L. Xi, A. Liu, M. Liu, S. Zhao; Beijing/CN ([email protected]) Purpose or Learning Objective: We aim to study the new morphological markers of Chronic Thromboembolic Pulmonary Disease (CTEPD) and chronic thromboembolic pulmonary hypertension (CTEPH) on co mputed tomography pulmonary angiography (CTPA).

Methods

or Background: We retrospectively enrolled CTEPH, CTEPD patients, and control group from January 2019 to Oc tober 2023 in our hospital. The morphological metrics including pulmonary blood volume, tortuosity, and fractal dimension (FD) on CTPA were automatically q uantified on an Artificial Intelligence workstation. We compared these metrics among three groups and assessed their correlation with hemodynamics.

Results

or Findings: A total of 190 participants (97 men, 56.2±10.9 year s old) including 116 CTEPH patients ,54 CTEPD patients and 20 control enrolled in this study. The pulmonary artery tortuosity in the control group, CTEPD group, and CTEPH group showed a gradually increased progre ssively (1.07 [1.06– 1.10] vs. 1.10 [1.07–1.14] vs. 1.14 [1.10–1.18], P< 0.01). There was a positive correlation between pulmonary artery tortuosity and mean pulmonary artery pressure (r=0.47, P<0.01), pulmonary vascular resis tance (r=0.44, P<0.01). Additionally, the volume of small and medium-sized pulmonary arteries was significantly higher in CTEPD patients compared to those with CTEPH (P0. 05).

Conclusion

Pulmonary arterial tortuosity on CTPA is a crucial imaging biomarker for distinguishing between CTEPH and CTEP D. The preservation of normal volumes in the small and medium-sized pulmon ary arteries observed in CTEPD patients implies that this feature could be a key determinant in maintaining normal resting pulmonary artery pressur e.

Limitations

First, it was a single-center investigation with a small cohort of CTEPD patients and controls, and the non-normal dis tribution limits the generalizability of our findings. Larger studies wi th more diverse populations are needed to validate our results. Secondly, there is a need for further refinement in the precision of pulmonary vessel seg mentation. Funding for this study: This work was supported by This study is supported by the National Natural Science Foundation of China (82272081), Chinese Academy of Medical Sciences Innovation Fund for Med ical Sciences (2021- I2M-1-049, 2022-I2M-C&T-B-109). All authors have no thing to disclose. Thursday Abstract-based Programme 105 Ethics committee - additional information: This single-center study was approved by the hospital’s Ethics Committee (2022-K Y-048) and was performed in accordance with the Declaration of Hel sinki. Author Disclosures: Min Liu: Nothing to disclose Wenqing Xu: Nothing to disclose Linfeng Xi: Nothing to disclose Anqi Liu: Nothing to disclose Shihua Zhao: Nothing to disclose Comparison of dual energy CT and V/Q SPECT in diagn osis of chronic thromboembolic pulmonary hypertension E. Pershina, D. Shchekochikhin, *A. Oganesyan*; Mos cow/RU ([email protected]) Purpose or Learning Objective: To assess diagnostic value of DECT versus V/Q SPECT for CTEPH detection in PAH patients in Mo scow Pulmonary Hypertension Center.

Methods

or Background: DECT with the calculation of iodine maps and V/Q SPECT were performed in 29 patients (f/m - 9/13; ag e 65 ±10) with PAH and risk factors for CTEPH. All patients underwent righ t heart catheterization for PAH confirmation. CTA analysis included the number and level of vessel occlusions together with the presence and size of l ung perfusion defects on iodine maps and CT signs of right heart failure.

Results

or Findings: 21 of 29 patients (72,4%) demonstrated CTA features of possible CTEPH such as intraluminal defects, enlarg ed pulmonary trunk, pulmonary mosaic patten. Iodine maps revealed perfu sion defects of 19 patients (86%). Two patients (7%) had intraluminal irregular defects without perfusion defects by DECT. V/Q SPECT determined per fusion defects in 20 patients (69%). There was one case (3, 4%) with mis match between iodine maps and V/Q SPECT. However, it was explained by th e enlarged pressure on the right heart chambers due to the cava-caval (cav a superior – cava inferior) venous anastomosis and hepatic veins abnormalities. Sensitivity for perfusion defects by DECT was 95% and specificity 100%. Furth ermore, the mean radiation dose was significantly lower for DECT vs. V/Q-SPECT (p = 0.006). Spreading of perfusion defect had strong correlatio n with severity of right heart failure (r=0.2, p<0.05).

Conclusion

Pulmonary CTA with iodine mapping improves the dete ction of CTEPH. Iodine maps demonstrate high sensitivity and specificity in comparison with V/Q SPECT. According to our preliminary data D ECT could be potentially implemented in future diagnostic PH algorithms at l east on par with V/Q- SPECT. Further research is needed.

Limitations

Pregnant Older 18 years GFR>30 ml/min Iodine allergic reaction Funding for this study: No funding Ethics committee - additional information: The study was approved by local ethical committee. Author Disclosures: Ekaterina Pershina: Nothing to disclose Anait Oganesyan: Other: Analysis datas Author: Has wrote study Speaker: presents study Dmitry Shchekochikhin: Nothing to disclose Development of a CT based prognostic predictive mod el in chronic thromboembolic pulmonary hypertension (CTEPH) *S. Gowda*, R. Kothari, V. Raj; Bengaluru/IN ([email protected]) Purpose or Learning Objective: Chronic thromboembolic pulmonary hypertension (CTEPH) is a severe complication of pu lmonary embolism characterized by persistent pulmonary arterial hype rtension due to unresolved thromboembolic obstructions. This study explores th e development and validation of a CT based prognostic predictive mode l designed to simulate hemodynamic parameters and guide management in CTEP H patients.

Methods

or Background: We retrospectively analysed the data of 1100 patients diagnosed with CTEPH. Integrated data cons isted of clinical history, echocardiogram reports, right heart catheterisation reports if any, CT pulmonary angiogram studies and proposed management strategies. CT clot burden score, CT parameter score and Perfusion defe ct score was obtained in all the patients. Patients were contacted to review the clinical outcomes.

Results

or Findings: Parameter scoring was seen to be useful in predicti ng hemodynamic parameters and outcomes. A cut-off of 3 is close to normal hemodynamic status (sPAP, accuracy of 76%), while a cut-off of 5 represents higher than normal values of mPAP. A score of ≥7 indicated increased risk of mortality.

Conclusion

The CT based prognostic predictive model represents a valuable tool in the management of CTEPH by integrating deta iled imaging with clinical factors. We suggest this model can be a standard co mponent of CTEPH patient evaluation and management especially when r ight heart catheterisation is contraindicated or not possible. Future studies should focus on prospective validation and integration into clinical practice t o further refine its utility and impact in the management. It can be used to avoid r ight heart catheterisation and ventilation-perfusion scans in selected patient s who meet the cut-off values, thereby reducing additional cost burden to the patients.

Limitations

Single centre study, which may also have an in-buil t case selection bias. Small sample size to ascertain like lihood of mortality Needs prospective validation to integrate into clin ical practice Funding for this study: No funding was provided for this study. Ethics committee - additional information: The study received institutional review board approval. Author Disclosures: Richa Kothari: Nothing to disclose Vimal Raj: Nothing to disclose Suraj Gowda: Nothing to disclose Performance of chest contrast-enhanced CT in pulmon ary hypertension clinical grouping: a dual-center, expert-blinded an alysis *L. Nardone*¹, L. Cereser¹, G. Agati¹, P. Ciolli², T. Nadarević³, C. Cicciò⁴, A. Borghesi², R. Girometti¹, C. Zuiani¹; ¹Udine/IT, ²Brescia/IT, ³Rijeka/HR, ⁴Negrar/IT ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the diagnostic accuracy of contrast-enhanced CT imaging in classif ying patients with pulmonary hypertension (PH) across international gu idelines-derived clinical groups I-V. The analysis focused on quantifying the utility of CT in a blinded setting, comparing the results with those from the PH multidisciplinary team meeting (PH-MDTM).

Methods

or Background: We retrospectively included 172 contrast-enhanced CT studies from patients with PH performed in two t ertiary referral centers. Three chest-devoted radiologists, blinded to the cl inical data, independently reviewed all the CTs, assigning probability percent ages for each of the five PH groups. A consensus grouping hypothesis was reached by averaging the probabilities across readers, and this was compared with the PH-MDTM grouping. Accuracy and Cohen’s Kappa (k)-derived in ter-reader agreement values with 95% confidence intervals (95% CI) were calculated. The readers’ discriminatory power between individual PH groups w as evaluated through areas under the receiver operating characteristic c urve (AUC) analyses.

Results

or Findings: The expert-blinded consensus diagnosis agreed with the PH-MDTM in 124/172 cases (accuracy, 72%; k, 0.62; 9 5%CI, 0.50-0.70). When including the second most probable group, the readers correctly grouped 148/172 cases (accuracy, 86%; k, 0.81; 95%C I, 0.74-0.88). Discriminatory power analysis for individual groups demonstrated good readers’ performance, with AUC values ranging from 0.79-0.90 depending on the group.

Conclusion

This study highlights the value of contrast-enhance d CT in classifying PH according to clinical groups, with e xpert readers achieving substantial-to-almost perfect agreement with the PH -MDTM grouping and good discriminatory power for individual PH groups. The results suggest that CT imaging can serve as a reliable tool in the clinica l work-up of PH patients, supporting its integration into multidisciplinary e valuations.

Limitations

Retrospective design, low number of group V PH case s. Funding for this study: No funding was received for this study. Ethics committee - additional information: The ethics committee notification can be found under the number IRB 250/2023. Author Disclosures: Andrea Borghesi: Nothing to disclose Chiara Zuiani: Nothing to disclose Rossano Girometti: Nothing to disclose Giorgio Agati: Nothing to disclose Pietro Ciolli: Nothing to disclose Tin Nadarević: Nothing to disclose Lorenzo Cereser: Nothing to disclose Luigi Nardone: Nothing to disclose Carmelo Cicciò: Nothing to disclose Thursday Abstract-based Programme 106 14:00-15:30 Research Stage 1 Research Presentation Session: Genitourinary RPS 1007 Imaging of renal and bladder malignancies: advances in diagnosis and characterisation Moderator N. Cowan; Portsmouth/UK Enhanced U-Net for Precise Auto-Segmentation of Bla dders and Tumors in CT Urography Imaging *L. Chen*, L. Mao, X. Li, X. Zhang, X. Bai, G. Zhan g, H. Xue, Z. Jin, H. Sun; Beijing/CN ([email protected]) Purpose or Learning Objective: This study developed a deep learning model for bladder and tumor segmentation from CT Urograph y images (CTU), integral to a system aiding bladder cancer management.

Methods

or Background: A 381-case dataset from two centers, approved by the Institutional Review Board, was utilized. It co mprised 280 training cases (275 with masses, 5 normal), 56 internal validation cases (54 with masses, 2 normal), and 45 external validation cases (44 with masses, 1 normal). The set included nephrographic phase CTU scans with 0.625mm and 1mm slice thicknesses for patients with pathologically confir med lesions. A reference standard with manual contours was provided by an ex perienced radiologist and reviewed by a senior one. The nnU-Net framework tra ined a U-Net-based segmentation model using an ensemble prediction in a five-fold cross- validation and test-time augmentation. Performance was assessed using DSC, 95% HD, and MSD on the testing set.

Results

or Findings: Results show that our approach achieves superb segmentation accuracy. In the internal validation s et, the U-Net-based model showed strong performance with a DSC of 97.9%, 95% HD of 0.48mm, and MSD of 3.43mm for bladder segmentation. It excelled in the external set with a DSC of 98.4%, 95% HD of 0.34mm, and MSD of 2.52mm. For tumor segmentation, the internal set results were a DSC o f 76.6%, 95% HD of 3.70mm, and MSD of 19.15mm, while the external set showed a slight decrease to a DSC of 74.4%, 95% HD of 3.73mm, and M SD of 22.08mm.

Conclusion

Though the tumor segmentation was less precise than bladder segmentation, the U-Net-based model still provided satisfactory accuracy for both, excelling in bladder delineation. This model proves valuable for detecting bladder cancer and evaluating treatment efficacy.

Limitations

The limited sample size of the external validation cohort limited the generalizability. Funding for this study: This work was supported by National High-Level Hospital Clinical Research Funding(2022-PUMCH-A-035 ), National High-Level Hospital Clinical Research Funding(2022-PUMCH-B-069 ), National High-Level Hospital Clinical Research Funding(2022-PUMCH-A-033 ), the Natural Science Foundation of China (Grant No.81901742), the Beijin g natural Science Foundation (Grant No. L232133), and the CAMS Innova tion Fund for Medical Sciences (2022-12M-C&T-B-019). Ethics committee - additional information: The Pecking Union College Hospital Institutional Review Board approval was ob tained (ethical approval number: I-22PJ887). Author Disclosures: Xiuli Li: Nothing to disclose Li Mao: Nothing to disclose Gumuyang Zhang: Nothing to disclose Hao Sun: Nothing to disclose Xin Bai: Nothing to disclose Zhengyu Jin: Nothing to disclose Li Chen: Nothing to disclose Xiaoxiao Zhang: Nothing to disclose Huadan Xue: Nothing to disclose Radiogenomics of renal cell carcinoma: using MRI tu mor features to predict mir-15a expression *Y. Mytsyk*¹, P. Kowal², Y. Kobilnyk³, I. Dutka¹, I . Komnatska¹, A. Górecki⁴; ¹Lviv/UA, ²Wroclaw/PL, ³Przemyśl/PL, ⁴Przeworsk/PL ([email protected]) Purpose or Learning Objective: The aim of this study was to evaluate the effectiveness of MRI tumor parameters in predicting tissue expression of miR- 15a in renal cell carcinoma (RCC) patients.

Methods

or Background: The study involved 64 patients with histologically confirmed conventional RCC, where miR-15a expressio n was measured, and preoperative contrast-enhanced MRI (1.5 T) was perf ormed. MiR-15a expression was determined using reverse transcripti on and real-time PCR. A polynomial regression model assessed associations b etween miR-15a expression and radiological tumor parameters, with accuracy evaluated by the Fisher method (adjusted R²).

Results

or Findings: It was found that radiological features of the cyst ic component, exophytic growth, necrosis, macroscopic fat, and nodular contrast enhancement of the tumor were observed in 29.69%, 2 3.44%, 32.81%, 20.31%, and 37.5% of patients, respectively. The me an levels of miR-15a expression in the presence of these features were 0 .35±1.02 U, 0.34±1.09 U, 4.01±3.42 U, 0.29±0.87 U, and 2.91±3.24 U, respectively. In the absence of these features, the mean expression of miR-15a was 2.01±2.93 U, 1.88±2.85 U, 0.82±1.85 U, 1.83±2.83 U, and 0.68±1.72 U, respectively (p<0.05). The highest miR-15a expression levels were observed wit h necrosis, and the lowest with macroscopic fat (p<0.05). Tumor size st rongly correlated with miR- 15a expression (r=0.724; p<0.001). Tumor size alone predicted miR-15a expression with an adjusted R² of 0.8281, and combi ning tumor size with other radiological features predicted 85% of miR-15a expr ession (R²=0.8336; p<0.001). The study developed a predictive formula for miR-15a expression based on RCC radiological features.

Conclusion

The findings suggest that MRI parameters can accura tely predict miR-15a expression, which holds diagnostic and prog nostic value in RCC.

Limitations

The main limitation was the inclusion of only conve ntional RCC. Funding for this study: No funding. Ethics committee - additional information: The study was approved by the Ethics Committee of the Danylo Halytsky Lviv Nation al Medical University (protocol No. 5 dated May 25, 2021). The work was c onducted in accordance with accepted standards for conducting research in the field of biology and medicine, based on the guidelines of the World Heal th Organization, the International Council of Medical Scientific Societi es, the International Code of Medical Ethics (1983), the Helsinki Declaration ado pted by the General Assembly of the World Medical Association, the Conv ention on Human Rights and Biomedicine of the Council of Europe (1997), an d the requirements and standards of ICH GCP (2002). In each specific case, patients or responsible individuals provided written consent for the surgic al intervention. Author Disclosures: Yulian Mytsyk: Nothing to disclose Iryna Komnatska: Nothing to disclose Yuriy Kobilnyk: Nothing to disclose Andrzej Górecki: Nothing to disclose Paweł Kowal: Nothing to disclose Ihor Dutka: Nothing to disclose An Artificial Intelligence Framework Based On Contr ast-Enhanced CT For Preoperative Predicting WHO/ISUP Nuclear Grade Of C lear Cell Renal Cell Carcinoma: A Multicenter Study *J. Han*¹, T. Liu², J. Li¹, Y. Zhang³; ¹Hohhot/CN, ²Guangzhou/CN, ³Zhuhai/CN ([email protected]) Purpose or Learning Objective: To determine whether the artificial intelligence integrated model based on automatic se gmentation of CT images can provide a robust prediction of clear cell renal cell carcinoma (ccRCC) ISUP/WHO grade.

Methods

or Background: Pretreatment CT scans were retrospectively acquired in patients with surgically proven ccRCC a t multiple centers from January 2017 to September 2023.The proposed framewo rk comprised five modules, including a 3D tumor segmentation model by 3D-UNet, a deep learning feature extraction module, a radiomic feat ure extraction module, a clinical-radiological feature screening module, and a fully-connected classification module that combines features from d ifferent sources to classify low-grade (I and II) and high-grade (III and IV) cc RCC. The Grad-CAM method and SHAP method are used to analyze the interpretab ility of the artificial intelligence model. Thursday Abstract-based Programme 107

Results

or Findings: The training data set was comprised of 335 patients from three centers, and 110 and 84 patients were in cluded in the two external test data sets. The average Dice coefficient of the 3D-UNet automatic segmentation network in the test sets was 0.86 and 0.82. Synchronous distant metastasis, Planned nephrectomy type, and tumor lon g axis as independent predictors of high-grade ccRCC. In the test sets, t he AUC and accuracy of integrated model were 0.85-0.92, 78-85%, respective ly, which were exceeded those of clinical-radiological feature model (0.85 vs0.75 [P = 0.039], 0.93 vs 0.76 [P = 0.043], 78% vs 65% [P < 0.001], 85% vs 71 % [P = 0.035].

Conclusion

An integrated model based on clinical features, rad iological features, radiomics features, and deep learning fea tures provided reliable prediction of WHO/ISUP grade for ccRCC, which outpe rformed the clinical- radiological feature model.

Limitations

This is a retrospective study and we only included ccRCC with a pathological diagnosis of WHO/ISUP grading after ne phrectomy , with some selection bias. Funding for this study: This work was supported in part by the National Natural Science Foundation of China under Grant Nos . 81801809, 82371917, 81830052, 81971691, 12126610, 62371476; in part by the Basic and Applied Basic Research Foundation of Guangdong Province und er Grant Nos. 2020A1515010572, and in part by the Zhuhai Basic an d Applied Basic Research Foundation under Grant Nos. ZH220170032000 01PWC. Ethics committee - additional information: This study was approved by the institutional review boards of the Fifth Hospital o f Sun Yat- sen University. Author Disclosures: Tao Liu: Nothing to disclose Yaqin Zhang: Nothing to disclose Junlin Li: Nothing to disclose Jiayue Han: Nothing to disclose Prevalence of Venous Extension in Malignant Adrenal Neoplasia: Identification of a Novel Imaging Sign L. Melges, C. Torres, F. Chahud, D. F. Maia, L. Col li, J. Elias, C. A. Fernandes Molina, M. Castro, *V. F. Muglia*; Ribeirao Preto/BR ([email protected]) Purpose or Learning Objective: Adrenal vein involvement is a common feature of adrenocortical carcinomas, but its preva lence in metastatic adrenal lesions remains unknown. Our goal was to assess the prevalence of adrenal vein involvement in primary and metastatic adrenal lesions and to determine if morphological changes in tumor shape precede venous extension.

Methods

or Background: This retrospective, single-center observational study evaluated 102 patients: 28 adrenal cortical c arcinoma (ACC) patients; and 74 non-ACC cancer patients that presented adren al metastasis (82 metastatic adrenal lesions). Two readers reviewed c ross-sectional imaging to assess tumor size, laterality, venous invasion, and the presence of the "edge sign." Surgical and histopathological confirmation was the reference standard for ACCs, while for metastases, sequential imaging or PET-CT results showing hypermetabolism were used when histopathology was u navailable.

Results

or Findings: Of the 28 ACC patients, 82.1% were female, with balanced laterality. Metastases primarily originate d from the lung (24.4%), colorectal (13.4%), and breast (12.2%) cancers and had a left-side dominance (61.7%). Venous extension was present in 14.6% of m etastases and 21.4% of ACCs. The "edge sign" was more frequently observed in metastatic lesions (26.8%) than in ACCs (17.8%). Interobserver agreeme nt was almost perfect for venous extension (κ = 0.9256) and substantial for the edge sign ( κ = 0.7844).

Conclusion

Venous extension was less prevalent in metastatic a drenal lesions compared to ACCs. The edge sign may precede venous extension, especially in metastatic cases. Although these find ings could impact clinical evaluation, prospective multicenter studies are nee ded to confirm the clinical significance of the edge sign.

Limitations

Retrospective, single-center study. Funding for this study: FAEPA - Foundation for the development of learning, assistance and research of Ribeirao Preto School of Medicine Hospital. Ethics committee - additional information: The study was approved by the Ethics and Research Committee of the Clinical Hospi tal at Ribeirao Preto Medical School under number CAAE 78221024.2.0000.54 40, with a waiver from the Informed Consent Form Author Disclosures: Jorge Elias: Nothing to disclose Margaret Castro: Nothing to disclose David Freire Maia: Nothing to disclose Fernando Chahud: Nothing to disclose Lais Melges: Nothing to disclose Carlos Augusto Fernandes Molina: Nothing to disclos e Leandro Colli: Grant Recipient: Bristol Myers Squib b - Immunotherapy for Renal Cell Carcinnoma Valdair Francisco Muglia: Nothing to disclose Cecilia Torres: Nothing to disclose IRM K01 study: Diagnostic Value of Multiparametric MRI for Small Solid Renal Tumors *E. Jambon*¹, N. Grenier¹, C. Marcelin¹, A. Crombé¹ , G. Margue¹, J-C. Bernhard¹, F. H. Cornelis²; ¹Bordeaux/FR, ²New York, NY/US ([email protected]) Purpose or Learning Objective: Small renal tumors are increasingly being detected incidentally, posing diagnostic challenges . Up to 23% of small renal tumors result in non-contributive biopsies. This st udy evaluates the diagnostic value of multiparametric MRI (mpMRI) in the charact erization of small solid renal tumors. The aim was to assess the diagnostic accuracy of mpMRI in differentiating malignant from benign small solid r enal tumors in patients with suspected malignancy but no evident signs of metast asis.

Methods

or Background: This is a prospective multicentric French study. A cohort of 387 patients in 17 centers with non-hered itary, solid renal masses between 1.5 and 4 cm in diameter was enrolled betwe en November 2018 and May 2022. MRI protocols included T1w, T2w, diffusio n-weighted imaging, and dynamic contrast-enhanced sequences. Radiologists p erformed blinded readings with a centralized review in case of disco rdance. The primary endpoint is the negative predictive value (NPV) of a dichotomized Likert scale score, targeting a 98% NPV.

Results

or Findings: The study found a 45% NPV for mpMRI, falling short of the expected 98% due to difficulties in distinguish ing clear cell renal cell carcinoma (ccRCC) from oncocytomas, which constitut ed 80% of the benign tumors in the cohort. Despite this, mpMRI influence d clinical management decisions, increasing "surveillance without biopsy" by 25% and reducing biopsies by 42%. However, it also led to a 25% incr ease in partial/total nephrectomies.

Conclusion

While mpMRI showed limited ability to accurately di stinguish certain benign lesions from malignant tumors under predefined criteria, it proved effective in identifying malignant cases. Th is led to a shift in clinical management favoring surgical interventions over bio psies. The study highlights the need for more objective imaging criteria and fu rther research into quantitative measures and radiomic analysis for bet ter tumor characterization.

Limitations

Predefined criteria Funding for this study: No funding Ethics committee - additional information: PHRC-K Approval from the French Ethics Committee (CPP) Author Disclosures: Jean-Christophe Bernhard: Nothing to disclose Gaelle Margue: Nothing to disclose François H Cornelis: Nothing to disclose Eva Jambon: Nothing to disclose Amandine Crombé: Nothing to disclose Clément Marcelin: Nothing to disclose Nicolas Grenier: Nothing to disclose ADC measurement may improve the diagnostic performa nce of bi- parametric bladder MRI in predicting detrusor muscl e invasion of bladder cancer *M. N. Tasdemir*, U. Eryürük, S. Aslan; Giresun/TR ([email protected]) Purpose or Learning Objective: To assess the diagnostic performance of a modified biparametric VIRADS (mbp-VIRADS), derived from a combination of ADC measurements and biparametric MRI (bp-MRI), in predicting detrusor muscle invasion in bladder cancer (BC).

Methods

or Background: Patients with histopathologically confirmed BC between June 2020 and May 2024 were analyzed retros pectively. Two image sets, biparametric MRI (set 1) and multiparametric (mp) MRI (set 2), were formed. Tumors were categorized using both the bp-V IRADS and mp-VIRADS systems. The optimal ADC value to differentiate mus cle-invasive bladder cancer (MIBC) from non-muscle-invasive bladder canc er (NMIBC) was determined using a receiver operating characteristi c (ROC) curve. To assess the mbp-VIRADS category, for cases with a bp-VIRADS score of 2-4, scores were upgraded for those below the ADC cut-off value and downgraded for those above it.

Results

or Findings: A total of 182 patients with BC met the study crite ria. Of these patients, 146 had NMIBC and 36 had MIBC. Comp aring VIRADS categories with MIBC detection, AUC of the ROC anal ysis was 0.896, 0.940, and 0.941 for the bp-MRI, mbp-MRI, and mp-MRI proto cols, respectively. The sensitivity for bp-VIRADS, mbp-VIRADS, and mp-VIRAD S scores (with a cutoff ≥4) were 78%, 88%, and 90%, specificity were 91%, 95 %, and 93%; and overall accuracy were 88%, 92%, and 93%, respective ly. Using bp-MRI, there were 12 false-positive and 9 false-negative cases f or predicting muscle invasion. With mp-MRI, false positives decreased to 9, and false negatives to 4. When using mbp-MRI, false positives further decr eased to 6, with 5 false negatives. Thursday Abstract-based Programme 108

Conclusion

By combining ADC measurements with bp- MRI features , the diagnostic performance of bp-MRI in predicting musc le invasion of bladder cancer can be significantly improved, approaching t hat of mp- MRI, while reducing the false-positive and false-negative rate s.

Limitations

This was a retrospective study, Funding for this study: None Ethics committee - additional information: The study was approved by the local ethics commitee. Author Disclosures: Uluhan Eryürük: Nothing to disclose Merve Nur Tasdemir: Nothing to disclose Serdar Aslan: Nothing to disclose Evaluating VI-RADS Score Performance in the Post-TU RBT setting: Exploring the Need for Modification *A. Dehghanpour*, M. Pecoraro, L. Laschena, M. Bicc hetti, C. Catalano, V. Panebianco; Rome/IT ([email protected]) Purpose or Learning Objective: The aim of this study was to assess the diagnostic accuracy of the VI-RADS score and its in dividual MRI categories, structural category (T2W), diffusion category (DWI) , and contrast-enhanced category (DCE), in patients who underwent diagnosti c transurethral resection of bladder tumor (TURBT). Additionally, we correlat ed the diagnostic accuracy of VI-RADS and its individual sequences with the ti me interval between TURBT and MRI, to suggest the optimal timing for MRI afte r TURBT. We also calculated the inter-reader agreement in scoring VI -RADS after TURBT.

Methods

or Background: This retrospective single-center study included 150 patients who underwent mpMRI after TURBT at varying intervals. Four experienced readers in bladder MRI, independently a nd blinded to clinicopathological information, evaluated the scan s, providing both VI-RADS scores and local staging. Each evaluation was perfo rmed twice: once with DWI as the dominant sequence and once with DCE as the d ominant sequence. The only exclusion criterion was prior systemic therapy . Histopathological results from therapeutic TURBT or radical cystectomy were u sed as the reference standard.

Results

or Findings: The AUC for VI-RADS in detecting muscle-invasive bladder cancer was 0.88 (95% CI 0.84-0.92) for the most experienced reader. On a per-sequence analysis, DWI showed the highest AUC (0.83 [95% CI 0.78-0.87]), followed by DCE (0.68 [95% CI 0.63-0.7 4]). The diagnostic accuracy of VI-RADS improved when the time between TURBT and MRI exceeded 2 weeks and became optimal after 4 weeks, regardless of whether DWI or DCE was the dominant sequence.

Conclusion

The earliest acceptable timing for MRI after TURBT is at least 2 weeks, with the optimal timing being after 4 weeks. In scoring VI-RADS after TURBT, DWI should be considered the dominant sequen ce, due to its high sensitivity and specificity.

Limitations

Retrospective design and readers being from the sam e center. Funding for this study: None. Ethics committee - additional information: Institutional Ethical Committee. Author Disclosures: Marco Bicchetti: Nothing to disclose Valeria Panebianco: Nothing to disclose Ailin Dehghanpour: Nothing to disclose Martina Pecoraro: Nothing to disclose Ludovica Laschena: Nothing to disclose Carlo Catalano: Nothing to disclose Uncertainty-Aware Interactive Deep Learning System for Predicting Pathological T3a Upstaging in Renal Cell Carcinoma Using CT Images *E. Yuan*¹, Q. Zhou², Y. Chen¹, C. He¹, L. Ye¹, J. Yao¹, B. Song¹; ¹Chengdu/CN, ²Shanghai/CN Purpose or Learning Objective: To develop and validate a deep learning system for predicting pathological T3a upstaging in renal cell carcinoma while integrating prediction uncertainty to facilitate mo re reliable clinical decision- making.

Methods

or Background: In this retrospective study, we collected pathology - confirmed RCC patients who underwent surgical resec tion from three tertiary academic medical centers. The data in first center were split into training and testing datasets. Three DenseNet-121 models were tr ained to predict the overall T3a invasion, the inner invasion, and the o uter invasion. The uncertainty was quantified by ensemble-based and vo ting-based methods. For uncertain cases, manual interpretation was performe d to obtain the final prediction. The performance of the pure model and u ncertainty-aware interactive system were evaluated and compared on t he testing dataset and two external datasets using area under the ROC curv e (AUC).

Results

or Findings: The data of 1329 patients (975:235:119) were collec ted and analyzed. The DL system performed worse in the uncertain group compared to the certain group of testing dataset (A UC 0.73 (95% CI: 0.61, 0.85) vs 0.81 (95% CI: 0.71, 0.90)), the external_1 dataset (AUC 0.50 (95% CI: 0.33, 0.68) vs 0.95 (95% CI: 0.90, 0.99)), and the external_2 dataset (AUC 0.82 (95% CI: 0.62, 1.00) vs 0.94 (95% CI: 0.89, 0. 99)). The net reclassification index for the DL system were 0.11, 0.18, and 0.13 i n testing, external_1, and external_2 datasets.

Conclusion

The uncertainty-aware interactive deep learning sys tem effectively predicts pathological T3a upstaging in renal cell carcinoma, with manual interpretation improving performance in unce rtain cases. This approach enhances diagnostic reliability, demonstra ting potential for improved clinical decision-making across multiple datasets.

Limitations

The sample size in external validation datasets wer e limited. Funding for this study: None Ethics committee - additional information: The written consent was waived for the retrospective design. Author Disclosures: Jin Yao: Nothing to disclose Qing Zhou: Nothing to disclose Chunlei He: Nothing to disclose Lei Ye: Nothing to disclose Enyu Yuan: Nothing to disclose Bin Song: Nothing to disclose Yuntian Chen: Nothing to disclose Enhancing Diagnostic Accuracy in Renal Tumor Identi fication: Impact of Structured Training on the Clear Cell Likelihood Sc ore *M. Cosenza*, G. Brembilla, G. Imperiale, A. Larche r, U. Capitanio, F. Montorsi, F. De Cobelli; Milan/IT ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the improvement in diagnostic performance of radiologis ts in identifying clear cell and papillary renal tumors using the Clear Cell Lik elihood Score (CCLS) before and after a structured training program.

Methods

or Background: This monocentric study analyzed 60 MRI scans, including 28 cases of clear cell carcinoma and 16 c ases of papillary carcinoma, all confirmed by histopathological examination. Fiv e radiologists evaluated the scans twice: the first assessment was based solely on their prior knowledge, using a cutoff of ≥4 for likelihood of clear cell carcinoma; the secon d assessment followed a training session that introdu ced the CCLS, where a score of ≥4 was assigned for clear cell tumors and ≤1 for papillary tumors. A washout period of 4 weeks was implemented between a ssessments.

Results

or Findings: Clear cell carcinoma overall sensitivity improved f rom 56% (95%CI: 47-64) pre-training to 86% (95%CI: 79-9 1) post-training, with a corresponding AUC for the ROC curve enhancing from 0.81 (95%CI: 0.76- 0.86) to 0.85 (95%CI:0.81-0.90). Papillary carcinom a sensitivity increased from 52% (95%CI:41-64) pre-training to 65% (95% CI: 54-7 5) post-training, with specificity rising from 90% (95%CI:85-93) to 95% (9 5%CI:91-97). The AUC for the ROC curve for papillary carcinoma rose signific antly from 0.79 (95%CI:73- 85) to 0.89 (95%CI: 85-94). Additionally, the agree ment improved for clear cell tumors, with a K of Conger increasing from 0.293 (9 5%CI: 0.181-0.406) to 0.594 (95%CI:0.469-0.718), and for papillary tumors , from 0.360 (95%CI:0.211-0.51) to 0.489 (95%CI:0.35-0.628).

Conclusion

Structured training and the application of the CCLS significantly enhance the diagnostic accuracy of radiologists in identifying clear cell and papillary renal tumors on MRI, underscoring the imp ortance of targeted education in improving radiological interpretations .

Limitations

Limitations include monocentric design, small sampl e size, limited radiologist cohort, and no longitudinal follow-up, impacting generalizability and sustainability. Funding for this study: None Ethics committee - additional information: IRB approved Author Disclosures: Francesco Montorsi: Nothing to disclose Michele Cosenza: Nothing to disclose Giulio Imperiale: Nothing to disclose Alessandro Larcher: Nothing to disclose Francesco De Cobelli: Nothing to disclose Giorgio Brembilla: Nothing to disclose Umberto Capitanio: Nothing to disclose Differentiating solid from friable tumor thrombus i n renal cell carcinoma using MRI ADC volumetric analysis P. Kowal¹, *Y. Mytsyk*², K. Ratajczyk¹, W. Bursiewi cz¹, M. Trzciniecki¹, K. Marek-Bukowiec¹, J. Rogala¹; ¹Wrocław/PL, ²Lviv/ UA ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the utility of first-order radiomic features derived from MRI appa rent diffusion coefficient (ADC) maps using volumetric analysis in distinguish ing solid from friable thrombus in patients with renal cell carcinoma (RCC ). Thursday Abstract-based Programme 109

Methods

or Background: A cohort of 27 patients with conventional histologi c subtype of RCC and tumor thrombus in the renal vein or inferior vena cava (IVC) was included. All patients underwent surgical intervention, comprising nephrectomy and thrombectomy, and received preopera tive abdominal MRI with diffusion-weighted imaging sequences at b-valu es of 50, 200, 800 s/mm². The ADC map was used for volumetric analysis, calcu lating various radiomic first-order features across the thrombus volume, in cluding ADC mean, median, range, 10th percentile, 90th percentile, interquart ile range, entropy, kurtosis, skewness, uniformity, and variance. Tumor thrombi w ere histologically classified as solid or friable, and associations be tween the radiomic features and thrombus consistency were analyzed.

Results

or Findings: Solid and friable tumor thrombi were identified in 51.9% and 48.1% of patients, respectively. Inverse associ ation noted between RCC thrombus cellularity and skewness (r=-0.799, p<0.00 1). No significant differences were observed in the mean values of ran ge, 90th percentile, interquartile range, kurtosis, uniformity, and vari ance between groups. For distinguishing solid from friable thrombus, the ADC mean, median, and entropy showed equal sensitivity (93%) and specificity (69% ), with entropy yielding the highest area under the curve (AUC) at 0.808. Skewne ss demonstrated a sensitivity of 86% and specificity of 92%, with an AUC of 0.931.

Conclusion

In RCC patients with tumor thrombus in the renal ve in or IVC, volumetric analysis of first-order radiomic feature s using ADC mapping facilitates accurate differentiation between solid and friable thrombus variants.

Limitations

The primary limitation of this study is that only c onventional histologic subtype of RCC was included in the analy sis. Funding for this study: No funding. Ethics committee - additional information: This study was approved by the Local Bioethical Committee in the Research and Deve lopment Center, Regional Specialist Hospital in Wroclaw (no. KB/12/ 2021). All procedures conducted followed the ethical guidelines set by th e institutional and/or national research committee, adhering to the principles outl ined in the 1964 Helsinki Declaration and its subsequent revisions, or equiva lent ethical standards. The written informed consent for enrolment in the study was signed by all patients. Author Disclosures: Yulian Mytsyk: Nothing to disclose Krzysztof Ratajczyk: Nothing to disclose Maciej Trzciniecki: Nothing to disclose Wiktor Bursiewicz: Nothing to disclose Joanna Rogala: Nothing to disclose Paweł Kowal: Nothing to disclose Karolina Marek-Bukowiec: Nothing to disclose Does the Bosniak 2019 classification really provide an objective assessment among the radiologists Ş. Evrimler, *E. Cigdem Karatayli*; Ankara/TR ([email protected]) Purpose or Learning Objective: The Bosniak Classification system was revised in 2019 to reduce subjectivity and prevent unnecessary nephron loss. In this plot study, we aimed to evaluate the inter- reader agreement in Bosniak classification among radiologists with different ex perience levels.

Methods

or Background: Out of 320 patients imaged between 01.2022 and 04.2024 in our hospital, 12 patients were randomly selected among those with pathology results, 22 patients were randomly select ed among those without pathology results, a total of 34 patients were sele cted. Eight residents, two radiology specialists, and one abdominal radiologis t evaluated abdominal CT and MRI scans of these patients. All participants w ere trained on the 2019 Bosniak classification prior to the assessment. The y assessed each criteria (septation presence, nodularity characteristics, co ntrast enhancement,etc.) separately and ultimately identified the Bosniak ty pe. The abdominal radiologist's classifications served as the referen ce for Kappa analysis. Participants also completed a questionnaire regardi ng difficulties in evaluating the classification and their use of objective value s.

Results

or Findings: Agreement with the reference increased with radiolo gy training duration (min kappa: 0.47, max kappa: 0.79 , p<0.01). Substantial agreement was noted between specialists and the ref erence (kappa: 0.68- 0.72, p<0.01). Agreement was better for non-measura ble parameters (e.g., presence of septation) compared to measurable ones (e.g., thickness of septation). Survey results indicated the most chall enging parameter was septation thickness, with 100% of participants unce rtain between categories 2 vs. 2F and 2F vs. 3. While 53% used objective measu rements during assessments, 81% relied on them as a guide rather t han exclusively.

Conclusion

Moderate-substantial agreement was found among radi ologists, improving with experience. Thickness of septations was particularly confusing, especially between certain categories.

Limitations

The study's small sample size suggests further rese arch with larger, diverse groups is needed. Funding for this study: No funding was provided for this study. Ethics committee - additional information: AEŞH-EK1-2024-54 numbered file, Ethical Committee, Etlik City Hospital, Ankar a Author Disclosures: Elif Cigdem Karatayli: Nothing to disclose Şehnaz Evrimler: Nothing to disclose 14:00-15:30 Research Stage 2 Research Presentation Session: Oncologic Imaging RPS 1016 Radiologic advances in hepatobiliary and pancreatic cancer Moderator S. De Vuysere; Leuven/BE ([email protected]) 3D Fractal Dimension Analysis of CT Imaging for Mic rovascular Invasion prediction in Hepatocellular Carcinoma B. Song, *F. Che*; Chengdu/CN ([email protected]) Purpose or Learning Objective: This study aimed to assess the potential role of 3-dimensional (3D) fractal dimension (FD) derive d from contrast-enhanced computed tomography (CT) images in predicting micro vascular invasion (MVI) in hepatocellular carcinoma (HCC) patients.

Methods

or Background: This retrospective study included 655 patients with surgically confirmed HCC from two medical centers ( training set: 406 patients; internal test set: 170 patients; external test set: 79 patients). Box-counting algorithms were employed to compute the 3D FD value s from portal venous phase images. Univariable and multivariable logisti c regression analyses were used to determine independent clinical and imaging predictors. Diagnostic performance of the model was assessed using receive r operating characteristic (ROC) curve analysis. Recurrence-fre e survival (RFS) and overall survival (OS) curves were generated via Kap lan-Meier method and compared using the log-rank test.

Results

or Findings: Patients with MVI-positive HCC exhibited significan tly higher FD values compared to those with MVI-negativ e HCC (P< 0.01). The combined model, integrating alpha-fetoprotein level , tumor size, number, and FD, demonstrated superior diagnostic performance fo r MVI prediction compared to the clinical model alone, with AUCs of 0.80 (95% CI: 0.75, 0.88) and 0.75 (95% CI: 0.67, 0.83) for the internal test set, and 0.83 (95% CI: 0.72, 0.92) and 0.74 (95% CI: 0.61, 0.85) for the externa l test set. Patients predicted to have high-risk MVI showed worse RFS and OS outco mes than those predicted to have low-risk MVI (all P<0.05).

Conclusion

The 3D FD value significantly differed between MVI- positive and MVI-negative HCC patients. Integration of FD into t he clinical model enhances MVI prediction accuracy and may identify patients a t high risk.

Limitations

Firstly, the retrospective nature of our study intr oduces inherent limitations. Secondly, our study only analyzed 3D f ractal features on the PVP phase. Funding for this study: None Ethics committee - additional information: Institutional Review Board approval was obtained by West China Hospital and He nan Provincial People’s Hospital Author Disclosures: Bin Song: Nothing to disclose Feng Che: Nothing to disclose The impact of pretreatment body composition on conv ersion surgery feasibility and survival in pancreatic cancer patie nts undergoing neoadjuvant therapy *H. Y. Chen*, B-B. Chen; Taipei/TW Purpose or Learning Objective: This study aims to evaluate the impact of pretreatment body composition, assessed via CT or M RI, on the likelihood of achieving conversion surgery and overall survival ( OS) in pancreatic cancer patients undergoing neoadjuvant therapy (NAT).

Methods

or Background: A retrospective analysis was conducted on 154 patients with pancreatic cancer receiving NAT befor e planned conversion surgery between May, 2018 and February, 2024. Clini cal and laboratory data, including carcinoembryonic antigen (CEA) and carboh ydrate antigen 19-9 (CA19-9), were collected. Body composition paramete rs such as skeletal Thursday Abstract-based Programme 110 muscle volume (SM), skeletal muscle index, subcutan eous adipose tissue (SAT), and visceral adipose tissue (VAT) at the L3 vertebral level were automatically calculated using the TotalSegmentator software. Independent predictors of conversion surgery were identified us ing multivariable logistic regression, and Kaplan-Meier curves and Cox regress ion models assessed the impact of these variables on OS.

Results

or Findings: Patients who underwent conversion surgery group had significantly smaller tumor sizes (P=0.033), lower rates of vascular invasion (P=0.042), lower log-transformed CA19-9 levels (P=0 .002), higher SM (P=0.049), and higher VAT (P=0.048). Multivariate a nalysis revealed that log- transformed CA19-9 (P=0.002) and higher SM (P=0.049 ) were independent predictors of conversion surgery. Conversion surger y (P<0.001) and higher SAT (P=0.001) were associated with better OS. SAT w as identified as an independent predictor of OS (P=0.04) after adjustin g for age, conversion surgery status, and TNM stage in the multivariate m odel.

Conclusion

Pretreatment body composition, as evaluated through imaging, may be associated with the feasibility of conversio n surgery and overall survival in patients with pancreatic cancer receivi ng NAT. Specifically, higher skeletal muscle volume and SAT appear to be favorab le prognostic factors, highlighting the importance of body composition in clinical decision-making and patient outcomes.

Limitations

The limitations of the study are single center stud y and relatively small patient group. Funding for this study: No funding was received for this study Ethics committee - additional information: Research Ethics Committee of National Taiwan University Hospital Author Disclosures: Bang-Bin Chen: Nothing to disclose Hsin Yu Chen: Nothing to disclose Iodine quantification and LI-RADS classification of hepatocellular carcinoma lesions in contrast-enhanced spectral CT studies *A. Celestino*, P. Marra, A. Barbaro, C. Gargiulo, R. Muglia, G. Muscogiuri, P. A. Bonaffini, S. Sironi; Bergamo/IT ([email protected]) Purpose or Learning Objective: The LI-RADS classification is universally employed for hepatocellular carcinoma (HCC) risk st ratification of liver nodules in cirrhosis, but it relies on a qualitative evalua tion. This study aims to investigate the potential role of material density (MD) parameters in the iodine maps of Spectral Computed Tomography (SCT), to disc riminate between LI- RADS categories in cirrhotic patients, therefore in creasing the radiologists’ confidence in LI-RADS class allocation.

Methods

or Background: Dual-energy SCT scans of cirrhotic patients with nodules between March 2022 and September 2023 were retrospectively reviewed. All the images were reviewed by trained r adiologists to classify nodules as LI-RADS 3, 4 or 5 by consensus. MD maps were generated in the hepatic arterial phase (HAP), portal venous (PVP) a nd equilibrium phase (EP). Iodine concentration density (ICD) of nodules (ICDn odule) and non-nodular liver parenchyma (ICDliver) were measured to calcul ate lesion-to-non-nodular liver ICD ratio (LNR), their differences (ΔICD) and ratios (rLNR). Results were correlated with LI-RADS categories.

Results

or Findings: 69 patients were included and 79 DECT exams were examined. 197 nodules (size 24.67 ± 23.11 mm, mean ± SD) were categorised under different LI-RADS classes, as follows: 44 as LI-RADS 3 (22.3%), 14 as LI-RADS 4 (7.1%), and 139 as LI-RADS 5 (70.6%). Art erial LNR, arterial ICDnodule, ΔICD and rLNR between HAP and PVP could discriminate between LI-RADS 3 and LI-RADS 4+5 nodules (p < 0.00 1). All the calculated MD parameters showed similar, or slightly higher di agnostic accuracy rates (all AUCs = 70-73%) compared to those previously reporte d by non-spectral CT (up to 70%).

Conclusion

MD parameters of liver nodules measured in SCT scan s are viable diagnostic tools that may increase the radio logist’s confidence in LI- RADS class allocation in cirrhotic patients.

Limitations

The lack of an adequate number of LI-RADS 4 nodules Funding for this study: None Ethics committee - additional information: None Author Disclosures: Antonio Celestino: Nothing to disclose Sandro Sironi: Nothing to disclose Paolo Marra: Nothing to disclose Pietro Andrea Bonaffini: Nothing to disclose Alessandro Barbaro: Nothing to disclose Riccardo Muglia: Nothing to disclose Carlotta Gargiulo: Nothing to disclose Giuseppe Muscogiuri: Nothing to disclose Simple Cystic Lesions of the Pancreas: Image Qualit y and Diagnostic Accuracy of Photon-Counting Detector Computed Tomog raphy *S. Rau*¹, T. Stein¹, A. Rau¹, C. Wilpert¹, F. B. P allasch¹, B. Bogner¹, S. Faby², J. Weiß¹; ¹Freiburg/DE, ²Forchheim/DE ([email protected]) Purpose or Learning Objective: To evaluate image quality and diagnostic accuracy of photon-counting detector (PCD) computed tomography (CT) for the detection of PCLs compared to conventional ener gy-integrating detector (EID) CT with MRI serving as reference standard.

Methods

or Background: In this prospective study, we included consecutive patients who underwent clinically indicated contras t-enhanced PCD-CT of the abdomen and for whom an additional abdominal EID-CT was available. Multiparametric MRI served as the reference standar d. CT images were assessed for the presence of PCLs by three radiolog ists independently in a blinded reading. In addition, image quality, lesion conspicuity, and diagnostic confidence were rated on a 5-point Likert scale (5= excellent). The coefficient- of-variation (CV) and the density difference (in Ho unsfield units [HU]) between PCLs and visually normal pancreatic parenchyma were calculated as quantitative imaging measures. Radiation dose was a ssessed using CTDIvol [mGy].

Results

or Findings: Among 106 included patients (age 62.7±12.6 years; 4 5 [42.5%] male), 46 had MRI-confirmed cystic lesions (mean size 8.7±7.4mm; range: 2-45 mm). Diagnostic accuracy for PCLs was s ignificantly higher for PCD-CT vs. EID-CT (area under the curve: 0.81 vs. 0 .74; p=0.002; sensitivity: 76.8% vs. 59.4%; specificity 84.4% vs. 88.3%, respe ctively). Image quality, lesion conspicuity and diagnostic confidence were r ated superior for PCD-CT vs. EID-CT (all p<0.001). Quantitative analyses rev ealed a significantly lower CV (0.19 vs. 0.24; p=0.002) and a higher density di fference (94.1 HU vs. 76.6 HU p<0.001) between PCLs and visually normal pancre atic parenchyma at lower radiation doses (7.13 vs. 8.68 mGy; p<0.001) for PCD-CT vs. EID-CT.

Conclusion

PCD-CT provided significantly higher diagnostic acc uracy and superior image quality for the detection of PCLs co mpared to conventional EID-CT at lower radiation dose.

Limitations

No long-term follow-up and/or histopathological cor relation of the detected PCLs were omitted. Funding for this study: None Ethics committee - additional information: The local Institutional Review Board (Ethics Committee of the University Medical C enter Freiburg, case number 21-2469) approved this prospective study and written informed consent was obtained from all patients prior to stu dy inclusion. Author Disclosures: Caroline Wilpert: Nothing to disclose Jakob Weiß: Nothing to disclose Alexander Rau: Nothing to disclose Balazs Bogner: Nothing to disclose Fabian Bernhard Pallasch: Nothing to disclose Sebastian Faby: Employee: Siemens Healthineers Thomas Stein: Nothing to disclose Stephan Rau: Nothing to disclose Prognostic Value of RECIST, mRECIST, and LI-RADS TR A Early Response in Predicting Survival in Hepatocellular C arcinoma Treated with Selective Internal Radiation Therapy M. Dupuis, A. Dupont, S. Pizza, V. Vilgrain, A. Ban do Delaunay, R. Lebtahi, M. Bouattour, M. Ronot, *J. Gregory*; Clichy/FR ([email protected]) Purpose or Learning Objective: This study evaluates the prognostic value of tumor response at three months on CT, assessed by R ECIST, mRECIST, and LI-RADS Treatment Response Algorithm (LR-TRA) in pa tients with hepatocellular carcinoma (HCC) treated with selecti ve internal radiation therapy (SIRT).

Methods

or Background: A retrospective analysis was conducted on 102 HCC patients treated with SIRT between 2018 and 202 0. RECIST, mRECIST, and LR-TRA were assessed at 3 months post-SIRT. Ove rall survival (OS) and progression-free survival (PFS) were assessed using Kaplan-Meier analysis and Cox proportional hazards models.

Results

or Findings: Median age was 71 years, most patients (90%) had advanced-stage tumors (BCLC-C). After a median foll ow-up of 32.0 months (95% CI: 16.8-60.9), 60/102 patients died (59%), an d 90/102 patients showed tumor progression (88%). Median OS was 20.4 months (95% CI: 15.4-33.0), and median PFS was 14.5 months (95% CI: 6.5-24.5); 1-year OS and PFS rates were 65.6% and 50.7%. Multivariable analysis revealed that early response according to RECIST 1.1 (HR 1.66, p=0.30), mRECIST (HR 1.40, p=0.215), and LR-TRA (HR 0.67, p=0.30) were not pre dictors of OS. Disease progression evaluated by RECIST (HR 2.55, p<0.001) and mRECIST (HR 2.53, p<0.001), bilirubin levels (HR 1.03, p<0.001) and prothrombin time (HR 0.98, p=0.005) were predictors of OS. For PFS, neit her RECIST nor mRECIST response, disease progression, nor LR-TRA viability were predictors. Thursday Abstract-based Programme 111

Conclusion

In this advanced-stage HCC population, early respon se assessed by RECIST, mRECIST, and LR-TRA criteria di d not predict OS or PFS after SIRT. However, early disease progression and liver function indicators were prognostic factors for OS.

Limitations

Several limitations exist in this single-center, re trospective study, including a small sample size, which may reduce the generalizability of the findings. Additionally, tumor heterogeneity, a key prognostic-factor for poor treatment response and shorter PFS, was not conside red. Funding for this study: None Ethics committee - additional information: This single-center retrospective clinical study was approved by the local Institutio nal Review Board (IRB 00006477), and informed consent was waived due to i ts retrospective nature. Author Disclosures: Axelle Dupont: Nothing to disclose Michel Dupuis: Nothing to disclose Mohamed Bouattour: Nothing to disclose Jules Gregory: Nothing to disclose Maxime Ronot: Nothing to disclose Silvia Pizza: Nothing to disclose Rachida Lebtahi: Nothing to disclose Valérie Vilgrain: Nothing to disclose Aurélie Bando Delaunay: Nothing to disclose Contrast-enhanced Ultrasound using Perfluorobutane for Diagnosing Small HCC (≤20mm) in High-risk Patients: Comparison with MRI LI -RADS Version 2018 *J. Zhou*, Y. Li, L. Lii; Guangzhou/CN ([email protected]) Purpose or Learning Objective: The sensitivity of Contrast-enhanced ultrasound (CEUS) for diagnosing hepatocellular car cinoma (HCC) is lower than MRI, especially in small liver nodules measuri ng 20mm or less. This study aimed to compare the diagnostic performance between CEUS with perfluorobutane (P-CEUS) and MRI Liver Imaging Repo rting and Data System (LI-RADS) version 2018 (v2018) for small liver nodu les in high-risk patients.

Methods

or Background: This multi-center retrospective study included high - risk patients with newly detected liver nodules mea suring 20mm or less from March 2020 to November 2023. Patients underwent CEU S with perfluorobutance (P-CEUS) and concurrent MRI at int ervals of no more than 1 month. The reference standard was pathological conf irmation or 24-month imaging follow-up (only for benign lesions). The di agnostic performance of CEUS LI-RADS v2017 for P-CEUS, modified criteria fo r P-CEUS, and MRI LI- RADS v2018 was calculated and compared. For the mod ified criteria for P- CEUS, LR-4 observations measuring 10 mm or larger w ith nonrim APHE and no washout were reclassified as LR-5 if aslo showin g a Kupffer defect; LR-M observations measuring 10 mm or larger with nonrim APHE and early washout were reclassified as LR-5 if aslo showing a mild Ku pffer defect.

Results

or Findings: A total of 367 participants (301 males; mean age, 5 4±11 years) with 401 observations (mean diameter, 15±4 m m) were included. Using LR-5 for HCC daignosis, MRI LI-RADS v2018 had highe r sensitivity (73% vs 57%, P0.05) in sensitivity (70% vs 73%) and specificity (92% vs 90%).

Conclusion

CEUS using perfluorobutane with modified criteria d emonstrated a diagnostic performance comparable to MRI in HCC d iagnosis for small liver nodules (≤20mm).

Limitations

Retrospective study Funding for this study: This study was not supported by any funding. Ethics committee - additional information: Approved Author Disclosures: Lingling Lii: Nothing to disclose Jianhua Zhou: Nothing to disclose Yu Li: Nothing to disclose Evaluation of dual-layer spectral CT compared with conventional CT for the diagnosis of hepatocellular carcinoma *M. Youssef Francis*, T. Broussaud, M. Wagner, J. B enzimra, N. Brillat-Savarin, O. Lucidarme; Paris/FR ([email protected]) Purpose or Learning Objective: To investigate whether the use of dual-layer spectral CT (DLSCT) improves the diagnostic accurac y of hepatocellular carcinomas (HCCs) compared with conventional CT.

Methods

or Background: 127 patients were included in this retrospective study. All patients underwent multiphase DLSCT (IQo n, Philips Healthcare) for the initial diagnosis of HCC. Arterial phase hypere nhancement (APHE) and portal and delayed washout of each lesion were qual itatively assessed by two radiologists independently using three-point Likert scales and the lesion-to-liver contrast-to-noise ratio (LLCNR) was quantified usin g ROI in conventional (CIs) and 40keV virtual monoenergetic images (VMIs). For qualitative analysis, interobserver agreement was assessed using the kapp a statistic. For quantitative analysis, LLCNRs were compared using W ilcoxon and Scheirer- Ray-Hare tests.

Results

or Findings: 163 HCCs were independently analyzed. The APHE was considered significantly higher for both observ ers in VMIs compared to CIs: mean 2.36 vs 1.32 p<0.001. The washout was con sidered significantly better seen in VMIs compared to CIs in the delayed phase: mean 1.99 vs 1.33, p<0.001; but not in the portal phase: mean 1.74 vs 1.36, NS. Interobserver agreement was good for APHE and washout in both CIs and VMIs. The LLCNR was significantly higher in the arterial phas e (6.18vs2.05;p<0.001) and significantly lower in the delayed phase (-2.02 vs -1.10; p<0.001) in 40 Kev- VMIs compared to CIs. As in the subjective analysis , LLCNR was not significantly different in the portal phase (-1.89 vs -1.52;NS).

Conclusion

Multiphasic DLCT with 40 keV VMIs increases the vis ibility of both APHE and washout in the delayed phase of focal liver lesions compared to CIs, leading to the final diagnosis of a higher number of HCCs.

Limitations

This is a single-center, retrospective study. DLSCT was not compared with MRI, which is currently the gold stan dard in the detection of HCC. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was IRB approved (number CRM-2304-339). Author Disclosures: Olivier Lucidarme: Nothing to disclose Marie Youssef Francis: Nothing to disclose Thomas Broussaud: Nothing to disclose Mathilde Wagner: Nothing to disclose Nina Brillat-Savarin: Nothing to disclose Julie Benzimra: Nothing to disclose Bone Mineral Density as a Prognostic Indicator for Overall Survival in Hepatocellular Carcinoma Patients with Portal Venou s Tumor Thrombus L. Müller¹, R. Kloeckner², L. Heim¹, F. Stoehr¹, F. Hahn¹, T. Bäuerle¹, A. Weinmann¹, D. Pinto Dos Santos³, *A. Mähringer-K unz*¹; ¹Mainz/DE, ²Lübeck/DE, ³Frankfurt/DE ([email protected]) Purpose or Learning Objective: Low bone mineral density (BMD) has recently emerged as a risk factor in hepatocellular carcinoma (HCC). However, its role in patients with HCC complicated by portal vein tumor thrombosis (PVTT) is unclear. This study explores the potentia l of BMD as a prognostic indicator within this subgroup, which is characteri zed by an exceptionally poor prognosis.

Methods

or Background: This retrospective study included 462 patients with HCC and PVTT diagnosis at our tertiary care center between January 2005 and December 2020. BMD was measured by mean Hounsfi eld units (HUs) at the midvertebral core of the first lumbar vertebra in computed tomography using the established cut-off of 160 HU. Analysis w as performed at two points in time: initial HCC diagnosis and PVTT onset. We a nalyzed the impact of BMD on median overall survival (OS) and conducted multi variate analysis with established survival predictors.

Results

or Findings: Median BMD was 136 HU (IQR, 113–160 HU) at HCC diagnosis and 134 HU (IQR, 109–159 HU) at PVTT diag nosis. At initial HCC diagnosis, patients with BMD ≥ 160 HU had a median OS of 10.4 months, compared to 5.5 months in those with BMD < 160 HU ( p < 0.001). At PVTT diagnosis, those with higher BMD had a median OS of 8.5 months, versus 4.7 months in patients with lower BMD (p < 0.001). BMD remained an independent prognostic factor in multivariate analysis, alongsi de growth type and ALBI grade.

Conclusion

BMD serves as an independent prognostic marker for survival in patients with HCC and PVTT. Incorporating BMD into existing classification and scoring systems could enhance the accuracy of s urvival predictions and inform clinical decision-making processes.

Limitations

The primary limitation is the retrospective study d esign, necessitating validation of these findings in a pro spective framework. Funding for this study: None Ethics committee - additional information: The study was approved by the responsible ethical body: Ethics committee of the M edical Association of Rhineland Palatinate, Mainz, Germany (Permit number : 15913). Thursday Abstract-based Programme 112 Author Disclosures: Daniel Pinto Dos Santos: Other: Chair of the ESR eH ealth and Informatics Subcommittee Lorena Heim: Nothing to disclose Fabian Stoehr: Grant Recipient: Research grant, cli nician scientist program, Else Kröner-Fresenius-Foundation Aline Mähringer-Kunz: Nothing to disclose Felix Hahn: Nothing to disclose Roman Kloeckner: Speaker: Astra Zeneca, Boston Scie ntific, , BTG, EISAI, Guerbet, Ipsen, Siemens, and SIRTEX Other: Chair of the ESR Audit and Standards Subcommittee Advisory Board: Boston Scien tific, Bristol-Myers Squibb, Guerbet, MSD Sharp & Dohme, Roche, and SIRT EX Lukas Müller: Grant Recipient: Research grant, clin ician scientist program, Else Kröner-Fresenius-Foundation Tobias Bäuerle: Nothing to disclose Arndt Weinmann: Nothing to disclose Solid pancreatic neoplasms: bridging radiology and cytopathology for accurate diagnosis *D. J. A. D. C. E. Aragão*, S. Santos, J. Nobre, L. M. Cabral; Lisboa/PT ([email protected]) Purpose or Learning Objective: To retrospectively evaluate a cohort of patients with solid pancreatic lesions who underwen t endoscopic ultrasound- guided fine-needle aspiration (EUS-FNA). To review the radiologic features of common solid pancreatic neoplasms and correlate the m with cytopathological findings.

Methods

or Background: A retrospective review was performed at our institution, encompassing 100 EUS-FNA procedures du ring 18 consecutive months. Patients with solid pancreatic lesions or l esions with a solid component (n=48) were included in the analysis. Cli nical data, imaging studies, and cytopathology reports were collected. The most illustrative cases were selected to demonstrate the radiologic and cytopath ologic characteristics of these neoplasms.

Results

or Findings: The most frequently diagnosed solid pancreatic neoplasm was adenocarcinoma (median age: 70 years; 61% male; 71% located in the pancreatic head), typically presenti ng as a hypovascular mass. Neuroendocrine tumors, often hypervascular, were th e second most common neoplasm. Rare cases included one case of pancreati c lymphoma and one case of metastatic disease involving the pancreas.

Conclusion

As expected, pancreatic adenocarcinoma, particularl y in elderly males and localized in the pancreatic head, was the most commonly diagnosed neoplasm. EUS-FNA combined with cytopatho logic analysis, along with multimodality imaging, remains essential in th e accurate diagnosis, staging, and management of solid pancreatic neoplas ms. The integration of these diagnostic tools ensures a comprehensive mult idisciplinary approach, leading to more effective patient care and treatmen t planning.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study is educational. Author Disclosures: Sofia Santos: Nothing to disclose Luís Monteiro Cabral: Nothing to disclose João Nobre: Nothing to disclose Diogo José Afonso Da Cruz E Aragão: Nothing to disc lose Improving Diagnostic Confidence in Assessing Pancre atic Tissue: Prospective Evaluation of Mechanical Elastography *V. Koch*¹, M. Cimprich¹, L. D. Grünewald¹, C. Booz ¹, T. Vogl¹, O. Darwish², J. Gotta¹, S. Mahmoudi¹, R. Sinkus³; ¹Frankfurt/DE, ²London/UK, ³Paris/FR Purpose or Learning Objective: This study aimed to evaluate 2D/3D magnetic resonance imaging (MRE) utilizing the grav itational transducer concept compared to the current acoustic product so lution (2D-MRE Resoundant) to further characterize pancreatic carc inoma and its potential to provide imaging biomarkers for outcome prediction.

Methods

or Background: In this prospective study, 40 patients with confirm ed pancreatic cancer undergoing MRI of the upper abdom en were enrolled between June 2023 and September 2024. Furthermore, 15 healthy volunteers were included as a healthy reference standard. All participants underwent two examinations using a 40Hz mechanical vibration freq uency (Aera 1.5T, Siemens Healthineers, Germany): initially with the acoustic MRE (Resoundant, 2D-MRE, SE-EPI sequence, 11s breath hold [BH]), and subsequently with the gravitational MRE (2D-MRE and 3D-MRE, GRE sequence, TE=9.53ms [in- phase], and fractional motion encoding at 30mT/m, 1 4s BH). Two experienced readers independently conducted data analysis. Addi tionally, superimposed analytic plane waves with known properties at vario us amplitudes and temporal noise levels were employed to investigate biases in stiffness reconstruction (2D/3D) and suggest Quality Indices for 2D/3D.

Results

or Findings: Significant differences were observed in stiffness values, shear wave speed, and phase angle between healthy v olunteers and patients with pancreatic cancer using both MRE approaches (p <.05). Additionally, patients who showed clinical response to chemothera py exhibited differences in stiffness (p<.05). However, the Bland-Altman plo ts exhibited a notable bias, with 2D-MRE tending to overestimate stiffness value s. 3D-MRE provided several imaging biomarkers that correlated with dis ease progression and response to therapy. Proposed Quality Indices enabl ed the identification of pixels exhibiting a deviation exceeding 10% from ac tual stiffness values in 3D- MRE.

Conclusion

Gravitational MRE proves to be an accurate techniqu e for noninvasively characterizing pancreatic tissue. In particular, 3D MRE can provide pertinent functional imaging markers, advan cing functional abdominal imaging.

Limitations

Single-center study. Funding for this study: This study has been funded by the Doktor Robert Pfleger Foundation. Ethics committee - additional information: The institutional ethical review board approved this prospective study. Written info rmed consent was obtained from all participants. Author Disclosures: Christian Booz: Nothing to disclose Omar Darwish: Nothing to disclose Thomas Vogl: Nothing to disclose Vitali Koch: Nothing to disclose Scherwin Mahmoudi: Nothing to disclose Ralph Sinkus: Nothing to disclose Leon David Grünewald: Nothing to disclose Jennifer Gotta: Nothing to disclose Marina Cimprich: Nothing to disclose Imaging features, management and outcomes of Solita ry Necrotic Nodule of the Liver: a case series and review of literatur e L. Asmundo, L. Giaccardi, A. Soro, C. Buonomenna, R . Vigorito, F. G. Greco, A. Casale, *M. Vaiani*; Milan/IT ([email protected]) Purpose or Learning Objective: Solitary Necrotic Nodule of the Liver (SNNL) is a rare, benign hepatic lesion frequently misdiag nosed as malignant, leading to unnecessary invasive procedures. This study pres ents a case series of patients diagnosed with SNNL, focusing on their ima ging characteristics, clinical management and outcomes.

Methods

or Background: This retrospective case series analyzed data from patients diagnosed with SNNL at a tertiary referral center. Collected data included demographics, imaging studies, and follow- up outcomes. A radiologist with 20 years of experience reviewed all imaging st udies. The reference standard was histological examination or follow-up imaging

Results

or Findings: Among 13 patients (54% female; median age 54 years) , MRI was the preferred imaging modality, with 3 to 5 scans performed over a median follow-up of 25 months (range 22-70). SNNL t ypically presented as a necrotic core surrounded by a fibrotic capsule, oft en with a nodular, elongated, or C-shaped appearance that mimicked biliary dilata tion. T1-weighted images showed isointensity in 53.8% of cases, with hypoint ensity of the core in 84.6%. T2-weighted images revealed isointensity (38.5%) or hypointensity (61.5%) of the core. Nodules remain hypovascular after contras t media injection. Diffusion-weighted imaging displayed no restricted diffusion. During follow-up, most nodules (92.3%) showed a reduction in size, wi th increased calcification on CT (from 38.5% to 69.2%).

Conclusion

SNNL presents diagnostic challenges due to its rese mblance to aggressive hepatic lesions, particularly cholangioc arcinoma and metastases. However, careful interpretation of MRI and CT findi ngs can prevent unnecessary invasive procedures.

Limitations

The small sample size and single-center design limi t the generalizability of the findings. While only a few cases had histological confirmation (30%), long-term imaging follow-up sup ports the benign nature of SNNL, reducing the need for biopsies and surgeries. Funding for this study: none. The authors did not receive support from any organization Ethics committee - additional information: Local ethic committee approved the study protocol Author Disclosures: Francesca Gabriella Greco: Nothing to disclose Luca Giaccardi: Nothing to disclose Ciriaco Buonomenna: Nothing to disclose Raffaella Vigorito: Nothing to disclose Alberto Soro: Nothing to disclose Luigi Asmundo: Nothing to disclose Marta Vaiani: Nothing to disclose Alessandra Casale: Nothing to disclose Thursday Abstract-based Programme 113 14:00-15:30 Research Stage 3 Research Presentation Session: Interventional Radiology RPS 1009 Advances in musculoskeletal and soft tissue interventions Moderator M. Szmygin; Lublin/PL ([email protected]) Efficacy and safety of image-guided bone biopsies: Insights gained from the German Society for Interventional Radiology and Minimally Invasive Therapy (DeGIR) registry data 2018-2022 *S. Zensen*¹, F. Behr¹, M. Opitz¹, D. Bos¹, M. Holt kamp¹, L. Salhöfer¹, J. Haubold¹, M. Moche², B. M. Schaarschmidt¹; ¹Esse n/DE, ²Leipzig/DE ([email protected]) Purpose or Learning Objective: Image-guided bone biopsies are critical for the diagnosis of a wide range of bone lesions, yet there is limited large-scale data on the technical success, diagnostic yield, an d safety of these procedures. This study aims to evaluate the efficac y and safety of image- guided bone biopsies using data from the German Soc iety for Interventional Radiology and Minimally Invasive Therapy (DeGIR, De utsche Gesellschaft für Interventionelle Radiologie und minimal-invasive Th erapie) registry.

Methods

or Background: In this retrospective observational study, 17,397 bone biopsies from 214 centers between 2018 and 202 2 were analyzed. Technical success was defined as the visually succe ssful placement of the biopsy needle within the target lesion. Complicatio ns were classified according to the Society of Interventional Radiology (SIR) gu idelines.

Results

or Findings: About one-third of biopsies were performed as outpatient procedures (34%,5,924/17,397). Most biop sies were conducted under local anesthesia (86.6%,15,072/17,397). CT-gu idance was used in the majority (68.7%,11,952/17,397). The technical succe ss rate was 98.9% (17,201/17,397), with histological representativene ss of 93.2% (10,316/11,073). Outpatient biopsies had a slightly higher technical success rate (99.32%,5,884/5,924) than inpatient biopsies ( 98.63%,11,316/11,473), though histological representativeness was lower (9 1.06%,1,284/1,410 vs. 93.48%,9,031/9,661, p=0.001). The overall complicat ion rate was low at 0.62% (108/17,397), with major complications in 23.1% (25 /108) of cases. Patients with abnormal coagulation parameters had higher com plication rates. Sub- /solid lesions had higher histological representati veness compared to necrotic- cystic lesions (94.01%,7,846/8,346 vs. 90.32%,1,558 /1,725, p<0.0001).

Conclusion

Image-guided bone biopsies are highly effective and safe, even in outpatient settings. The data support their cont inued use as a minimally invasive diagnostic tool, with low complication rat es and high diagnostic accuracy.

Limitations

This study is limited by the nature of registry dat a, which is subject to reporting biases and lacks external vali dation. Incomplete data from some centers also restricted the scope of certain a nalyses. Funding for this study: This research received no specific grant from any funding agency in the public, commercial, or not-fo r-profit sectors. Ethics committee - additional information: Ethical approval for this retrospective registry study was granted by the eth ics committee of the University of Duisburg-Essen, Germany (22-10893-BO) . Author Disclosures: Denise Bos: Nothing to disclose Benedikt Michael Schaarschmidt: Nothing to disclose Johannes Haubold: Nothing to disclose Sebastian Zensen: Nothing to disclose Luca Salhöfer: Nothing to disclose Florian Behr: Nothing to disclose Michael Moche: Nothing to disclose Marcel Opitz: Nothing to disclose Mathias Holtkamp: Nothing to disclose Early results of a prospective study on palliative arterial embolization for bone metastases: the EMBONEMET study *N. Papalexis*, G. Peta, S. Quarchioni, M. Di Carlo , L. Campanacci, M. Carta, M. Miceli, G. Facchini; Bologna/IT ([email protected]) Purpose or Learning Objective: To evaluate the clinical and radiological effect of arterial embolization using N-butyl cyano acrylate (NBCA) as palliation for bone metastases.

Methods

or Background: This study analyzes the early results of a prospective study “EMBONEMET”, designed to prospect ively evaluate the safety and efficacy of palliative arterial emboliza tion for bone metastases. Thirty-three patients were enrolled from June 2023 to June 2024. The primary goal was pain control, measured in VAS score at 3, 6 and 12 months follow-up. The secondary goal was the size reduction of the le sion. Data on technical success were also collected. All embolization-relat ed complications were evaluated according to the CIRSE classification sys tem for complications.

Results

or Findings: The average follow-up was 9 months (range 2 to 12 months). Baseline VAS scores averaged 5,3 (SD 3,19) , decreasing to 3.56 (SD 3,1) at three months. Progressive reduction was obs erved at 6 and 12 months with scores of 3,52 (SD 3,33), and 2,8 (SD 2,7) res pectively. Metastatic tumor size was reduced from a mean of 196.8 cm3 (range 39 .5 to 486.4 cm3) pre- embolization to a mean of 179.4 cm3 (range 38.2 to 458.6 cm3) at the 6-month follow-up (p<0.05). Twenty-nine patients experience d post embolization-related pain, that resolved within 15 days. Three patients experienced sensory loss of the lower leg, paresthesia and pain in the femoral catheter access point.

Conclusion

The preliminary results are promising, suggesting t hat arterial embolization could be a safe and effective tool for pain management and disease control in metastatic bone disease.

Limitations

Small sample size, lack of control group Funding for this study: None Ethics committee - additional information: Prospective study approved by the local ethics committee of Emilia Romagna, Italy . Author Disclosures: Marco Miceli: Nothing to disclose Michela Carta: Nothing to disclose Laura Campanacci: Nothing to disclose Giancarlo Facchini: Nothing to disclose Giuliano Peta: Nothing to disclose Simone Quarchioni: Nothing to disclose Maddalena Di Carlo: Nothing to disclose Nicolas Papalexis: Nothing to disclose Percutaneous cryoablation of progressing extra-abdo minal desmoid tumours *A. Vanzulli*, L. V. Sciacqua, L. Saggiante, T. Cas cella, C. Colombo, E. Palassini, C. Morosi, S. Stacchiotti, A. Gronchi ; Milan/IT ([email protected]) Purpose or Learning Objective: To evaluate the safety and efficacy of percutaneous cryoablation for the treatment of extr a-abdominal desmoid tumours (DT) progressing after active surveillance/ first-line treatments or threatening to life/function/quality-of-life.

Methods

or Background: We retrospectively evaluated baseline and post- treatment MR and CT imaging of 27 consecutive proce dures performed at our Institution between May 2021 and July 2024. Treatme nt response was assessed both with standard and modified (m-) RECIS T 1.1 (employing also T2WI and DWI to define tumour viability), with the underlying premise that dimensional reduction alone does not adequately cap ture tissue viability and may incompletely depict responses to local treatmen ts. Three different timepoints were considered: approximately 40 days a fter treatment, 5-10 months after treatment and 11-16 months after treat ment.

Results

or Findings: The study cohort comprised 25 patients (21 females and 4 males; median age at treatment: 36 years; age ran ge 13-59 years) affected by extra-abdominal treatment eligible for local abl ative therapies. Tumour locations included the head & neck (3), the thoraci c wall (4), the abdominal wall (19) and the lower extremities (1), with an average maximum diameter of 81,2 mm (range 32-162,3 mm). At approximately 40 days af ter treatment, mRECIST responses were: 2/19 (10,5%) Stable Disease (SD), 6 /19 (31,6%) Partial Response (PR) and 11/19 (57,9%) Complete Response ( CR). At 5-10 months after treatment, mRECIST responses were: 9/18 (50%) SD, 2/18 (11,1%) PR and 7/18 (38,9%) CR. At 11-16 months after treatmen t, mRECIST responses were: 6/16 (37,5%) SD, 4/16 (25%) PR and 6/16 (37,5 %) CR. At 11-16 months after treatment, standard RECIST responses were: 1/ 16 (6,3%) Progressive Disease (PD), 14/16 (87,5%) SD and 1/16 (6,3%) PR.

Conclusion

Percutaneous cryoablation represents a feasible tre atment for extra-abdominal DT requiring treatment. mRECIST 1.1 outperform standard RECIST1.1 in this clinical scenario.

Limitations

Retrospective Thursday Abstract-based Programme 114 Funding for this study: Nothing to disclose Ethics committee - additional information: Not applicable Author Disclosures: Lorenzo Saggiante: Nothing to disclose Lucilla Violetta Sciacqua: Nothing to disclose Elena Palassini: Nothing to disclose Chiara Colombo: Nothing to disclose Alessandro Gronchi: Nothing to disclose Andrea Vanzulli: Nothing to disclose Tommaso Cascella: Nothing to disclose Carlo Morosi: Nothing to disclose Silvia Stacchiotti: Nothing to disclose Transarterial Embolization for Adhesive Capsulitis: outcome assessment using MRI *B. Wang*¹, K-W. Liang², H. Y. Lin²; ¹Tainan/TW, ²T aichung/TW ([email protected]) Purpose or Learning Objective: To assess the efficacy of transarterial embolization (TAE) for adhesive capsulitis (AC) by evaluating clinical outcomes and changes in inflammatory status using magnetic r esonance imaging (MRI).

Methods

or Background: Patients diagnosed with AC, undergoing TAE, and with baseline and 3-month contrast-enhanced MRI eva luations, were included. MRI results were analyzed to assess periarticular c apsule/ligament inflammation. Clinical assessments included pain sc ores using the Numeric Rating Scale (NRS) and functional scores using the Quick Disabilities of the Arm, Shoulder, and Hand (Quick DASH) questionnaire.

Results

or Findings: Twenty-five patients with AC were included. Signifi cant reductions in average NRS pain scores, and signific ant improvements in Quick DASH scores and ROM, including anterior flexion and abduction, were observed at 1, 3, and 6 months after TAE (all P < 0 .001). MRI analyses revealed that TAE significantly decreased the grade of axillary recess (AR) capsule enhancement, the grade of rotator interval (RI) capsule T2 signal intensity, and the grade of RI capsule enhancement (all P ≤ 0.004).

Conclusion

TAE is a promising and safe therapeutic approach fo r AC, improving pain alleviation and functional recovery. The observed MRI findings suggest that the mechanism of TAE for AC may involv e the reduction of inflammation and the elimination of angiogenesis.

Limitations

First, it was a single-arm study without a control group, meaning that clinical outcomes such as pain and functional scores were subjective and could be influenced by a placebo effect, necessitat ing cautious interpretation. Second, the study involved a relatively small numbe r of patients. Third, the classification of hypersignality and enhancement in tensity is arbitrary. Finally, different MRI machines with varying magnetic field strengths (1.5T and 3T) were used across patient examinations, introducing potential discrepancies in image interpretation. Funding for this study: Nil Ethics committee - additional information: Institutional Review Board of National Cheng Kung University Hospital: IRB No: B- ER-113-038 Author Disclosures: Bow Wang: Nothing to disclose Keng-Wei Liang: Nothing to disclose Hsuan Yin Lin: Nothing to disclose A novel treatment for persistent symptoms following spinal surgery: percutaneous ct-guided trans-facetal screw fixation *K. Desalos*, V. Sala, P-A. Ranc, L. J. Pavan, T. V ivarrat-Perrin, N. Amoretti; Nice/FR ([email protected]) Purpose or Learning Objective: Segmental spinal instability after laminectomy and adjacent segment disease (ASD) foll owing arthrodesis often requires repeated surgical interventions, usually b y complex surgical procedures such as surgical arthrodesis under gener al anesthesia which can be often demanding in patients with associated como rbidities. Trans-facetal fixation (TFF) under local anesthesia and CT guidance is a minimally invasive technique which involves placement of percutaneous screws through the posterior facet joints to improve spinal stability. Our retrospective study is aimed at evaluating the efficacy of pain reduction  and improvement of daily activities following CT guided TFF, in patients wit h symptoms related to focal instability secondary to laminectomy or ASD.

Methods

or Background: TFF were performed in 43 symptomatic patients with previous history of spinal surgery (laminectom y and/or surgical arthrodesis) at Nice University Hospital between 20 17 and 2024. The pre and postoperative pain and disability levels were colle cted prospectively, using the visual analogue scale (VAS) and the Oswestry Disabi lity Index (ODI), at 6- months and 1-year. Long term outcomes were assessed by telephone consultations.

Results

or Findings: There was a mean decrease of VAS by 3.4 points at 6 months and by 3. 6 points at 1 year. Mean decrease of ODI was 17.1 points at 6 months (47.7 +/- 13.4 versus 30.6 +/- 17.5, P=0.0 009)). All the screws were satisfactorily positioned and the pain tolerance un der local anesthesia was very good. The average duration of the procedure was 46 minutes, without any significant complications.

Conclusion

TFF under CT guidance and local anesthesia resulted in a significant pain reduction and improvement in daily activities, without any complications. TFF under CT guidance and local anes thesia is a safe and effective alternative for symptomatic instability after previous surgical laminectomy and/or arthrodesis.

Limitations

Small cohort Monocentric study Funding for this study: CHU Nice Ethics committee - additional information: Not applicable Author Disclosures: Luca Jacopo Pavan: Nothing to disclose Kevin Desalos: Nothing to disclose Thomas Vivarrat-Perrin: Nothing to disclose Nicolas Amoretti: Nothing to disclose Paul-Alexis Ranc: Nothing to disclose Vincent Sala: Nothing to disclose Genicular Artery Embolization for the Treatment of Symptomatic Knee- OA using resorbable particles: a pilot study of 66 patients *F. N. N. Fleckenstein*, B. Gebauer, F. Collettini; Berlin/DE ([email protected]) Purpose or Learning Objective: Genicular artery embolization (GAE) is an innovative, minimally invasive therapy for patients with symptomatic knee osteoarthritis (OA) that is unresponsive to conserv ative treatments. Despite its potential, there is ongoing debate regarding the op timal embolic material for this procedure. This study assesses both the safety and efficacy of GAE using a novel, resorbable particle in treating symptomati c knee OA.

Methods

or Background: A single-center study was conducted at our institution. Participants were aged between 40 and 90 years, all presenting with moderate to severe knee OA (Kellgren-Lawrence grade s 2 to 4) and a history of failed conservative treatments. Baseline pain wa s measured using the visual analog scale (VAS), and symptoms were assessed usin g the Knee Injury and Osteoarthritis Outcome Score (KOOS). After femoral arterial access was achieved with a 4 Fr sheath, embolization was perfo rmed using Nexsphere-F particles (100-300 µm). Target vessels were identif ied through digital subtraction angiography, correlating with patients' pain locations. Adverse events and symptom improvements were evaluated at 6 weeks, 3 months, and 6 months following the procedure.

Results

or Findings: 66 patients were enrolled, with a median age of 68y . OA severity was grade 2 (14%), 3 (44%), and 4 (42%). T echnical success was 100%. Skin discoloration and mild knee pain, were r eported in 9% of cases. No major complications occurred. At the 6-month, the K OOS quality-of-life index showed an 89% improvement, while VAS score indicate d an 82% reduction in pain, median baseline 52/100 and 8/10, respectively .

Conclusion

This study demonstrates that using resorbable parti cles for GAE is both safe and effective in alleviating symptoms associated with OA that do not respond to conservative treatments.

Limitations

This is a single-center study without a control gro up, yet the size of the cohort is considerably large. Funding for this study: No Ethics committee - additional information: Ethics approved Author Disclosures: Bernhard Gebauer: Nothing to disclose Federico Collettini: Nothing to disclose Florian Nima Nima Fleckenstein: Nothing to disclose

Results

of US-guided Hyaluronic Acid in patients wi th ankle osteoarthritis and osteochondral lesions of the talus *M. De Albert De Delas-Vigo*, E. A. Vargas Meouchi, I. Benegas, A. Sallent, G. Duarri, Y. Lara Taranchenko, S. Roche, M. Veinte millas, J. Alonso; Barcelona/ES ([email protected]) Purpose or Learning Objective: Describe the effects regarding pain relief and possible complications of hyaluronic acid (HA) inje ction in patients with ankle OA and OLT.

Methods

or Background: Observational study of patients with ankle OA and OLT that had an ultrasound guided HA injection by t he radiology department between January 1st 2020 and December 1st 2023 in o ur center and a minimum follow-up of 6 months. Patients' visual ana log scale (VAS) at baseline, 3 months and 6 months after injection wer e recorded. Exclusion criteria were administration of another medication (except mepivacaine) or biological therapy and incomplete data collection d uring follow-up. Thursday Abstract-based Programme 115

Results

or Findings: 137 patients were referred to the radiology departm ent for an ultrasound guided injection. Fifty-eight pat ients (63 ankles) that received HA injection for ankle OA and OLT were identified. Thirty-seven (42 ankles) were included. Baseline VAS score was 7.98 ± 1.37; 5.76 ± 2.14, and 6.64 ± 2.07 at 3 and 6 months post injection respectively (p<0.05). Patients reported a mean of 7.9 ± 7.81 months with some pain relief. Patients mean age was 59.6 (range 32-83) and mean follow-up was 18.8 ± 12.1 mo nths. Eleven (26.2%) patients received a second HA injection and 7 (16.6 %) underwent surgery during follow-up. No complications were recorded in this series.

Conclusion

HA injections is a safe treatment that may provide a temporary pain improvement in patients with ankle OA and OLT.

Limitations

Retrospective studio Funding for this study: None Ethics committee - additional information: Prospective clinical study Author Disclosures: Yuri Lara Taranchenko: Nothing to disclose Sarai Roche: Nothing to disclose Matias De Albert De Delas-Vigo: Nothing to disclose Gemma Duarri: Nothing to disclose Maite Veintemillas: Nothing to disclose Jaime Alonso: Nothing to disclose Enrique A. Vargas Meouchi: Nothing to disclose Iker Benegas: Nothing to disclose Andrea Sallent: Nothing to disclose Percutaneous Hydrodissection Technique For Anterior Cervical Approach of the spine I. Ben Rejeb¹, J. Lavigne¹, S. Grange², J-B. Noel³, *J-B. Pialat*¹, N. Stacoffe¹; ¹Pierre-Bénite/FR, ²Saint-Étienne/FR, ³Lyon/FR ([email protected]) Purpose or Learning Objective: Hydrodissection aims to reduce the risks associated with traditional open surgery by enablin g minimally invasive access to the cervical spine. To evaluate the feasibility, effectiveness, and safety of using hydrodissection in percutaneous anterior cerv ical spine procedures.

Methods

or Background: A retrospective analysis was conducted on 32 hydrodissection procedures (30 patients) performed between 2020 and 2024 in three medical centers. The patient cohort included individuals with advanced oncological conditions, benign tumors, trauma, and infections. The hydrodissection technique involved gradual injectio n of a mixture of normal saline and contrast medium under CT guidance, displ acing the jugulo-carotid structures and creating a safe space for interventi onal procedures such as odontoid osteosynthesis, cementoplasty, biopsy, and thermoablation.

Results

or Findings: The procedure was successfully completed in all 30 patients. The right-sided approach was predominantl y used to avoid esophageal injury. The mean volume of dissection fl uid used was around 250 ml, with continuous hydrodissection being essential to maintain the created space. A few patient with multiple level procedure reach to 500 ml. The extubation following the procedure was shortly dela yed to avoid any compression risk. No perioperative or postoperative complications were reported.

Conclusion

The use of hydrodissection in anterior cervical spi ne procedures offers a safe and effective alternative to open sur gery, particularly for high-risk patients. This technique minimizes neurovascular co mplications and ensures safe procedural access, with no recorded complicati ons in this short serie.

Limitations

Further studies with larger cohorts are warranted t o standardize the technique and confirm its safety and efficacy. Funding for this study: None Ethics committee - additional information: Actually evaluated by a comittee to be approved for multiple center retrospective in clusion Author Disclosures: Sylvain Grange: Nothing to disclose Nicolas Stacoffe: Nothing to disclose Ilyess Ben Rejeb: Nothing to disclose Joris Lavigne: Nothing to disclose Jean-Baptiste Pialat: Nothing to disclose Jean-Baptiste Noel: Nothing to disclose Patient reported outcomes and return to work after CT-guided percutaneous lumbar discectomy: a prospective study *P-A. Ranc*, N. Amoretti; Nice/FR ([email protected]) Purpose or Learning Objective: To evaluate the efficacy percutaneous lumbar discectomy (PLD) under computed tomography ( CT) guidance on pain, functional capacities and to estimate the speed of recovery by assessing the time before return to work.

Methods

or Background: Patients treated by PLD were prospectively included between December 2019 and April 2021. Data collected consists in pain, duration of the symptoms, analgesia intakes, time of absence from work and the Oswestry Disability Index (ODI). Patients w ere followed-up during six months. Duration of hospitalization and time before return to work were reported. The Fisher test was used to compare nomin al variables, the Kruskal- Wallis test for ordinal variables, and the Student test to compare quantitative continuous variables.

Results

or Findings: A total of 87 patients were evaluated (median age, 56; interquartile range [IQR], 42.5-66). The median ODI decreased from 44 (IQR, 33-53) to 7 (IQR, 2-16.5) at six months (p<0,001). The median visual analog scale (VAS) decreased from 8 (IQR, 8-9) to 2 (IQR, 0-3) within six months (p<0,001). In 96,5% of cases, patients were dischar ged on the day of the procedure, and 3,5% the following day. No severe ad verse events were reported according to the society of interventional radiology (SIR) classification system. Out of the 57 patients at work, 50 were abl e to return to work during the follow-up with a median time of 8 days (IQR, 0- 20).

Conclusion

Symptomatic lumbar disc herniations can be successf ully treated by PLD, resulting in significant improvement of sym ptoms, functional capacities, and a fast return to work.

Limitations

The limitations of the study are the lack of a cont rol group, so no real comparison among patients treated with surgica l methods. Full follow-up was not obtained for 20 patients who had to be excl uded from the analysis. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study received approval from the institutional review board and reference o n the clinicaltrials.gov database. Author Disclosures: Nicolas Amoretti: Author: Last Author Paul-Alexis Ranc: Author: First Author Immediate imaging findings after positioning of a n ew percutaneous interspinous process spacer *L. J. Pavan*, P-A. Ranc, T. Vivarrat-Perrin, K. De salos, V. Kilani, E. Dien, N. Amoretti; Nice/FR ([email protected]) Purpose or Learning Objective: To evaluate the immediate changing in spine imaging after positioning of a percutaneous intersp inous process spacer (IPS) for symptomatic degenerative lumbar spinal stenosis (DLSS).

Methods

or Background: All patients treated in our Centre from January 202 1 to December 2023 with a new percutaneous IPS (Lobst er®-Diametros Medical®) were retrospectively reviewed. Patients u nderwent this procedure for treating a symptomatic DLSS nonresponding to lu mbar epidural injection. All procedures were performed with combined CT-scan and fluoroscopy guidance under general anesthesia. For each patient neural foramina area on sagittal plane, as well as zygapophyseal articular space on axial plane, were independently measured by two operators on the trea ted level of preoperative and postoperative CT-scans.

Results

or Findings: Thirty-four consecutive patients were treated in th e considered period, and all were retrospectively inc luded in the study (mean age 79.2 ± 8.3 years [72-92], 27 males, 17 females). Mean neural foramina area increased from 73 ± 19 to 93 ± 24 mm2 (+ 27%; p < 0.01). Mean facet joint articular space increased from 2,2 ± 0,9 to 3,1 ± 1,1 mm (+ 40%; p < 0.01). Inter-observer reliability was very good (Cr onbach's alpha = 0.9). No procedural complication was reported.

Conclusion

These imaging changes may explain the clinical effe cts of IPS. The widening of foramina may be related to a decrea se pression on the middle column and the posterior portion of the interverteb ral disk, with reduction of its protrusion. The widening of the zygapophyseal space is linked to a stretching of flavum ligaments with a consequent reduction of its bulging into the spinal canal. Both factors will play a major role in reduc ing clinical symptoms of DLSS.

Limitations

The retrospective design of the study. Funding for this study: None Ethics committee - additional information: Not neede since observational and retrospective Author Disclosures: Victor Kilani: Nothing to disclose Luca Jacopo Pavan: Nothing to disclose Kevin Desalos: Nothing to disclose Emmanuel Dien: Nothing to disclose Thomas Vivarrat-Perrin: Nothing to disclose Nicolas Amoretti: Nothing to disclose Paul-Alexis Ranc: Nothing to disclose Thursday Abstract-based Programme 116 Evaluation of the therapeutic value of conventional lymphography for the treatment of inguinal lymphatic fistulas after lymp hadenectomy *C. Wolfram*, T. J. Vogl, K. Eichler, T. Gruber-Rou h; Frankfurt/DE ([email protected]) Purpose or Learning Objective: There is critics that advancements in technology, particularly in cross-sectional imaging techniques such as computed tomography (CT) and magnetic resonance ima ging (MRI) are better than conventional lymphography. The aim of the pres ent study is to prove that a lipiodol-based conservative lymphography seals pe rsistent lymphatic fistulas, providing a safe and effective alternative to conse rvative therapies and surgical interventions.

Methods

or Background: A group of 39 patients who underwent lymphadenectomy resulting into inguinal lymphatic f istulas between 2003 and 2023 was selected. Participants were eligible if th ey had persistent lymphatic leakage after inguinal lymphadenectomy and were unr esponsive to conservative treatment. Of these 39 patients, four could not be statistically evaluated in our retrospective study due to various technical problems. The remaining 35 patients were evaluated. Lipiodol lymp hography was performed using transpedal lymphatic vessel cannulation. Data were collected through clinical records (RIS/PACS) and imaging follow-ups. Statistical analyses included the Wilcoxon–Mann–Whitney test using BiAS software. Success was defined as the complete occlusion of lymphatic leak age, and data on complications and additional interventions were col lected.

Results

or Findings: Therapeutic success was achieved in 22 patients (62.86%), with a mean resolution time of 7.13 days. For four patients there was no data on further course, while 13 required additi onal interventions (three surgical, six radiotherapy). Statistical analysis s howed no significant correlation between the volume of administered iodized oil and therapeutic success (p = 0.51), nor lymphatic drainage volume (p = 0.69).

Conclusion

Our results highlight that conventional lymphograph y is a successful therapy. These findings could inform fut ure studies aiming to refine patient selection criteria and optimize treatment p rotocols for complex lymphatic conditions.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee notification can be found under the number UID 2023-1444. Author Disclosures: Christian Wolfram: Nothing to disclose Kathrin Eichler: Nothing to disclose Thomas J. Vogl: Nothing to disclose Tatjana Gruber-Rouh: Nothing to disclose 14:00-15:30 Research Stage 4 Research Presentation Session: Breast RPS 1002 Exploring the role of artificial intelligence in breast imaging Moderator I. Thomassin-Naggara; Paris/FR ([email protected]) A 10-year image-derived AI risk model for use in pr imary prevention of breast cancer *M. Eriksson*¹, K. Czene¹, C. Scott², P. Hall¹, C. Vachon²; ¹Stockholm/SE, ²Rochester, MN/US ([email protected]) Purpose or Learning Objective: Image-derived artificial intelligence (AI) risk models have shown significant potential in enhancin g breast cancer (BC) screening through short-term risk assessment. A lon g-term image-derived AI risk model for primary prevention has yet to be dev eloped and externally validated.

Methods

or Background: This study utilized a case-cohort approach, including women aged 35-94 recruited between 2009-2 017 from population- based screenings in Olmsted County, Minnesota (U.S. ), and the KARMA cohort in Sweden. Median follow-up was 10 years, wi th BCs diagnosed before 5/2022. An image-derived AI risk model, initially d eveloped in a Swedish population, was validated independently in the Olms ted/KARMA cohorts. At study entry, 10-year absolute risks were estimated. Time-dependent discriminatory performance (AUC(t)) and expected-to -observed event ratios (E/O) were calculated.

Results

or Findings: The combined Olmsted/KARMA cohorts included 8,721 women, with a mean age of 54.4 years (±10.6) in the subcohort and 1,633 incident BC cases with a mean age of 57.0 years (±10.6). The AI-derived 10- year average risks were 3.83% and 3.14%, with E/O r atios of 0.99 (95%CI 0.94-1.05) in Olmsted and 0.99 (95%CI 0.91-1.08) in KARMA. The 10-year AUC(t) values were 0.70 (95%CI 0.68-0.73) for Olmst ed and 0.73 (95%CI 0.69-0.77) for KARMA. Using the U.S. Preventive Ser vices Task Force (USPSTF) guidelines, 41% of cases in KARMA were ide ntified as high-risk, compared to 15% with Tyrer-Cuzick-v8 and 5.1% with BCSC-v3 (p<0.01). Under the National Institute for Health and Care Ex cellence (NICE) guidelines, these figures were 31%, 7.4%, and 0.2%, respectivel y.

Conclusion

The 10-year image-derived AI risk model demonstrate d strong predictive performance in both U.S. and Swedish cas e-cohorts, outperforming traditional clinical risk models in KARMA. This AI model holds significant potential for clinical application in primary preve ntion, targeting up to 40% of BCs.

Limitations

The study population was mainly White women. Funding for this study: Swedish Research Council Ethics committee - additional information: Mayo Clinic and Olmsted Medical Center Institutional review board and the S wedish Ethical Review Authority Author Disclosures: Mikael Eriksson: Patent Holder: Patent on "system a nd method for assessing breast cancer risk using imagery" with a licence to iCAD medical, Nashua, NH, U.S. Patent on "compositions and methods for monito ring the treatment of breast disorders" with a licence to Atossa Therapeu tics, Seattle, WA, U.S. Celine Vachon: Nothing to disclose Christopher Scott: Nothing to disclose Kamila Czene: Patent Holder: Patent on "system and method for assessing breast cancer risk using imagery" with a licence to iCAD medical, Nashua, NH, U.S. Patent on "compositions and methods for monito ring the treatment of breast disorders" with a licence to Atossa Therapeu tics, Seattle, WA, U.S. Per Hall: Patent Holder: Patent on "system and meth od for assessing breast cancer risk using imagery" with a licence to iCAD m edical, Nashua, NH, U.S. Patent on "compositions and methods for monitoring the treatment of breast disorders" with a licence to Atossa Therapeutics, S eattle, WA, U.S. Cracking the Code: Predicting Pathogenic Mutations in Breast Cancer with Ultrasound Radiomics *R. M. Pintican*, N. Antone; Cluj-Napoca/RO ([email protected]) Purpose or Learning Objective: To evaluate the potential of US-based in the prediction of pathogenic mutational status of breas t cancer patients, relevant to prophylactic mastectomy recommendations.

Methods

or Background: This retrospective study included 73 breast cancer patients tested with multigene panel tests includin g all seven pathogenic mutations (BRCA1, BRCA2, TP53, PTEN, CDH1, PALB2, a nd STK11 mutations). US images were acquired prior to any tr eatment and tumoral and peritumoral areas were used to extract radiomics da ta. The study population was divided into testing and validation group, each with pathogenic- and non- pathogenic mutation population. Radiomics features were analyzed using machine learning models, alone and in combination w ith clinical features ( ki67%).

Results

or Findings: We observed significant differences in radiomics features between pathogenic- and non-pathogenic mut ation driven tumors. Using a three-step feature selection process we dev elop the prediction models (The Mann-Whitney U test, Spearman Correlation and LASSO Regression); the Rad-score 1 ( tumor) achieved an accuracy of 78 .6% in identifying pathogenic mutation carriers, while Rad-score 2 (tu mor+peritumoral) increased the model's accuracy to 85%. The Rad-Clin 1 and Rad -Clin 2 achieved 83% and 95% acuracy in predicting mutational status. On validation cohort we obtained the following AUCs: Rad-score 1 = 66%; Rad -score 2 = 91%; Rad- Clin 1 = 58%; Rad-clin 2 = 83%.

Conclusion

Radiomics models based on US images of breast tumor s may provide a promising alternative for predicting path ogenic mutation status in BC patients. The highest accuracy was reached when we combined radiomics data extracted from the tumor and peritumoral area. This approach could reduce dependence on costly genetic testing and exp edite the diagnostic process.

Limitations

Small sample size Unicentric study Funding for this study: No funding Ethics committee - additional information: Retrospective study - the informed consent was waived. Author Disclosures: Roxana Maria Pintican: Nothing to disclose Nicoleta Antone: Nothing to disclose Thursday Abstract-based Programme 117 Do we still need to double read the most suspicious screening mammograms when using AI for decision support? A su b-analysis from the AITIC breast cancer screening prospective trial *E. Elías Cabot*¹, A. Rodriguez Ruiz², J. L. Raya P ovedano¹, S. Romero Martin¹, M. Álvarez Benito¹; ¹Cordoba/ES, ²Nijmegen/NL ([email protected]) Purpose or Learning Objective: To evaluate the differences between single and double reading of the most suspicious mammogram s after the introduction of AI in breast cancer screening.

Methods

or Background: This was a sub-analysis of the AITIC paired prospective trial in the breast cancer screening pr ogram in Cordoba, Spain. In this trial, between March 2022 and January 2024, 31 ,301 women (age 50-71) were included and imaged with either DM or DBT base d on equipment availability. Two reading strategies were independe ntly applied to each exam: Double blind and non-consensual reading of all exam s (standard strategy) and AI-based triaging (AI strategy), where an AI system (Transpara v1.7, ScreenPoint Medical) evaluated the cancer risk of a ll exams. Cases identified by AI as Low risk were automatically assessed as ne gative, while exams with Intermediate or Elevated risk were double read with concurrent AI-support. For the latter group, cancer detection (CDR) and false positive rates (FPR) were compared between single and double reading. P value s using McNemar and binomial confidence intervals (CI) were computed.

Results

or Findings: The AI strategy, double reading only 36% of the tot al screening mammograms, resulted in 228 screen-detect ed cancers (CDR=7.3/1000, CI: 6.4-8.2/1000) and 1,723 recalls (FPR=4.8%, CI: 4.5- 5.0%). Should these exams have been single read wit h AI support, there would have been 190 screen-detected cancers (CDR = 6.0/10 00, CI: 5.2-7.0/1000), and 1,082 recalls (FPR = 2.9%, CI: 2.7-3.0%), a -17 % (P<0.05) and -42% reduction (P<0.05) with respect to double reading. The standard strategy resulted in 1,501 recalls (FPR=4.2%, CI: 4.0-4.4%) and 198 cancers (CDR=6.3/1000, CI: 5.5-7.2/1000).

Conclusion

After introduction of AI for triage and decision su pport in screening, increased cancer detection rates were ac hieved in comparison to standard of care by still double reading a subgroup of the most suspicious exams.

Limitations

Single-site. Funding for this study: None. Ethics committee - additional information: Local IRB approval. Author Disclosures: Alejandro Rodriguez Ruiz: Employee: ScreenPoint Med ical Marina Álvarez Benito: Nothing to disclose Sara Romero Martin: Nothing to disclose Esperanza Elías Cabot: Nothing to disclose Jose Luis Raya Povedano: Nothing to disclose Benefits and risks of AI use for reviewing negative screening mammograms *C. De Wolf*¹, K. Brändle², J-L. Bulliard²; ¹Geneva /CH, ²Lausanne/CH ([email protected]) Purpose or Learning Objective: Introduction: Breast cancer remains a global health concern, with artificial intelligence (AI) offering promising advancements in improving screening accuracy. Traditional method s, requiring high-volume readings, often lead to fatigue and reading errors. AI addresses these

Limitations

by providing fatigue-free, reproducible results. This study assesses the benefits and costs of AI in detecting high-risk lesions in mammograms initially classified as negative by radiologists.

Methods

or Background: Methods: Risk scores (Transpara® version 1.7.3) were calculated for 54’300 mammograms from a public Swiss screening program (2018–2021). Data included screen detected (n=321) and interval cancers (n=94), lesion location, and double-blind r adiologist readings. We included risk score thresholds considered as elevat ed risk (61 to 90). Key outcomes included additional workload (additional m ammograms in consensus conference), avoided false-negative interval cancer s (FN-IC, n=39), and increased false-positive (FP) rates. Multivariable logistic regression was used to predict the rise in FP cases across thresholds.

Results

or Findings: Results: The FN-IC rate reduction ranged from 8.3% (threshold 90) to 31.3% (threshold 61), with an add itional workload of 2 to 67 extra mammograms per 1,000 participants. Avoiding o ne FN-IC case required 28 to 242 extra readings, resulting in 12 to 86 add itional false positives (FP). FP rates rose by 2.1% to 59.3%, with the workload i ncreasing by a third for every 5-point threshold drop up to 75. With an AI t hreshold set to 85, the false positive rate increased by 2.3‰ (from 40.5‰ to 42.8 ‰) and the workload would increase by 6 mammograms /1000 participations .

Conclusion

AI assistance may enhance mammography sensitivity. However, this comes with a relatively high cost in terms of FP results and additional readings. Therefore, determination of the critical threshold must be context- specific to achieve optimal benefit – risk ratio.

Limitations

Retrospective design. Funding for this study: No external funding Ethics committee - additional information: All women signed an informed consent that their anonymized screening data could be used for quality assurance purposes. Author Disclosures: Jean-Luc Bulliard: Nothing to disclose Karen Brändle: Nothing to disclose Christophorus De Wolf: Nothing to disclose Patient perceptions and attitudes towards the use o f AI in the symptomatic breast unit *S. Singh*, R. P. Crean, H. Briody, R. Bruen, N. Ha mbly, M. Bambrick, D. Duke, M. Mullooly, N. Healy; Dublin/IE ([email protected]) Purpose or Learning Objective: Artificial intelligence (AI) has been evaluated in a number of breast screening settings with favou rable results. While there are limited studies looking at patient attitudes to AI in breast screening none have examined perceptions of AI in the symptomatic setting. The aim of this study was to determine attitudes towards AI among p atients attending the symptomatic breast unit.

Methods

or Background: An anonymous 15 question, voluntary questionnaire was given to all patients attending t he symptomatic breast clinic imaging department of Beaumont Hospital from 01/07/ 2024 to 30/09/2024.

Results

were collated in a password protected Excel database and descriptive statistics performed. Likert responses were numeric ised so that mean of 1 denotes strong agreement and 5 denotes strong disag reement.

Results

or Findings: Of the 1500 patients who were surveyed, most were aged 40–59 years (62.1%). Almost one-quarter had ei ther a personal (364/1500) or family history of breast cancer (360/ 1500). 62% (927/1500) had some or strong interest in AI. Regarding the use of AI in healthcare, 46% agreed it was a good idea, 8% disagreed and 46% wer e indifferent. There was support for AI assisting radiologists in reading ma mmograms (Mean (M)=2.43,95% CI:2.39-2.48) but disapproval of AI be ing the sole reader (M=3.82,95% CI:3.77-3.87). Respondents strongly pre ferred human radiologists over AI for reading mammograms, even i f AI were more efficient (M=1.95,95% CI:1.90-1.99) or more accurate (M=2.17, 95% CI:2.13-2.22). 75% of patients would blame both the AI developer a nd the human radiologist for an incorrect result. All results were statistic ally significant (p<0.001).

Conclusion

Respondents hold favourable views towards the use o f AI in healthcare. They welcome use of AI as an adjunct fo r radiologists but disagree with AI being the only reader of their mammogram.

Limitations

N/A Funding for this study: RCSI seed funding Ethics committee - additional information: Approval has been obtained from the hospital audit committee (CA2024/126). Formal e thical approval was not deemed necessary as this is an anonymised, voluntar y study. Author Disclosures: Niamh Hambly: Nothing to disclose Maeve Mullooly: Nothing to disclose Deirdre Duke: Nothing to disclose Marie Bambrick: Nothing to disclose Richard Bruen: Nothing to disclose Hayley Briody: Nothing to disclose Rory Peter Crean: Nothing to disclose Nuala Healy: Nothing to disclose Sneha Singh: Nothing to disclose The effect of an artificial intelligence decision s upport system on radiologists’ screening mammography performance and visual search patterns *J. Gommers*, S. D. Verboom, M. Broeders, I. Sechop oulos; Nijmegen/NL ([email protected]) Purpose or Learning Objective: To investigate the effect of using a commercial artificial intelligence (AI) decision su pport system on the diagnostic performance and visual search patterns of radiologi sts interpreting screening mammograms.

Methods

or Background: A multi-reader, multi-case study was performed with 12 Dutch screening radiologists interpreting 1 50 screening mammography examinations (75 normal, 75 malignant). Radiologist s read the examinations without and with AI support while an eye tracker re corded their eye movements. AI classified the examinations as low (m aximum region scores:<40), intermediate (40-59), medium-high (60- 79), or very-high risk (80- 100). Radiologists provided a probability of malign ancy score (0-100) and recall decision for each examination. The performan ce under the two reading conditions was compared using the area under the re ceiver operating characteristics curve (AUC), sensitivity, and speci ficity through mixed-model analysis of variance. Reading time and eye tracking outcomes were compared by bootstrap resampling (n=20,000). Thursday Abstract-based Programme 118

Results

or Findings: The average AUC increased significantly from 0.93 without AI support to 0.97 with AI support (P<.001) . There was no evidence of a significant change in sensitivity (81.7% vs 87.2% , P=.06) or specificity (89.0% vs 91.1%, P=.46), although sensitivity tende d to increase for AI- classified high-risk examinations (medium-high: 54. 9% vs 61.8%, very-high: 89.5% vs 95.6%). Overall reading time did not chang e significantly (29.4 vs 30.8 seconds, P=.32), but decreased for AI-classifi ed low-risk examinations (25.1 vs 20.1 seconds, P<.001). When using AI, radi ologists covered less of the breast area with fixations (11.1% vs 9.5%, P=.0 05), while spending more time fixating in lesion areas (4.0 vs 5.1 seconds, P<.001).

Conclusion

Reading with an AI decision support system increase d radiologists’ screening performance and allowed the m to focus more on lesion- specific areas without increasing overall reading t ime, indicating a more efficient search.

Limitations

Enriched case set and one AI system only. Funding for this study: aiREAD financed by KWF Dutch Cancer Society and the Dutch Research Council (NWO) Domain Applied and Engineering Sciences (AES), as part of their joint strategic re search program Technology for Oncology II. The collaboration project is co-fu nded by the PPP Allowance made available by Health-Holland, Top Sector Life S ciences & Health, to stimulate public-private partnerships. Ethics committee - additional information: The need for ethical approval for this retrospective multi-reader multi-case study wa s waived by the Research Ethics Committee of Radboud University Medical Cent er (registration number 2021–13186). Author Disclosures: Mireille Broeders: Research/Grant Support: Screenpo int Medical, Sectra Benelux, Hologic, Volpara Solutions, Lunit, iCAD Sp eaker: Siemens Healthcare, Hologic Sarah Delaja Verboom: Nothing to disclose Jessie Gommers: Nothing to disclose Ioannis Sechopoulos: Research/Grant Support: Siemen s Healthcare, Canon Medical, ScreenPoint Medical, Sectra Benelux, Volpa ra Solutions, Lunit, iCAD Speaker: Siemens Healthcare, Canon Medical Mammographic features of false positive AI markings on screening mammograms from BreastScreen Norway *M. A. Martiniussen*¹, M. B. Bergan², J. Gjesvik², M. Undrum Kristiansen¹, S. Hofvind²; ¹Graalum/NO, ²Oslo/NO ([email protected]) Purpose or Learning Objective: False positive AI markings are an expected challenge when implementing artificial intelligence (AI) in mammographic screening and might contribute to an unsustainable increase in the workload for the radiologists. The aim of this study was to gain knowledge about false positive AI markings from two AI systems on screeni ng mammograms.

Methods

or Background: In this retrospective study, 129 385 screening examinations from BreastScreen Norway, performed at Ostfold Hospital Trust, 2008-2018, were run through two AI systems. System A was Lunit INSIGHT MMG version 1.1.7.2, and system B was a non-commerc ial system, developed by the Norwegian Computing Center and the Cancer Re gistry of Norway. Each model provided a score on a scale from 0-100, and m arked the most suspicious areas. Higher score indicated higher ris k of cancer. Two radiologists performed a consensus-based informed review of exam inations among those with the 5% highest AI score from both systems, int erpreted negative at index screening and without cancer diagnosed at index and two consecutive screening rounds. Mammographic features correspondi ng to the AI markings were classified according to the Breast Imaging Rep orting and Data System (BI-RADS). The results were analyzed using descript ive statistics.

Results

or Findings: Among the examinations that met the inclusion crite ria (n=252), 120 examinations from 120 women were rando mly selected for review. The mammographic feature corresponding to t he AI markings was calcifications for 71.7% (86/120) for system A and 67.5% (81/120) for system B, a mass for 12.5% (15/120) for system A and 14.2% (17/120) for system B, while asymmetry accounted for 10.8% (13/120) for sy stem A and 11.7% (14/120) for system B.

Conclusion

Calcifications was the main mammographic feature in screening mammograms with high AI score without diagnosed can cer.

Limitations

No limitations were identified. Funding for this study: The South-Eastern Norway Regional Health Authority Ostfold Hospital Trust Ethics committee - additional information: The study was approved by the Regional Committees for Medical and Health Research Ethics (#13294, #11022) Author Disclosures: Jonas Gjesvik: Nothing to disclose Marie Burns Bergan: Nothing to disclose Marit Almenning Martiniussen: Nothing to disclose Solveig Hofvind: Nothing to disclose Merete Undrum Kristiansen: Nothing to disclose Re-attendance in BreastScreen Norway after a false positive screening

Result

*M. Larsen*, N. Moshina, J. Gjesvik, S. Sagstad, Å. S. Holen, M. B. Bergan, T. E. Nilsen, S. Hofvind; Oslo/NO ([email protected]) Purpose or Learning Objective: Higher risk of breast cancer after a false positive versus a negative screening result has bee n reported. We aimed to compare re-attendance for women with a false positi ve versus negative screening result using more than 25 years of screen ing data.

Methods

or Background: BreastScreen Norway invites women aged 50-69 to biennial screening. The study sample included 3 990 388 screening examinations from 921 309 women where an invitation to the subsequent screening round was available (eligible for re-atte ndance). Attendance in the subsequent screening round was analysed using mixed logistic regression with age at screening and screening history as covariate s and screening outcome as exposure. Predicted probabilities (re-attendance ) and 95% confidence intervals (CI) were calculated using average margin al effects.

Results

or Findings: Having a false positive result after the prevalent screening examination resulted in a re-attendance r ate of 88.3%. For women with a negative result, re-attendance was 90.3% aft er the prevalent examination. Having a false positive or negative re sult in the 9th screening round, gave a re-attendance rate of 89.0% and 91.1% , respectively. Predicted re-attendance rate was 88.9% (95% CI: 88.9-89.0) af ter a false positive result and 88.1% (95% CI 88.0-88.3%) after a negative resu lt. Using negative result, false positive without invasive procedure or false positive with invasive procedure as exposure variable, the predicted proba bilities of re-attendance were 88.9% (95% CI: 88.9-89.0), 88.4% (95% CI: 88.2 %-88.6%) and 87.6% (95% CI: 87.3%-88.0%), respectively.

Conclusion

Despite small differences in re-attendance after a false positive versus negative screening result, we consider the d ifference clinically important. Women should be informed about the impor tance of re-attending the screening programme after a false positive result.

Limitations

We do not have patient reported data on reasons for non- attendance. Funding for this study: No funding was received for this study. Ethics committee - additional information: Programme quality assurance is covered by the Cancer Registry Regulations. Author Disclosures: Nataliia Moshina: Nothing to disclose Jonas Gjesvik: Nothing to disclose Marie Burns Bergan: Nothing to disclose Åsne Sørlien Holen: Nothing to disclose Silje Sagstad: Nothing to disclose Tom Erik Nilsen: Nothing to disclose Solveig Hofvind: Nothing to disclose Marthe Larsen: Nothing to disclose The application of artificial intelligence to enhan ce the identification of previously missed non-palpable breast carcinomas S. A. Mansour, R. M. Kamal, S. Hussien, M. Emara, Y . Kassab, S. Taha, M. M. Gomaa, *Y. M. Nada*; Cairo/EG ([email protected]) Purpose or Learning Objective: To investigate the impact of artificial intelligence (AI) on digital mammograms in increasi ng the chance of detection of missed breast cancer, study the early morphology indictors detected by AI and overlooked by the radiologist and correlate wit h the missed cancer pathological types.

Methods

or Background: Screening and diagnostic mammograms (done in 2020-2023) presenting breast carcinomas (n = 1998) were analyzed in concordance with prior one-year-ago (2019-2022) ass umed negative or benign) mammograms. Present mammograms were reviewe d for the mammographic descriptors: asymmetry, distortion, ma ss, and microcalcifications. The AI analyzed mammograms and presented abnormalities by overlaying color hue and scoring p ercentage for the degree of suspicion of malignancy.

Results

or Findings: Artificial intelligence detected 555 (54%) lesions in the prior mammograms, and in present mammograms (year 2 020-2023) targeted 904 (88%) carcinomas. The descriptor proportion of asymmetry was the common presentation of missed breast carcinoma (n=3 56/555, 64.1%) in the prior mammograms and the AI highest detection rate presented by distortion (100%) followed by grouped microcalcifiactions (80% ). AI performance to predict malignancy in previously assigned negative or benign mammograms showed a sensitivity of 73.4%, a specificity of 89% , and an accuracy of 78.4%.

Conclusion

Reading mammograms with artificial intelligence enh anced the detection of early cancerous changes. AI detection rate is not correlated with certain pathological types of breast cancer. Close follow-up is required for AI abnormality scoring of low values to minimize the p otential for missed breast carcinoma. Thursday Abstract-based Programme 119

Limitations

The study is being limited by the retrospective stu dy design; and that it was a two institutional-based study, so mul tiple institutional-based studies are recommended. Funding for this study: The study has no source of funding Ethics committee - additional information: The study has been ethically approved by the research center of the affiliated i nstitute Author Disclosures: Sahar Abdelkhalek Mansour: Nothing to disclose Samar Hussien: Nothing to disclose Mohammed Mohamed Gomaa: Nothing to disclose Yasmin Mohamed Nada: Nothing to disclose Youmna Kassab: Nothing to disclose Mostafa Emara: Nothing to disclose Sherif Taha: Nothing to disclose Rasha Mohamed Kamal: Nothing to disclose AI-assisted Breast Mass Classification in Digital B reast Tomosynthesis (DBT): Applicability and insights from a single aca demic centre *G. Cura Curà*¹, G. Bartoli², M. Costa², E. Regini² , E. Puglisi², F. Piccione², F. Schettini², M. Durando², P. Fonio²; ¹Vercelli/IT , ²Torino/IT ([email protected]) Purpose or Learning Objective: In previous research, we trained a deep- learning model to classify benign and malignant mas ses identified on DBT images (convolutional neural network: efficientNetB 0; dataset: 448 masses, size < 6 cm, 221 malignant, 227 benign; accuracy 94 %, sensitivity 95.6%, specificity 91.7%). The aim of this study is to eva luate its applicability on breast mass lesions diagnostic work-up in clinical practic e.

Methods

or Background: In this single-centre multireader study, we prospectively collected DBT images from patients wi th biopsy-proven breast masses (size < 6 cm). For each case, masses were ma nually delineated with orthogonal axes on the best focused slice in both D BT standard views. A preliminary set of 64 DBTs (46 benign, 18 malignant ) was reviewed by three independent dedicated breast radiologists with diff erent experience, then assessed with the AI model. The software provides t he benign/malignant classification combined with a prediction confidenc e score. The response of the software was compared to biopsy results, focusi ng on BI-RADS classification, error rates, inter-reader agreement , and reading time.

Results

or Findings: In 7% of cases, there was inter-reader disagreement on the software prediction. The model correctly classi fied 84% of masses, confirming the 92.8% of lesions categorized as BI-R ADS 3 as benign. Software-assisted reading did not modify the readin g time compared to conventional methods.

Conclusion

In these preliminary results, the highest agreement between radiologists and the AI model was observed with BI- RADS 3 lesions, highlighting the benefit of software-assisted chara cterization of benign masses. However, variability in inter-reader agreement on t he software’s predictions limits its reliability in real practice. Further in vestigations and model refinement are necessary to improve the model robustness.

Limitations

Small sample size and single-centre study Funding for this study: No funding was provided for this study Ethics committee - additional information: Non applicable Author Disclosures: Manuela Durando: Nothing to disclose Gaia Cura Curà: Nothing to disclose Matilde Costa: Nothing to disclose Francesca Schettini: Nothing to disclose Eugenia Puglisi: Nothing to disclose Elisa Regini: Nothing to disclose Federica Piccione: Nothing to disclose Germana Bartoli: Nothing to disclose Paolo Fonio: Nothing to disclose Multi-site validation of an image-based AI breast c ancer risk model for mammography to drive personalized screening after a negative screening *A. D. Lauritzen*¹, A. Rodriguez-Ruiz², N. Karsseme ijer², C. De Wolf³, R. Mann², M. Nielsen¹, I. Vejborg⁴, M. Lillholm¹; ¹Copenhagen/DK, ²Nijmegen/NL, ³Geneva/CH, ⁴Gentofte/DK ([email protected]) Purpose or Learning Objective: To validate the performance of an image- based AI breast cancer risk model to stratify women attending screening after a negative screening.

Methods

or Background: Exams from women attending two European screening programs (Denmark and Switzerland) and fr om a public U.S. database (EMBED) were consecutively sampled. All ex ams were screen- negative (cancer-free for 180 days) and had follow- up information of between two and six years. Mammography exams were processed by an AI breast cancer risk model (Transpara Risk, ScreenPoint Medi cal, trial version for research). The risk model computes three image biom arkers (suspicious findings, volumetric breast density, breast texture ), and combined with age, it generates a five-year breast cancer risk score per exam. All exams were fully independent from the development of the risk model. Risk model AUCs were computed for each cohort along with sensitivity for women with the highest 10% risk and breast density, respectively.

Results

or Findings: In total, 98,084 exams were included (31,349, 17,44 5, and 49,290 from Switzerland, US, and Denmark, respe ctively) with 1,336 breast cancers diagnosed within 5 years from screen ing. Images were acquired with machines from four manufacturers (Hol ogic, Siemens, GE, Philips). The AUCs of the AI risk model were 0.73 ( 95% CI: 0.69-0.76), 0.74 (95% CI: 0.69-0.79) and 0.74 (95% CI: 0.73-0.76) fo r Switzerland, US, and Denmark, respectively. When simulating using risk t o offer supplemental imaging to 10% of women, after a negative screening , sensitivity was 37% (95% CI: 34%-39%), in comparison to 15% (95% CI: 13 -17%) when using density alone.

Conclusion

An image-based AI breast cancer risk model shows hi gh accuracy and robustness to stratify women attending screening according to risk and could support personalized screening with higher sensitivity than breast density.

Limitations

The retrospective study design is a limitation of t his study. Funding for this study: Supported in part by Eurostars (grant E9714 IBSCREEN) Ethics committee - additional information: The Danish Patient Safety Authority and Danish Data Protection Agency approve d this retrospective study and the use of relevant Danish data, and waived the need for informed consent (ref. 3–3013–2118, addendum 2019/2023). Author Disclosures: Andreas David Lauritzen: Nothing to disclose Alejandro Rodriguez-Ruiz: Employee: ScreenPoint Med ical Mads Nielsen: Nothing to disclose Martin Lillholm: Nothing to disclose Nico Karssemeijer: Employee: ScreenPoint Medical Chris De Wolf: Nothing to disclose Ilse Vejborg: Nothing to disclose Ritse Mann: Nothing to disclose Replacing one radiologist with AI for independent d ouble reading in mammographic screening *M. B. Bergan*, M. Larsen, J. Gjesvik, N. Moshina, S. Sagstad, T. Hovda, H. W. Koch, M. A. Martiniussen, S. Hofvind; Oslo/NO ([email protected]) Purpose or Learning Objective: The aim of this study was to explore how replacing one radiologist with artificial intellige nce (AI) for independent double reading in mammographic screening would affect canc er detection.

Methods

or Background: This study sample consisted of 1,027,430 screening examinations, including 5786 screen-detec ted cancers, that were independently interpreted by two radiologists in Br eastScreen Norway, 2004- 2018. The radiologists scored each breast from 1, n egative for abnormality, to 5, high suspicion of malignancy, and score ≥2 was considered positive. All examinations were processed by the AI system Lunit INSIGHT MMG version 1.1.7.2, assigning a continuous malignancy score fr om 0, no risk, to 100, very high risk. Cancer detection was presented for the c ombination of one radiologist and AI at various AI thresholds for pos itive examinations.

Results

or Findings: Of all screen-detected cancers, 86.9% (5028/5786) w ere classified as positive (score ≥2) by one radiologist. When defining 10% of the examinations with the highest AI score as positive by AI, 79.9% (4622/5786) of the screen-detected cancers and 7.5% (134/1783) of the interval cancers would be detected. When 5% with the highest AI scor es were considered positive, 75.5% (4348/5786) of the screen-detected and 5.7% (102/1783) of the interval cancers would be detected. In a scenar io where 1% of the examinations were classified as positive by AI, 58. 2% (3369/5786) of the screen-detected and 2.4% (42/1783) of the interval cancers would be detected.

Conclusion

At an AI threshold of 5%, replacing one of the radi ologists with AI in independent double reading of screening mammogra ms will reduce the reading volume by 50% at the cost of missing 24.5% of screen-detected cancers, but with the possibility of detecting 5.7% of the interval cancers.

Limitations

We assume that all cancers classified as positive b y the radiologist and AI were detected. Funding for this study: Funding was provided by the Norwegian Cancer Society (Pink Ribbon) Ethics committee - additional information: The study was approved by the Regional Committees for Medical and Health Research Ethics (#2018/2574). Thursday Abstract-based Programme 120 Author Disclosures: Nataliia Moshina: Nothing to disclose Jonas Gjesvik: Nothing to disclose Marie Burns Bergan: Nothing to disclose Henrik Wethe Koch: Nothing to disclose Tone Hovda: Nothing to disclose Marit Almenning Martiniussen: Nothing to disclose Silje Sagstad: Nothing to disclose Solveig Hofvind: Nothing to disclose Marthe Larsen: Nothing to disclose 16:00-17:30 Research Stage 1 Research Presentation Session: Cardiac RPS 1103 Exploring cardiac imaging through quantitative MRI Moderator N. Fink; Munich/DE Improved and automated detection of papillary muscl e infarction using joint bright- and black-blood LGE MRI *T. Richard*, V. Nogues, T. Boulle, V. De Villedon De Naide, K. Narceau, B. Durand, S. Sridi, H. Cochet, A. Bustin; Bordeaux /FR ([email protected]) Purpose or Learning Objective: Papillary muscle infarction (PMI) has been linked to significantly increased mortality, and is a source of ventricular arrhythmias and mitral regurgitation. Bright-blood LGE (PSIR) imaging is the clinical gold standard for myocardial fibrosis char acterization. However, the limited contrast at the blood-scar interface makes PMI visualization often challenging. Black-blood LGE imaging overcomes this limitation by improving scar-to-blood contrast. Here, we aim to develop a n ovel co-registered joint bright- and black-blood LGE technology (SPOT) that could improve visual PMI detection (visu-PMI) , while allowing an automated PMI detection algorithm (auto-PMI).

Methods

or Background: Short-axis 2D whole-heart PSIR and SPOT images were collected on a 1.5T Siemens Aera system under breath-hold 12min post gadolinium injection. Auto-PMI included image acqui sition, statistics-based slice selection, left ventricular endocardium segme ntation, blood pool preprocessing, and fibrosis detection. 198 patients participated to the study and were divided into an optimization dataset for a uto-PMI parameters selection, and a testing dataset to evaluate visu-P MI and auto-PMI performances. Two radiologists assessed PMI on PSIR and SPOT images. A consensus reading was used as reference standard. N umber of patients with PMI were compared. Sensitivity and accuracy of both sequences and auto-PMI were measured. Inter- and intra-observer reproducib ility were reported.

Results

or Findings: Radiologists detected significantly more PMI with S POT (average increase: 30%). SPOT outperformed average PSIR sensitivity (93% vs. 75%) and accuracy (93% vs. 86%). Average inter- and intra- reproducibility increased with SPOT (79% vs. 74%, 97% vs. 88%). Aut o-PMI outperformed PSIR sensitivity (87%), while the accuracy equaled the SPOT average (86%).

Conclusion

Co-registered joint bright- and black-blood SPOT im aging allows for improved PMI detection and opens a new door for automated PMI assessment.

Limitations

Further validation in larger cohort is warranted. V isu- and auto- PMI reliability depends mostly on contrast selectio n. Funding for this study: This research was supported by funding from the French National Research Agency under grant agreeme nt ANR-22-CPJ2- 0009-01, and from the European Research Council (ER C) grant "SMHEART" under the European Union’s Horizon 2020 research an d innovation programme (grant agreement No101076351). Ethics committee - additional information: The study was approved by the Biomedical Research Ethics Committee and all partic ipants provided informed consent for participation. Author Disclosures: Victor De Villedon De Naide: Nothing to disclose Soumaya Sridi: Nothing to disclose Aurelien Bustin: Nothing to disclose Hubert Cochet: Nothing to disclose Kalvin Narceau: Nothing to disclose Thibault Boulle: Nothing to disclose Victor Nogues: Nothing to disclose Théo Richard: Nothing to disclose Baptiste Durand: Nothing to disclose Optimizing Static B1+ Shimming for Cardiac MRI at 7 Tesla: Impact on Image Quality and Myocardial Strain *A. A. Peters*, K. Fischer, M. Hundertmark, C. Scha ub, G. Bonanno, S. Schmitter, D. Günsch, C. Gräni, B. Jung; Bern/CH Purpose or Learning Objective: The purpose of this work was to: 1) establish an efficient workflow for additional 7T-specific st atic B1+ shimming procedure as short as possible, 2) investigate whether a sing le set of B1+ shim values calculated at the beginning of the exam provides re liable image quality over a region of interest (ROI), 3) investigate whether sp atial resolution or field strength has an influence on volumetric and myocard ial strain parameters.

Methods

or Background: Ten healthy volunteers underwent cine imaging at 7T and 3T on the same day. The B1+ shimming process at 7T used relative B1+ maps to minimize the coefficient of variation ( CV) within a ROI covering the heart, with the constraint to maximize excitati on homogeneity. Image quality was assessed by two experienced readers usi ng a 4-point Likert scale, and quantitative measures such as left and right ve ntricular volumes and strain parameters were evaluated.

Results

or Findings: Results showed that B1+ shimming significantly improved homogeneity, reducing the CV from 61.5% to 23.3%, while increasing transmit efficiency. Image quality at 7T exhibited more inhomogeneities compared to 3T, but these did not i mpact the clinical assessment of myocardial function. Quantitative ana lysis revealed higher myocardial mass and smaller ventricular volumes at 7T, though these differences were minimal and insignificant regardin g clincial assessment. Strain parameters were comparable between 3T and 7T .

Conclusion

In conclusion, this study demonstrates that a fast and efficient workflow for B1+ shimming at 7T can achieve diagnos tic image quality and accurate functional analysis comparable to 3T MRI.

Limitations

- Small sample size - Small number of readers Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Moritz Hundertmark: Nothing to disclose Alan Arthur Peters: Nothing to disclose Kady Fischer: Nothing to disclose Christoph Gräni: Nothing to disclose Sebastian Schmitter: Nothing to disclose Gabriele Bonanno: Nothing to disclose Christof Schaub: Nothing to disclose Dominik Günsch: Nothing to disclose Bernd Jung: Nothing to disclose Changes of myocardial extracellular volume fraction measurements in acute versus chronic disease in a large animal infa rct model *M. C. Halfmann*¹, L. Van Der Meulen², M. W. Smulde rs², H. M. J. M. Nies², F. Prizen², C. Mihl², A. Varga-Szemes³, R. J. Holta ckers², T. S. Emrich¹; ¹Mainz/DE, ²Maastricht/NL, ³Charleston, SC/US Purpose or Learning Objective: Cardiac MRI derived myocardial extracellular volume fraction (ECV) is a reproducible imaging bio marker for myocardial fibrosis. However, well-controlled evidence on the influence of the timing of the scan in relation to the contrast injection is scarc e. Therefore, this study aimed to compare ECV measurements at different time point s after contrast injection in a large animal infarct model.

Methods

or Background: Cardiac MRI was performed at 1.5T while the animals were ventilated and under general anesthesi a. Hematocrit levels were drawn directly prior to the scan. Native short-axis T1-maps of the left ventricle were acquired. 7 and 30 minutes following an iv-adm inistration of 0.2 mg/kg gadobutrol, post-contrast T1-maps at identical slic e locations were acquired. ECV was calculated for both global and separately f or the infarcted and remote myocardium. Results were compared using Pearson’s c orrelation and paired sample Student’s t-tests. Thursday Abstract-based Programme 121

Results

or Findings: A total of 13 Yorkshire pigs with balloon catheter- induced myocardial infarction were included in this prospective study. However, 7 animals died before the MRI due to sever e arrhythmias and two animals did not undergo the scan due to instability . Thus, MRI was successful in four pigs. Median time between infarction and ca rdiac MRI was 8 days [IQR 8–8]. While there was a strong correlation between measurements at both time points (r=0.94), ECV was significantly higher at 30 vs. 7 minutes (32.2±5.0% vs. 27.8±4.2%,P=.015). This was confirmed when asse ssing infarcted (56.9±11.4% vs. 43.7±9.1%,P=.018) and remote myocardium (28.4±2.8% vs. 25.5±3.2%,P=.010) separately.

Conclusion

Myocardial ECV by cardiac MRI significantly increas es with increasing time after contrast injection in a large animal infarct model. A similar effect was observed in regions with only remote myo cardium and only infarcted myocardium.

Limitations

The limitation of the study is the limited number o f animals. Funding for this study: No funding was received for this study. Ethics committee - additional information: Animal handling complied with the Dutch Law on Animal Experimentation and the Eur opean Directive on the Protection of Animals used for Scientific Purposes (2010/63/EU). This study was approved by the Experimental Animal Committee o f Maastricht University (DEC2016-002). Author Disclosures: Casper Mihl: Nothing to disclose Fritz Prizen: Nothing to disclose Robert J. Holtackers: Nothing to disclose Tilman Stephan Emrich: Advisory Board: Siemens Heal thineers Speaker: Siemens Healthineers Moritz Christian Halfmann: Nothing to disclose Lara Van Der Meulen: Nothing to disclose Martin W. Smulders: Nothing to disclose Akos Varga-Szemes: Nothing to disclose Hedwig M. J. M. Nies: Nothing to disclose CMR e' as a novel diagnostic biomarker of Asymptoma tic Left Ventricular Diastolic Dysfunction (ALVDD) *N. Mcveigh*¹, D. T. Ryan¹, F. Ryan², M. Ferre², J. Mccambridge², M. Ledwidge², K. Mcdonald², J. Dodd¹; ¹Dublin 4/IE, ²Dublin/IE Purpose or Learning Objective: Evaluate a novel cardiac MRI biomarker of diastolic dysfunction, CMRe', in pre-clinical patie nts at risk of heart failure(HF).

Methods

or Background: 236patients from the PARABLE trial (NCT04687111) underwent CMR. Mitral annular relaxat ion velocity(CMRe’) was measured at four mitral annular anchor points a nd compared with feature tracking analysis of radial, circumferential and lo ngitudinal diastolic strain rate and velocity as the gold-standard. Comparison were made with a control group of 25 age/gender-matched subjects. Comparisons were made with independent t-test, diagnostic accuracy was perform ed with receiver operator curve analysis and predictors of diastolic dysfunct ion were analysed using logistic regression.

Results

or Findings: LAVimax, LVEDVi, LVESVi and cardiac mass all demonstrated significant increases between patient and control groups (p<0.001 for all). Peak diastolic longitudinal velo city was the only significant feature tracking variable that differed between gro ups (p<0.001). LAVimax did not correlate with any measured feature tracking pa rameter when adjusted for clinical, left ventricle and left atrial parameters . In similar multivariate analysis, CMRe’ correlated with diastolic radial, circumferen tial and longitudinal strains rates, as well as radial and longitudinal diastolic velocity measurements (p<0.001). It also correlated with echo e’ (r=0.195 ,p=0.0069), LV mass (r=-.18,p=0.008), LAVimax (r=-.18,p=0.008) and BNP (r=-0.30,p<0.0001). LAVimax and total CMR e’ both exhibited high accura cy as independent predictors of diastolic dysfunction (AUC:0.89, 0.76 ,p<0.001 for both). Combined model (LAVImax and CMR e’ total) predicted diastolic dysfunction with an AUC = 0.99. LAVimax, CMR e’ and peak diasto lic longitudinal velocity were independent predictors of diastolic dysfunctio n (p<0.001 for all), adjusted for clinical and standard CMR parameters.

Conclusion

CMRe' is a precise imaging biomarker for ALVDD. Int egrating LAVimax and CMRe' holds promise in optimizing CMR m ethodologies for identifying patients at risk of diastolic dysfuncti on.

Limitations

Lack of BNP and Echo markers for the control group. Funding for this study: This trial was supported by the Health Research Board of the Government of Ireland, the European Co mmission Framework Programme 7, the Heartbeat Trust CLG, and Novartis (the manufacturer of sacubitril/valsartan). Under the terms of the grant from Novartis, the study was an investigator-led, Heartbeat Trust–sponsored clin ical trial. Ethics committee - additional information: SVUH Ethics Committee Author Disclosures: Jonathan Dodd: Nothing to disclose Kenneth Mcdonald: Nothing to disclose Fiona Ryan: Nothing to disclose David Thomas Ryan: Nothing to disclose Maria Ferre: Nothing to disclose Niall Mcveigh: Nothing to disclose Joe Mccambridge: Nothing to disclose Mark Ledwidge: Nothing to disclose Right Ventricular Function Predicts Outcome in Hear t Failure with Preserved Ejection Fraction: Strain Analysis Derive d from MR Feature- Tracking *L. Zhu*, J. He, S. Zhao, M. Lu; Beijing/CN Purpose or Learning Objective: To evaluate the association between right ventricular (RV) strain parameters derived from car diac magnetic resonance feature tracking (CMR-FT) and adverse outcomes in p atients with heart failure with preserved ejection fraction (HFpEF).

Methods

or Background: Patients with HFpEF who underwent CMR examination from January 2010 to December 2018 were retrospectively enrolled. The primary endpoint was all-cause mortal ity.

Results

or Findings: A total of 1019 consecutive patients with HFpEF (ag e 56.9 ± 12.3 years; 70% male) were enrolled in this study. During a median follow-up of 7.8 years, 103 (10.1%) patients reache d the primary endpoint. In multivariable Cox regression analysis, both RV glob al longitudinal and circumferential strain were independent predictors of the primary endpoint (HRadj per 1% increase, 1.07 [95% CI: 1.02, 1.12; P = .005] and 1.13 [95% CI: 1.05, 1.21; P < .001], respectively). The full mode l based on clinical, conventional imaging, and RV strain variables for t he primary endpoint improved the model discrimination (C-index = 0.794) compared with the baseline model based solely on clinical variables ( C-index = 0.716) and the model incorporating clinical and conventional imagi ng variables (C-index = 0.760). In receiver operating characteristic analys is for the primary endpoint, the addition of CMR-specific variables including la te gadolinium enhancement and FT RV strain yielded an improved area under the curve for the baseline models (all P < .001).

Conclusion

RV global longitudinal and circumferential strain d erived from CMR-FT were independent predictors of adverse clini cal outcomes in patients with HFpEF, providing incremental prognostic value over traditional clinical and CMR-derived risk markers.

Limitations

This was a single-center, retrospective study. Echo cardiographic parameters, including E/e’, were excluded from the Cox regression analysis due to missing values exceeding 50%. Funding for this study: The Beijing Natural Science Foundation (grant no. 7242110) Ethics committee - additional information: This study was approved by the institution ethics review board of Fuwai Hospital. Author Disclosures: Minjie Lu: Nothing to disclose Leyi Zhu: Nothing to disclose Shihua Zhao: Nothing to disclose Jian He: Nothing to disclose Non-compaction Cardiomyopathy and Diastolic dysfunc tion *S. S. D. Dereli Bulut*, S. N. Emir; Istanbul/TR ([email protected]) Purpose or Learning Objective: Non-compaction cardiomyopathy (NCCMP) is an uncommon disorder marked by increased trabecu lation of the ventricular wall and the existence of non-compacted myocardial regions. These anatomical alterations may hinder the ventricle's r elaxing capacity. Following the preliminary evaluation by echocardiogram (TTE), cardiac magnetic resonance imaging (CMR) is conducted for an in-dept h assessment. A non- compacted to compacted myocardium ratio (N/C) excee ding 2.3 substantiates the diagnosis. This study aims to assess individual s with NCCMP who underwent CMR for signs of diastolic dysfunction (D D) and to explore the correlation between disease severity and DD.

Methods

or Background: This retrospective, single-center study comprised 82 patients initially diagnosed with NCCMP based on TTE data. Cardiac Magnetic Resonance imaging was conducted utilising a 1.5 Tesla MRI scanner (Avanto, Siemens). Morphological and functional eva luations including left ventricular (LV) and right ventricular (RV) volume quantifications, cardiac output (CO), ejection fraction (EF), LV mass, peak ejection rate (PER), and peak filling rate (PFR). Correlation analysis was p erformed among these metrics. Thursday Abstract-based Programme 122

Results

or Findings: A substantial relationship was seen between the elevated N/C ratio and both PER and PFR (p<0.05; r= 0.31, r=0.21, respectively). A positive association was noted bet ween the N/C ratio and variations in LV mass assessed during the average a nd end-diastolic phases (r=0.35, p0.05), and no additional significant relationshi ps were detected.

Conclusion

The structural alterations in NCCMP may hinder vent ricular relaxation, adversely impacting diastolic function, which can be accurately identified using CMR. Timely identification of DD i s essential for enhancing patients' quality of life.

Limitations

The patient population was small, the study was pla nned retrospectively Funding for this study: The authors declared that this study has received n o financial support. Ethics committee - additional information: This study was approved by the Ethics Committee of the University Hospital Author Disclosures: Sevde Nur Emir: Nothing to disclose Safiye Sanem Dereli Dereli Bulut: Nothing to disclo se Impact of Formalin Fixation on Biventricular Parame ters in Diffusion Tensor CMR: Insights from a Miniature-Swine Model *L. Zhu*, J. Xu, H. Zhang, C. Cui, P. Sun, S. Zhao, M. Lu; Beijing/CN Purpose or Learning Objective: To examine the impact of formalin fixation on biventricular parameters derived from diffusion ten sor cardiac magnetic resonance (CMR) in a miniature-swine model, using h istological findings as the

Reference

standard.

Methods

or Background: High-resolution ex-vivo diffusion tensor CMR data of one healthy miniature-swine were acquired at bas eline, and at 5- and 9-days after the fixation with 10% neutral buffered formal in. Diffusion tensor CMR parameters were estimated using the AHA 16-segment model for the left ventricular (LV) wall, and an 8-segment model for t he right ventricular (RV) wall. Histology with hematoxylin and eosin staining was performed at 10 days following formalin fixation to assess helix angles (HA) and HA gradients.

Results

or Findings: Diffusion tensor CMR data at baseline, and 5- and 9 - days following formalin fixation were head-to-head analyzed. Subepicardial HAs became much more negative after fixation in bot h LV and RV walls, and endocardial HAs showed a positive increase, which l ed to a significant elevation in HA gradients. In the LV wall, mean dif fusivity values were slightly reduced during the first 5 days of fixation, follow ed by a marked decrease over the subsequent 4 days; while in the RV wall, these values also reduced during the first 5 days of fixation but did not change sig nificantly over the next 4 days. HAs derived from diffusion tensor CMR exhibited exc ellent consistencies with those assessed with histology, among which baseline HAs yielded the highest interclass correlation coefficient of 0.953.

Conclusion

Formalin fixation had an impact on both fiber orien tations and diffusion properties derived from diffusion tensor CMR, and baseline fiber orientations assessed before fixation showed the be st consistency with histology findings.

Limitations

This study requires larger sample sizes to enhance the robustness of the findings. Funding for this study: The Beijing Natural Science Foundation (grant no. 7242110) Ethics committee - additional information: Ethics approval was obtained from the Ethics Committee for Animal Study of Fuwai Hospital. Author Disclosures: Minjie Lu: Nothing to disclose Leyi Zhu: Nothing to disclose Peng Sun: Nothing to disclose Shihua Zhao: Nothing to disclose Jing Xu: Nothing to disclose Huaying Zhang: Nothing to disclose Chen Cui: Nothing to disclose Automated myocardial scar segmentation on joint bri ght- and black- blood late gadolinium enhancement images *T. Génisson*¹, V. De Villedon De Naide¹, B. Durand ¹, K. Narceau¹, J-D. Maes¹, P. Gut², H. Cochet¹, M. Stuber², A. Bus tin¹; ¹Bordeaux/FR, ²Lausanne/CH ([email protected]) Purpose or Learning Objective: Bright-blood sequences are used to retrieve heart anatomy information, while black-blood late g adolinium enhancement has shown promise for scar detection. However myoca rdial scar assessment, crucial for accurate patient prognostic, is time-co nsuming, operator-dependent and labor-intensive. Here, we propose an artificial intelligence-based method for automated scar segmentation on joint bright- an d black-blood LGE (SPOT) images.

Methods

or Background: A cohort of 70 patients (21% female, age range 28- 81yo) with known or suspected ischemic heart diseas e was divided into a training (70%), validation (10%) and testing (20%) set. Breath-held short-axis 2D whole-heart single-shot co-registered bright- an d black-blood SPOT, and

Reference

phase-sensitive inversion recovery (PSIR) images were collected on a 1.5T (Siemens Aera) 12min post-contrast injection . An experienced radiologist manually performed scar segmentation on SPOT and PSIR images. A transformer-based model automatically segmented l eft ventricular wall on SPOT bright-blood images and the contours were prop agated onto black-blood images. Then, a U-net automatically segmented the s car within these contours. Scar segmentation accuracy was assessed. Another ex perienced radiologist graded the scar segmentation clinical quality (Like rt scale: 0=redo; 1=major adjustments; 2=minor adjustments; 3=no adjustments needed). Concordance between scar size assessed with manual PSIR and aut omated SPOT processing was evaluated. Scar segmentation times w ere recorded.

Results

or Findings: Scar segmentation was automatically achieved on SPOT in 0.14s per slice, reaching a global Dice of 76.1%. Scar segmentations were rated 3 in 62%, 1 or 2 in 33% and 0 in only 5% of the cases. No significant differences between scar sizes were fou nd when comparing with manual PSIR processing (P<0.05).

Conclusion

The proposed method allows for fast, accurate and a utomated scar segmentation on SPOT images, achieving clinica l quality needed to better help guide therapy.

Limitations

Validation in larger cohort is warranted. Funding for this study: This research was supported by funding from the French National Research Agency under grant agreeme nt ANR-22-CPJ2- 0009-01, and from the European Research Council (ER C) grant "SMHEART" under the European Union’s Horizon 2020 research an d innovation programme (grant agreement No101076351). Ethics committee - additional information: The study was approved by the Biomedical Research Ethics Committee and all partic ipants provided informed consent for participation. Author Disclosures: Victor De Villedon De Naide: Nothing to disclose Pauline Gut: Nothing to disclose Aurelien Bustin: Nothing to disclose Hubert Cochet: Nothing to disclose Jean-David Maes: Nothing to disclose Thaïs Génisson: Nothing to disclose Kalvin Narceau: Nothing to disclose Baptiste Durand: Nothing to disclose Matthias Stuber: Nothing to disclose Impact of Reduced Spatial Resolution on Cardiac Str ain Using Deformable Registration and Feature-Tracking: A Pil ot Comparison *C. G. Glessgen*¹, T. Chitiboi², J. Wetzl², J-P. Va llee¹; ¹Geneva/CH, ²Erlangen/DE Purpose or Learning Objective: Accelerated cardiac MRI cine sequences are essential for real-time imaging and for patients un able to hold their breath but come with reduced spatial resolution. A comparison of strain measurements from high-resolution (HR) and low-resolution (LR) c ine data is presented, using two strain methods: deformable registration analysi s (DRA) and feature- tracking (FT).

Methods

or Background: Twenty patients with normal cardiac MRI findings who underwent standard SSFP cines (HR; 0.8×0.8mm2) and compressed- sensing accelerated single-shot SSFP cines (LR; 2.4 ×2.4mm2) at 1.5 T were retrospectively analyzed. Breath-hold short-axis im ages were processed using DRA (TrufiStrain v2.4, Siemens Healthineers) and FT (cvi42 v6.1, Circle CVI). Global radial (GRS) and circumferential strain (GCS ) were calculated for each image/software pair. Correlations between HR and LR images and between software were assessed using Pearson's R coefficien t; concordance was evaluated using Bland-Altman analysis.

Results

or Findings: GCS correlation between HR and LR was stronger for DRA (r = 0.93, p < 0.05) than for FT (r = 0.68, p < 0.05). GRS correlations were similar for DRA and FT (r = 0.63 and r = 0.65, respectively, p < 0.05). Bland- Altman analysis showed a mean GCS difference betwee n HR and LR of 0.02 (LoA: -3.14 to 3.19) for DRA and -4.35 (LoA: -7.00 to -1.69) for FT; for GRS, a mean difference of 18.37 (LoA: 0.60 to 36.15) for D RA and 13.81 (LoA: 6.23 to 21.38) for FT.

Conclusion

Cardiac strain measurements from LR and HR cines ar e well correlated but exhibit significant limits of agreem ent, which should be considered in clinical practice. DRA showed superio r correlation and narrower limits of agreement for GCS, while FT demonstrated narrower but significant limits for GRS.

Limitations

This pilot work is limited by its sample size and t he absence of pathological data. Funding for this study: None Ethics committee - additional information: Waiver for informed consent (ID: 01574) Thursday Abstract-based Programme 123 Author Disclosures: Jean-Paul Vallee: Nothing to disclose Carl Guillaume Glessgen: Nothing to disclose Teodora Chitiboi: Employee: Siemens Healthineers Jens Wetzl: Employee: Siemens Healthineers Precision prediction of heart failure events in pat ients with NDLVC using multi-parametric cardiovascular magnetic resonance *M. Jiang*, M. Lu; Beijing/CN ([email protected]) Purpose or Learning Objective: To assess whether left ventricular (LV) global longitudinal strain (GLS), derived from card iovascular magnetic resonance (CMR), is associated with (i) major heart failure (HF) events, and (ii) sudden cardiac death (SCD) in patients with non-dil ated left ventricular cardiomyopathy (NDLVC).

Methods

or Background: We conducted a retrospective observational cohort study of patients with NDLVC assessed by CMR, inclu ding feature-tracking to assess LV GLS and late gadolinium enhancement (LGE) . Long-term adjudicated follow-up included (i) HF hospitalizati on, heart transplantation or HF death, and (ii) SCD or aborted SCD (aSCD).

Results

or Findings: Of 386 patients with NDLVC (mean age 45 years, 258 men [66.8%], median LVEF 49% [46–54]) followed up f or a median 6.2 years, 68 patients (17.6%) experienced HF events and 15 (3 .9%) experienced SCD or aSCD. Following adjustment in a multivariable mo del, the presence of LGE and LV GLS predicted the HF events (HR 1.95; 95% CI 1.17-3.27; p=0.011 vs. per % HR 1.14, CI 1.07–1.22, p<0.001). However, LV GLS was not associated with SCD/aSCD, whereas LGE presence still was (unad justed HR 5.36, 95% CI 1.20–23.99, p=0.028). LVEF was neither associate d with HF events nor SCD/aSCD.

Conclusion

Multi-parametric CMR has utility for precision prog nostic stratification of patients with NDLVC. LV GLS strat ifies risk of progressive HF, while LGE stratifies both HF and SCD risk.

Limitations

T1 mapping and extracellular volume calculation, wh ich were more sensitive in detecting subtle myocardial alter ation and fibrosis, were not not performed systematically in patients. Additiona l investigations are required to assess the potential diagnostic and prognostic s ignificance of mapping techniques within this particular context. Funding for this study: None Ethics committee - additional information: The Institutional Review Board approved this study, and written informed consent w as waived due to the retrospective nature of the study. Author Disclosures: Minjie Lu: Nothing to disclose Mengdi Jiang: Nothing to disclose 16:00-17:30 Research Stage 2 Research Presentation Session: Emergency Imaging RPS 1117 Emergency radiology: new technologies and workload challenges Moderator A. Platon; Geneva/CH ([email protected]) Non-invasive bullet characterization by material de composition in Photon counting CT *B. M. Schaarschmidt*¹, J. Hegmanns¹, J. Wulff¹, V. Haase², S. Faby², C. Bäumer¹, S. Zensen¹, J. Haubold¹, B. Hartung¹; ¹ Essen/DE, ²Forchheim/DE Purpose or Learning Objective: Gunshot deaths are a worldwide health concern. Especially in patients with lodged bullets or forensic analyses, image- based bullet characterization is of major interest. Therefore, the present study investigated bullet differentiation from different materials using photon counting computed tomography (PCCT).

Methods

or Background: Using a new research scan mode, six lead and three brass bullets were analyzed on a NAEOTOM Alph a PCCT system (Siemens Healthineers, Forchheim, Germany). For eac h scan, a set of four images was reconstructed using four different energ y thresholds of the detector (20, 55, 72, and 90 keV). Two independent readers placed three circular regions of interest (ROI) on the 20 keV th reshold images on an OsiriX Workstation (Pixmeo SARL, Bernex, Switzerland). The se ROIs were then automatically duplicated to the other threshold ima ges. Dual energy indices (DEIs) were computed for the energy threshold pairs of 20/90 keV, 55/90 keV, and 72/90keV based on the measured HUmean and HUmax values.

Results

or Findings: DEIs of lead and brass projectiles differed signifi cantly, most notably for the 20/90 keV DEI HUmean (Qr40): l ead: -0.085±0.021, brass: 0.024±0.048, p<0.001; HUmax (Qr40): lead: -0.093±0.011, brass: 0.023±0.057, p<0.001). For the 55/90 keV and 72/90 keV DEIs, differences between the two projectile materials decreased, but remained statistically significant.

Conclusion

In the DEIs derived from different energy threshold images obtained by PCCT, significant differences could be observed between lead and brass bullets. Therefore, PCCT might be a potential technique for bullet

Material

analysis in both clinical and forensic ima ging.

Limitations

Phantom study only. Funding for this study: The study was performed at Siemens Healthineers facilities in Forchheim, Germany, with support from Viktor Haase and Sebastian Faby (employees of Siemens Healthineers A G). Ethics committee - additional information: As no human or animal subjects were investigated in this study, no approval by the local ethics committee was necessary. Author Disclosures: Benno Hartung: Nothing to disclose Viktor Haase: Employee: Siemens Healthineers AG Benedikt Michael Schaarschmidt: Nothing to disclose Johannes Haubold: Nothing to disclose Sebastian Zensen: Nothing to disclose Christian Bäumer: Nothing to disclose Jan Hegmanns: Nothing to disclose Jörg Wulff: Nothing to disclose Sebastian Faby: Employee: Siemens Healthineers AG Virtual monoenergetic images from photon-counting d etector CT in thoracic trauma: Improved discriminability of sever e lung injury and atelectasis at low-keV energy levels H-L. Kaatsch, *B. V. Becker*, D. Dillinger, J. Piec hotka, C. Schreyer, R. Schwab, D. Overhoff, S. Waldeck; Koblenz/DE Purpose or Learning Objective: The aim of this study was to assess the usability of virtual monoenergetic images (VMIs) de rived from photon-counting detector CT (PCD-CT) for discriminability of severe lung injury and atelectasis after thoracic trauma.

Methods

or Background: We retrospectively selected 20 polytraumatized patients, who underwent contrast-enhanced whole-bod y PCD-CT and showed simultaneous presence of trauma-associated atelecta sis and pulmonary injury. VMIs were reconstructed from 40 to 120 keV at 10 ke V increments. Quantitative image analysis was performed based on density measurements and the calculation of injury-to-atelectasis contra st-to-noise ratio (CNR)). Three radiologists rated subjective discriminability, noi se perception and overall image quality by means of a 5-point Likert scale.

Results

or Findings: CT values for severe lung injury and atelectasis di ffered significantly at each keV level (p < 0.001) with a gradual decrease for atelectasis from 342 ± 97 HU at 40 keV to 69 ± 15 HU at 120 keV and a near constant behavior for severe lung injury from 42 ± 49 HU at 40 keV to 44 ± 22 at 120 keV. In line with this, the highest injury-t o-atelectasis CNR was achieved at 40 keV (3.97) with a continuous decline down to 120 keV (1.21). CNR reached no significant differences between 40 and 5 0 keV as well as 110 and 120 keV (p >0.05), whereas all other pairings were significantly different (p<0.05). The best ratings for subjective discrimin ability were reported for VMIs at 40 keV, whereas VMIs at 60-70 keV provided the o ptimal noise perception and overall image quality.

Conclusion

Low-keV VMIs at 40-50 keV from PCD-CT considerably improved the discriminability of severe lung injury and atel ectasis after thoracic trauma.

Limitations

Lack of a validated reference standard for lung inj ury vs. atelectasis, retrospective study design and small s ample size. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study is a retrospective single-centre analysis that has been approved by th e local ethics committee of the chamber of physicians Rhineland-Palatinate in M ainz, Germany (number 2022-16314). Author Disclosures: Joel Piechotka: Nothing to disclose Daniel Overhoff: Nothing to disclose Hanns-Leonhard Kaatsch: Nothing to disclose Robert Schwab: Nothing to disclose Benjamin Valentin Becker: Nothing to disclose Christof Schreyer: Nothing to disclose Stephan Waldeck: Nothing to disclose Daniel Dillinger: Nothing to disclose Thursday Abstract-based Programme 124 Body composition parameters in initial CT imaging o f mechanically ventilated trauma patients: Single-centre observati onal study *H-J. Meyer*, T. Dermendzhiev, T. Denecke, M. Struc k; Leipzig/DE ([email protected]) Purpose or Learning Objective: Body composition parameters provide relevant prognostic significance in critical care c ohorts and cancer populations. Published results regarding polytrauma patients are inconclusive to date. The goal of this study was to analyse the role of body composition parameters in severely injured trauma patients.

Methods

or Background: All consecutive patients requiring emergency tracheal intubation and mechanical ventilation befo re initial computed tomography (CT) at a level-1 trauma centre over a 1 2-year period (2008-2019) were reanalysed. The analysis included CT-derived b ody composition parameters based upon whole-body trauma CT as progn ostic variables for 30- day mortality, intensive care unit length of stay ( ICU LOS) and mechanical ventilation duration.

Results

or Findings: Four hundred seventy-two patients (75% male) with a median age of 49 years, median injury severity scor e of 26 and 30-day mortality rate of 22% (104 patients) met the inclus ion criteria and were analysed. Regarding body composition parameters, 23 1 patients (49%) had visceral obesity, 75 patients had sarcopenia (16%) and 35 patients had sarcopenic obesity (7.4%). After adjustment for sta tistically significant univariable predictors age, body mass index, sarcop enic obesity, visceral obesity, American Society of Anesthesiologists clas sification ≥3, injury severity score and Glasgow Coma Scale ≤ 8 points, the Cox proportional hazard model identified sarcopenia as significant prognostic fac tor of 30-day mortality (hazard ratio 2.84; 95% confidence interval 1.38-5. 85; P = 0.004), which was confirmed in Kaplan-Meier survival analysis (log-ra nk P = 0.006).

Conclusion

In a multivariable analysis of mechanically ventila ted trauma patients, CT-defined sarcopenia was significantly a ssociated with 30-day mortality whereas no associations of body compositi on parameters with ICU LOS and duration of mechanical ventilation were obs erved.

Limitations

It is a single-center study with a retrospective de sign. Funding for this study: None Ethics committee - additional information: Approval by the ethics committee at the Medical Faculty, Leipzig University, Leipzig , Germany (IRB00001750, project ID 441/15ek, September 14, 2020) Author Disclosures: Timm Denecke: Nothing to disclose Manuel Struck: Nothing to disclose Tihomir Dermendzhiev: Nothing to disclose Hans-Jonas Meyer: Nothing to disclose AI Denoising Enhances Image Quality and Diagnostic Accuracy While Reducing Radiation Exposure in Prospective LDCT Sca ns for Acute Abdomen *A. S. Brendlin*, U. Schmid, S. Afat; Tübingen/DE Purpose or Learning Objective: To identify the optimal low-dose CT protocol that minimizes radiation exposure while preserving high image quality and diagnostic accuracy in patients presenting with acu te abdomen.

Methods

or Background: A prospective, randomized study was conducted with 180 patients assigned to one of three CT proto cols. Each protocol included high-dose (HD) and low-dose (LD) scans rec onstructed using Iterative Reconstruction (IR2) and AI Denoising (AID). Subjec tive image quality was assessed by radiologists evaluating diagnostic conf idence, contrast, and sharpness. Objective image quality metrics, includi ng noise and contrast-to- noise ratio (CNR), were measured. Diagnostic accura cy was evaluated through sensitivity and specificity calculations for detect ing abdominal pathologies.

Results

or Findings: Protocol 2 LD achieved the lowest Size-Specific Dos e Estimate (SSDE) at 4.83 mGy, significantly reducing radiation dose compared to Protocols 1 and 3 (P < 0.001). AID significantly enhanced subjective image quality in LD scans across all protocols, with Prot ocols 2 and 3 achieving ratings comparable to HD scans with IR2. Objective assessments showed that AID substantially reduced image noise and increased CNR in LD scans, with Protocol 2 LD exhibiting the highest CNR. In outcom e analysis, both Protocols 1 and 2 demonstrated 100% sensitivity and specifici ty in LD scans with AID, eliminating false negatives and matching the diagno stic performance of HD scans. Protocol 3 maintained high diagnostic accura cy across all doses and reconstruction methods.

Conclusion

Protocol 2 with AID emerges as the most effective l ow-dose CT strategy, offering significant radiation dose reduc tion while maintaining superior image quality and diagnostic accuracy.

Limitations

- Single-Center Study with Limited Sample Size - Po tential Observer Bias in Subjective Assessments - Lack of L ong-Term Outcome Data - Equipment and Protocol Specificity Funding for this study: None Ethics committee - additional information: University Hospital Tuebingen Author Disclosures: Andreas Stefan Brendlin: Nothing to disclose Saif Afat: Nothing to disclose Ulrich Schmid: Nothing to disclose Can perfusion-derived cerebral CT angiography repla ce routine cerebral CT angiography by using artificial intelligence ite rative reconstruction for acute ischemic stroke patients? J. Xie¹, *T. Wang*², G. Zhang², J. Huang¹, M. Wang¹ , Y. Lin¹; ¹Taizhou/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: To investigate the feasibility and realiability of replacing routine cerebral CT angiography (CTA) with CT perfusion (CTP)- derived cerebral CTA by using artificial intelligen ce iterative reconstruction (AIIR) for acute ischemic stroke (AIS) patients.

Methods

or Background: Forty-nine patients (33-93 y, male: 35) with AIS undergoing a cerebral CTP and a routine CTA were pr ospectively collected. CTA images derived from CTP at the arterial phase ( 100kVp/150mAs) were reconstructed with hybrid iterative (Group A1) and AIIR (Group A2), whereas routine CTA images (100kVp/ref. 200mAs) were obtain ed with hybrid iterative reconstruction (Group B). Two radiologists independ ently located the responsible vessels, with digital subtraction angio graphy (DSA) as reference standard. They further graded the image noise, shar pness of the vascular edge, small vessel visibility, and overall diagnosa bility using a five-point Likert scale (1: poor, 5: excellent). Objective parameters , including the SNR and CNR of the internal carotid artery, the middle cerebral artery, and the basilar artery, were also calculated.

Results

or Findings: The diagnosis of responsible vessels was consistent by both radiologists, where the diagnostic accuracy of Groups A2 and B were comparable (47/49, 95.92%) and higher than that in Group A1 (43/49, 87.76%). The inter-observer agreement was excellent (κ = 0.84).All subjective scores were significantly higher in Group A2 than t hose in Groups A1 and B (all p < 0.017), especially for the small vessel vi sibility (4.9 ± 0.2 vs.2.8± 0.4 vs. 4.2 ± 0.5). AIIR significantly reduced noise, leading to a significantly higher SNR and CNR for all arteries in Group A2 than those in Groups A1 and B (all p < 0.017).

Conclusion

Perfusion-derived cerebral CTA by use of the AIIR a chieves comparable diagnostic image quality to that of the routine cerebral CTA for AIS patients.

Limitations

Not applicable Funding for this study: No funding was received for this study. Ethics committee - additional information: Not applicable Author Disclosures: Jicheng Xie: Nothing to disclose Tiantian Wang: Nothing to disclose Guozhi Zhang: Nothing to disclose Minke Wang: Nothing to disclose Jinbiao Huang: Nothing to disclose Youyou Lin: Nothing to disclose Real life performance of a commercially available A I for post-traumatic intracranial haemorrhage detection on CT-scans: a s upportive tool *L. Mabit*, G. Herpe; Poitiers/FR ([email protected]) Purpose or Learning Objective: Investigate the real-world performance of qER.AI, an artificial intelligence-based CT haemorr hage detection tool, in a post-traumatic population.

Methods

or Background: Retrospective monocentric observationnal study of a dataset of consecutively acquired head CT scans a t the emergency radiology unit to explore a brain trauma. AI performance was compared to groundtruth determined by expert consensus. A subset of nighshi ft cases with radiological report of junior resident was compared to AI result s and groundtruth.

Results

or Findings: 682 head CT scan were analyzed. AI demonstrated a sensitivity of 88.8% and specificity of 92.1% overa ll, with a positive predictive value of 65.4% and a negative predictive value of 9 8%. AI's performance was comparable to junior residents in detecting ICH, wi th the latter showing a sensitivity of 85.7% and a high specificity of 99.3 %. Interestingly, the AI detected two out of three ICH cases missed by junio r residents. When AI and residents performances were combined, the sensitivi ty improved to 95.2%, and the overall accuracy reached 98.8%.

Conclusion

This study shows a better performance of AI and rad iologist resident associated than each one alone. These resu lts are encouraging to rethink the radiological workflow and the future of triage of this large population of brain traumatised patients in emergency unit. Thursday Abstract-based Programme 125

Limitations

The limitations of the study are exclusion of some CTs due to logistics issues (potential selection bias), and a low number of positive ICH cases in the nightshift subset. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was reviewed and approved by the Institutional Review Board of CERIM (CRM-2401-400). Author Disclosures: Guillaume Herpe: Nothing to disclose Léo Mabit: Nothing to disclose Increasing workload in emergency radiology: A decad e-long trend analysis using Large Language Models *M. T. Meyer*, E. M. Merkle, T. Heye; Basel/CH Purpose or Learning Objective: This study aims to identify trends explaining the significant rise in CT examinations ordered by the emergency department (ED) of a tertiary hospital in the last decade (201 4 to 2023) using large language models (LLM).

Methods

or Background: Radiology Information System data were extracted for CT scans performed on ED patients between 2014- 2023. The open-source Llama 3.1 405B LLM was used to classify each report into YES (expected pathology confirmed), NO (no relevant pathology), a nd OTHER (unexpected but relevant pathology found). Trends were analyzed by body-region, focusing on patients older than 65 years, as this group show ed the highest increase in imaging.

Results

or Findings: Over the past decade, the number of CT scans increased 2.2 times for ED patients, with a more pr onounced rise for patients over 65 years (2.4). Preliminary results show that the LLM achieved an accuracy of 84% compared to a manual review of 500 randomly sampled reports. Scans of abdomen/pelvis (YES in 49-60%) an d thorax/abdomen/pelvis (YES in 54-63%) remained stable with minor annual f luctuation. CT scans of the skull in polytrauma patients showed steady incr ease in NO findings from 43% (2016) to 59% (2022). CT of the skull showed a gradual increase in NO findings from 57% (2015) to 66% (2022). CT of the t horax remained relatively stable (YES 41-46% from 2014-2019), with notable pe aks during the COVID- pandemic (52% in 2021).

Conclusion

Diagnostic yield across different organ groups vari es greatly with high proportions of NO findings in skull CTs (typic ally to rule-out bleeding). Over time, unremarkable CT scans, particularly of t he skull, have increased, while trends for other body regions (e.g., thorax/a bdomen/pelvis), were less distinct.

Limitations

The results are preliminary. Accuracy of the LLM is limited, which might lead to over-/underestimation of trends. Funding for this study: None Ethics committee - additional information: The study is retrospective and General Consent is available. Author Disclosures: Tobias Heye: Nothing to disclose Elmar M. Merkle: Nothing to disclose Manfred Tobias Meyer: Nothing to disclose Increasing On-Call Workload for Radiology Trainees: A Five-Year Analysis in a Tertiary Referral Centre *P. Rohan*, H. Briody, C. Mccarthy, M. M. Morrin; D ublin/IE ([email protected]) Purpose or Learning Objective: The demand for acute diagnostic radiology services during on-call hours continues to increase , placing strain on radiology trainees. This study aims to assess the on-call wor kload for Radiology Specialist Registrars (SpRs) over a five-year perio d at a model 4 tertiary referral centre in Ireland.

Methods

or Background: A retrospective review was conducted, analysing the volume of computed tomography (CT) scans perfor med during on-call hours (weekdays 17:00-08:00, weekends, and public h olidays) across three months—January, July, and November—from 2019 to 202 3. Data were categorized by study type, focusing on Emergency De partment (ED) requests and key scan types, including non-contrast brain CT (NCB), CT abdomen/pelvis (CTAP), CT thorax/abdomen/pelvis (CT TAP), and cerebral stroke protocol ("FAST"). Statistical analysis incl uded percentage changes and paired t-tests.

Results

or Findings: Between 2019 and 2023, there was a 25% increase in the total number of on-call CTs performed (p = 0.09 8), with a significant 46% rise in ED-related studies (p < 0.05). Significant increases were observed in NCBs (38%, p < 0.05) and CT TAPs (220%, p < 0.05). The number of CTs performed after midnight increased by 82% (p = 0.05 5), while FAST protocol CTs rose by 41% (p = 0.056), reflecting the growing incidence of stroke in Ireland.

Conclusion

The increasing on-call workload for Radiology SpRs, particularly driven by ED requests and the rising demand for str oke imaging, underscores the need for careful planning and resource allocati on to manage future demand. Addressing trainee workflows is essential t o sustaining radiology services while maintaining high standards of care.

Limitations

This retrospective study is limited to on-call CT w orkload. Other modalities like ultrasound, MRI, and X-rays, as wel l as communication burdens with clinical teams, were not assessed and should b e explored in future studies. Funding for this study: No funding was received for this study. Ethics committee - additional information: Ethics committee approval not required. Author Disclosures: Cathal Mccarthy: Nothing to disclose Martina M Morrin: Nothing to disclose Hayley Briody: Nothing to disclose Pat Rohan: Nothing to disclose Trends in CT examination utilization in the emergen cy department during and after the COVID-19 pandemic *F. Kempter*, D. Jäschke, J. Vosshenrich, B. Ceresa , T. Heye; Basel/CH Purpose or Learning Objective: To investigate the impact of COVID-19 pandemic-related measures on trends and volume in C T examinations requested in the emergency department.

Methods

or Background: CT examinations of the head, chest, and/or abdomen-pelvis (n=161,008), and chest radiographs ( n=113,240) performed at our tertiary care hospital between 01/2014- 12/2023 were retrospectively analyzed. CT examinations (head, chest, abdomen, du al- region and polytrauma) and chest radiographs requested by the emergency department during (03/2020-03/2022) and after the COVID-19 pan demic (04/2022- 12/2023) were compared to a pre-pandemic control pe riod (02/2018-02/2020). Analyses included CT examinations per emergency dep artment visit, and prediction models based on pre-pandemic trends and inpatient data. A regular expressions text search algorithm determined the mo st common clinical questions.

Results

or Findings: The usage of dual-region and chest CT examinations were higher during (+116,4% and +115.8%, respective ly; p<0.001) and after the COVID-19 pandemic (+88,4% and +70.7%, respectiv ely; p<0.001), compared to the control period. Chest radiograph us age decreased (-54.1% and -36.4%, respectively; p<0.001). The post-pandem ic overall CT examination rate per emergency department visit inc reased by 4.7%. The prediction model underestimated (p<0.001) the growt h (dual-region CT: 22.3%, chest CT: 26.7%, chest radiographs: -30.4%), and th e rise (p<0.001) was higher compared to inpatient data (dual-region CT: 54.8%, chest CT: 52.0%, CR: -32.3%). Post- pandemic, the number of clinical questions to rule out “pulmonary infiltrates”, “abdominal pain” and “infe ction focus” increased up to 235.7% compared to the control period.

Conclusion

Following the COVID-19 pandemic, chest CT and dual- region CT usage in the emergency department experienced a dis proportionate and sustained surge compared to pre-pandemic growth.

Limitations

Single-center, retrospective design limits generali zability. CT as standard imaging for suspected COVID-19 pneumonia ( 2020–2022) may explain the shift. Only clinical questions, not dia gnoses, were analyzed. Despite a large sample size, data mislabeling is po ssible. Funding for this study: Not applicable. Ethics committee - additional information: The local ethics committee of northwestern and central Switzerland (EKNZ, project ID 2022-01016) approved this study. Author Disclosures: Tobias Heye: Nothing to disclose Dominik Jäschke: Nothing to disclose Felix Kempter: Nothing to disclose Benjamin Ceresa: Nothing to disclose Jan Vosshenrich: Nothing to disclose An 18-year Retrospective Analysis of Urgent Inpatie nt and Emergency CT Reporting at a UK Local General Hospital *A. Gmati*, Z. Foster, M. Mobley; Warwick/UK ([email protected]) Purpose or Learning Objective: Radiology has transformed how acute medicine is practiced, with growing imaging demands placing significant strain on radiology departments in the UK. This study retr ospectively examines 18 years of urgent inpatient and emergency CT reportin g at South Warwickshire University NHS Foundation Trust (SWFT) to assess tr ends in workload, staffing, and operational changes over this period. Thursday Abstract-based Programme 126

Methods

or Background: Data were extracted from the Radiology Information System (RIS) at SWFT, encompassing approximately 14 0,000 urgent scans performed since 2007. These were analysed to show c hanges over time such as seasonal variability and peaks in workload throu ghout the week, as well as the overall trend for increased imaging.

Results

or Findings: Urgent CT workload at SWFT has significantly increa sed over the study period, reflecting national trends. This timescale covers a period of rapid growth and operational change at SWFT, suc h as the introduction of weekend shifts and the implementation of outsourced teleradiology for out-of- hours reporting. Typical daily reporting numbers ha ve increased from 12 scans per day in 2007 to over 50 in 2024. Staffing has no t kept pace, with WTE consultant numbers dropping from 12 in 2019 to 9.5 in 2023. Despite outsourcing, many departments continue to struggle with growing diagnostic demands.

Conclusion

Across the UK, CT usage surged from 1 million scans in 1997 to over 6.7 million in 2021. Rising CT demand at SWFT reflects broader national trends, with increasing pressure on radiology servi ces. The UK’s Royal College of Radiologists predicts a shortage of 1,669 radiol ogists by 2025, highlighting a national critical staffing issue. Operational chang es such as weekend working may provide temporary relief but place increasing d emands on the same pool of radiologists. Systemic reforms and increased inv estment are essential to meet future diagnostic demands.

Limitations

N/A Funding for this study: N/A Ethics committee - additional information: N/A Author Disclosures: Zoe Foster: Nothing to disclose Mark Mobley: Nothing to disclose Aimen Gmati: Nothing to disclose Is population aging behind the increasing workload in emergency radiology? *J. Sarnecki*, M. T. Meyer, E. M. Merkle, T. Heye; Basel/CH ([email protected]) Purpose or Learning Objective: To investigate the trends in imaging in emergency patients over the last decade.

Methods

or Background: 190'028 emergency CT examinations from 1/1/2015 to 31/12/2023 of patients aged 18-100 year s, performed within 24h of the request, were included. The examination time wa s stratified into dayshift (8-17:30h), late-shift (17:30-22h), night-shift (22 -8h) and weekend-dayshift (Saturday/Sunday 8-17:30h). The patients were divid ed into two age groups (18-64 versus 65-100 yrs.) and binned using 5-year increments. The percentage increase in imaging numbers was calculat ed based on 2015.

Results

or Findings: A sharp overall increase in CT imaging numbers was observed in 2020 (18-64 yrs.: 127.0% vs. 65-100 yrs .: 155.3%), peaking in 2022 (18-64 yrs.: 182.4% vs. 65-100 yrs.: 235.6%). The mean age in the 18-64 yrs. group increased from 44.8 (2015) to maximally 46.3 (2022), but did not increase for 65-100 yrs. (79.2 vs. 79.3 yrs.). The day-shift imaging numbers showed a linear steady increase, whereas the night- shift examinations showed the strongest increase, peaking in 2023 with 267.5% (18-64 yrs) and 403.0% (65-100 yrs) followed by changes in weekend and lat e-shift CT numbers (18-64 yrs.: 119.8%;127.1% vs. 65-100 yrs.: 173.1%; 139.0% ). Between 2020-2023 for the younger age group, age-bins 55-59 (199.5-26 0.9%) and 60-64 (140.8- 200.6%) showed the largest relative increase compar ed to 2015 for late-, night- and weekend-dayshift CTs combined (all age bins 18- 64yrs: 30-112.2%). In the 65-100 yrs. group, the largest relative increas e was seen for patients 95- 100 years old (318.2-360.0%) compared to other 5-ye ar age bins (145.7- 276.8%). Chest, Chest-Abdomen-Pelvis and Polytrauma CT showed the greatest increase.

Conclusion

The results support that a change in imaging was tr iggered and sustained by the COVID-19 pandemic with a dispropor tionate increase in CT imaging volume overall and especially for older pat ients during non-day-shifts.

Limitations

Retrospective, single-center study. Funding for this study: No funding. Ethics committee - additional information: Not applicable Author Disclosures: Tobias Heye: Nothing to disclose Elmar M. Merkle: Nothing to disclose Jędrzej Sarnecki: Nothing to disclose Manfred Tobias Meyer: Nothing to disclose The increase and inappropriate requests for cranial CT scans in emergency departments contribute to overuse and dec rease test accuracy *A. Martínez López*, G. Pagán Vicente, E. Otón Gonz ález, H. Ortiz Mayoral, E. C. Cotillo Ramos, M. S. Canales, B. Molina-Lozan o, J. Plasencia Martínez, I. Santiago Suárez; Murcia/ES ([email protected]) Purpose or Learning Objective: The number of non-traumatic urgent cranial computed tomography (NT-UCCT) is exponentially incr easing but limited research has been conducted on the quality of clini cal justification. We aimed (1) to assess how clinical information in the elect ronic NT-UCCT request agreed with that provided in the emergency departme nt discharge summary and (2) to analyze the potential effect of those di screpancies on the NT-UCCT overload.

Methods

or Background: Patients undergoing NT-UCCT in 2017-2021 were randomly selected for this retrospective study. Sig ns and symptoms (S/S) in electronic request and emergency department dischar ge summary, acute and relevant computed tomography (CT) findings, and fin al diagnosis at emergency department discharge summary were collected. Concor dance between both and their association with CT findings and final di agnosis were analyzed.

Results

or Findings: We recruited 156 patients: 80 men; mean age, 55. Acute, relevant CT findings were detected in 28 cas es (17.9%). The final diagnosis was neurological disease 46 (29.5%), non- neurological disease 58 (37.2%), and no definitive diagnosis 51 (32.7%). Fu ll agreement between the electronic request and emergency department dischar ge summary occurred in 36 patients (23.1%). Motor deficit was the most fre quent false positive S/S (18; 11.54%), having low positive predictive value (30.3 0%; 95%CI 15.59-48.71%) and worst association with acute relevant CT findin gs than when true positive (OR 2.54; 95%CI 0.04-6.21 vs. OR 6.26, 95%CI 2.21-1 7.78). Nausea/vomiting was the third most common false negative S/S (13; 1 0.26%) and reduced the likelihood of acute relevant CT findings (OR 0.126; 95%CI 0.016-0.971; p = 0.020).

Conclusion

Discrepancies between electronic request and emerge ncy department discharge summary were observed in >75% of patients, leading to unnecessary NT-UCCT tests.

Limitations

Differences in how clinical information is recorded , the variability among physicians, results not applicable to large h ospitals, small sample size limited detailed analysis. Funding for this study: No Ethics committee - additional information: The study was approved by the ethics committee. Author Disclosures: Estefania Corina Cotillo Ramos: Author: Author Elena Otón González: Author: Author Isabel Santiago Suárez: Author: Author Marta Sánchez Canales: Author: Author Juana Plasencia Martínez: Advisory Board: Author Gonzalo Pagán Vicente: Author: Author Belén Molina-Lozano: Author: Author Andrea Martínez López: Author: Author Herminia Ortiz Mayoral: Author: Author 16:00-17:30 Research Stage 3 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 1105 Artificial intelligence and planet radiology: the green machine Moderator R. Mirón Mombiela; Herlev/DK ([email protected]) How do radiology department carbon footprints contr ibute to climate change? *S. D. Jagadeesha*¹, R. Botchu²; ¹Mysuru/IN, ²Birmi ngham/UK ([email protected]) Purpose or Learning Objective: The purpose of this study is to investigate paper usage in the radiology department of a single hospital institution over the last three years to forecast paper usage up to 2050 . Thursday Abstract-based Programme 127

Methods

or Background: This retrospective study was performed in the radiology department of our tertiary orthopedic hos pital. The study included forms used for diagnostic and interventional proced ures in various departmental modalities. Diagnostic procedures requ ire one to three forms and interventional procedures require three forms each. Based on the established ratio that 1.2 trees are cut for every 10,000 paper s used, the study calculated the number of trees cut annually over the past thre e years and projected paper usage and tree loss until 2050

Results

or Findings: Paper usage was distributed between diagnostic and interventional procedures, with 67% used in diagnos tics and 33% in interventions. The corresponding number of trees cu t during this period amounted to 53.7 trees, with 47.4 trees for diagnos tic procedures and 6.4 trees for interventional procedures. A total of 57.8 tree s for diagnostic procedures and 11.7 trees for interventional procedures were f orecasted to be cut annually from 2024 to 2050, cumulatively being 1227 trees by the year 2050.

Conclusion

Our individual department had a significant contrib ution from paper usage in the carbon footprint of the departme nt. Adoption of digitalized appointment, prescribing and patient records is imp ortant in reducing this and achieving NHS net-zero targets.

Limitations

The use of paper for forms, there are other signifi cant sources of paper consumption within the department. For exampl e, extensive paper packaging used for interventional consumables, and tissue paper used for various applications, such as covering ultrasound a nd CT couches, are also contributing to the overall paper usage in the radi ology department. This has been excluded in the study. Funding for this study: Not applicable Ethics committee - additional information: This has obtained ethical committee clearance from the Hospital. Author Disclosures: Rajesh Botchu: Nothing to disclose Sushmitha Devihalli Jagadeesha: Nothing to disclose AI-driven green gains: Enhancing efficiency with en vironmental benefits in Imaging *P. Strouhal*¹, N. Khan², A. Heathcote¹, M. Darwish ³, S. Persichini³, B. Miles¹, M. Trumann⁴, I. Farid³; ¹Warwick/UK, ²Dubai/AE, ³Chalfont St G iles/UK, ⁴Freiburg/DE ([email protected]) Purpose or Learning Objective: Alliance Medical Ltd (AML) provides diagnostic imaging for 800,000 NHS patients annuall y via networked facilities. Growing concerns over operational and energy effici encies in 2022 prompted AML’s implementation of GE HealthCare’s Imaging360 solution. 18 months on, we showcase how such Artifical Intelligence (AI) dr iven solutions are pivotal in refining patient flows, scheduling, staffing, energ y usage and logistics management within imaging services.

Methods

or Background: Integration comprised 6 separate Imaging360 components utilising data from various sources, inc luding HL7, DICOM, Business Intelligence software and CSV extracts; mo bile and static CT and MRI scanners were incorporated from multiple sites across England (with PET- CT scanners now being onboarded).

Results

or Findings: Using predictive analytics, AML reduced missed appointments from 17% to 3% per week, improving res ource utilisation. Optimising protocols and schedules done on-cloud al lowed reduced senior staff travel (approx. 380-480 km/month) and time (3 7.5hr/week) to manage scanner protocols; and significantly increased scan ner efficiency: - MRI: Throughput rose from 21 to 27 scans per day (+33%), with kWh/patient reduced from 15.5 to 11.8. This saved 3.7 kWh per e xam —enough to power 45 average households annually. One MRI site increa sed throughput by 43%, achieving 410 exams per month increase. - CT: scann er throughput improved by 256 scans per month average, cutting idle time; and reducing protocol variability for CT chest, abdomen, pelvis from 47 t o 15 standardised protocols, with related radiation doses lowered from 500 to 35 0 mGy.cm. Increased throughput was achieved with no extra staff or equi pment.

Conclusion

Integrating AI into radiology workflows allows tran sformative changes not only of operational efficiencies and co st savings, but improved sustainable practice. Going forward, further eco-fr iendly innovations could enhance both performance and sustainability across the healthcare imaging sector.

Limitations

Imaging360 optimised for GE HealthCare scanners Funding for this study: GE HealthCare supporting implementation of AI platform Ethics committee - additional information: N/A Author Disclosures: Naeem Khan: Employee: GE HealthCare Peter Strouhal: Board Member: Alliance Medical Brad Miles: Employee: Alliance Medical Stefano Persichini: Employee: GE HealthCare Mariam Darwish: Employee: GE HealthCare Imran Farid: Employee: GE HealthCare Markus Trumann: Employee: GE HealthCare Ann Heathcote: Employee: Alliance Medical Balancing Sustainability and Performance: Evaluatin g Energy Use, Carbon Footprint and Task Performance of Locally ru n Large Language Models for Radiology Report Simplification *A. Gupta*, R. Dheeka, R. Kumar, A. Rastogi, H. Mal hotra, K. Rangarajan; New Delhi/IN ([email protected]) Purpose or Learning Objective: To investigate tradeoffs between performance and energy use when using different loc ally-run large language models (LLMs) and prompts for patient-centric simpl ification of radiology reports.

Methods

or Background: This study evaluated three different open-source LLMs (Meta’s Llama 3.1-8B, Microsoft’s Phi-3.5-Mini and Mistral-7B) using five different prompts to simplify 50 computed tomograph y report impressions, collected from our tertiary-care oncology centre. M odels were run on a local workstation with graphic processing unit. Energy us e (in watt-hours) and carbon emissions (in grams) for each inference, wer e measured using an open-source tool (CodeCarbon). Readability of origi nal and generated simplified reports was quantitatively assessed usin g an average score of four readability indices. LLM performance for simplifica tion task was measured as difference in readability scores between original r eports and LLM-generated reports. Energy efficiency ratios (performance per watt-hour) and carbon footprint (performance per gram of emissions) were calculated for each model- prompt combination.

Results

or Findings: Llama-prompt 5 (multi-shot learning) demonstrated t he highest task performance (7.36), best energy effici ency ratio (31.89/Wh), and least carbon footprint (44.70/g). Phi-prompt 5 achi eved high simplification (6.14) and energy efficiency (25.87/Wh). For Mistra l, prompt 1 (no context) was optimal (2.15/Wh and 3.01/g), but performance (1.16 ) lagged behind Llama and Phi. Friedman test revealed significant differe nces among readability scores (p < 0.001), with post-hoc Wilcoxon tests sh owing significant improvements for Llama and Phi over the original an d Mistral, and Llama outperforming Phi (adjusted p < 0.0033).

Conclusion

Different LLM-prompt combinations showed variabilit y in energy use, carbon emissions, and simplification task perf ormance. These results highlight the importance of LLM-prompt combination selection for medical applications, balancing sustainability and performa nce.

Limitations

Development of test prompts has inherent potential for subjectivity. Apart from prompt engineering, we did not use other accuracy improving techniques like retrieval augmented gener ation. Funding for this study: None Ethics committee - additional information: Study approved by the Institute Ethics Committee All India Institute of Medical Sci ences, New Delhi (Ref. No. - IEC-343/15.06.2023) Author Disclosures: Krithika Rangarajan: Nothing to disclose Rohit Kumar: Nothing to disclose Ashish Rastogi: Nothing to disclose Amit Gupta: Nothing to disclose Hema Malhotra: Nothing to disclose Rahul Dheeka: Nothing to disclose Greenhouse gas emissions due to long-term data stor age of reformatted CT series and strategies for mitigation Y. Jia¹, M. Deng¹, *R. Burger*¹, S. L. Sheard¹, K. Hanneman², M. Drucker Iarovich², R. Illing¹, A. G. Rockall¹; ¹ London/UK, ²Toronto, ON/CA ([email protected]) Purpose or Learning Objective: Image data storage and associated greenhouse gas (GHG) emissions is accelerating, yet strategies to minimise this are limited. Reducing the average file size of CT studies by reducing the number of reformats stored could help reduce emissi ons. This study aims to estimate GHG emissions associated with storage of C T reformats by modelling measurements from endometrial cancer baseline stagi ng CT. Secondary aims were to model the findings comparing cloud storage emissions and assess the hypothetical GHG mitigation impact of a data retent ion policy Thursday Abstract-based Programme 128

Methods

or Background: Baseline staging CT chest, abdomen, and pelvis (CT-CAP) in 183 endometrial cancer patients in a UK cancer centre between 2013-2016 were analysed (Cohort A). The number of s tored multiplanar reformats, maximum intensity projections images and lung reconstructions were recorded. The file size of each reformat was n oted for 30 studies (Cohort B). Comparison was made with an external dataset of 100 baseline CT-CAP from Canada between 2018-2023 (Cohort C). Mitigatio n of GHG emissions was projected for different storagescenarios.

Results

or Findings: Reformatted series were present in cohort A (97%, 179/183), cohort B (97%, 29/30) and cohort C (100% ,100/100). Of the total file size of cohort B (25590mb), 65% (16685mb) was refor mats and/or duplicate series. On-premise storage of all reformats for cum ulative new UK endometrial cancer cases from 2020-2040 would produce 349 metri c tonnes CO2 equivalent (MTCO2e). Over 20 years, projected reduc tions in MTCO2e were 69%(107/349) for storing only acquired axial slices , 80%(70/349) for switching to cloud storage, and 36%(222/349) for implemented a data retention policy.

Conclusion

A significant number of studies contained unnecessa ry reformats, increasing average file size. A strategy to revise CT data storage protocols can substantially lower radiology GHG emissions, withou t compromising patient care.

Limitations

Small selective patient dataset. Funding for this study: None Ethics committee - additional information: Institutional approval was obtained for a quality improvement project. Author Disclosures: Kate Hanneman: Nothing to disclose Yifan Jia: Nothing to disclose Michael Deng: Nothing to disclose Rebecca Burger: Nothing to disclose Moran Drucker Iarovich: Nothing to disclose Rowland Illing: Nothing to disclose Sarah Lucy Sheard: Nothing to disclose Andrea Grace Rockall: Nothing to disclose Ultrasound's Hidden Environmental Cost: Linens and Disposables C. L. Thiel¹, J. Leschied², D. Carver², *J. R. Sche el*², R. Omary², M. Vigil-Garcia, Phd³, P. Gehrels³, C. Meijer³, O. Struk³; ¹Madison, WI/US, ²Nashville, TN/US, ³Amsterdam/NL ([email protected]) Purpose or Learning Objective: To understand the sources of environmental impact of ultrasound imaging in a US-based, adult d iagnostic radiology service.

Methods

or Background: A life cycle assessment (LCA) was used to evaluate the environmental impact of 2 ultrasound machines a nd their surrounding resource needs, including production, use and dispo sal of other capital equipment, linens, disposable supplies, pharmaceuti cals, and data storage. A sensitivity analysis was performed to assess the im pact of low-carbon electricity sources.

Results

or Findings: Contrary to expectations, linens and disposable sup plies emerged as the major contributors to ultrasound's g reenhouse gas (GHG) emissions, each accounting for approximately 30% of its total impact. Energy use from the ultrasound units themselves was compar atively lower, at 7%, along with the production of the ultrasound units ( 7%), and the production and use of workstations (11%). The study also noted tha t ultrasound equipment spent 30-45% of the time in non-scanning mode. The sensitivity analysis showed the use of photovoltaics as an electricity s ource would reduce US’s GHGs by 9%; however, it would not shift the major s ources of GHGs away from linens and disposable supplies.

Conclusion

Reducing linen use, adopting reusable alternatives for disposable supplies, and encouraging textile and supply manufa cturers and laundering facilities to develop lower carbon alternatives are essential for improving the sustainability of ultrasound practices.

Limitations

The study, while providing valuable insights into t he environmental impact of US, has limitations due to its single-center focus; exclusion of mammography, nuclear medicine, and int erventional radiology; a one-month data collection period; and some data and modeling limitations. Funding for this study: No external funding was received for this study. Philips and VUMC independently contributed to this study through in-kind labor. Ethics committee - additional information: This study was deemed non- human subjects research. Author Disclosures: Marta Vigil-Garcia, Phd: Employee: Philips Cecilia Meijer: Employee: Philips Reed Omary: Nothing to disclose Pien Gehrels: Employee: Philips Jessiva Leschied: Nothing to disclose Olesya Struk: Employee: Philips Diana Carver: Nothing to disclose John R. Scheel: Nothing to disclose Cassandra L Thiel: Consultant: Philips CT's Carbon Footprint: Energy and Disposables S. Pruthi¹, C. L. Thiel², D. Carver¹, J. R. Scheel¹ , R. Omary¹, M. Vigil-Garcia³, P. Gehrels³, C. Meijer³, *O. Struk*³; ¹Nashville, T N/US, ²Madison, WI/US, ³Amsterdam/NL ([email protected]) Purpose or Learning Objective: To understand the sources of environmental impact of CT scanning within a US-based, adult diag nostic radiology setting.

Methods

or Background: A life cycle assessment (LCA) was conducted, evaluating the production, use, and end-of-life of CT scanners, including energy consumption, production and use of other cap ital equipment, disposable supplies, linens, pharmaceuticals, and d ata storage. A sensitivity analysis was performed assessing the impact of a lo w-carbon electricity source.

Results

or Findings: Energy use and disposable supplies were identified as major contributors to CT's greenhouse gas (GHG) emi ssions, accounting for 42% and 20%, respectively. The production of CTs co ntributed 17% to GHG emissions. Furthermore, the study revealed a 50% di fference in GHG emissions between CT scanners of the same model, su ggesting opportunities for optimization. CT scanners were also found to sp end between 44-72% of the time in a non-scanning mode. Sensitivity analysis s howed that using low- carbon electricity could significantly decrease CT' s energy emissions, shifting the major sources of emissions to the production of CT imaging equipment and disposable supplies.

Conclusion

Optimizing energy use, minimizing disposable suppli es, and ensuring efficient equipment utilization are crucia l for reducing CT's environmental impact.

Limitations

The study, while providing valuable insights into t he environmental impact of CT, has limitations due to its single-center focus; exclusion of mammography, nuclear medicine, and int erventional radiology; a one-month data collection period; and some data and modeling limitations. Funding for this study: No external funding was received for this study. Philips and VUMC independently contributed to this study through in-kind labor. Ethics committee - additional information: This study was deemed non- human subjects research. Author Disclosures: Sumit Pruthi: Nothing to disclose Marta Vigil-Garcia: Employee: Philips Cecilia Meijer: Employee: Philips Reed Omary: Nothing to disclose Pien Gehrels: Employee: Philips Olesya Struk: Employee: Philips Diana Carver: Nothing to disclose John R. Scheel: Nothing to disclose Cassandra L Thiel: Consultant: Philips The Environmental Cost of MRI: A Life Cycle Assessm ent D. Carver¹, C. L. Thiel², J. R. Scheel¹, *R. Omary* ¹, M. Vigil-Garcia³, P. Gehrels³, S. Thornander³, C. Meijer³, O. Struk³; ¹Nashville, TN/US, ²Madison, WI/US, ³Amsterdam/NL ([email protected]) Purpose or Learning Objective: To understand the sources of environmental impact of MRI within a US based diagnostic radiolog y department.

Methods

or Background: A life cycle assessment (LCA) was conducted that evaluated the production, use, and end-of-life of 3 MRI scanners in an adult diagnostic radiology department. Other model inputs included the production and energy use of other capital equipment, disposab le supplies, linens, pharmaceuticals, and data storage. A sensitivity an alysis assessed the impact of using a low-carbon electricity source.

Results

or Findings: Energy consumption emerged as the dominant source o f MRI's greenhouse gas (GHG) emissions, representing 79% of its total impact. Notably, the 3T MRI demonstrated 1.4 times higher e nergy use and 1.9 times higher production emissions compared to the 1.5T. A dditionally, MRI scanners were found to be in low-power or ready-to-scan mode for 72-75% of the time, indicating potential for energy optimization. Sensi tivity analysis revealed that decarbonizing the electricity grid could lead to an 87% reduction in energy- related GHG emissions from MRI. In this scenario, t he production of imaging equipment itself would become the largest contribut or to MRI's GHG emissions.

Conclusion

Improving energy efficiency through measures such a s optimizing scan protocols, developing automation of scanner ef ficiency modes, and transitioning to renewable energy sources are cruci al steps in reducing MRI's environmental footprint. If changing the grid is no t possible, other opportunities include reducing scan times via AI (e.g. Smart spee d) or optimized scheduling.

Limitations

The study, while providing valuable insights into t he environmental impact of MRI, has limitations due to its: single-center focus; exclusion of mammography, nuclear medicine, and int erventional radiology; one-month data collection period; and some data and modeling limitations. Thursday Abstract-based Programme 129 Shifting to a low-carbon electricity grid highlight s the additional need to address emissions associated with the production of MRI equ ipment itself. Funding for this study: No external funding was received for this study. Philips and VUMC independently contributed to this study through in-kind labor. Ethics committee - additional information: This study was deemed non- human subjects research. Author Disclosures: Marta Vigil-Garcia: Employee: Philips Cecilia Meijer: Employee: Philips Reed Omary: Nothing to disclose Pien Gehrels: Employee: Philips Olesya Struk: Employee: Philips Diana Carver: Nothing to disclose John R. Scheel: Nothing to disclose Cassandra L. Thiel: Consultant: Philips Sophie Thornander: Employee: Philips AI-Powered MRI: Time, Energy, and Emission Savings for a Greener Future *T. Polidori*, M. Zerunian, D. De Santis, F. Puccia relli, B. Masci, A. Del Gaudio, F. Fanelli, D. Caruso, A. Laghi; Rome/IT ([email protected]) Purpose or Learning Objective: The study aimed to assess the energy and greenhouse-gas (GHG) emission savings feasible usin g artificial intelligence (AI) in multi-district MRI-protocols, including lum bar-spine-MRI, cardiac-MRI, and upper abdomen-MRI. We evaluated the impact of A I-algorithms applied on MRI acquisition on scan time reduction, energy cons umption, and CO2 emissions per patient, providing insights into the potential benefits of AI in routine clinical practice.

Methods

or Background: This retrospective study analyzed 148 patients, including 45 upper abdomen-MRI, 53 cardiac-MRI, and 50 lumbar spine- MRI.MRI scans were acquired both without and with A I assistance applied to specific 2D and 3D sequences.The Air Recon-DL (GE H ealthcare) was used for T2 and DWI sequences in upper abdomen-MRI, as w ell as T1, T2, and STIR sequences in lumbar spine-MRI. The Sonic-DL (G E Healthcare) was applied to SSFP sequences specifically for cardiac- MRI.The outcomes measured were time savings per patient, reduced ene rgy consumption (kW/h), and the corresponding reduction in CO2-equivalent e missions.

Results

or Findings: The application of AI across all three districts st udied resulted in significant time savings per patient co mpared to non-AI protocol (p<0.01): 5’11’’ (58%) for upper abdomen-MRI, 1’30’ ’ (52%) for cardiac-MRI, and 6’ (50%) for lumbar spine-MRI.These time reduct ions corresponded to significant energy savings of 1.39kW/h, 0.40kW/h, a nd 1.68kW/h per patient (p<0.05), respectively. The equivalent reduction in CO2 emissions was 0.57kg for upper abdomen-MRI, 0.16kg for cardiac-MRI, and0 .69 kg for lumbar spine- MRI (p<0.05).

Conclusion

The implementation of AI in MRI protocols significa ntly reduces scan time, energy consumption, and GHG emissions, h ighlighting its potential for enhancing the sustainability of medical imaging practices.Integrating AI into routine clinical protocols could offer considerable environmental benefits, contributing to the reduction of the healthcare sec tor’s carbon footprint.

Limitations

Limitations include a small patient cohort and the use of a single vendor for MRI-protocols. Funding for this study: No-funding was received for this study. Ethics committee - additional information: No Author Disclosures: Damiano Caruso: Nothing to disclose Benedetta Masci: Nothing to disclose Francesco Pucciarelli: Nothing to disclose Marta Zerunian: Nothing to disclose Federica Fanelli: Nothing to disclose Domenico De Santis: Nothing to disclose Tiziano Polidori: Nothing to disclose Andrea Laghi: Nothing to disclose Antonella Del Gaudio: Nothing to disclose Automated scout-image based estimation of contrast agent dosing: a deep learning approach *R. T. Schirrmeister*, P. S. Friemel, M. Reisert, F . Bamberg, J. Weiß, A. Rau; Freiburg/DE ([email protected]) Purpose or Learning Objective: To develop and test a deep learning algorithm for approximation of contrast agent dosag e based on CT scout images.

Methods

or Background: We prospectively enrolled 817 patients undergoing clinically indicated CT imaging, predominantly of t he chest and/or abdomen. Patient weight was collected 1) manually and 2) sel f-reported prior to the examination by study staff. Based on the scout imag es, we developed an EfficientNet convolutional neural network pipeline to estimate the optimal contrast agent dose based on patient weight and pro vide a browser-based user interface as a versatile open-source tool to a ccount for different contrast agent compounds We additionally analyzed the body-w eight-informative CT features using a weight-conditional variational aut oencoder.

Results

or Findings: The training cohort consisted of 218 chest, 51 abdominal, 511 whole-body, and 37 CT scans of vario us other anatomical regions. Self-reported patient weight was statistic ally significantly lower than manual measurements (75.02 kg vs.76.92 kg; p < 10 ⁻⁵, Wilcoxon signed-rank test). Our pipeline predicted patient weight with a mean absolute error of 4.74 ± 0.14 kg in 5-fold cross-validation and is publicly available at https://nora- imaging.org/ct-scout-weight/. Interpretability anal ysis revealed that both larger anatomical shape and higher overall Hounsfield unit s were predictive of body weight.

Conclusion

Our open-source deep learning pipeline allows for a utomatic estimation of accurate contrast agent dosing based on scout images in routine CT imaging studies. This approach has the potential to streamline contrast agent dosing workflows, improve efficiency, and enh ance patient safety by providing quick and accurate weight estimates witho ut additional measurements or reliance on potentially outdated re cords.

Limitations

The model's performance may vary depending on patie nt positioning and scout image quality and the approac h requires validation on larger patient cohorts and other clinical centers. Funding for this study: Funded by an unrestricted research grant from Siemens Healthineers. Ethics committee - additional information: The study was approved by an ethics committee. Written informed consent was obta ined from each participant. Author Disclosures: Paul Simeon Friemel: Nothing to disclose Marco Reisert: Nothing to disclose Jakob Weiß: Nothing to disclose Alexander Rau: Advisory Board: Bayer Robin Tibor Schirrmeister: Nothing to disclose Fabian Bamberg: Nothing to disclose Cross-Modality Image Conversion from non-contrast C ardiac Magnetic Resonance to contrast-enhanced Computed Tomography Angiography using Diffusion Models *E. Almar Munoz*, C. G. Colintenorio, C. Kremser, M . Haltmeier, A. Mayr; Innsbruck/AT ([email protected]) Purpose or Learning Objective: Transcatheter Aortic Valve Implantation (TAVI) is the preferred treatment for patients with severe aortic stenosis at high to intermediate surgical risk. The gold-standard pr eoperative imaging modality is contrast-enhanced CTA; however, non-contrast CMR is an alternative for patients with contraindications to contrast agents despite its limitations in detecting calcifications. We propose diffusion mode ls to improve CMR-to-CTA conversion, facilitating comprehensive TAVI plannin g and predicting valve calcifications without contrast.

Methods

or Background: Our pipeline integrates Denoising Diffusion Probabilistic Models (DDPMs) and Stochastic Differe ntial Equation (SDE) models. This pipeline was evaluated using an in-hou se dataset consisting of 39 paired CTA and CMR scans. The image pairs were alig ned using rigid registration techniques. To improve the registratio n process, we utilized aorta segmentation masks obtained using nnUNet for CMR sc ans and TotalSegmentator for CTA scans.

Results

or Findings: Regarding the aorta segmentation, we obtained Dice values of 0.987±0.006 for CMR and 0.980±0.005 for CTA. The Dice Score obtained in the rigid registration was above 0.87. Regarding the image conversion, our results demonstrate that the overal l synthesized CTA images exhibit high fidelity to their real counterparts, v alidated by metrics including the Structural Similarity Index Measure (SSIM) and Peak Signal-to-Noise Ratio (PSNR), both exceeding 0.80 and 22, respectively. F ocusing on the valve's calcifications, some are accurately converted into CTA-calcified regions but are not always consistent or repeatable.

Conclusion

This study highlights the potential of diffusion mo dels in medical imaging, offering a promising solution for patients unable to receive contrast agents, thereby improving the safety and efficacy o f TAVI planning.

Limitations

Firstly, the model encounters difficulties in repli cating small details in the CTA, including calcifications. Secon dly, the diffusion models applied are very sensitive to image training; both data modalities must present low noise levels or artifacts. Funding for this study: Fund provided by FWF-DOC-110 Ethics committee - additional information: Nothing to declare Thursday Abstract-based Programme 130 Author Disclosures: Markus Haltmeier: Nothing to disclose Enrique Almar Munoz: Nothing to disclose Agnes Mayr: Nothing to disclose Christian Kremser: Nothing to disclose Carmen Guadalupe Colintenorio: Nothing to disclose Diffusion Model for Non-contrast MR to Aid Diagnosi s of Focal Liver Lesions: A Multi-Center Study *S. Dong*, Z. Shen, F. Yan, R. Li; Shanghai/CN ([email protected]) Purpose or Learning Objective: To develop a diffusion model for generating virtual dynamic contrast-enhanced MRI (DCE-MRI) ima ges from non-contrast T1-weighted scans and assess its efficacy in FLL di agnosis.

Methods

or Background: Gadolinium-based contrast agents (GBCAs) in DCE-MRI are crucial for characterizing focal liver lesions (FLLs), but their use increases risks for patients with renal impairment and adds to imaging costs. Virtual contrast-enhanced images from non-contrast T1-weighted scans could reduce these risks and streamline diagnostics. FLLs ≥1 cm, identified through DCE-MRI, were included, with lesion types such as H CC, ICC, liver metastases, cysts, hemangiomas, and FNH. A diffusio n model was trained on non-contrast T1-weighted and corresponding multipha se DCE-MRI images (arterial, portal venous, and delayed phases). Trai ning occurred at Center 1 (2018–2023) with a 3:1 split for training and inter nal testing. External validation used data from three other centers (2018–2024). A d iagnostic model for FLLs was also trained on synthetic DCE-MRI images. Norma lized mean absolute error (NMAE), peak signal-to-noise ratio (PSNR), an d structural similarity index measure (SSIM) were used for evaluation. Three radi ologists scored image quality on a three-point scale. The human machine c omparison was conducted with six radiologists in different experience.

Results

or Findings: The study included 1187 patients in the training se t (mean age, 51 ±12), with 395 internal (52 ±16) and 347, 271, and 219 external patients (57 ±11, 56 ±12, 58 ±11). The model showed strong similarity between virtual and real DCE-MRI images, with NMAE 0.021–0. 038, PSNR 28.9–31.8 dB, and SSIM 0.881–0.927. Diagnostic accuracy was 9 3% for the internal and 91% for external sets, outperforming three junior r adiologists (P < .001) and matching three senior radiologists (P = .19).

Conclusion

The diffusion model provides a safe, cost-effective alternative to traditional DCE-MRI, maintaining high diagnostic ac curacy for FLLs.

Limitations

None reported. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Ruokun Li: Nothing to disclose Shunjie Dong: Nothing to disclose Fuhua Yan: Nothing to disclose Zhehan Shen: Nothing to disclose 16:00-17:30 Research Stage 4 Research Presentation Session: Neuro RPS 1111 Neuro interventions and beyond Moderator D. Ozretić; Zagreb/HR ([email protected]) Cerebral metabolic rate of oxygen on admission MRI may predict infarct growth in hyperacute ischemic stroke patients treat ed successfully with thrombectomy: a retrospective observational study *A. Bani Sadr*, J. Fournel, M. Hermier, N. Nighogho ssian, Y. Berthezene; Lyon/FR ([email protected]) Purpose or Learning Objective: Despite successful thrombectomy, most stroke patients experience infarct growth which neg atively affect functional outcomes. Advances in oxygen metabolism mapping on admission dynamic- susceptibility contrast MRI have shown promise in a ssessing the viability of diffusion-weighted imaging (DWI) lesions. We aimed to assess the utility of oxygen metabolism mapping at the voxel scale in det ermining the fate of diffusion-mismatch regions following successful thr ombectomy.

Methods

or Background: We conducted a retrospective analysis of the HIBISCUS-STROKE cohort (NCT: 03149705), a single-ce nter, observational study enrolling patients treated with thrombectomy between 2016 and 2022. Admission DSC-MRI was used to generate time-to-maxi mum (Tmax), cerebral blood volume (CBV), cerebral metabolic rate of oxyg en (CMRO2), and oxygen extraction fraction (OEF) maps. In patients with su ccessful reperfusion (modified Thrombolysis in Cerebral Infarction [mTIC I] score ≥2B), Tmax ≥6s voxels excluding DWI abnormalities were analyzed on day 6 T2-fluid attenuated inversion recovery MRI. Semi-quantitativ e measurements of CMRO2, CBV, and OEF were extracted from regions ide ntified as either necrotic or salvaged at follow-up.

Results

or Findings: Among the 321 patients enrolled, 134 (41.7%) met inclusion criteria (median age 71.0 years; 58.2% ma le; median NIHSS score 15.0). In the training cohort, ROC analysis identif ied optimal thresholds for predicting necrosis: CBV (0.99), CMRO2 (0.64), and OEF (1.59). In the validation cohort, CMRO2 achieved an area under the curve (AUC) of 73.6% (95% confidence interval [CI]: 65.0–82.1), signific antly outperforming CBV (AUC: 63.5%, 95% CI: 53.8–73.1; P=0.003) and OEF (A UC: 55.0%, 95% CI: 45.2–64.7; P=0.0005). Multivariable logistic regres sion revealed that CMRO2 <0.64 was independently associated with necrosis in diffusion-perfusion mismatch regions (OR: 6.0, 95% CI: 3.2–11.6, P<0.00 01).

Conclusion

In acute stroke patients achieving successful throm bectomy, a CMRO2 < 0.64 in regions of diffusion-perfusion mism atch.

Limitations

DSC-MRI dervied oxygen metabolism mapping is not ye t validated against PET Funding for this study: This work was supported by the RHU MARVELOUS (ANR-16-RHUS-0009) of Université de Lyon, within th e program “Investissements d'Avenir” operated by the French N ational Research Agency. Ethics committee - additional information: The local ethics committee (Institutional Review Board No: 00009118) approved this study, and all participants or their relatives gave provided infor med consent. Author Disclosures: Yves Berthezene: Nothing to disclose Marc Hermier: Nothing to disclose Norbert Nighoghossian: Nothing to disclose Julien Fournel: Nothing to disclose Alexandre Bani Sadr: Nothing to disclose Why we fail: Factors leading to unsuccessful mechan ical thrombectomies *R. Bruen*¹, H. Briody¹, S. Singh¹, N. Healy², M. T . Crockett¹, J. Müller¹, S. O' Reilly¹, J. Thornton¹, P. Nicholson¹; ¹Dublin 9/IE, ²Dublin/IE ([email protected]) Purpose or Learning Objective: Mechanical thrombectomy (MT) has revolutionized acute ischemic stroke care, demonstr ating superior functional outcomes compared to intravenous t-PA alone. A subs tantial minority of patients fail to achieve successful recanalization. This study analyses the causes of unsuccessful MT in a large tertiary refer ral centre over an 11-year period.

Methods

or Background: We retrospectively reviewed a prospective stroke registry at our institution, identifying all patien ts with acute ischemic stroke who underwent MT between January 2012-2023. Unsuccessfu l MT was defined as a post-interventional modified Thrombolysis in Cere bral Infarction (mTICI) score < 2b. We collected demographic data, NIHSS, A SPECTS, occlusion location, pre-MT alteplase administration and speci fic reasons for MT failure, categorizing them into three groups: Target not rea ched, Target reached but recanalization failed, and non-technical reasons.

Results

or Findings: Of 2620 MT procedures performed, 259 (9.9%) were unsuccessful. The median patient age was 75 years ( IQR 65-83), 48% (n=124) were female and the median NIHSS score was 16 (IQR 10-19). Occlusions were located in the anterior circulation in 98% of cases, with a median ASPECTS of 9. In Category 1 (n=48, 19%), non-reachi ng of the target was attributed to cervical artery tortuosity (n=23, 48% ), challenging aortic arch anatomy (n=7, 14.6%) and inability to traverse a ce rvical occlusion (n=18, 37.5%). Category 2 failures were the most common (n =146, 56%) and were caused by unsuccessful microcatheter advancement be yond the occlusion (n=8, 5.4%), stent retriever and aspiration cathete r failed recanalization (n=39, 26.7%), and spontaneous/iatrogenic re-occlusion (n= 99, 67.8%). Category 3 (non-technical failures; n=62, 24%) were less commo n and were mainly due to patient neurological decline.

Conclusion

Unsuccessful mechanical thrombectomy was encountere d in 9.9% of cases in our cohort. The most common reason was spontaneous/iatrogenic re-occlusion.

Limitations

Study was performed in a single site tertiary refer ral centre. Funding for this study: None Ethics committee - additional information: Retrospective data. No ethics required. Thursday Abstract-based Programme 131 Author Disclosures: Patrick Nicholson: Nothing to disclose Richard Bruen: Nothing to disclose John Thornton: Nothing to disclose Hayley Briody: Nothing to disclose Matthew Thomas Crockett: Nothing to disclose Sean O' Reilly: Nothing to disclose Nuala Healy: Nothing to disclose Sneha Singh: Nothing to disclose Jennifer Müller: Nothing to disclose Comparison of Endovascular Thrombectomy Outcomes Be tween In- Hospital and Out-of-Hospital Stroke *S. Singh*, P. Rohan, C. Leneghan, R. Bruen, M. T. Crockett, A. O'Hare, S. Power, J. Thornton, P. Nicholson; Dublin/IE ([email protected]) Purpose or Learning Objective: To compare patient characteristics and outcomes following endovascular thrombectomy (EVT) for ischemic stroke between patients experiencing in-hospital stroke (I HS) and out-of-hospital stroke (OHS).

Methods

or Background: A single-center, retrospective observational cohort study was conducted using data from the institution al EVT database at Beaumont Hospital. Patients were categorized as IHS or OHS, and their baseline characteristics and outcomes were compared .

Results

or Findings: Of 2619 patients undergoing EVT, 383 (14.6%) experienced IHS (median age 72 years, 57% male) and 2235 (85.4%) experienced OHS (median age 72 years, 54% male). OH S patients had higher pre-stroke modified Rankin Scale (mRS) scores (p<0. 0001) and were significantly more likely to receive intravenous th rombolysis (p<0.0001). IHS patients had higher median National Institutes of H ealth Stroke Scale (NIHSS) scores at day 1 (11 vs. 8, p=0.14) and day 5 (6 vs. 4, p=0.0586), although these differences were not statistically significan t. IHS patients also had higher mRS scores at day 30 (p<0.0001) and day 90 (p<0.000 1). OHS patients tended to have better Thrombolysis in Cerebral Infa rction (TICI) scores post- revascularization (p=0.0535). Treatment was faster in the IHS group: onset to groin puncture (210 vs. 308 minutes, p<0.0001), ons et to first reperfusion (243 vs. 341 minutes, p<0.0001). There was no significan t difference in groin puncture to first reperfusion time (23 vs. 21 minut es, p=0.1367). IHS patients had a longer median EVT procedure time (37 vs. 32 m inutes, p=0.0363).

Conclusion

Despite shorter time intervals to intervention, pat ients with IHS experienced worse functional outcomes after EVT com pared to patients with OHS.

Limitations

N/A Funding for this study: N/A Ethics committee - additional information: Approved by the local ethics committee at Beaumont Hospital, Dublin, Ireland Author Disclosures: Patrick Nicholson: Nothing to disclose Richard Bruen: Nothing to disclose John Thornton: Nothing to disclose Alan O'Hare: Nothing to disclose Matthew Thomas Crockett: Nothing to disclose Sarah Power: Nothing to disclose Caoimhe Leneghan: Nothing to disclose Sneha Singh: Nothing to disclose Pat Rohan: Nothing to disclose How predictive is CT angiography source image ASPEC TS (CTA-SI ASPECTS) score on initial CT for futile mechanical thrombectomy? An ongoing study *A. Tsaoulia*, M. Mantatzis, L. Kougias, A. Stofori adi, P. K. Prassopoulos; Thessaloniki/GR ([email protected]) Purpose or Learning Objective: The standard of care for patients with large vessel occlusion (LVO) is mechanical thrombectomy, with or without preceding thrombolysis. Patients with a large infarction core have a higher risk of reperfusion edema or hemorrhage. Alberta stroke pro gram early computed tomography (ASPECTS) score is widely used to evalua te the extent of acute ischemic stroke at the middle cerebral artery terri tory. An extention of it is CTA- SI ASPECTS (CTAsp), which shows the collateral circ ulation and discriminates hypoperfused areas. However, it is not established whether this hypoperfusion depicts core or penumbra and if CTAsp is more accur ate for patient selection. We aim to find a possible correlation between CTAsp and infarct core

Methods

or Background: We analyzed the initial CT/CTAs of seven patients with LVO that had complete first pass recanalizatio n and the CTAsp scores were calculated. All CT/CTAs were performed on a 16 -slice SIEMENS Emotion-16. A non-contrast CT scan was obtained aft er thrombectomy, 24 and 48 hours later. These scans were performed on a 128 -slice GE OPTIMA. Ten ROIS were applied in each hemisphere in areas defin ed by ASPECTS, to find density differences.

Results

or Findings: The patients were divided into two groups based on CTAsp scores; 3 patients had a score of 6-7 and 4 w ith a score of 8-10. In two patients a discrepancy between ASPECTS and CTAsp wa s found. The final infarct core was associated with CTAsp in 6/7 patie nts, while in one, the hypodense area in CTAsp was normalized in follow up .

Conclusion

CTAsp proves to be helpful for attempting to predic t which patients will have a good outcome after endovascula r treatment.

Limitations

Small number of cases-preliminary results Funding for this study: None Ethics committee - additional information: The study is based on an imaging score from emergency obtained CT scans. Author Disclosures: Michalis Mantatzis: Nothing to disclose Panos K. Prassopoulos: Nothing to disclose Aikaterini Tsaoulia: Nothing to disclose Anatoli Stoforiadi: Nothing to disclose Leonidas Kougias: Nothing to disclose Comparison of antithrombogenic coated and uncoated flow-diverters in ruptured and unruptured cerebral aneurysms *D. Weiß*, M. Vach, V. L. Ivan, S. Muhammad, B. Hof mann, M. Neyazi, B. Turowski, M. Kaschner; Düsseldorf/DE ([email protected]) Purpose or Learning Objective: Flow-diversion has become a key treatment option for complex intracranial aneurysms. Recent a dvancements include coated flow-diverters (FD), designed to potentially reduce the need for dual antiplatelet therapy thereby removing the associate d secondary risks, while maintaining patency and low complication rates. Com paring coated and uncoated FDs may offer insights into long-term outc omes and treatment optimization.

Methods

or Background: In this retrospective single-center study, we investigated the data of 21 consecutive patients wi th cerebral aneurysms, treated between 2021 and 2023 with the coated Deriv o 2heal Embolization Device (D2H) and the uncoated Derivo Embolization D evice (DED) (both Acandis, Pforzheim, Germany). We described the proc edure and analyzed clinical and radiological data, along with long-ter m outcomes after 18 months of follow-up.

Results

or Findings: Nine patients (42.9%) had incidental, while 12 (57. 1%) had symptomatic aneurysms, including ten with WFNS IV subarachnoid hemorrhages. Aneurysm locations included mostly the internal carotid (n=9) and the vertebral artery (n=7). All FDs were succes sfully deployed: 11 patients received the coated device and 10 the uncoated devi ce. After 18 months, 73.3% patients had favorable outcomes (mRS 0-2). On e coated FD occluded asymptomatically after six months, and one uncoated FD occluded immediately but could be recanalized.

Conclusion

We observed favorable occlusion rates for both coat ed and uncoated FDs. The role of dual antiplatelet therapy remains debated. Large multicenter studies are essential to evaluate the p atency of coated compared to uncoated FDs and determine whether they can redu ce thrombogenicity, potentially allowing for less or no antiplatelet th erapy in emergencies.

Limitations

Unequal distribution of emergency and elective trea tments. Limited by both the number of patients. The study d oes not address whether monotherapy would be sufficient for drug-coated ste nts. Funding for this study: None Ethics committee - additional information: Local ethics committee of medical faculty Author Disclosures: Marius Kaschner: Nothing to disclose Marius Vach: Nothing to disclose Milad Neyazi: Nothing to disclose Sajjad Muhammad: Nothing to disclose Daniel Weiß: Nothing to disclose Björn Hofmann: Nothing to disclose Vivien Lorena Ivan: Nothing to disclose Bernd Turowski: Nothing to disclose Evaluation the Effect of CT Black Blood Technique i n Post-treatment Follow-up of Intracranial Aneurysms Treated with Fl ow-diverting Stents *D. Xie*, Z. Lai, H. Ma, R. Xu, J. Wu, J. Zhao; Gua ngzhou/CN ([email protected]) Purpose or Learning Objective: To assess the feasibility and effect of CT black blood technique in the post-treatment follow- up of intracranial aneurysms treated with flow-diverting stents.

Methods

or Background: A retrospective analysis was conducted on 18 patients who underwent treatment with flow-divertin g stents for intracranial aneurysms and subsequently underwent digital subtra ction angiography (DSA) Thursday Abstract-based Programme 132 follow-up. All patients underwent CT angiography (C TA) examination before DSA. The contrast-enhancement boost technique was e mployed to process CTA images to obtain CT black blood images. Two rad iologists independently rated the image quality and diagnostic confidence f or stent-related stenosis on both conventional CTA and CT black blood images usi ng a 4-point scale. The performance of conventional CTA and CT black blood in diagnosing stent- related stenosis was compared, and the examination time and radiation dose of CTA and DSA were recorded and compared.

Results

or Findings: Subjective ratings of image quality and diagnostic confidence for stent-related stenosis were signific antly higher for CT black blood images compared to conventional CTA images (3 .94±0.23 vs. 3.06±0.62 and 3.89±0.31 vs. 2.83±0.6, respectively; all p＜0.01). DSA detected 2 cases of distal stenosis and 5 cases of overall stenosis within the stent. Compared to conventional CTA (sensitivity: 100%, specificity: 3 6.4%, accuracy: 61.1%), CT black blood demonstrated significantly improved per formance in detecting stent-related stenosis, with diagnostic sensitivity , specificity, and accuracy all reaching 100%, and perfect inter-observer agreement (k=1.0). Regarding radiation dose, the average radiation dose for conv entional CTA was (67.85±8.31) mGy, whereas the radiation dose required for DSA significantly increased to (516.81±193.83) mGy (p＜0.001). The examination times for CT and DSA were (3.53±0.74) minutes and (9.57±8.26) minutes, respectively(p＜ 0.01).

Conclusion

The CT black blood technique shows potential as the preferred

Method

for post-treatment follow-up of intracranial aneurysms treated with flow- diverting stents.

Limitations

Sample size was relatively small. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Jiale Wu: Author: author Dingxiang Xie: Nothing to disclose Rulin Xu: Nothing to disclose Zhiman Lai: Nothing to disclose Hui Ma: Nothing to disclose Jing Zhao: Author: author Impact of Automatically Assessed Collateral Circula tion and Infarct Core on Functional Outcome in Acute Ischemic Stroke Pati ents treated with Endovascular Thrombectomy *I. Požar*¹, F. F. Bajrovi ć², L. Umek², K. Šurlan Popović²; ¹Izola/SI, ²Ljubljana/SI Purpose or Learning Objective: This study aimed to evaluate the predictive value of automatically assessed collateral circulat ion (CC) and infarct core for functional outcome in acute ischemic stroke (AIS) p atients treated with endovascular thrombectomy (EVT).

Methods

or Background: We conducted a retrospective cohort study of 208 patients with anterior large vessel occlusion treat ed with EVT. Two AI-powered software were used to automatically assess CC and i nfarct core. Comparative analyses included patient demographics, clinical an d imaging data, and functional outcome. Univariate and multivariable lo gistic regression analyses were conducted to predict the 90-day functional out come. A favorable outcome was defined as a modified Rankin scale (mRS) score ≤2.

Results

or Findings: Among the 208 patients, 114 (54.8%) were women and 94 were men, with a mean age of 71.4±13.3 years. Patients with higher collateral score (CS) exhibited lower infarct core volumes (p<0.001) and better mRS score at 90 days (p=0.008). Among patients with a favorable outcome, the mean infarct core volume was lower compared to those with poor outcomes (5 mL vs. 8.6 mL, p=0.003). In univariate logistic regression, both infarct core (OR 0.94, p=0.005) and CC (OR 1.84, p= 0.014) were predictors of favorable outcome. However, in multivariable models , only infarct core remained a significant independent predictor [AORs of 0.95 (p=0.021) and 0.96 (p=0.039)].

Conclusion

Automatically assessed infarct core is a robust pre dictor of functional outcome in AIS patients post-EVT, while CC's predictive value diminishes when adjusted for infarct core. These fi ndings support the integration of AI-powered evaluations in clinical s ettings to improve prognosis and treatment strategies for AIS.

Limitations

Our study's limitations include using a 40-slice CT scanner, which affects data acquisition speed, particularly in per fusion; non-blinded assessments potentially introducing bias; and relia nce on a single CS evaluation, which could lead to inaccurate data. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by the National Medical Ethics Committee of the Republic o f Slovenia (No. 0120- 377/2019/4). Author Disclosures: Lan Umek: Nothing to disclose Fajko F. Bajrović: Nothing to disclose Katarina Šurlan Popović: Nothing to disclose Ingrid Požar: Nothing to disclose Higher Relative Brain Age of stroke patients treate d with mechanical thrombectomy is associated with poor outcomes *M. Guettier*, H. Biegalski, R. Lopes, J. Dumont, W . Gorwood, H. Henon, N. Bricout, G. Kuchcinski, M. Bretzner; Lille/FR ([email protected]) Purpose or Learning Objective: Relative brain age (RBA) is a novel MRI- derived biomarker that quantifies brain health rela tive to other patients within a cohort. It has previously been associated with poor er outcomes in untreated ischemic stroke patients. However, its impact on po ststroke outcomes in patients treated with mechanical thrombectomy (MT) remains unclear.We investigated the clinical determinants of RBA and i ts association with poststroke outcomes in patients treated with MT

Methods

or Background: We conducted a retrospective analysis of clinical and imaging data from stroke patients admitted to L ille University Hospital for anterior circulation MT between 2015 and 2020. Admi ssion axial T2 FLAIR MRI images were used to obtain brain parcellation d ata. A modified brain age prediction pipeline was applied to estimate brain a ge and derive RBA. Linear regression was used to identify cardiovascular risk factors associated with higher RBA, while logistic regression was employed to assess the impact of RBA on post-stroke outcomes.

Results

or Findings: A total of 1,296 patients were included, with a mea n age of 70 years; 54% were women. Patients with a histor y of diabetes mellitus and smoking had significantly higher RBA, indicating ol der-appearing brains (p=0.001 and p=0.010, respectively). Univariate ana lysis showed that patients with higher RBA, reflecting poorer brain health, we re less likely to achieve favorable functional outcomes after stroke (p=0.027 ). In multivariate analysis, several factors, including age, RBA, admission NIHS S score, intravenous thrombolysis, successful MT, glycemia, time from im aging to MT, and MT duration, were significantly associated with post-s troke outcomes (adjusted odds ratios: 0.48, 0.80, 0.48, 2.04, 5.72, 0.75, 0. 81, and 0.69, respectively).

Conclusion

Our study highlights the influence of smoking and d iabetes on brain aging and the detrimental effects of poor bra in health on post-stroke outcomes, building on decades of clinical knowledge .

Limitations

Retrospective data Funding for this study: This study has been funded by the ESR/EIBIR 2022 Seed Grant. Ethics committee - additional information: The ethical committee (Comité de protection des personnes Nord-Ouest IV) classifi ed the study as observational on March 9, 2010, and the committee p rotecting personal information of the patient approved the study by De cember 21, 2010 (n°10.677). Anonymized data supporting the findings of this study are available from the corresponding author upon reasonable reque st. Author Disclosures: William Gorwood: Nothing to disclose Julien Dumont: Nothing to disclose Martin Bretzner: Nothing to disclose Grégory Kuchcinski: Nothing to disclose Mélanie Guettier: Nothing to disclose Hugo Biegalski: Nothing to disclose Hilde Henon: Nothing to disclose Nicolas Bricout: Nothing to disclose Renaud Lopes: Nothing to disclose Silent Brain Infarcts Post-Interventional Cardiac C atheterization: Insights from High-Resolution DW-MRI in a Randomized Study *N. Tan*, X. Zhou; Kunming/CN ([email protected]) Purpose or Learning Objective: To analyze the incidence and anatomic distribution of acute cerebral embolism and identif y associated risk factors in patients undergoing interventional cardiac catheter ization.

Methods

or Background: We conducted a prospective study of patients from two cohorts between January 2023 and April 2024. Di ffusion-weighted magnetic resonance imaging (DW-MRI) was used to det ect silent brain infarcts (SBIs) preoperatively and within 48 hours postopera tively. For transcatheter aortic valve implantation (TAVI) patients, a cerebr al embolic protection device (CEPD, model CEP016F) was randomly employed to mini mize the risk of small emboli entering the cerebral vasculature. In atrial fibrillation ablation procedures, some patients underwent Vein of Marshal l Ethanol Infusion (VOMEt) to decrease the recurrence of atrial fibril lation and prevent the formation of new embolic events. Independent risk f actors were identified through multivariate logistic regression analysis. Thursday Abstract-based Programme 133

Results

or Findings: Silent brain infarcts were detected in 34 of 48 pat ients (70.8%) within 48 hours postoperatively. Among the patients who used CEPDs (n = 12), 7 (56.8%) found SBIs. Patients who used C EPDs showed a trend toward smaller infarct volumes ( 5.32 cm³ vs. 8.11 cm³; P = 0.07). In patients who underwent VOMEt procedures (n = 20), 13 (65.0%) developed SBIs; however, neither the incidence nor the volume of SB Is showed significant differences compared to those who did not undergo V OMEt. Multivariate logistic regression analysis identified operation t ime as an independent positive predictor of SBIs (odds ratios, 6.190 and 13.564; b oth P < .001).

Conclusion

Silent brain infarcts were detected in 70.8% of pat ients undergoing interventional cardiac catheterization, predominantly affecting the parietal lobes. These findings highlight the import ance of procedural optimization to reduce cerebral embolic risk.

Limitations

The study was with a relatively small simple size, and conducted at a single center. Funding for this study: No Ethics committee - additional information: Kunming Yan’an Hospital Ethics Committee Author Disclosures: Na Tan: Nothing to disclose Xinyan Zhou: Nothing to disclose Prognostic Value of CT Contrast Staining after Endo vascular Therapy in Basilar Artery Occlusion Stroke P. Reidler, O. Öcal, J. Ricke, D. Puhr-Westerheide, *M. P. Fabritius*; Munich/DE Purpose or Learning Objective: Contrast staining (CS) signifies prolonged tissue absorption of iodinated contrast media follo wing endovascular therapy (EVT) for large vessel occlusion stroke, indicating blood-brain barrier disruption. With EVT becoming the standard for trea ting basilar artery occlusion (BAO) stroke, our study aimed to determin e the prognostic significance of post-interventional CS in BAO strok e cases.

Methods

or Background: We included BAO patients who received postinterventional noncontrast CT within 24h after EVT. Expert radiologists confirmed the presence of CS on CT and its volume w as quantified. Functional outcomes were assessed on the modified Rankin (mRS) scale at 90 days and unfavorable outcome was defined as mRS ≥4. A multivariable LASSO- penalized logistic regression analysis was used to determine association of CS and other clinical and imaging parameters with func tional outcome.

Results

or Findings: 42 patients fulfilled the inclusion criteria (15 female, 35,7%). CS on postinterventional CT was present in 18 patients (42.9%) with a median [interquartile range / IQR] volume of 7.9 mL [3.7-14.6]. Patients with CS had a worse outcome with higher mRS after 90 day s (median [IQR]: 6 [4-6] vs. 2 [1-4], p<0.001). Multivariable LASSO analysis revealed significant and strongest association of CS with clinical outcome.

Conclusion

CS on postinterventional CT after EVT for BAO is an independent predictor of unfavorable functional outcome, outper forming other pre- and post- interventional imaging parameters.

Limitations

Retrospective, small sample size Funding for this study: None Ethics committee - additional information: LMU Munich Author Disclosures: Matthias Philipp Fabritius: Nothing to disclose Daniel Puhr-Westerheide: Nothing to disclose Osman Öcal: Nothing to disclose Paul Reidler: Nothing to disclose Jens Ricke: Nothing to disclose EVT in young adults with stroke: Outcomes and proce dural considerations *C. Leneghan*¹, D. Leneghan², P. Nicholson¹, S. Sin gh¹, J. Thornton¹, M. T. Crockett¹, A. O'Hare¹, J. Müller¹, P. Fearon¹ ; ¹Dublin/IE, ²Lucerne/CH ([email protected]) Purpose or Learning Objective: This study compares clinical presentations, outcomes, and procedural aspects of endovascular th rombectomy (EVT) in young adults (18-49 years) versus older adults ( ≥50 years) with acute ischemic stroke due to intracranial artery occlusion (IAO).

Methods

or Background: We analysed data from a prospectively maintained registry of patients treated with EVT in a large te rtiary referral centre between 2012 and 2022. Young (18-49) and older (≥50) patients were compared regarding baseline characteristics, 30- and 90-day modified Rankin Scale (mRS), 24-hour and 5-day NIHSS, reperfusion success (mTICI), post-EVT ASPECTS, and 90-day mortality.

Results

or Findings: Of 2201 patients, 243 (11%) were young adults. Youn g patients presented with lower median pre-treatment ASPECTS (8 vs. 9, p=0.0008). Older patients had higher NIHSS at 24 ho urs (OR 1.63, 95% CI 1.07-2.46), but no difference was observed at 5 day s. Functional dependence (mRS ≥3) was more frequent in older patients at 30 (OR 1. 92, 95% CI 1.45- 2.53) and 90 days (OR 2.22, 95% CI 1.66-2.97). Mort ality at 90 days was lower in younger patients (OR 0.45, 95% CI 0.28-0.70).

Conclusion

Younger patients undergoing EVT for IAO have lower initial ASPECTS but demonstrate faster neurological recover y and improved functional outcomes compared to older patients. Thi s suggests that aggressive EVT is warranted even in younger patients with lowe r ASPECTS.

Limitations

Retrospective design, single-centre study. Funding for this study: No specific funding was received for this study. Ethics committee - additional information: The study was approved by the local clinical audit committee. Author Disclosures: Patricia Fearon: Nothing to disclose Patrick Nicholson: Nothing to disclose John Thornton: Nothing to disclose Darren Leneghan: Nothing to disclose Alan O'Hare: Nothing to disclose Matthew Thomas Crockett: Nothing to disclose Caoimhe Leneghan: Nothing to disclose Sneha Singh: Nothing to disclose Jennifer Müller: Nothing to disclose Could be venous MT safe and effective for dural sin us thrombosis? *M. T. Contaldo*, A. Cervo, A. Macera, C. Rollo, A. Vitiello, G. Pero, G. Schwarz, M. Sessa, M. Piano; Milan/IT ([email protected]) Purpose or Learning Objective: Cerebral venous thrombosis (CVT) is a rare cause of stroke, that tends to affect young people and the role of endovascular treatment (EVT) remains debated. This study aims to evaluate the efficacy and safety of mechanical thrombectomy (MT) performed at our center for dural sinus thrombosis.

Methods

or Background: Within a retrospective observational analysis conducted over a 6-year period, data from 62 patien ts referred to the stroke unit for CVT were analyzed. Among them, 32 patients , classified with severe CVT, underwent EVT. We assessed safety by examining intraprocedural and periprocedural adverse events (asymptomatic, mild, or severe). Clinical outcomes were evaluated at baseline and discharge. Efficacy was determined by analyzing the recanalization rate (complete, par tial, or absent) at the end of the procedure.

Results

or Findings: A total of 32 patients received EVT, accounting for 36 procedures. EVT with MT was performed within 6 hour s from onset in 21 out of 32 cases. In most cases, MT was performed as primar y treatment, alongside best medical therapy. An intravenous bolus of hepar in was administered in the angio-suite to patients who had not yet received an ticoagulants or thrombolytics. Successful recanalization (complete or partial without cortical venous drainage delay) was achieved in 91.3% of pro cedures. CVT recurrence occurred in 3 out of 36 procedures. Mortality rate was 3.1%.

Conclusion

This is one of the largest series of patients treat ed with MT in combination with best medical therapy, underscoring the favorable safety and efficacy profile of EVT. MT could be a first-line o ption for multiple dural sinus thrombosis, rapid deterioration, venous hypertensio n, or anticoagulant failure.

Limitations

This study is a single center experience with no co ntrol-group and a retrospective design, with a heterogeneous case-b y-case patient selection. Funding for this study: No funding was received for this study. Ethics committee - additional information: Not applicable Author Disclosures: Amedeo Cervo: Nothing to disclose Maria Teresa Contaldo: Nothing to disclose Antonio Macera: Nothing to disclose Alessio Vitiello: Nothing to disclose Maria Sessa: Nothing to disclose Guglielmo Pero: Nothing to disclose Mariangela Piano: Nothing to disclose Ghil Schwarz: Nothing to disclose Claudia Rollo: Nothing to disclose 134 Friday, February 28 Friday Abstract-based Programme 135 08:00-09:00 Research Stage 1 Research Presentation Session: Oncologic Imaging RPS 1216 New perspectives in breast and gynaecological cancer Moderator G. Ivanac; Zagreb/HR ([email protected]) Prognostic role of Whole-body MRI (WB-MRI) in patie nts with metastatic breast cancer receiving systemic anti-cancer therap y *C. Pizzi*¹, C. Sattin¹, F. Arnone¹, D. Berloco¹, P . Hoxha¹, P. Summers¹, R. Maggioni¹, A. R. R. Padhani², G. Petralia¹; ¹Mil an/IT, ²Northwood/UK ([email protected]) Purpose or Learning Objective: To investigate the potential of the response assessment category (RAC) from MET-RADS-P guideline s as prognostic biomarker in metastatic breast cancer (MBC) patient s.

Methods

or Background: We enrolled MBC patients who underwent whole- body MRI at baseline and at each time point (every 12 weeks disease until progression) after systemic anti-cancer therapy (SA CT). We correlated the maximum RAC at time point 1 (TP1) with overall surv ival (OS). Patients were divided in two groups: those with a maximum RAC 1-2 (highly likely or likely to be responding, respectively) and those with a maxim um RAC 3-4-5 (stable disease, likely or highly likely to be progressing) at TP1. Survival curves were depicted in Kaplan-Meier plots and compared via a l og-rank test and hazard ratio (HR) using Cox regression model, with point c omparisons of three-year survival and median survival duration, using R.

Results

or Findings: Out of 45 MBC patients enrolled, a higher OS was observed in patients with a maximum RAC 1-2 (N=18) than in those with a maximum RAC 3-4-5 (N=27) at TP1 (log-rank test p=0. 007). Because more than 50% of the maximum RAC 1-2 patients are still living, the median survival duration could not be determined, median survival i n the maximum RAC 3-4-5 group was 36 months (upper limit of 95%CI not avail able). The HR for the maximum RAC 3-4-5 patients was 2.28 (95%CI 1.24 – 3 .33). Three years OS was 88.9% for RAC1-2 vs 42.6% for RAC 3-4-5; for a difference of 46.2% (95%CI 12.7%-79.8%, p=0.0068).

Conclusion

Our observations support the potential of RAC after TP1 as a prognostic biomarker in MBC patients undergoing SAC T.

Limitations

Retrospective and monocentric study. Funding for this study: No funding was received for this study. Ethics committee - additional information: Not applicable Author Disclosures: Roberta Maggioni: Nothing to disclose Francesca Arnone: Nothing to disclose Giuseppe Petralia: Nothing to disclose Caterina Pizzi: Nothing to disclose Paolo Hoxha: Nothing to disclose Paul Summers: Nothing to disclose Anwar R. R Padhani: Nothing to disclose Caterina Sattin: Nothing to disclose Donatello Berloco: Nothing to disclose Dual-energy CT machine learning model to characteri ze lymph nodes in patients with breast cancer *P. Morrone*, C. Zampieri, C. Esposito, E. Barone, I. Capitoni, F. Gentili, G. Bagnacci, S. Guerrini, M. A. Mazzei; Siena/IT ([email protected]) Purpose or Learning Objective: To identify a machine learning (ML) model with morphological and dual-energy (DE) data, to ch aracterize lymph node’s (LN) status during breast cancer (BC) staging.

Methods

or Background: From a cohort of 636 patients who undergone whole-body DE-CT and subsequent surgery with axilla ry lymphadenectomy between April 2015 to July 2023, 117 patients were included. Exclusion criteria: previous ipsilateral breast or axillary s urgery, or chemo-radiotherapy; poor quality CT; lack of anatomopathological data. For the morphological analysis, the main diameter of the neoplasm and loc ation, long and short axis and morphological features (fat hilum, cortical are a status, extranodal extension-ENE) of the ipsilateral axillary LNs were recorded. For quantitative analysis regions of interest (ROIs) were placed on the neoplasm and axillary LNs encompassing an area of post-contrast enhanceme nt as large and homogeneous as possible. An attempt was made to pla ce the ROIs on the entire LN excluding the fat hilum and surrounding s tructures, setting a HU displaying threshold to suppress negative HU values . For each ROI, mean attenuation value at 40 and 70keV, iodine concentra tion (IC), water concentration (WC) and effective-Z value were recor ded.

Results

or Findings: 116 BC and 375 LNs were analyzed, 180 pathological and 195 non-pathological. On univariate analysis th e attenuation (HU) at 40 and 70keV, slope, IC, WC, long and short LNs axis s howed statistically significant differences between histologically prov en pathological and non- pathological LNs (p<0.001). There were statisticall y significant differences (p<0.001) according to the cortical status and ENE. The logistic regression- based ML model included IC, short axis, fat hilum, cortical status and ENE; the ROC curve showed an AUC of 0.881, demonstrating goo d model accuracy.

Conclusion

The ML model provides a good discriminatory ability to differentiate pathological from non-pathological ax illary LNs in patients with BC.

Limitations

Not applicable Funding for this study: No funding was received for this study. Ethics committee - additional information: Waived from our etical committe due to the retrospective nature of this study. Author Disclosures: Chiara Esposito: Nothing to disclose Maria Antonietta Mazzei: Nothing to disclose Iacopo Capitoni: Nothing to disclose Paola Morrone: Nothing to disclose Francesco Gentili: Nothing to disclose Cecilia Zampieri: Nothing to disclose Elisa Barone: Nothing to disclose Susanna Guerrini: Nothing to disclose Giulio Bagnacci: Nothing to disclose Associations between ADC histogram analysis values and tumor-micro milieu in uterine cervical cancer *H-J. Meyer*¹, A-K. Höhn¹, A. Surov²; ¹Leipzig/DE, ²Minden/DE ([email protected]) Purpose or Learning Objective: The complex interactions of the tumor micromilieu could be reflected by diffusion-weighte d imaging (DWI) derived from the magnetic resonance imaging (MRI). The pres ent study investigated the association between apparent diffusion coeffici ent (ADC) values and histopathological features in uterine cervical canc er.

Methods

or Background: This retrospective study used the prebiopsy MRI to analyze histogram ADC-parameters. The biopsy specim ens were stained for Ki 67, E-cadherine, vimentin and tumor-infiltrating ly mphocytes (TIL, all CD45 positive cells). Tumor-stroma ratio (TSR) was calcu lated on routine H&E specimen. Spearman’s correlation analysis and recei ver-operating characteristics curves were used as statistical ana lyses.

Results

or Findings: The patient sample comprised 70 female patients (ag e range 32-79 years; mean age 55.4 years) with squamo us cell cervical carcinoma. The interreader agreement was high rangi ng from intraclass coefficient (ICC)=0.71 for entropy to ICC=0.96 for ADCmedian. Several ADC- histogram parameters correlated strongly with the T SR. The highest correlation coefficient achieved p10 (r=-0.81, p<0.0001). ADCme an can predict tumors with high TSR, AUC: 0.91, sensitivity: 0.91 (95%CI 0.77;0.96), specificity: 0.91 (95%CI 0.78;0.97). Also, several ADC-histogram para meters correlated slightly with the proliferation index Ki 67. No associations were found with TIL, E- Cadherin and vimentin. In well and moderately diffe rentiated cancers, ADC histogram values showed stronger correlations with Ki 67 and TSR than in poorly differentiated tumors.

Conclusion

ADC values are strongly associated with tumor-strom a ratio. ADC mean can be used for prediction of tumors with high TSR. Associations between histopathology and ADC values depend on tum or differentiation. ADC values show only weak associations with Ki 67 and n one with TIL, vimentin and E-Cadherin.

Limitations

First, it is a retrospective study with known inher ent bias. Second, the patient sample is comprised from a single cente r. Funding for this study: None Ethics committee - additional information: Ethics commitee University of Leipzig (Ethical code: 012/13–28012013) Author Disclosures: Alexey Surov: Nothing to disclose Anne-Kathrin Höhn: Nothing to disclose Hans-Jonas Meyer: Nothing to disclose Friday Abstract-based Programme 136 Low-dose pre-operative CT of ovarian tumor with art ificial intelligence iterative reconstruction for diagnosing peritoneal invasion *X. Cai*¹, J. Han², G. Zhang², F. Yang¹, Y. Wang¹, J. Liu¹, R. Li¹; ¹Shijiazhuang/CN, ²Shanghai/CN Purpose or Learning Objective: To test the feasibility of low-dose abdominopelvic CT with an artificial intelligence i terative reconstruction (AIIR) for diagnosing peritoneal invasion in pre-operative imaging of ovarian tumor.

Methods

or Background: In this prospective study, 88 patients with pathology-confirmed ovarian tumors were enrolled, w here the routine-dose CT scan at portal venous phase (120 kV/ref. 200 mAs) w as followed immediately with a low-dose scan (120 kV/ref. 40 mAs). Images a t routine dose were reconstructed with hybrid iterative reconstruction (HIR) and images at low dose were reconstructed with AIIR. Two radiologists inde pendently diagnosed the peritoneal invasion using a 5-point confidence scal e (1: definitely absent, 5: definitely present). In case of disagreement, the c onsensus was obtained through a third radiologist. The diagnostic perform ance was assessed using receiver operating characteristic (ROC) analysis wi th pathological results serving as the reference. The inter-observer agreem ent was assessed by Cohen’s Kappa test.

Results

or Findings: The 88 patients consisted of 37 patients with benign/borderline ovarian tumors and 51 patients wi th ovarian carcinomas. The effective dose of low-dose CT at portal venous phas e was 79.8% lower than that of routine-dose scan (2.64 ± 0.46 mSv vs. 13.04 ± 2.25 mSv, p < 0.001). In diagnosing peritoneal invasion, the area under the ROC curve (AUC) of low- dose AIIR and routine-dose HIR images was 0.961 and 0.960, respectively (p = 0.734). The sensitivity, specificity, and accurac y were 86.1%, 92.3%, and 89.8%, respectively, for low-dose AIIR images, and 86.1%, 90.4%, and 88.6%, respectively, for routine-dose HIR images. The inte robserver agreement was good for diagnosing peritoneal invasion (κ = 0.694).

Conclusion

In low-dose pre-operative CT of ovarian tumor with 80% dose reduction, AIIR delivers similar diagnostic accurac y for peritoneal invasion as compared to routine abdominopelvic CT.

Limitations

Not applicable. Funding for this study: No funding was received for this study. Ethics committee - additional information: The ethics committee notification can be found under the number 2024KS138. Author Disclosures: Jing Liu: Nothing to disclose Ruxun Li: Nothing to disclose Guozhi Zhang: Nothing to disclose Yaning Wang: Nothing to disclose Xiaojia Cai: Nothing to disclose Fan Yang: Nothing to disclose Jintao Han: Nothing to disclose Developing a machine learning model for the differe ntiation of uterine leiomyosarcoma from leiomyomas using clinical and M RI radiomics features *K. Shapriya*, A. Jackson, X. Li, S. Das, N. Bharwa ni, A. G. Rockall; London/UK ([email protected]) Purpose or Learning Objective: Preoperative differentiation between leiomyosarcoma (LMS) and atypical benign leiomyoma (LM) is diagnostically challenging. This study aims to develop and validat e a machine learning (ML) model using MRI-based clinical and radiomic feature s to distinguish LMS from LM.

Methods

or Background: This retrospective study included 214 patients with atypical myometrial lesions who underwent surgery b etween 2013 and 2023. All subjects had preoperative full blood count (FBC ) and MR imaging. Among 214 cases, 193 were LM and 21 were LMS. T2-weighted sagittal MRI sequences were manually segmented then optimized us ing nonuniformity correction method to ensure image stability. 4114 r adiomic features were extracted using TexLab (version 2) and IBSI complia nt MATLAB™ software. These radiomic features and 11 clinical variables ( including age and FBC) were incorporated into several ML models, with the dependent variable being LMS (binary). Data was split 70:30 into training an d testing sets. To address data imbalance, an ensemble classification model wa s employed with unequal classification costs, penalizing the misclassificat ion of LMS. The model’s performance was evaluated using area under the curv e (AUC), sensitivity, specificity, accuracy, F1 score, and confusion matr ix.

Results

or Findings: The final ensemble model included four clinical and six radiomic features. The test dataset demonstrated a sensitivity, specificity, accuracy and AUC of 1, 0.8, 0.84 and 0.90, and F1 s core of 0.50.

Conclusion

This study presents a promising ML model for preope rative differentiation of LMS from LM, achieving high accu racy (84%). As sarcoma subjects are uncommon, the ML model was developed t o take data imbalance into account. High sensitivity was achieved, but wi th some loss of specificity. Future research will focus on validating this model using larger datasets to enhance its reliability and clinical application.

Limitations

None Funding for this study: No funding Ethics committee - additional information: HRA and Health and Care Research Wales (HCRW)

Reference

20/HRA/4925 Author Disclosures: Xingfeng Li: Nothing to disclose Kavita Shapriya: Nothing to disclose Andrea Grace Rockall: Nothing to disclose Saranya Das: Nothing to disclose Alastair Jackson: Nothing to disclose Nishat Bharwani: Nothing to disclose 08:00-09:00 Research Stage 2 Research Presentation Session: Musculoskeletal RPS 1210 Imaging of body composition Moderator G. Guglielmi; Foggia/IT ([email protected]) Clinical validation of a deep learning based automa ted HUAC analysis for improved sarcopenia assessment V. K. Venugopal¹, V. Rengan², *S. Ingole*¹; ¹New De lhi/IN, ²Chennai/IN ([email protected]) Purpose or Learning Objective: To assess the validity and reliability of an automated sarcopenia estimation approach using a de ep-learning based ensemble psoas segmentation and Hounsfield Unit Ave rage Calculation (HUAC) model in comparison to traditional manual me asurements.

Methods

or Background: This study retrospectively analyzed 149 CT scans, comparing sarcopenia assessments between manual HUA C measurements and those derived from an automated TransUNet-based system. The AI model combined convolutional neural networks with Transfo rmer blocks to enhance feature extraction and contextual understanding of muscle tissue, crucial for precise sarcopenia evaluation. The HUAC was calcula ted by measuring the area and mean Hounsfield Units (HU) of the left and right psoas muscles at the L3 vertebra level. Statistical analysis included me an, standard deviation, correlation, paired t-tests, Bland-Altman plots, an d advanced validation metrics such as Intersection over Union (IoU) and Dice coef ficient to evaluate the model's segmentation accuracy.

Results

or Findings: The AI model produced a mean HUAC of 19.66, slightl y higher than the 18.03 from manual assessments, with corresponding standard deviations of 4.27 and 4.54, respectively. The corr elation coefficient of 0.78 indicated strong agreement between the two methods. The model achieved an IoU of 90% and a Dice coefficient of 0.90, demonstr ating high precision in muscle segmentation. The systematic bias observed ( mean difference of -1.63 HUAC) highlights areas for further calibration of t he AI model.

Conclusion

The integration of AI in sarcopenia assessment thro ugh HUAC calculations offers a promising alternative to manu al measurements, providing speed, reproducibility, and precision. Despite some variance, the AI method aligns closely with traditional approaches, suggest ing that with further refinement, it could become a standard tool in clin ical settings.

Limitations

Small sample set Funding for this study: Nil Ethics committee - additional information: IRB Waiver Author Disclosures: Sarang Ingole: Nothing to disclose Vasantha Kumar Venugopal: Consultant: Carpl.ai Vinayak Rengan: Founder: Curium life technologies Qualitative and quantitative CT evaluation of abdom inal fat and muscle tissue in patients with ankylosing spondylitis and investigation of their possible effects on biological agent treatment resp onse N. Kaştan, *I. Erdem Toslak*, S. Bakırcı, A. Yavuz; Antal ya/TR ([email protected]) Purpose or Learning Objective: Ankylosing spondylitis (AS) is a chronic inflammatory disease affecting the axial skeleton. Inflammatory cytokines like TNFalpha and interleukins increase in AS, along wit h adipose tissue and Friday Abstract-based Programme 137 muscle catabolism. Sarcopenia in AS correlates with higher inflammation, greater disease activity, and reduced muscle perfor mance. We hypothesize that the sarcopenia index and quantitative measures of muscle and fatty tissues may relate to the response to biological ag ent treatment

Methods

or Background: This retrospective study involved 62 adults diagnosed with AS who underwent CT at the L2 verteb ra level before any treatment and received biological agent therapy. CT measurements included visceral and subcutaneous abdominal adipose tissue cross-sectional area (VAT cm², SAT cm²), total abdominal muscle area (TA MA), psoas muscle volume (PsoA), sarcopenia index (SMI), visceral and subcutaneous abdominal adipose tissue attenuation (VAT HU, SAT HU), and ps oas muscle attenuation (Pso HU). BASDAI score changes were assessed at the first post-treatment visit to evaluate disease activity. Comparisons and correlations were performed between SMI, adipose and muscle tissue me asurements, and clinical parameters.

Results

or Findings: TAMA, PsoA, and SMI values were significantly highe r in patients with full recovery compared to those wi th partial recovery (p0.05). This study demonstrates that SMI, TAMA, and psoas muscle volume can serve as prognost ic markers for biological treatment response in AS patients.

Conclusion

Our study is an example of opportunistic-quantitati ve imaging

Methods

and it has been shown that SMI value, TAMA and psoas muscle volume value can be used as prognostic markers in r esponse to biological treatment in AS patients.

Limitations

Retrospective design and small sample size were the limitations of our study. Funding for this study: No external funding. Ethics committee - additional information: Local IRB Author Disclosures: Iclal Erdem Toslak: Nothing to disclose Alpaslan Yavuz: Nothing to disclose Nazmi Kaştan: Nothing to disclose Sibel Bakırcı: Nothing to disclose Opportunistic Osteoporosis Assessment from Routine CT - Effect of Intravenous Contrast Agents on Absolute Values, T-S cores, and Derived Classifications in Single- and Dual-Energy CT L. D. Grünewald, V. Koch, S. Mahmoudi, J-E. Scholtz , *J. Gotta*, S. Martin, C. Booz, I. Yel, T. Vogl; Frankfurt/DE Purpose or Learning Objective: To evaluate the impact of intravenous contrast agents on osteoporosis assessment via rout ine CT in arterial and venous phases and identify mitigation strategies us ing dual-energy CT (DECT).

Methods

or Background: 288 patients (154 men, 134 women; median age 62 years) who underwent abdominal DECT scans in non-co ntrast, late-arterial, and portal venous phases between January 2018 and D ecember 2023 were retrospectively analyzed. Trabecular HU values were measured in all phases, including 90kV and 150kV DECT series, using automat ic segmentation. T- scores were calculated to classify patients as oste oporotic, osteopenic, or normal. Changes in HU values, T-scores, and classif ications due to contrast were compared to non-contrast images, with effects quantified using Cohen’s d.

Results

or Findings: Median trabecular HU at L1 was 147 (IQR 116–185). Contrast in late arterial and portal venous phases increased HU values by +14.4 (+11.2%) and +25.7 (+20.7%), respectively. Us ing 150kV DECT reduced these changes to -20.5 (-12.2%) for arterial and -2 3.15 (-12.6%) for venous phases. Cohen’s d was lowest for normal arterial ph ase (+0.55) and highest for 90kV arterial phase (+1.9). Based on T-scores, 120 patients were classified as healthy, 108 as osteopenic, and 60 as osteoporotic. The lowest number of reclassifications occurred in arterial (n=92) and v enous (n=104) phases. For arterial phase, 44 patients shifted from osteoporos is to osteopenia; for venous phase, 52 shifted similarly. High-kV acquisition re duced these reclassifications (n=24 arterial, n=32 venous) but increased shifts f rom healthy to osteopenia.

Conclusion

Intravenous contrast significantly affects HU-based osteoporosis assessment, leading to reclassifications, especiall y from osteopenia to healthy. Using 150kV DECT can partially reduce these reclass ifications, though it may incorrectly shift healthy cases toward osteopenia.

Limitations

Modifying kV settings is not immediately possible w ithout dedicated equipment Funding for this study: No funding was received for this study Ethics committee - additional information: Consent waived due to retrospective nature Author Disclosures: Simon Martin: Nothing to disclose Christian Booz: Nothing to disclose Ibrahim Yel: Nothing to disclose Thomas Vogl: Nothing to disclose Jan-Erik Scholtz: Nothing to disclose Vitali Koch: Nothing to disclose Scherwin Mahmoudi: Nothing to disclose Leon David Grünewald: Nothing to disclose Jennifer Gotta: Nothing to disclose Femoral osteoporosis prediction model using autoseg mentation and machine learning analysis with PyRadiomics on abdom en-pelvic computed tomography (CT) *H. Ha*¹, H. Lim², M. Park¹; ¹Anyang-Si/KR, ²Seoul/ KR ([email protected]) Purpose or Learning Objective: This study aimed to assess the diagnostic performance of osteoporosis prediction by the combi nation of autosegmentation of the proximal femur and machine learning analysis with a

Reference

standard of dual-energy X-ray absorptiome try (DXA)

Methods

or Background: Abdomen-pelvic CT scans were retrospectively analyzed from 1,122 patients who received both DXA and abdomen-pelvic computed tomography (APCT) scan from January 2018 t o December 2020. The study cohort consisted of a training cohort and a temporal validation cohort. The left proximal femur was automatically s egmented, and a prediction model was built by machine-learning analysis using a random forest (RF) analysis and 854 PyRadiomics features. The technica l success rate of autosegmentation, diagnostic test, area under the r eceiver operator characteristics curve (AUC), and precision recall c urve (AUC-PR) analysis were used to analyze the training and validation co horts.

Results

or Findings: The osteoporosis prevalence of the training and validation cohorts was 24.5%, and 10.3%, respective ly. The technical success rate of autosegmentation of the proximal femur was 99.7%. In the diagnostic test, the training and validation cohorts showed 78 .4% vs. 63.3% sensitivity, 89.4% vs. 98.1% specificity. The prediction perform ance to identify osteoporosis within the groups used for training an d validation cohort was high and the AUC and AUC-PR to forecast the occurrence o f osteoporosis within the training and validation cohorts were 90.8% [95% confidence interval (CI), 88.4–93.2%] vs. 78.0% (95% CI, 76.0–79.9%) and 94.6 % (95% CI, 89.3– 99.8%) vs. 88.8% (95% CI, 86.2–91.5%), respectively .

Conclusion

The osteoporosis prediction model using autosegment ation of proximal femur and machine-learning analysis with P yRadiomics features on APCT showed excellent diagnostic feasibility and te chnical success.

Limitations

The limitation of this study was that there was an imbalance in the sex ratio of osteoporosis patients and that this wa s a single-center study. Funding for this study: This work was supported by the Central Medical Service (CMS) Research Fund. The specific grant num ber was not assigned by the company or funder (Central Medical Service Comp any, Ltd., Seoul, Korea). Ethics committee - additional information: The study was approved by the institutional review board of Hallym University Sac red Heart Hospital (No. HALLYM 2020-12-015), and the need for informed cons ent was waived due to the nature of the retrospective analysis. Author Disclosures: Hongil Ha: Nothing to disclose Minsu Park: Nothing to disclose Hyunkyung Lim: Nothing to disclose Radiomic Analysis of Thigh Fat Fraction Maps to Ide ntify Patterns in Neuromuscular Disorders *G. Vignati*, M. Moscatelli, R. Fabrizio, R. Pascuz zo, C. Foschini, F. Doniselli, D. Aquino, F. Mazzi, L. M. Sconfienza; Milan/IT ([email protected]) Purpose or Learning Objective: To analyze radiomics features extracted from thigh fat fraction (FF) maps in order to ident ify common patterns in neuromuscular disorders across different patients.

Methods

or Background: Radiomics features of the classes “shape”, “first- order”, and “gray-level co-occurrence matrix” (GLCM ) were extracted from the thigh FF maps of all patients for each of the 13 VO Is using PyRadiomics, with a fixed bin size of 32. A preliminary feature selec tion step was necessary due to the large number of features extracted (n=1305) relative to the limited number of patients (n=25). After feature selection, the sparse K-means clustering algorithm was applied: it is a clusterin g approach that identifies relevant features while performing clustering. Fina lly, Uniform Manifold Approximation and Projection (UMAP) was used to vis ualize the selected features and statistical analyses were done with R (version 4.3.1) using the caret, sparcl, and umap packages. Friday Abstract-based Programme 138

Results

or Findings: The sparse K-means algorithm identified two cluster s of 14 and 11 patients, respectively, based on 60 selec ted radiomic features from 8 muscles. Clinical diagnosis of patients affected by neuromuscolar disorders is compared with cluster assignment and distinctive features were observed between genetic and inflammatory/autoimmune myopath ies.

Conclusion

This study successfully utilized radiomics features from thigh fat fraction (FF) maps to identify distinct patterns in neuromuscolar disorders across different patients suggesting that radiomic analysis could be a valuable tool for understanding and classifying muscle disor ders in clinical settings.

Limitations

Limitations of this study are the small sample size (only 25 patients), the large number of initial radiomic fea tures (n=1305) and the feature selection process, which could introduce bias. The study also relied on manually segmented regions of interest (VOIs), whic h may introduce variability in the analysis Funding for this study: None Ethics committee - additional information: None Author Disclosures: Domenico Aquino: Nothing to disclose Marco Moscatelli: Nothing to disclose Fabio Doniselli: Nothing to disclose Federica Mazzi: Nothing to disclose Giacomo Vignati: Nothing to disclose Luca Maria Sconfienza: Nothing to disclose Chiara Foschini: Nothing to disclose Riccardo Pascuzzo: Nothing to disclose Renato Fabrizio: Nothing to disclose Multiparametric MRI at 3 and 7 T for characterizati on of skeletal muscle pathology in patients with filamin-C, desmin and LB D3 mutations *C. S. Mathy*, L. V. Gast, T. Gerhalter, M. Türk, T . Bäuerle, A. Doerfler, M. Uder, A. M. Nagel, R. Schröder; Erlangen/DE Purpose or Learning Objective: To characterize patterns of skeletal muscle changes in myofibrillar myopathy (MFM), a group of rare neuromuscular diseases with desmin-positive aggregates and myofib rillar degeneration, using multiparametric MRI.

Methods

or Background: Less affected lower leg of nine patients with genetically confirmed MFM due to FLNC (n=5), DES (n =2) and LDB3 (n=2) mutations (50.9±8.6 years, 6m, 3f) and 10 healthy controls (50.0±11.0 years, 6m, 4f) were examined. 1H-MRI at 3 T included T1-we ighted and T2-weigthed STIR for qualitative assessment of fatty replacemen t/edema, Dixon-type sequence for proton-density fat fraction (PDFF) qua ntification and diffusion- tensor imaging (DTI) for characterization of (micro -)structural changes. 39K/23Na-MRI acquisition-weighted Stack-of-Stars se quences at 7 T allowed after partial-volume and relaxation correction quan tification of apparent tissue sodium/potassium concentrations (aTSC/aTPC).

Results

or Findings: Muscular fatty replacement and edema-like alteratio ns were highly variant intermuscular and interindividu al. 35/63 of elevated muscle compartments of patients with MFM were highly fatty replaced (PDFF >50%). Calculated apparent diffusion coefficients (ADCs) f rom DTI were reduced in gastrocnemius lateralis (GL), peroneus (PER) and ex tensor digitorum longus (EDL) muscles (p = 0.04 – 0.03) when excluding high ly fatty replaced muscles, but simulations showed that this behavior could pri marily be attributed to increasing PDFF. Fat-corrected aTSC were increased in all muscle regions (mean all muscles: 55.6±16.3 mM vs 23.2±5.5 mM, p <0.001), aTPC decreased in all regions but GL and PER (mean all m uscles: 75.4±13.3 mM vs 108.9±9.9 mM, p <0.001).

Conclusion

Alterations of 39K/23Na ion homeostasis that go bey ond changes caused by fatty-replacement in irreversible disease stages could be proved in patients with MFM. This could form the ba sis for a novel biomarker for determining early disease extent and disease re sponse to therapies. (Micro-)structural changes were indistinguishable f rom mere fatty replacement changes.

Limitations

Low number of participants (MFM prevalence <1:100.0 00). Funding for this study: C.S.M. and T.B. were founded by the Deutsche Forschungsgemeinschaft (DFG, German Research Founda tion) – 493624887 (Clinician Scientist Program NOTICE). Funding by th e DFG is gratefully acknowledged (project 500888779 / RU5534 MR biosign atures at UHF). Ethics committee - additional information: Approved by local ethic comittee of Friedrich-Alexander University Erlangen-Nurember g Author Disclosures: Tobias Bäuerle: Nothing to disclose Teresa Gerhalter: Nothing to disclose Claudius Sebastian Mathy: Nothing to disclose Matthias Türk: Nothing to disclose Michael Uder: Nothing to disclose Rolf Schröder: Nothing to disclose Lena V. Gast: Employee: Siemens Healthineers Armin M. Nagel: Nothing to disclose Arnd Doerfler: Nothing to disclose The Top 100 Most Cited Articles on Musculoskeletal Radiology: A Bibliometric Analysis *L. Moore*; Dublin/IE ([email protected]) Purpose or Learning Objective: To identify and characterize the most influential publications relating to the musculoske letal system and radiology. The number of citations an article receives is refl ective of its impact in the scientific community.

Methods

or Background: The top 100 most cited articles were identified using the Web of Science database. Data pertaining to the year of publication, publishing journal, journal impact factor, authorsh ip, country of origin and institution were collected.

Results

or Findings: The number of citations per article for the top 100 list ranged from 149 to 709 (median 208; mean 240). The average number of citations per year, per article, ranged from 5 to 6 0 (median 12, mean 26). The United States was the most common country of origin (n=74). The Journal with the greatest number of articles was Radiology (n=34 ). The University of California contributed the most articles (n=11).

Conclusion

This study presents a detailed analysis of the top 100 most-cited articles published in musculoskeletal radiology. It provides clinicians and researchers with insight into the current influenti al research papers in this field and the characteristics of those studies. It also h ighlights research trends and areas that may benefit from further research.

Limitations

The use of citation count is a source of bias; the more time that has passed since the publication of an article, the more likely it is to be cited over time. In order to mitigate this source of bias , the average citation count per year was also used. Some articles may have been inadvertently excluded as a result of search criteria used. In addition, using journal IF from one particular year (2024) does not allow for temporal fluctuations in IF. Furthermore, the potential bias of self-citation wa s not accounted for in this study. Funding for this study: None. Ethics committee - additional information: This article does not require ethics committee approval. Author Disclosures: Lucia Moore: Nothing to disclose 08:00-09:00 Research Stage 3 Research Presentation Session: Physics in Medical Imaging RPS 1213 Evolution of CT: a key to its sustainability Moderator N. Saltybaeva; Lucerne/CH ([email protected]) Validation of a novel CBCT reconstruction algorithm for treatment planning and IGRT in neoadjuvant radiotherapy of lo cally advanced rectal cancer patients *M. C. Daniotti*¹, S. Trivellato², L. De Sanctis¹, V. Pisoni², J. Stancanello³, J. Mason³, R. Pellegrini³, S. Arcangeli², E. De Pon ti²; ¹Milan/IT, ²Monza/IT, ³Stockholm/SE ([email protected]) Purpose or Learning Objective: A new CBCT reconstruction algorithm based on poli-energetic quantitative (Polyquant) method e mpowered with a convolutional neural network scatter correction has been recently proposed. This study aimed to validate the use of the Polyqua nt CBCTs (pCBCTs) for image-guided radiotherapy (IGRT) and planning for l ocally advanced rectal cancer (LARC).

Methods

or Background: Translational shifts obtained on all axes with the registration of pCBCTs to CT were compared to the c linical version of Elekta XVI CBCT ones and statistical significance was inve stigated with the t-test and ANOVA-test. pCBCTs were calibrated with a populatio n-based curve (pop-CC) elaborated coupling pCBCT gray levels to the CT rel ative electron density (RED) for ten pelvic patients. pop-CC was validated by comparing dose calculations on pCBCT and bulk density pseudo-CT us ing 1%/3mm local gamma-analysis. Five LARC patients treated on Elekt a VersaHD were selected. The RED difference between CT and first s ession-pCBCT were assessed on a voxel-to-voxel basis, on a contour ba sis, and on a dosimetric basis using 1%/3mm local gamma-analysis to compare pCBCT and pseudo- CT calculations. Friday Abstract-based Programme 139

Results

or Findings: The translational differences between IGRT results of CBCT and pCBCT were always 95%. For LARC patients, voxel- to-voxel and structure-based analysis showed no rel evant RED discrepancies between pCBCT and CT. Residual RED differences resu lted dosimetrically negligible compared with dose distributions calcula ted on pseudo-CT, with gamma passing rates > 95%.

Conclusion

The optimized pCBCTs were successfully RED calibrat ed and validated for IGRT and planning for LARC radiothera py. The results suggest that pCBCTs could be exploited in the clinical work flow for adaptive radiotherapy in LARC patients,

Limitations

Further investigation for their extended use might still be necessary. Funding for this study: No funding Ethics committee - additional information: Not applicable Author Disclosures: Jonathan Mason: Employee: Elekta AB Sara Trivellato: Nothing to disclose Martina Camilla Daniotti: Nothing to disclose Stefano Arcangeli: Nothing to disclose Elena De Ponti: Nothing to disclose Valerio Pisoni: Nothing to disclose Joseph Stancanello: Employee: Elekta AB Lorenzo De Sanctis: Nothing to disclose Roberto Pellegrini: Employee: Elekta AB The impact of detector coverage on motion artefacts in photon-counting CT imaging using a motion phantom *E. Verelst*¹, G. Van Gompel¹, D. Crotty², H. Linde r³, P. D. Deak⁴, J. De Mey¹, N. Buls¹; ¹Brussels/BE, ²Cork/IE, ³Stockholm/SE, ⁴Münsingen/CH Purpose or Learning Objective: Reducing motion-induced artefacts is an increasingly important aspect of ultra high-resolut ion (UHR) photon-counting CT (PCCT). Using a motion-controlled phantom, this study evaluates the benefit of using wider detector coverage in PCCT to reduce motion artefacts.

Methods

or Background: A tissue-simulating cuboid phantom, 80-mm in length, was mounted onto a custom-built motion-cont rolled phantom. A 3-mm diameter stent (Superia, Nano-Therapeutics, India) was inserted into a commensurate hole inside the phantom. Programmed to simulate pulsatile motion, helical images were acquired during motion by a prototype Silicon- based PCCT (Si-PCCT) using 40- and 80-mm detector c overage, representing table speeds of 128.6-mm/s and 257.1-mm/s, respecti vely. Motion-induced image artefacts were evaluated by measuring the vol umetric error relative to the nominal phantom volume. A 5-point Likert rating evaluated stent appearance for both detector coverages against a re ference static image (1-no similarity, 5-similar). Differences were assessed u sing a paired sample t-test and Wilcoxon signed-rank test, respectively. P-valu es < 0.05 indicated statistical significance.

Results

or Findings: Relative to the nominal phantom volume (1571-mm3), the 80-mm detector coverage statistically significa ntly reduced volumetric error (mean 81-mm3, SD 18-mm3) compared to the error gene rated with 40-mm coverage (254-mm3, 48-mm3), p=0.006. Stent appearan ce under wider detector coverage was likewise assessed to better m atch the reference image, with average Likert scores for 40 mm and 80mm of 1 [1–1.5] and 4 [3.5–3.75], respectively, p=0.043.

Conclusion

In high-resolution PCCT imaging, to combat motion-i nduced artefacts, it is important to combine wide-detector UHR CT acquisitions with higher table speeds. This study demonstrates the po tential of using a prototype wide-coverage Si-PCCT system with fast tables speed s to reduce such artefacts. While this study specifically evaluated pulsatile motion, an 80-mm detector coverage may reduce additional body motion -induced artefacts, such as peristalsis.

Limitations

This is an ex-vivo phantom study. Funding for this study: Flemish Research Foundation (FWO), personal grant, nr: 1SH1Z24N. Ethics committee - additional information: Ethical approval was not required for this study Author Disclosures: Johan De Mey: Nothing to disclose Emma Verelst: Nothing to disclose Dominic Crotty: Employee: GE Healthcare Gert Van Gompel: Nothing to disclose Paul D. Deak: Employee: GE Healthcare Nico Buls: Nothing to disclose Hugo Linder: Employee: GE Healthcare 3D-printed Anthropomorphic Head Phantom Featuring W hite and Gray Matter Structures for Evaluating CT Imaging K. Mei¹, L. Roshkovan¹, S. Sharma², S. Ross², J. Wo o¹, S. S. Halliburton³, *L. Liu*¹, R. Thompson³, P. Noël¹; ¹Philadelphia, P A/US, ²Vernon Hills, IL/US, ³Mayfield Village, OH/US Purpose or Learning Objective: To develop a 3D-printed, patient-specific brain phantom for assessing performance of non-cont rast CT head imaging.

Methods

or Background: Unenhanced T1 MRI scan (best available gold standard) of a healthy brain (21y/o,F) was retrospe ctively collected and converted into CT Hounsfield unit image to generate a realistic phantom using PixelPrint technique. The brain phantom was created as a 30 mm thickness section including both left and right hemispheres ( approximately 157 x 120 mm) at 1:1 scale. Additionally, a separate skull ph antom was printed from the same patient images using calcium-doped filament. T he brain phantom, with surrounding skull phantom, was scanned with CT at 1 20 kVp and 24.4 mGy. Images were reconstructed with and without iterativ e denoising at 0.5 mm pixel spacing. Attenuation values were measured in gray a nd white matter.

Results

or Findings: Gray and white matter were clearly distinguishable in an appropriate CT examination window, with and without denoising. Line profile was plotted along the center of the phantom. Realis tic CT values of approximately 45 HU for gray matter and 25 HU for w hite matter were observed. Maximum density observed in the skull rea ched approximately 750 HU, which was limited by the density of the 3D-prin ting filament used. Image noise, estimated by standard deviation, ranged betw een 3 and 4.5 HU across both sets of denoised images.

Conclusion

The PixelPrint 3D-printed brain phantom successfull y depicts realistic tissue attenuationfor white and gray matt er, demonstrating potential as a valuable tool for evaluating CT head imaging perf ormance.

Limitations

This study converts MR images to CT numbers and sim ulates non-contrast CT scans. Funding for this study: This work was partly supported by Canon Medical Systems Corporation (Otawara, Japan). Ethics committee - additional information: University of Pennsylvania Author Disclosures: Steven Ross: Employee: Canon Medical Research USA Kai Mei: Nothing to disclose Peter Noël: Nothing to disclose Richard Thompson: Employee: Canon Healthcare USA Leening Liu: Nothing to disclose Sandra Simon Halliburton: Employee: Canon Healthcar e USA Leonid Roshkovan: Nothing to disclose Shobhit Sharma: Employee: Canon Medical Research US A John Woo: Nothing to disclose Towards functional lung color K-edge imaging enable d by spectral photon-counting CT in combination with dedicated co ntrast agents: a phantom study *A. J. Gutwinska*¹, D. Rosario², A. Pang², C. A. He rnandez-Fajardo¹, R. Coulibaly¹, A. Robert¹, S. Rit¹, D. P. Cormode², S. A. Si-Mohamed³; ¹Lyon/FR, ²Philadelphia, PA/US, ³Bron/FR ([email protected]) Purpose or Learning Objective: To evaluate the image quality of color K- edge imaging for contrast agents based on 8 differe nt elements using a clinical prototype spectral photon-counting CT (SPCCT).

Methods

or Background: A SPCCT with a field-of-view of 500mm was used (Philips; Israel). An anthropomorphic thoracic phan tom (QRM GmbH) with twelve 1.5mL K-edge solutions (gadolinium-Gd, holmi um-Ho, ytterbium-Yb, hafnium-Hf, tantalum-Ta, tungsten-W, gold-Au, bismu th-Bi) ranging from 0 to 2mg/mL was scanned at 120kVp and 50/75/150mAs. Five acquisitions per agent were performed using dedicated energy thresho lds. Conventional images in Hounsfield units and color K-edge images in mg/mL were reconstructed with isotropic voxels of 0.7mm3. Nois e, mean relative error (MRE) between prepared and measured concentrations, signal-to-noise ratio (SNR) were measured on color K-edge images. Contras t-to-noise ratio (CNR) on conventional and color K-edge images were measur ed and compared.

Results

or Findings: Mean noise ranged from 0.04 to 0.13mg/mL among all samples with a lowest value for Gd at 75mAs (0.04±0 .01 mg/mL) and highest for Ta at 50mAs (0.13±0.03mg/mL). Overall MRE was 2 9.7% with higher accuracy for Yb (e.g., 7.0% at 150mAs), and lower f or Ta (55.7% at 150mAs). SNR increased as function of concentrations with a factor per mg of 16.7, 14.6, 10.9, 9.5, 9.3, 9.3, 8.8, 4.7 for Gd, Yb, Ho, Hf, A u, W, Ta and Bi, respectively, at 75mAs. CNR in color K-edge images increased as f unction of concentrations, and were higher in comparison to CN R in conventional images (e.g., 1075%, 957%, 677%, 658%, 601%, 590%, 536%, 4 11%, for Gd, Au, W, Ta, Yb, Ho, Bi and Hf, respectively, at 75mAs).

Conclusion

Image quality of color K-edge imaging in an anthrop omorphic thoracic phantom demonstrated high performances for 8 color K-edge agents using SPCCT whilst outperforming sensitivity in com parison to conventional imaging. Friday Abstract-based Programme 140

Limitations

Phantom study. Funding for this study: The ERC starting Grant "KOLOR SPCCT Imaging" (N°101118079). Ethics committee - additional information: Not applicable. Author Disclosures: Antoine Robert: Nothing to disclose Salim Aymeric Si-Mohamed: Nothing to disclose David Peter Cormode: Nothing to disclose Christian Alejandro Hernandez-Fajardo: Nothing to d isclose Simon Rit: Nothing to disclose Agnieszka Joanna Gutwinska: Nothing to disclose Derick Rosario: Nothing to disclose Ramata Coulibaly: Nothing to disclose Amanda Pang: Nothing to disclose Task-based image quality evaluation of ultra-high r esolution color K-edge imaging enabled by spectral photon-counting CT: a p hantom study *R. Coulibaly*¹, A. Robert¹, A. Houmeau¹, M. N. Ant onuccio², P. C. Douek¹, S. Rit¹, J. Greffier³, S. A. Si-Mohamed¹; ¹Lyon/FR, ²Paris/FR, ³Nimes/FR ([email protected]) Purpose or Learning Objective: To evaluate the image quality of color K- edge imaging with a gadolinium agent using spectral photon-counting CT (SPCCT) in a phantom with a mixture of contrast age nts.

Methods

or Background: A clinical prototype SPCCT system (FOV 500mm, Philips; Israel) was used to scan custom-made cylin drical phantom of 27cm (Color iQCT). Three inserts of the phantom were fil led up with agents as follows: iodine only, mixture of iodine and gadolin ium, gadolinium only. Two configurations were considered, one with 0.5mg/mL o f each contrast agent and another with 2mg/mL. For each configuration, two se ries of nine helical scans (120kVp) were acquired at 75mAs and 150mAs. Spectra l K-edge images of gadolinium were obtained by doing material decompos ition using 3 basis (water/iodine/gadolinium), using an iterative recon struction algorithm at 3 levels (iDose 0, 6, 11) were compared between inserts with gadolinium using iQMetrix-CT software.

Results

or Findings: Despite the presence of iodine, color K-edge imagin g enable specific differentiation of the gadolinium, showing a concentration difference of 1.1% between inserts of gadolinium on ly and mixture (150mAs, i11, 2mg/mL). NPS peak was observed at the same spa tial frequency in all configurations (i.e., 0.053±0.018 mm⁻¹), whereas the noise magnitude decreased when the dose and the iDose4 levels incre ased (-19.65±0.01% between iDose0 and iDose11). TTF values at 50% (f50 ) were similar between inserts with gadolinium only and the mixture, regar dless of the dose or iDose levels (e.g., 0.213±0.090 mm⁻¹ vs 0.237±0.061 mm⁻¹, at 150mAs, i11, 0.5mg/mL). f50 values were improved with increasing dose, iDose levels, and with higher concentrations.

Conclusion

Color K-edge imaging of a gadolinium contrast agent using SPCCT demonstrated high spatial resolution and low noise magnitude at low concentrations, even in the presence of iodine.

Limitations

Phantom study. Funding for this study: ERC starting Grant "KOLOR SPCCT Imaging" (N°101118079). Ethics committee - additional information: No ethics was required for this phantom study. Author Disclosures: Antoine Robert: Nothing to disclose Salim Aymeric Si-Mohamed: Nothing to disclose Angele Houmeau: Nothing to disclose Philippe Charles Douek: Nothing to disclose Simon Rit: Nothing to disclose Joel Greffier: Nothing to disclose Ramata Coulibaly: Nothing to disclose Maria Nicole Antonuccio: Employee: Philips X-ray phase contrast imaging, moving beyond traditi onal X-ray imaging

Methods

a first pilot in intra-operative specimen imaging *G. Havariyoun*; London/UK ([email protected]) Purpose or Learning Objective: Several surgical procedures benefit from the ability to image resected tissue samples in real ti me, e.g. to ensure no margin involvement. Micro-CT or tomosynthesis have great p otential, but suffer from limited soft-tissue sensitivity of X-rays. X-ray ph ase contrast imaging (XPCI) provides soft tissue sensitivity and increased cont rast through exploitation of phase effects. This work is presented on behalf of the UCL AXIm team.

Methods

or Background: XPCI was initially restricted to specialized facili ties such as synchrotrons. Our group has developed a met hod that has enabled creation of a pre-commercial prototype compatible w ith surgical and radiology workflows. This has been used to image >100 breast tissue samples from breast conserving surgery both in vitro and in real time. Images were compared to standard specimen radiography and histo pathology. The system also allows higher resolution (~10 micrometre) imag ing in slower scans for e.g. digital histology.

Results

or Findings: System optimization (which also included size reduc tion) led to clinically acceptable scan times, which were verified by trialling the system in a real intra-operative context. XPCI imag ing resulted to sensitivity and specificity values of 83% (95% CI 69-92%) and 8 3% (95% CI 70-92%), respectively. Standard specimen radiography resulte d to sensitivity and specificity values of 32% (95% CI 20-49%) and 86% ( 95% CI 73-93%), respectively.

Conclusion

XPCI has a specificity comparable to standard speci men radiography but a significantly higher sensitivity. This would lead to significant reduction in re-excision rates and in turn a reduct ion in patient stress, surgical times, healthcare costs and improved cosmetic outco mes.

Limitations

Comparisons with standard specimen radiography were made as this is the most commonly used tool in the clinical setting. Comparison with more advanced techniques will be made in the future . Funding for this study: This work is funded by the Wellcome Trust (Grant 200137/Z/15/Z). Alessandro Olivo (AXIm lead) is fun ded by the Royal Academy of Engineering under their “Chairs in Emerg ing Technologies” scheme (CiET1819/2/78). Ethics committee - additional information: The Breast Cancer Now Tissue Bank (Approval No. 15/EE/0192) provided the ethical ly approved samples, the authors thank the patients who have generously cons ented to donate their tissues which have been utilised in this work. Author Disclosures: Glafkos Havariyoun: Nothing to disclose A workflow to harmonize CT abdomen protocols beyond dose equalization *J. Vignero*, B. Miseur, J. Binst, H. Bosmans; Leuv en/BE ([email protected]) Purpose or Learning Objective: A radiologist had raised concerns about excessive noise in CT abdomen images of scanner A, while another scanner (B) of the same model, using similar protocol setti ngs, produced images of acceptable quality. This discrepancy led to a study aimed at harmonizing protocol settings to ensure consistent image qualit y across all patient sizes, moving beyond standard dose equalization methods. C urrently, protocol adjustments—based on dose level, scan task, and tub e current modulation strength—are often made intuitively.

Methods

or Background: Clinical CT scan data, collected through a dose monitoring platform (DOSE, Qaelum), was analysed us ing water equivalent diameter (WED), global noise level (GNL), kVp and C TDI. Three phantoms with ellipsoid cross-sections (WED: 22, 33 and 43cm ) and iodine/calcium inserts were scanned to map protocol settings to sc an parameters (kVp, mAs, CTDI) and GNL for each scanner and phantom. The rec onstruction settings were kept fixed.

Results

or Findings: Initially, scanner A and B produced similar radiati on doses for the same WED groups, but scanner A had on average a 13% higher GNL, with a maximum of 18% difference for the small est WED group. The radiologist could define GNL upper limits for each kVp. Using the phantom scan maps, protocol settings were chosen to achieve the desired image quality. Post-optimization, 87% of scans met the GN L criteria, compared to 70% before. To achieve this, radiation doses were i ncreased with 17%.

Conclusion

Identical scanner models and protocol settings do n ot guarantee consistent image quality. Quality measures should b e included in optimization efforts. We present a procedure using new phantoms and dedicated metrics to improve protocol harmonization.

Limitations

The pipeline has only be verified on two scanners. Funding for this study: In part funded by the i-Violin project that is co-f unded under the EU4Health Programme 2021-2027, grant agre ement no. 101056832. Ethics committee - additional information: Retrospective, technical study Author Disclosures: Hilde Bosmans: Shareholder: Qaelum Janne Vignero: Nothing to disclose Joke Binst: Nothing to disclose Bram Miseur: Nothing to disclose Friday Abstract-based Programme 141 Impact of Acquisition Parameters on Quantitative Im aging Using Rapid kVp-Switching Spectral CT *O. Sandvold*¹, A. Perkins², H. Daerr³, T. Koehler³ , R. Proksa¹, R. Manjeshwar², P. Noël¹; ¹Philadelphia, PA/US, ²Cl eveland, OH/US, ³Hamburg/DE Purpose or Learning Objective: To investigate the quantitative effects of varying the ratio of high and low kVp tube voltage durations on spectral CT

Results

with a rapid kVp-switching X-ray tube.

Methods

or Background: Rapid kVp-switching offers excellent spectral separation, but experimental research on acquisitio n parameters is limited. This study addresses the gap. A rapid kVp-switching X-ray tube (Philips Healthcare) on a spectral CT bench system was opera ted at 500 mAs alternating between 140 and 80 kVp. The total integ ration period (IP) containing high and low kVp IPs was 1 ms. We varied the ratio of 140 kVp duration to total IP time from 0.15 to 0.85 and cor respondingly adjusted the low kVp IP time. A 3D-printed plastic phantom (20 cm di ameter) was rotated at 1 Hz and contained four tissue-mimicking inserts: iod ine (2.0, 5.0 mg/ml), iodine 4.0 mg/ml + human equivalent (HE) blood, and HE blo od (Sun Nuclear). We performed reconstruction and two-material decomposi tion without applying denoising. The noise and contrast-to-noise ratio (C NR) of the known concentration inserts were measured in photoelectri c material images. Dose was estimated from the reference detector.

Results

or Findings: Dose comparisons showed 0.15, 0.25, 0.33, and 0.67 ratio scans used 47%, 51%, 63%, and 91% of the high est dose scan (0.85 ratio). Measured noise values appeared to follow a quadratic trend as the ratio increased, with 0.33 containing the lowest average noise. The 0.33 ratio image dose normalized CNR was approximately 1.24x, 1.14x, 1.11x, and 1.68x greater than the dose normalized CNR in 0.15, 0.25, 0.67, and 0.85 ratio images respectively.

Conclusion

Dose normalized CNR depends strongly on the ratio b etween high and low kV durations. This ratio should be con sidered for optimizing the spectral acquisition.

Limitations

None Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Heiner Daerr: Employee: Philips Innovative Technolo gies Thomas Koehler: Employee: Philips Innovative Techno logies Olivia Sandvold: Nothing to disclose Peter Noël: Nothing to disclose Roland Proksa: Nothing to disclose Ravindra Manjeshwar: Employee: Philips Healthcare Amy Perkins: Employee: Philips Healthcare 08:00-09:00 Research Stage 4 Research Presentation Session: Breast RPS 1202 Improving the clinical impact of contrast- enhanced mammography Moderator R. Alcántara; Barcelona/ES Author Disclosures: Rodrigo Alcántara: Advisory Board: GE Healthcare; R esearch Grant/Support: GE Healthcare; Speaker: GE Healthcare, BD, Bayer CEM for the assessment of screening recalls: diagno stic performance at three-year follow-up *S. Marziali*¹, A. Cozzi¹, M. Fanizza¹, V. Magni¹, L. Menicagli¹, A. Benedek¹, G. Di Giulio², F. Sardanelli¹; ¹Milan/IT, ²Pavia/IT ([email protected]) Purpose or Learning Objective: To evaluate the diagnostic performance of contrast-enhanced mammography (CEM) for the assessm ent of screening recalls.

Methods

or Background: Recalled women were prospectively enrolled at two centres to undergo CEM alongside standard assessmen t (SA) through additional views, tomosynthesis, and/or ultrasound between January 2019 and July 2021. Exclusion criteria were symptoms, implan ts, allergy to contrast agents, renal failure, and pregnancy. SA and CEM we re independently evaluated by one of six radiologists, who recommend ed biopsy or ≥3-year follow-up. Diagnostic performance of CEM (low-energ y plus recombined images) was calculated considering histopathology f or lesions biopsied and/or surgically removed as well as ≥3-year follow-up for both breasts.

Results

or Findings: Between January 2019 and July 2021, 220 women were enrolled, 207 of them (median age 56.6 years) with 225 suspicious findings, 135 of them referred for biopsy (4 by rCEM alone, 2 /4 being one DCIS and one invasive carcinoma). During ≥3-year follow-up, 4 interval cancers were reported: one mucinous invasive carcinoma and one D CIS at the site of previous assessment with biopsy; one DCIS in a diff erent quadrant of the same breast; and one invasive carcinoma NST at the contr alateral breast (found at the third year after CEM). No woman was lost at fol low-up. The overall CEM performance was: sensitivity 80/84 (95.2%, 95% CI 8 8.3−98.7%); s pe c ific ity 108/123 (87.8%, 95% CI 80.7−90.3%9).

Conclusion

The role of CEM in the assessment of recalls is con firmed at over 3-year follow-up in terms of both sensitivity and s pecificity. The two cases of interval cancers at the site of previous biopsy hig hlight the need of radiologic- pathologic correlation.

Limitations

Limited sample size. Funding for this study: GE Healthcare Ethics committee - additional information: San Raffaele Hospital, Milan, Italy Author Disclosures: Andrea Cozzi: Nothing to disclose Sara Marziali: Nothing to disclose Francesco Sardanelli: Speaker: Bayer AG, Siemens He althineers Advisory Board: Bayer AG, Bracco imaging, GE healthcare Rese arch/Grant Support: Bayer AG, Bracco imaging, GE healthcare, Marianna Fanizza: Nothing to disclose Giuseppe Di Giulio: Nothing to disclose Veronica Magni: Nothing to disclose Adrienn Benedek: Nothing to disclose Laura Menicagli: Nothing to disclose COntrast enhanced Mammography in women with previou s Breast cancer Operated with conserving surgery (COMBO TRIA L): interim

Results

of a prospective intraindividual study *G. Vatteroni*¹, N. Turri¹, F. Fici¹, M. Filippini² , N. Basla³, G. Pinna¹, G. Pruneddu¹, R. M. Trimboli¹, D. Bernardi¹; ¹Milan /IT, ²Brescia/IT, ³Pavia/IT ([email protected]) Purpose or Learning Objective: To present interim results from the 'COMBO TRIAL, a prospective intraindividual study evaluati ng the performance of Contrast-Enhanced Mammography(CEM) vs Digital Mammo graphy(DM) for the surveillance of women with a personal history of br east cancer(BC).

Methods

or Background: Between January 2023 and April 2024, women who underwent breast-conserving surgery for BC were inv ited to undergo CEM for routine surveillance. Exclusion criteria included: suspicious symptoms of BC, allergy to iodinated contrast agents, renal failure , breast implants. For each patient, one reader reported CEM while a second rea der, independent and blinded, evaluated only LE images equivalent to DM. The reference standard was 1-year follow-up for negative cases and biopsy/ surgery for BI-RADS 4/5. Cancer Detection(CD) rate for both DM and CEM and i ncremental CD rate for CEM, sensitivity, specificity, PPV, NPV and accurac y were calculated. We evaluated differences in diagnostic performance bet ween DM and CEM using McNemar test(p<0.05 significant).

Results

or Findings: Overall, 600 women were included in the analysis wi th a recall rate of 9.8%. Among them, 14 cases of BC wer e detected: 8 (5 DCIS +3 invasive) were identified by both DM and CEM, while CEM detected 6 additional cases (1 DCIS+5 invasive). Three cases w ere missed by both DM and CEM but subsequently detected by US, resulting in a global recurrence rate of 2.8%. The CD rate for CEM was 23 per 1000, compared to 13 per 1000 for DM, indicating an incremental CD for CEM of 10 per 1000 (p=0.014). Compared to DM, CEM demonstrated significantly high er sensitivity(82.4% vs. 47.1%), slightly lower specificity(96.4% vs. 97.6%) , slightly higher PPV(40.0% vs. 36.4%), slightly higher NPV(99.5% vs. 98.4%), a nd similar accuracy(96.0% vs. 96.2%).

Conclusion

Implementation of CEM in BC surveillance was associ ated with a significant increased detection of invasive cancers .

Limitations

n/a Funding for this study: this study received research support by Siemens Helthineers Ethics committee - additional information: The ethics commitee approved this study Author Disclosures: Rubina Manuela Trimboli: Nothing to disclose Federica Fici: Nothing to disclose Giulia Pruneddu: Nothing to disclose Daniela Bernardi: Nothing to disclose Nicolò Turri: Nothing to disclose Giulia Vatteroni: Nothing to disclose Marco Filippini: Nothing to disclose Nicoletta Basla: Nothing to disclose Giulia Pinna: Nothing to disclose Friday Abstract-based Programme 142 Evaluation of lesion conspicuity on contrast-enhanc ed mammography improves the performance in the assessment of malig nancy *M. Conti*, R. Rella, S. Palma, S. Amodeo, N. Di Ca taldo, D. Moretti, M. Costantini, O. Tommasini, P. Belli; Rome/IT ([email protected]) Purpose or Learning Objective: Aim of the study is to assess the performance of Lesion Conspicuity (LC) in Contrast Enhanced Mammography (CEM) in the prediction of malignancy.

Methods

or Background: This is an observational retrospective study involving 153 women (median age, 44.1 years, IQR: 3 6-52) who underwent CEM and subsequent histological assessment at Polic linico Universitario A. Gemelli IRCCS (April 2021-October 2023). Two radiol ogists (with 2 and 7 years of experience in breast imaging, independentl y) evaluated low-energy (LE) images and LC (categorizing it as absent, low, moderate or high) and assigned a BIRADS category of suspicion to the lesi on basing on both. Diagnostic performance of LE images and LC interpre tation together was calculated using histological results of the biopsy as gold standard. Subgroup analyses based on mammographic breast density, back ground parenchymal enhancement (BPE) on CEM and type of lesions were a lso performed.

Results

or Findings: The interpretation of LE images together with the v alue of LC showed a sensitivity (SE) of 96.8% (95%CI: 92 .1%-99.1%) and a specificity (SP) of 66.7% (95%CI: 46.0%-83.5%) vs. a SE=90.7% (95%CI: 82.5%-95.9%) and a SP=52.6% (95%CI: 28.8%-75.5%) fo r LE images evaluation alone. Diagnostic performance of LE imag es + LC evaluation was better than LE images alone both in dense (SE=96.4% and SP=89.8% vs. SE=93.3% and SP=60.0%) and adipose (SE=97.3% and SP =66.7% vs. SE=89.3% and SP=44.4%) breasts. Diagnostic performa nce of LE images + LC evaluation is better than LE images alone both i n minimal/mild and moderate/marked BPE and in all types of lesions.

Conclusion

The evaluation of LE images joint with the value of LC demonstrated a better performance than LE images al one in predicting the malignancy of lesions.

Limitations

Limitations of the present study include its retros pective and monocentric design. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by the local Institutional Review Board (ID: 6476) Author Disclosures: Nicola Di Cataldo: Nothing to disclose Rossella Rella: Nothing to disclose Oscar Tommasini: Nothing to disclose Simone Palma: Nothing to disclose Marco Conti: Nothing to disclose Melania Costantini: Nothing to disclose Silvia Amodeo: Nothing to disclose Delia Moretti: Nothing to disclose Paolo Belli: Nothing to disclose Diagnostic Accuracy of Contrast-Enhanced Mammograph y (CEM) in Preoperative Staging of Breast Tumors: A Comparativ e Study with Histology and Mammography *M. Balbino*¹, F. Masino², M. Montatore³, S. Surian o⁴, F. A. Carpagnano⁴, G. Capuano³, G. Guglielmi⁵; ¹Triggiano/IT, ²Bari/IT, ³Barletta/IT, ⁴Foggia/IT, ⁵Andria/IT ([email protected]) Purpose or Learning Objective: To evaluate the diagnostic accuracy of CEM in detecting and measuring breast tumors, comparing it with histological findings.

Methods

or Background: This study evaluated the diagnostic accuracy of Contrast-Enhanced Spectral Mammography (CEM) in 69 breast cancer patients treated at the “Santa Maria” Hospital (Bar i, Italy) from January 2018 to September 2023. Patients ranged from 33 to 86 years old, with an average age of 55.8 years. All underwent CEM and subsequent bio psy or cytology for confirmation. Exclusion criteria included deep lesi ons, post-biopsy hematomas, or renal insufficiency. CEM was performed using a G E Healthcare system with two post-injection projections: Cranio-Caudal and M edio-Lateral-Oblique. Three experienced radiologists analyzed the scans, comparing lesion sizes from mammography, CEM, and histology.

Results

or Findings: Results showed that CEM underestimated lesion size in 28 cases, while mammography did so in 21. CEM ident ified 92.3% of multifocal masses, compared to 58.3% with mammography. In hist ology, 49.27% of patients had mass-like neoformations, with 33.3% sh owing homogeneous contrast enhancement. High contrast enhancement cor related with higher Ki67 proliferation indices (50-65%). The majority of neo plasms were Infiltrating Ductal Carcinomas. Statistical analysis revealed th at CEM was more accurate than mammography, with the Wilcoxon signed-rank tes t showing no significant difference between CEM and histological measurement s (p=0.9928). CEM showed high diagnostic accuracy in preoperative sta ging, supporting its use over MRI, particularly due to its lower cost, faste r acquisition time, and better patient tolerance. The study concludes CEM is a val uable, cost-effective alternative to MRI for assessing tumors before surg ery, particularly in preoperative staging and the identification of mult ifocal lesions.

Conclusion

CEM demonstrated high accuracy in assessing breast tumor size and extent, proving to be a valid alternative to MR I due to its lower cost, faster acquisition time, and better patient tolerability.

Limitations

No Funding for this study: No Ethics committee - additional information: No Author Disclosures: Marina Balbino: Nothing to disclose Giuseppe Guglielmi: Nothing to disclose Francesca Anna Carpagnano: Nothing to disclose Silvia Suriano: Nothing to disclose Manuela Montatore: Nothing to disclose Federica Masino: Nothing to disclose Giulia Capuano: Nothing to disclose Optimizing microcalcification assessment: the role of contrast-enhanced mammography on reducing unnecessary biopsies *A. Santonocito*, T. H. Helbich, P. Clauser, P. A. Baltzer; Vienna/AT ([email protected]) Purpose or Learning Objective: Microcalcifications are frequently observed in screening mammography, with malignancy rates ran ging from 6% to 82%. Their characterization and work-up are a major chal lenge for the radiologist. This study aimed to assess the role of contrast-enh anced mammography (CEM) in managing microcalcifications.

Methods

or Background: This retrospective, single-centre, IRB-approved, study included consecutive patients underwent CEM f or BI-RADS 4 microcalcifications between October 2018 and Septem ber 2024. Patients without a standard of reference were excluded. The standard of reference was histology from biopsy or surgery, or a one-year fol low-up for non-suspicious cases. Morphology, distribution, density, and densi ty heterogeneity of microcalcifications were assessed on LE images; enh ancement type and lesion conspicuity were assessed on RC images accor ding to the CEM lexicon. Microcalcification and enhancement characteristics were analysed through Spearman correlation and chi-square tests to evalua te associations with histological outcomes.

Results

or Findings: A total of 210 lesions in 197 patients (mean age 55 ±11 years old) were analysed. Of these, 124 (63%) were benign and 72 (37%) were malignant. Lesion conspicuity correlated with morphology (r=0.429), pleomorphism (r=0.514), distribution (r=0.235), and density heterogeneity (r=0.204), while no correlation was found with dens ity (r=-0.111). Chi-square test showed significant differences between benign and malignant calcifications for lesion type (p<0.001), morpholog y (p<0.001), grade of pleomorphism (p<0.001), distribution (p<0.001), den sity (p<0.001), enhancement type (p<0.001), lesion conspicuity (p0.05).

Conclusion

Our findings suggest that CEM may help differentiat e between benign and malignant microcalcifications, potential ly reducing unnecessary biopsies.

Limitations

Retrospective study; small cohort Funding for this study: None Ethics committee - additional information: Number: 1391/2022 Author Disclosures: Pascal A.T. Baltzer: Nothing to disclose Thomas H. Helbich: Nothing to disclose Ambra Santonocito: Nothing to disclose Paola Clauser: Nothing to disclose Contrast Enhanced Mammography (CEM) in the manageme nt of locally advanced breast cancer receiving neoadjuvant therap y (NAT) *S. Vidali*, F. Di Naro, D. De Benedetto, G. Bicchi erai, E. Vanzi, C. Bellini, C. Boeri, V. Miele, J. Nori; Florence/IT ([email protected]) Purpose or Learning Objective: Among contrast-enhanced mammography (CEM)'s indication, is the evaluation of locally ad vanced breast cancer(LABC)'s response to neoadjuvant therapy (NAT). CEM has pote ntial not only in the assessment of radiologic complete response (rCR) bu t also in predicting tumour response based on biological features and en hancement patterns.

Methods

or Background: We retrospectively analysed post-NAT CEMs of 141 patients diagnosed with LABC between 2016-2024 and correlated enhancement patterns (rCR, residual enhancement, RE ) with diagnostic biopsy biologic features and post-surgical pathology data (pCR, residual disease RD). Friday Abstract-based Programme 143

Results

or Findings: CEM showed RE in 92 patients (68 with RD and 24 wit h pCR) was negative in 49 (17 with RD and 32 with pCR ), with resulting mean sensitivity of 80% and specificity of 57%. Reclassi fying DCISs as RD instead of pCR, specificity raised to 65%: these performance d ata are adequate and comparable to those reached by CE-MRI, the gold sta ndard. Based on biological subtypes, distinct enhancement patterns resulted coherent with the expected therapeutic responses: luminal A tumours ( 46 patients) were less responsive to NAT (11% pCR, 5 patients, 4 of them w ith rCR, 80%), while HER2+ (37 patients) and triple negative(TN)BCs (20 patients), the more aggressive forms were more likely to achieve pCR an d rCR: respectively 59% pCR (22 patients) in the HER2+ group of which 64% ( 14 patients) with rCR, 50% pCR in the TNBCs (10 patients) of which 60% (6 patients) with rCR. Luminal B lesions (38 patients) achieved pCR in 45% of cases (17 patients) of which 13 with rCR (76%).

Conclusion

CEM is adequate and reliable in assessing disease r esponse to NAT: enhancement patterns demonstrated correlation with biological features, forecasting CEM’s potential as a prognostic and man agement tool for increasing conservative therapies and diagnostic fo llow-up.

Limitations

Sample's numerosity and heterogeneity. Funding for this study: No funding was received for this study. Ethics committee - additional information: Regione Toscana, Comitato Etico Area Vasta Centro, reference number: SPE_16.2 51 Author Disclosures: Giulia Bicchierai: Nothing to disclose Cecilia Boeri: Nothing to disclose Jacopo Nori: Nothing to disclose Sofia Vidali: Nothing to disclose Vittorio Miele: Nothing to disclose Chiara Bellini: Nothing to disclose Ermanno Vanzi: Nothing to disclose Diego De Benedetto: Nothing to disclose Federica Di Naro: Nothing to disclose Contrast Enhanced Mammography Screening in Women wi th Dense Breasts *J. M. Net*, J. Spoont, S. Stamler, C. Pluguez-Turu ll, N. Brofman, A. Hamedi-Sangsari, M. Yepes; Miami, FL/US ([email protected]) Purpose or Learning Objective: To determine outcomes of screening Contrast Enhanced Mammography (CEM) in women with B I-RADS density 3 or 4, more specifically to determine outcomes of CEM i n this population who did not undergo screening breast ultrasound.

Methods

or Background: HIPAA compliant and IRB approved retrospective study evaluated all medical records of patients who underwent CEM at our institution between 8/2019-8/2024. Reports were rev iewed and scored for breast density, additional work up (additional imag ing, US, MRI), final BI- RADS, biopsy results and whether target represented CEM finding, if high risk lesion or malignant case was scored based on final surgical pathology. Final

Results

were tallied to include overall CDR, PPV3, and interval cancer rate.

Results

or Findings: 1215 CEM studies were performed between 2020-2024, 1079 were performed for screening of which 740 were assigned BIRADS density 3 or 4. 87 patients (11.7%) were referred f or additional work up including US in 75, and MRI in 12. Of these, 46 wer e referred for biopsy which confirmed malignancy in 16 patients translating int o a CDR of 21.6/1000 and PPV3 of 35.5%. Of the remaining 653 patients (88.2% ) without additional work up, 46 underwent annual MRI within 12 months of CEM confirming the only 2 interval cancers in this study (0.3%), one stage 0 DCIS/linear 1.3 cm NME on staggered 6 month MRI and the other a 1.2 cm TNBC ( T1Nitc) on staggered 6 month MRI - neither was seen in retrospect.

Conclusion

The low interval cancer rate combined with high CDR suggest that CEM screening in women with dense breasts can potentially replace supplemental breast US for adjunct screening.

Limitations

Limitations include modest sample size, retrospecti ve design, and inclusion of cases from a single institution which limits generalizability. Funding for this study: None Ethics committee - additional information: IRB approved study. Author Disclosures: Antoine Hamedi-Sangsari: Nothing to disclose Jose Miguel Net: Nothing to disclose Monica Yepes: Nothing to disclose Nicole Brofman: Nothing to disclose Sarah Stamler: Nothing to disclose Cedric Pluguez-Turull: Nothing to disclose Jamie Spoont: Nothing to disclose 09:30-11:00 Research Stage 1 Research Presentation Session: Vascular RPS 1315 Imaging of the aorta, pulmonary, and coronary arteries Moderator M. Cejna; Feldkirch/AT ([email protected]) Low-energy virtual monochromatic CT with deep-learn ing image reconstruction to improve detection of endoleak *T. Higashigawa*¹, Y. Ichikawa¹, K. Nakajima², T. K obayashi¹, K. Domae¹, A. Yamazaki¹, N. Kato¹, H. Sakuma¹; ¹Tsu/JP, ²Ise/J P Purpose or Learning Objective: To evaluate the diagnostic performance of low-energy virtual monochromatic CT imaging (VMI) c ombined with deep- learning image reconstruction (DLIR) for the detect ion of endoleaks.

Methods

or Background: A cohort of 71 patients after endovascular aortic repair who underwent dynamic contrast-enhanced CT b etween March 2022 and August 2023 were studied. Raw data were reconst ructed using three different methods: 70-keV VMI using conventional hy brid iterative reconstruction (HIR [ASiR-V50%]), and 40- and 70-ke V VMI using DLIR (TrueFidelity-H). Contrast-to-noise ratio (CNR) of the endoleaks on venous phase CT was calculated. Three observers assessed t he presence or absence of endoleak on a 5-point scale, taking into account the confidence level: score- 1, endoleaks are definitely not present; score-2, p robably not present; score-3, may be present; score-4, probably present; score-5, definitely present. A score of 3 or higher was considered positive for endoleak .

Results

or Findings: Endoleaks were observed in 41 (58%) of 71 subjects. The CNRs of endoleaks were significantly higher in 40-keV DLIR (17.1±9.8) compared to 70-keV HIR (6.4±3.8; P<0.001) and 70-ke V DLIR (10.2±6.2; P<0.001). ROC analysis for endoleak detection showe d that AUC for 40-keV DLIR (0.92-0.99) was the largest for all observers (70-keV DLIR, 0.91-0.97; 70- keV HIR, 0.88-0.96). The percentage of patients wit h endoleaks who were correctly identified with a confidence level of ≥ score-4 in 40-keV VMI with DLIR was significantly higher compared to those in 70-keV VMI with HIR in one observer (Observer1, 85%(35/41) vs 73%(30/41), P=0. 02; Observer2, 85%(35/41) vs 78%(32/41), P=0.20; Observer3, 98%(40 /41) vs 90%(37/41), P=0.10, respectively).

Conclusion

The 40-keV VMI combined with DLIR reconstruction me thod improves the CNR of endoleaks and may help to corre ctly identify endoleaks with higher confidence compared to 70-keV VMI with HIR.

Limitations

The limitations of the study is the relatively smal l study population. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by institutional review board (approval number; H2019- 207). Author Disclosures: Kensuke Domae: Nothing to disclose Takatoshi Higashigawa: Nothing to disclose Hajime Sakuma: Nothing to disclose Yasutaka Ichikawa: Nothing to disclose Noriyuki Kato: Nothing to disclose Tatsuhiro Kobayashi: Nothing to disclose Akio Yamazaki: Nothing to disclose Ken Nakajima: Nothing to disclose Image Quality Improvement of Ultra-low Dose CT Pulm onary Angiography Using Deep Learning Reconstruction Algo rithm: Two-center Prospective Study *J. Lu*, L. Shen, Z. Zhao, Z. Bi, M. Zeng, M. M. Wa ng; Shanghai/CN ([email protected]) Purpose or Learning Objective: To investigate the effects of deep learning reconstruction (DLR) on the image quality in ultra- low dose CT pulmonary angiography (CTPA), compared to hybrid iterative re construction (HIR) at routine dose.

Methods

or Background: This study prospectively included 130 patients with suspected pulmonary embolism (PE) who underwent CTP A examination in two hospitals from April to July 2024. The noise index of routine dose (RD) group and ultra-low dose (ULD) group was set to 10 and 20 , respectively. The CT images of RD group were reconstructed using HIR, wh ile ULD group were reconstructed with HIR and DLR. Pulmonary CT value, signal-to-noise ratio Friday Abstract-based Programme 144 (SNR), and contrast-to-noise ratio (CNR) were asses sed as quantitative criteria of image quality. Two senior radiologists independe ntly evaluated the overall image noise, pulmonary artery visibility, and diagn ostic confidence based on a 5‑point Likert scale (5, best; 1, worst). The Mann-Wh itney U test and the Wilcoxon signed rank test was used for statistical analysis.

Results

or Findings: No statistically significant difference was found i n the clinical data between the two groups (p>0.05). The ULD-DLR group exhibited higher SNRs and CNRs in all seven pulmonary arterie s compared to the RD- HIR group (both p<0.05). The overall image noise an d diagnostic confidence of the ULD-DLR images were significantly better than t hat in the RD-HIR images and ULD-HIR images (both p<0.001). The effective do se in the RD group and ULD group were 2.84±0.49mSv and 0.70±0.21mSv, respectively, representing a reduction of approximately 75% in the ULD group ( p<0.001).

Conclusion

DLR can significantly reduce the radiation dose of CTPA examination without compromising the diagnosis of P E. Even at ultra-low radiation dose, its image quality is still better t han HIR at routine dose.

Limitations

Not applicable. Funding for this study: No. Ethics committee - additional information: Shanghai Geriatrics Medical Center Ethics Committee (B2024-009) Author Disclosures: Mengsu Zeng: Nothing to disclose Zicheng Zhao: Nothing to disclose Mingliang Mingliang Wang: Nothing to disclose Leilei Shen: Nothing to disclose Zhenghong Bi: Nothing to disclose Jinjuan Lu: Nothing to disclose Patient tailored contrast volume for preoperative C T angiography of the aorta: a prospective study based on patient heart r ate and body surface area *M. Dewilde*, W. Coudyzer, A. Laenen, H. Bosmans, G . Maleux; Leuven/BE ([email protected]) Purpose or Learning Objective: To prospectively compare aortic image quality by adapting contrast volume and kiloVoltage (kV) in patients referred for preoperative aortic computed tomography (aCT).

Methods

or Background: Eighty prospectively included patients were assigned into 3 groups: 50% of the contrast dose ca lculated on body surface area (BSA) and heart rate (HR) (group 1, n=56); 50% of the contrast dose calculated on BSA and HR and additional kV reductio n (group 2, n=11); 50% of contrast dose calculated on BSA and HR and addition al contrast dilution 80% contrast & 20% NaCl (group 3, n=13). Quantitative ( measurement of Hounsfield units) analysis at different anatomical aortic levels and qualitative image analysis by 2 radiologists using a visual sco re (1 = inadequate; 5 = excellent) was performed.

Results

or Findings: Mean contrast dose injected was 46.1 ml, 28.3 ml an d 35.0 ml for group 1, 2 and 3 respectively, with a s ignificant difference between group 1 and 2 (P=<0.001) and between group 1 and 3 (P=<0.001); no difference between group 2 and 3 (P=0.072). Mean qu alitative scores were 4.35/5, 2.82/5 and 3.46/5 for group 1, 2 and 3 resp ectively. No patient needed repeat imaging for inadequate aortic CT-imaging. In terobserver agreement was moderate for group 1 and 3 (0.577 and 0.576 resp.) and fair for group 2 (0.282) with consistent difference in scoring.

Conclusion

Meaningful contrast dose reduction in preoperative aCT while maintaining diagnostic efficacy is feasible through utilization of a contrast injection algorithm incorporating patient’s HR and BSA, coupled with adjusting kV values.

Limitations

Limited number of patients (n=80). Only patients wi th a pre operative angio CT were included. Funding for this study: No funding was received for this study. Ethics committee - additional information: The institutional ethics committee approved this study (S58042). Author Disclosures: Walter Coudyzer: Nothing to disclose Annouschka Laenen: Nothing to disclose Hilde Bosmans: Nothing to disclose Miloud Dewilde: Nothing to disclose Geert Maleux: Nothing to disclose AI-based reconstruction algorithm applied to low-kV and low contrast medium volume CT for TAVI planning compared with lo w dose CT with Model-Based algorithm: image quality and radiation dose exposure *C. R. G. L. O. M. Talei Franzesi*, D. Ippolito, C. Maino, P. N. Franco, D. G. Gandola, R. Corso; Milan/IT ([email protected]) Purpose or Learning Objective: To evaluate image quality and radiation dose reduction of deep learning reconstruction algo rithm in CT angiography (CTA) studies performed for TAVI planning, compared with low dose CTA reconstructed with hybrid iterative algorithm

Methods

or Background: Fifty six patients candidates for TAVI were enrolle d in this study and 26 patients (study-group) were ex amined with 128 MDCT scanner, with 80 kV, automated mAs dose-modulation and 50 mL of contrast media (CM), combined with a new deep learning recon struction algorithm (Precise Image); while a control group of 32 patien ts were evaluated with 256 MDCT (100 KV; automated mAs; 50 mL of CM) reconstru cted with hybrid iterative reconstruction algorithm (iDose4). Subjec tive (using a 4-point Likert scale) and objective image quality (vascular enhanc ement, SNR and CNR in different aortic levels and in the iliac arteries) were evaluated and the radiation dose exposure of both groups (CTDIvol and DLP) was calculated

Results

or Findings: Study group with deep learning algorithm demonstrat ed significantly higher mean attenuation values (p<.05 ) in all the measurements compared to the control group with model based algo rithm (aortic root 621HU vs 314 HU; external iliac arteries 537HU vs 335HU). Mean DLP and CTDI of study group was significantly lower than in control group (DLP: 395 mGy*cm vs 1600 mGy*cm, p<0.001; CTDI: 8.03 mGy vs 23.5 mGy, p <0.001), with an overall radiation dose reduction of about 75%. Furt hermore, study group showed a significant decrease of image noise with a n increase of image quality

Conclusion

Deep learning based CT reconstruction algorithm com bined with low Kv setting allows to significantly reduce radia tion dose exposure and increase the image quality in CTA protocol for TAVI planning, in comparison with low dose CTA reconstructed with hybrid iterati ve algorithm

Limitations

None Funding for this study: None Ethics committee - additional information: None Author Disclosures: Davide Giacomo Gandola: Nothing to disclose Cesare Maino: Nothing to disclose Cammillo Roberto Giovanni Leopoldo Oreste Massimili ano Talei Franzesi: Nothing to disclose Rocco Corso: Nothing to disclose Paolo Niccolò Franco: Nothing to disclose Davide Ippolito: Nothing to disclose Determining elasticity of the thoracic aorta in pat ients with giant cell arteritis using non-contrast-enhanced magnetic reso nance imaging at 1.5 T *M. Both*, C. Jochum, J. H. Schirmer, E. A. Strathm ann, P. Langguth, S. Sandra Freitag-Wolf, C. Von Der Burchard, O. Jan sen, M. Salehi Ravesh; Kiel/DE ([email protected]) Purpose or Learning Objective: The application of imaging techniques for early detection of thoracic aortic aneurysms in pat ients with giant cell arteritis (GCA), including the identification and monitoring of subgroups at high risk for this condition, is still the subject of debate. We investigated whether aortic stiffness could be quantified based on MRI and used as a potential biomarker for post-inflammatory damage.

Methods

or Background: Ten GCA patients in clinical remission and 36 healthy volunteers (HVs) were examined using non-co ntrast-enhanced cine- balanced steady-state free precession (bSSFP) MRI t echnique to determine the distensibility and diameter of the ascending (A Ao), descending (DAo), and arch (AArch) segments of the thoracic aorta. In add ition, changes in aortic diameters during follow-up in GCA patients and the impact of demographic and clinical characteristics on the aortic elasticity w ere investigated.

Results

or Findings: Distensibility was significantly higher in the AArc h (p=0.039) and in the DAo (p=0.004) than in the AAo in HVs, but not in GCA patients. Aortic distensibility was significantly l ower in patients than in HVs in the AArch (0.89 vs. 2.15, p=0.035). Age was an addi tional predictor of aortic stiffening in the AAo (p=0.029) and DAo (p=0.001) o f HVs. In patients with GCA, the diameter increased at an above-average rat e in all aortic segments (AAo 1.04 mm/year, AArch 1.12 mm/year, DAo 0.95 mm/ year) compared to baseline MRI.

Conclusion

The bSSFP MRI technique revealed functional and str uctural differences in the thoracic aorta of patients with GCA as a potential marker for weakness of the thoracic aortic wall. Friday Abstract-based Programme 145

Limitations

The small size of our patient group is the main lim itation due to its single-center design. Another limitation relates to the fact that most of our study patients presented with predominantly cranial symptoms, some without proof of aortitis on MRI. Funding for this study: None Ethics committee - additional information: The Ethics Committee at the Faculty of Medicine of Kiel University approved thi s study (No. D577/18). Author Disclosures: Claus Von Der Burchard: Nothing to disclose Mona Salehi Ravesh: Nothing to disclose Patrick Langguth: Nothing to disclose Marcus Both: Nothing to disclose Chiara Jochum: Nothing to disclose Jan Henrik Schirmer: Nothing to disclose Sandra Sandra Freitag-Wolf: Nothing to disclose Olav Jansen: Nothing to disclose Eike Andreas Strathmann: Nothing to disclose Accelerating Coronay CT Angiographies via Improved Patient Preparation *A. M. C. Boehner*, B. Salam, A. Jacob, A. Isaak, C . C. Pieper, D. Kütting; Bonn/DE ([email protected]) Purpose or Learning Objective: Coronary Computed Tomography Angiography (CCTA) often requires extensive prepara tion, contributing to prolonged in-room time. This study aims to assess t he impact of pre- examination preparation, including the administrati on of IV beta-blockers and ECG lead placement outside the examination room, on reducing in-room time for coronary CT scans.

Methods

or Background: A prospective study with 139 patients was conducted, comparing standard in-room preparation ( control cohort) with receiving preparation outside (intervention cohort) and mostly omitted preparation, scanned via a spiral acquisition proto col (spiral cohort). Patients' heart rates were regulated in the intervention grou p before entering the examination room. Key measures included: patient en tering the examination- room, installation of patient monitoring, heart rat e adjustment, first scan, heart rate during scanning and image quality.

Results

or Findings: The intervention cohort demonstrated significantly (P<0.0001) reduced in-room time compared to the con trol cohort (984±347s vs. 704±308s). The spiral cohort performed best and displayed the lowest variability (583±103s, P<0.0001). The heart rates of the spiral cohort was highest with 69±21bpm (P<0.04), but the interventio n cohort did not differ from the control cohort (60±7bpm vs. 59±6bpm, P=0.88). The rate of non-diagnostic segments remained low across all groups (control: 3 .5%, intervention: 1.5%, spiral: 2.0%).

Conclusion

Pre-examination preparation outside of the examinat ion room, including installation of patient monitoring and ad ministration of IV beta- blockers, significantly reduces in-room time for CC TA without compromising image quality. Alternatively, a spiral image acquis ition protocol allows for the omission of most preparatory steps. Both approaches offer a feasible strategy to streamline workflow and enhance efficiency in ca rdiac imaging departments.

Limitations

The limitation regarding the spiral cohort is the n eed to preselect patients with a coronary-calcium-score <400. Some r adiological departments may lack the premises to implement our approach. Funding for this study: The study was conducted in collaboration with Siemens Healthineers. Ethics committee - additional information: Administrational Author Disclosures: Alexander Isaak: Nothing to disclose Alexander Marc Christian Boehner: Nothing to disclo se Alice Jacob: Nothing to disclose Daniel Kütting: Nothing to disclose Claus Christian Pieper: Nothing to disclose Babak Salam: Nothing to disclose Optimizing HU Thresholds for Accurate Calcium Scori ng in Contrast- Enhanced CT: Robust Alternatives to the Agatston Sc ore *L. D. Grünewald*, V. Koch, S. Mahmoudi, J. Gotta, P. Reschke, J-E. Scholtz, S. Martin, C. Booz, T. Vogl; Frankfurt/DE Purpose or Learning Objective: To approximate the Agatston score in contrast-enhanced CTs using a volumetric approach w ithout distortions from contrast agents.

Methods

or Background: The aorta of 1276 patients (886 men, 390 women; median age 67 years; interquartile range 57-76) wit hout prior surgical interventions who underwent contrast-enhanced multi -phase CT between January 2018 and December 2023 were retrospectively analyzed. For all patients, the Agatston score was derived from unenh anced CT scans for the thoracic and abdominal aorta. The number and volume of plaques were assessed for the thoracic and abdominal aorta in ar terial and venous contrast phases using thresholds ranging from 100 to 1000 to assess the influence of contrast agents. Correlations with the Agatston sco re were calculated, and linear regression was used to identify the optimal threshold.

Results

or Findings: Median aortic enhancement was 46 HU (unenhanced), 323 HU (arterial), and 120 HU (venous). In venous p hases, a threshold of 300 HU yielded the highest correlations with the Agatst on score (thoracic: r=0.91; abdominal: r=0.93; p<0.001). In arterial phases, a threshold of 900 HU provided the best correlation (thoracic: r=0.72; ab dominal: r=0.65; p<0.001). Linear regression confirmed these thresholds, but s howed only moderate predictive power (R²=0.66 for venous, R²=0.52 for a rterial phases). Dynamic thresholding resulted in poor correlation (r=0.26; p<0.001) and low predictive value (R²=0.07).

Conclusion

Plaque volume assessment using optimized HU thresho lds can reliably approximate the Agatston score in contrast -enhanced CTs, offering a robust assessment without contrast-induced distorti on. This approach is particularly valuable when non-contrast images are unavailable, such as in staging or pre-TAVR evaluations.

Limitations

Patient collective with high plaque burden. Funding for this study: No funding was received for this study. Ethics committee - additional information: Waiver due to retrospective nature Author Disclosures: Simon Martin: Nothing to disclose Christian Booz: Nothing to disclose Thomas Vogl: Nothing to disclose Jan-Erik Scholtz: Nothing to disclose Vitali Koch: Nothing to disclose Scherwin Mahmoudi: Nothing to disclose Philipp Reschke: Nothing to disclose Leon David Grünewald: Nothing to disclose Jennifer Gotta: Nothing to disclose Performances of spectral CT for the detection and c haracterization of communications between the true and the false lumen in aortic dissections A. Janin-Manificat¹, M. Sigovan¹, L. Boussel¹, A. M illon¹, P. C. Douek¹, *S. Boccalini*²; ¹Lyon/FR, ²Villeurbanne/FR ([email protected]) Purpose or Learning Objective: To assess the performance of conventional CT (conv-CT) and spectral CT (spectral-CT) for the detection and characterization of communications between the true (TL) and false lumen (FL) in aortic dissections, using 4D-flow MRI as the ref erence.

Methods

or Background: 18 patients with type A and B aortic dissection who underwent 4D-flow MRI, conv-CT and spectral-CT were included. For each patient, the exams closest in time, without any sur gical or endovascular intervention in-between were retrieved and subjecti vely analysed by two observers, independently for conv-CT and in consens us for MRI and spectral- CT. Communications between the two lumens were iden tified as: intimal tears on conv-CT; focal alterations in velocities corresp onding to jet flows on 4D- flow; both intimal tears and focal changes of contr ast concentration corresponding to jet flows on spectral-CT. The numb er, size, and location of communications were noted. Additionally, the direct ion of the flow was assessed for spectral-CT and MRI.

Results

or Findings: Of the 176 communications detected with 4D-flow, spectral-CT allowed visualisation of 122 (69%) comp ared to 58 (33%) for Obs1 and 38 (22%) for Obs2 for conv-CT, yielding an accu racy twice as high (63% vs. 29-30%). On spectral CT, in only 45 cases (26%) the size of the communications could be assessed, in all other case s only jet flows were detected without visible intimal tears. The flow wa s unidirectional TL-FL in 2 cases for both MRI and spectral-CT and bidirectiona l in 5 and 3 cases for the two modalities. In all other cases the flow was in the direction TL-FL.

Conclusion

Spectral-CT outperformed conv-CT for the detection of communications between TL and FL in aortic dissecti ons. Spectral-CT allows direct visualization of flow jets, and their direction, through intimal tears.

Limitations

Low number of patients; time in-between different e xams Funding for this study: No Ethics committee - additional information: Approved Author Disclosures: Antoine Janin-Manificat: Nothing to disclose Philippe Charles Douek: Speaker: Philips Loïc Boussel: Speaker: Philips Sara Boccalini: Speaker: Philips Monica Sigovan: Nothing to disclose Antoine Millon: Nothing to disclose Friday Abstract-based Programme 146 Comparison of artificial intelligence and inexperie nced physicians in pulmonary embolism detection at deep learning recon struction-based ultra-low radiation dose CT pulmonary angiography *J. Lu*, L. Shen, Z. Zhao, Z. Bi, M. Zeng, M. M. Wa ng; Shanghai/CN ([email protected]) Purpose or Learning Objective: To assess the performance of artificial intelligence (AI) software and inexperienced physic ians in diagnosing pulmonary embolism (PE) at deep learning reconstruc tion-based ultra-low dose (ULD) CT pulmonary angiography (CTPA).

Methods

or Background: This prospective two-center study contained 210 patients with suspected pulmonary embolism (PE) who underwent CTPA examination, randomizing into two groups with equal proportion of patients. Images in the routine-dose (RD) group were reconstr ucted using hybrid iterative reconstruction (HIR, AIDR 3D, FC08), whil e ULD images were reconstructed using HIR and deep learning reconstru ction (DLR, AiCE), respectively. A subset of 74 participants (1:1 PE t o non-PE ratio) was randomly selected and evaluated by two inexperienced physici ans and AI software (Discover PE, uAI). Reference standard was establis hed by expert consensus. The diagnostic accuracy (sensitivity and specificit y) of the AI or reader interpretations were compared between methods by bo otstrapping.

Results

or Findings: There was no statistically significant difference i n the patient demographics between two groups. ULD-DLR im ages exhibited significantly higher objective and subjective image quality compared to both RD-HIR and ULD-HIR images. The AI software exhibite d near-perfect accuracy in both ULD-HIR and ULD-DLR sets (sensitivity: 97.3 0 %, specificity: 100 %). In comparison, two physicians showed a mean sensiti vity of 75.68% and specificity of 93.75% in ULD-HIR sets, and a mean s ensitivity of 94.59% and specificity of 100.00% in ULD-DLR sets. Inter-obser ver agreement was moderate for HIR (κ = 0.75) and good for DLR (κ = 0.81). The effective dose of ULD group was significantly lower than the RD group (2.74±0.47 mSv vs. 0.73±0.25 mSv, p<0.001).

Conclusion

DLR can significantly reduce the radiation dose of CTPA examination without compromising the diagnosis of p ulmonary embolism even at ultra-low radiation dose. AI software outperform s inexperienced physicians in interpreting ULD images.

Limitations

Not applicable. Funding for this study: No. Ethics committee - additional information: Shanghai Geriatrics Medical Center Ethics Committee (B2024-009) Author Disclosures: Mengsu Zeng: Nothing to disclose Zicheng Zhao: Nothing to disclose Mingliang Mingliang Wang: Nothing to disclose Leilei Shen: Nothing to disclose Zhenghong Bi: Nothing to disclose Jinjuan Lu: Nothing to disclose 09:30-11:00 Research Stage 2 Research Presentation Session: Interventional Radiology RPS 1309 Interventions in malignant liver disease Moderator L. Novosel; Zagreb/HR ([email protected]) Oncologic Ablation in Germany: 2018-2023 data from the German Society of Interventional Radiology Registry *J. Uhlig*¹, L. Biggemann¹, J. Nadjiri², T. Kroenck e³; ¹Göttingen/DE, ²Munich/DE, ³Augsburg/DE ([email protected]) Purpose or Learning Objective: To assess the current utilization, technical approaches and complications of oncologic ablation in Germany.

Methods

or Background: The German Society of Interventional Radiology (“DeGIR”) registry was queried for patients receivi ng ablation treatments between 2018-2023. Patient demographics, indication s, and technical ablation parameters were descriptively assessed.

Results

or Findings: N=9157 patients receiving oncologic ablation were included (34.3% female; median age 67yo). Between 2 018-2023, annual ablation number remained approximately constant at 1000 cases/year. Ablations were performed in the liver (71.4%), kidn ey (11.3%), musculoskeletal system (7.3%) and lung (5.4%), mainly with curative intent (64%) or for symptomatic treatment / palliation (30.6%). N=7371 patients were imaged with CT before ablation (80.5%), 4176 with MRI (45.6%), and 173 with PET (1.9%; not mutually exclusive). Ablation guidance was achi eved using CT (89.6%), ultrasound (5.7%), MRI (2.8%), cone-beam CT or fluo roscopy (0.9%, each). Ablation procedures were mainly performed under gen eral anesthesia (74.6%) or analgosedation (14.5%). Microwave ablation was p erformed in most cases (69.4%), followed by radiofrequency (23.6%) and cry oablation (2.8%), often combined with tract ablation (51.2%). Only 48 proce dures (0.5%) were preemptively terminated, mainly due to anatomical d ifficulties (n=16) or uncooperative patients (n=11). During or within the first 24h after ablation, 4.8% of patients experienced any complications, the majority being low-grade. Another 45 patients (0.5%) experienced delayed comp lications 24h or later after ablation, mainly infections/abscesses (n=21).

Conclusion

Oncologic ablations are routinely performed in Germ any with low procedural complication rates, mostly using CT-guid ed microwave or radiofrequency ablation for hepatic or renal tumors .

Limitations

Since participation in the DeGIR registry is not ma ndatory, there could be selection bias of included cases and parti cipating sites, limiting the generalizability of results. Funding for this study: Not applicable. Ethics committee - additional information: Not applicable - retrospective anonymised registry data. Author Disclosures: Thomas Kroencke: Nothing to disclose Johannes Uhlig: Nothing to disclose Lorenz Biggemann: Nothing to disclose Jonathan Nadjiri: Nothing to disclose Enhancing Neoadjuvant Immunotherapy Efficacy throug h Partial Cryoablation in a Hepatocellular Carcinoma Mouse Mo del *T. Kao*¹, E. Meister¹, J. Santana², J. Israel², A. Shewarega², J. Tefera², D. C. Madoff², L. J. Savic¹, J. Chapiro²; ¹Berlin/D E, ²New Haven, CT/US ([email protected]) Purpose or Learning Objective: Hepatocellular carcinoma (HCC) exhibits an immunosuppressive microenvironment which can be agg ravated by incomplete tumor ablation. Immune checkpoint inhibitors (ICIs) such as anti-PD-1 are guideline-approved therapies for advanced HCC. Comb ining ablation with ICIs could potentially strengthen anti-cancer immunity, but supporting evidence is limited. We aim to evaluate the effect of neoadjuva nt systemic anti-PD-1 on the local immune response in residual tumors following partial cryoablation in a TIB-75 murine HCC model.

Methods

or Background: Forty-eight male and female BALB/c mice aged 6- 12 weeks underwent orthotopic inoculation of TIB-75 cells to induce a solitary HCC lesion. After 7 days, mice were randomized into 4 treatment groups: (a) control, (b) anti-PD-1, (c) partial cryoablation, a nd (d) anti-PD-1 followed by partial cryoablation. The percentage of positively stained T-cell subsets and tumor-associated macrophages within the tumor was a ssessed in paraffinized liver tissue samples using immunohistochemistry (CD 3+, CD4+, CD8+, CD68+, CD206+, FOXP3+) and quantified on digitized slides. Treatment groups were compared using unpaired Mann-Whitney U and Kruskal- Wallis test with Dunn correction.

Results

or Findings: Mice treated with anti-PD-1 (n=12, group b) showed greater tumoral infiltration of CD3+, CD4+ and CD8+ T-cells than control (CD3+: mean 21.4% vs. 6.7%; P=<0.0001, CD4+: mean 2 1.3% vs. 6.0%; P=<0.0001, CD8+: mean 7.5% vs. 3.8%; P=0.005). Part ial cryoablation alone (n=12) had greater infiltration of CD206+ M2-like m acrophages than control (mean 32.4% vs. 14.6%; P=0.007). Anti-PD-1 combined with partial cryoablation (n=12) showed significantly more infil tration of CD3+ T-cells (mean 13.7% vs. 6.1%; P=0.002) and fewer CD206+ M2- like macrophages (mean 26.1% vs. 32.4%; P=0.3474) than partial cryoa blation alone (n=12).

Conclusion

Immune evasion following partial cryoablation can b e counteracted with neoadjuvant anti-PD-1, suggesting effective combination therapy to treat both early-stage and advanced-stag e HCC.

Limitations

The model was inoculated in healthy, non-cirrhotic mouse liver. Funding for this study: NIH grant 2R01CA206180 Ethics committee - additional information: All experimental procedures were approved by the Yale University Institutional Animal Care and Use Committee (IACUC protocol number: 2022-20262). Author Disclosures: Joshua Israel: Nothing to disclose Julius Chapiro: Nothing to disclose Jonathan Tefera: Nothing to disclose Jessica Santana: Nothing to disclose Annabella Shewarega: Nothing to disclose David Craig Madoff: Nothing to disclose Ellen Meister: Nothing to disclose Tabea Kao: Nothing to disclose Lynn Jeanette Savic: Nothing to disclose Friday Abstract-based Programme 147 MRI-based risk stratification for viable Hepatocell ular Carcinomas post- Transarterial Chemoembolization: Correlation with p athological outcomes and prognostic implications *W. Wang*, Y-C. Wang; Nanjing/CN ([email protected]) Purpose or Learning Objective: Accurate risk stratification of viable hepatocellular carcinomas (HCC) following transarte rial chemoembolization (TACE) is essential for the development of individu alized treatment strategies and enhancing the accuracy of prognosis predictions .

Methods

or Background: This multi-center, retrospective study includes HCC patients who received TACE as their initial and sol e treatment from February 2015 to October 2022 as training set (203 viable tu mors). Additionally, a dataset from a multicenter clinical trial (NCT03113 955) was subject to secondary analysis as test set (102 viable tumors). The final pathological validation set consists of a separate center, inclu ding individuals who had liver resection post-first TACE (120 viable tumors). All participants in both the training and test cohorts underwent contrast-enhanc ed MRI scans at baseline, and at one and six months after TACE. In the traini ng set, univariate and multivariate logistic regression analysis was perfo rmed to identify clinical, laboratory and imaging variables to include in the predictive model.

Results

or Findings: The predictive model incorporated five key imaging features: Mild-moderate T2 hyperintensity, T2-weigh ted peritumoral hyperintensity, Diffusion restriction, Irregular sh ape, and Heterogeneity. The model achieved areas under the curve (AUCs) of 0.85 (95% confidence interval [CI] 0.79 to 0.90) for the training cohort and 0.88 (95% CI 0.81 to 0.95) for the external test cohort. The risk model effect ively distinguished high-risk from low-risk groups in the test cohort, with signi ficant differences in progression-free survival (PFS) (P = 0.004) and two -year overall survival (OS) (P = 0.028). In the pathology cohort, the model cor related with microvascular invasion (MVI) grades (P = 0.003) and liver capsule invasion (P = 0.007).

Conclusion

This risk model based on imaging features for viabl e HCCs post- TACE exhibits robust predictive power for tumor via bility at six months and for long-term survival outcomes.

Limitations

Not applicable Funding for this study: This study has received funding by National Natural Science Foundation of China (NSFC, No. 82271978, 92 359304, 82330060) and Zhongda Hospital Affiliated to Southeast Univer sity, Jiangsu Province High-Level Hospital Pairing Assistance Construction Funds (No. zdyyxy09). Ethics committee - additional information: This multicenter, retrospective study was reviewed and approved by IEC for clinical research of the Zhongda Hospital, Southeast University, approval number [20 22ZDSYLL410-P01], and conducted following the ethical principles outlined in the Helsinki Declaration of 1964 and its subsequent amendments, or other ethica l standards with equivalent requirements. All patients and their fam ilies signed informed consent forms prior to surgery. Author Disclosures: Yuan-Cheng Wang: Nothing to disclose Weilang Wang: Nothing to disclose Transarterial Embolization Alone Versus Drug-Elutin g Beads Chemoembolization for HepatocellularCarcinoma (RAD- 18-TAcE): a Randomized Clinical Trial *M. Taninokuchi Tomassoni*¹, M. Renzulli¹, S. Zanel la¹, A. Doriguzzi Breatta², P. Marra³, F. De Cobelli⁴, C. Mosconi¹; ¹Bologna/IT, ²Turin/IT, ³Bergamo/IT, ⁴Milan/IT ([email protected]) Purpose or Learning Objective: This randomized clinical trial aims to compare transarterial embolization (TAE)and drug-el uted beads transarterial chemoembolization (DEB-TACE) in the treatment of he patocellular carcinoma (HCC).

Methods

or Background: Patients diagnosed with unresectable HCC were randomly assigned to either theTAE or DEB-TACE grou p. The primary endpoint was time to progression (TTP), and seconda ry endpoints included overall survival, cost-effectiveness, tumor respons e rates, and adverse events.

Results

or Findings: A total of 111 patients were enrolled, with 56 in t he TAE group and55 in the DEB-TACE group. Baseline charact eristics were balanced between the two groups. The primary endpoint analys is showed that TAE was not different from DEB-TACE in termsof TTP (average of 12.13 and 10.87 months respectively, p=0.432). Overall survival, tu morresponse rates, and adverse events were also similar between the two gr oups. The cost- effectiveness ratio of DEB-TACE vs. TAE was evaluat ed considering that, with equaleffectiveness of the two treatments highlighte d by the previous points, there being nostatistically significant difference in terms of days of hospitalization between DEB-TACEand TAE (average of 4.62 days and 5.20 days respectively, p=0.638).

Conclusion

In this randomized clinical trial, TAE showed compa rable outcomes to DEB-TACE in the treatment of unresectab le hepatocellular carcinoma. These findings suggest that TAE could be considered as an alternative for treating HCC with no differences in terms of safety and efficacy.

Limitations

The relatively small sample size and short-term fol low-up period may limit the generalizability of the findings and the ability to detect subtle differences in outcomes between TAE and DEB-TACE. Funding for this study: This study was funded by the Italian Ministry of Health. Ethics committee - additional information: The trial was conducted in accordance with ethical standards and received appr oval from the institutional review boards. The study was registered at www.clin icaltrials.gov (NCT04803019). Author Disclosures: Andrea Doriguzzi Breatta: Nothing to disclose Paolo Marra: Nothing to disclose Matteo Renzulli: Nothing to disclose Cristina Mosconi: Nothing to disclose Sara Zanella: Nothing to disclose Makoto Taninokuchi Tomassoni: Nothing to disclose Francesco De Cobelli: Nothing to disclose MR guided catheter-based radiotherapy/brachytherapy of liver tumours – first experience and feasibility *M. P. Fabritius*, A. Haghpanah, O. Dietrich, D. Pu hr-Westerheide, V. F. Schmidt, S. Corradini, J. Ricke, O. Öcal, M. Seidensticker; Munich/DE Purpose or Learning Objective: To show feasibility and safety of MR guided catheter-based radiotherapy/brachytherapy of primar y or secondary liver tumours

Methods

or Background: Between June 2023 and April 2024, 27 patients with 54 liver lesions were treated within a prospec itve single-center trial on MR- guided catheter-based radiotherapy (MR BRIGHT trial ). Treatments were performed under conscious sedation and local anesth esia using a 1.5T MRI system (Magnetom Solafit, Siemens) with a 15 cm loo p coil. Gadoxetic acid (0.1 mmol/kg) was administered for contrast enhance ment, followed by insertion of an 18G coaxial needle and navigation t o the lesion via real-time gradient-echo fluoroscopy sequences (iMRI UI Extens ion, Research Software Package). The needle was exchanged for a 6F hydroph ilic angiography sheath with a brachytherapy catheter. 3D T1-weighted seque nces were sent to the radiation department for brachytherapy with an IR19 2 high-dose-rate (HDR) afterloading unit. Target doses ranged from 15 to 2 5 Gy, depending on tumor type (HCC, CRC, GIST, NET, and other metastases). C atheters were removed after BT, and the radiation tract sealed with gelat in sponge.

Results

or Findings: The average lesion diameter was 13 ± 6 mm, whereas the average clinical target volume (CTV) was 3.0 ± 2.9 cm3. The average room time was 74 ± 35 minutes, the average time for catheter placement was 19 ± 11 minutes. The mean dose administered per lesion ( D100) was 18.9 ± 3.6 Gy. Complications during and after BT were generally ra re with only 2 patients (7.4 %.) having a minor bleeding without need for blood transfusion or intervention.

Conclusion

Overall, MR-guided catheter-based radiotherapy for liver tumours is feasible and safe, particularly for small lesion s. With low complication rates and precise dosimetry achieved through advanced ima ging, this approach holds promise for effective tumour management.

Limitations

n/a Funding for this study: None Ethics committee - additional information: LMU Munich Author Disclosures: Stefanie Corradini: Nothing to disclose Olaf Dietrich: Nothing to disclose Matthias Philipp Fabritius: Nothing to disclose Alireza Haghpanah: Nothing to disclose Vanessa Franziska Schmidt: Nothing to disclose Max Seidensticker: Nothing to disclose Daniel Puhr-Westerheide: Nothing to disclose Osman Öcal: Nothing to disclose Jens Ricke: Nothing to disclose Evaluation of Pain and Satisfaction in Patients wit h Liver Tumor treated with CT guided High Dose Rate Brachytherapy under A nalgosedation– Preliminary Results *M. Z. Erforth*, L. K. Segger, U. Fehrenbach, F. Co llettini, B. Gebauer, T. A. Auer; Berlin/DE ([email protected]) Purpose or Learning Objective: To evaluate feasibility for CT guided high dose rate (HDR) brachytherapy under analgosedation performed by interventional radiologists in patients with liver tumors.

Methods

or Background: In this prospective single-center study (EA/122/23) , 97 patients who received CT-guided HDR brachytherap y along with analgosedation using fentanyl and midazolam were en rolled and 77 were included in the final analysis. At the outset, a pe rsonality profile (from the EORTC catalog) related to their pain experience was also recorded for each patient. Structured questionnaires were employed to assess the patients' pain Friday Abstract-based Programme 148 levels and satisfaction both at and after the inter vention. Three months later, the patients were recontacted, and a follow-up surv ey was conducted. The

Results

were recorded by means of a numeric analog scale and presented as categorical variables.

Results

or Findings: First, pain was measured (0: no pain; to 10: maximu m pain) 1. at catheter placement; 2. at the radiation ; 3. after the radiation (catheter removal). At catheter placement, 75.5% (5 8/77) rated the pain as low (0-2), 18.0% (14/77) as moderate (3-6), and 6.5% (5 /77) as severe (7-10). At the radiation and afterwards pain levels were rated as low in 83.0% (64/77) and 79.0% (61/77), as moderate in 11.5% (9/77) and 17.0% (13/77) and as severe in 5.5% (4/77) and 4.0% (3/77), respectively . Second, patient satisfaction was measured (1: completely dissatisfi ed; to 10: completely satisfied). In 1.3% (1/77) the lowest score was rec orded while in 98.7% (76/77) a score ≥7 was recorded.

Conclusion

CT guided HDR brachytherapy under analgosedation is feasible and can be performed by interventional radiologists themselves without general anesthesia.

Limitations

Limitations include the short observation period an d the small cohort of patients with a heterogeneous clinical hi story, neoplasm histology, and location, as well as inhomogeneity regarding pr evious treatments. Funding for this study: None Ethics committee - additional information: Institutional Review Board approved prospective study (EA1/122/23). Author Disclosures: Laura Katharina Segger: Nothing to disclose Mo Zelda Erforth: Nothing to disclose Uli Fehrenbach: Nothing to disclose Bernhard Gebauer: Nothing to disclose Timo Alexander Auer: Nothing to disclose Federico Collettini: Nothing to disclose Dual-phase Cone-Beam CT (DP-CBCT) role as imaging n avigation guidance in HCC lesions treatment with trans-arteri al chemoembolization (TACE): a single centre experience *N. Rossini*¹, C. Floridi¹, M. Macchini¹, L. M. Cac ioppa¹, A. Felicioli¹, C. Mincarelli², R. Candelari¹, A. Giovagnoni¹; ¹Anc ona/IT, ²Macerata/IT ([email protected]) Purpose or Learning Objective: To evaluate how intra-procedural DP-CBCT navigation guidance influences TACE success rate in terms of residual disease in follow-up imaging.

Methods

or Background: This retrospective analysis includes all patients w ith HCC treated with TACE (cTACE or DEB-TACE) in our ce ntre between January 2017 and January 2024 with at least 1 month of imag ing (CT or MRI) follow-up available. All patients had a recent baseline CT or MRI before TACE. Patients were divided in two groups, the first one included patients with DP-CBCT performed intra-procedurally during TACE (DP-CBCT g roup), the second included patients with no CBCT performed during tre atment (no-DP-CBCT group). The two groups were similar in vascular ana tomy, lesions number, morphology and localization. Response to treatment was evaluated in imaging follow-up with mRECIST criteria. The two groups wer e compared for treatment response after TACE in terms of residual disease in follow-up imaging.

Results

or Findings: 152 patients were included in the study (M:F 112:40 ). 82 patients were included in the DP-CBCT group whereas 70 in the no-DP-CBCT group. Residual disease was of 26.9% in DP-CBCT gro up and of 63.0% in no- DP-CBCT group. A significant difference in terms of residual disease was observed between the two groups (p<0.05). Significa nt lower cases of residual disease were present in DP-CBCT group.

Conclusion

DP-CBCT imaging navigation guidance improves signif icantly success rate in TACE, ensuring a better visualizati on of HCC lesions feeding vessels and a consequent higher possibility of comp lete treatment of the nodules without residual disease.

Limitations

The main limitation of this study is the brief foll ow up imaging that should be extended in future studies. Funding for this study: This research received no external funding. Ethics committee - additional information: All the procedures performed in studies involving human participants were in accord ance with the ethical standards of the institutional and/or national rese arch committee and with the 1964 Declaration of Helsinki and its later amendmen ts or comparable ethical standards. This study obtained the approval of the Internal Review Board (IRB) of University Politecnica Delle Marche. Author Disclosures: Cinzia Mincarelli: Nothing to disclose Alessandro Felicioli: Nothing to disclose Chiara Floridi: Nothing to disclose Laura Maria Cacioppa: Nothing to disclose Marco Macchini: Nothing to disclose Roberto Candelari: Nothing to disclose Andrea Giovagnoni: Nothing to disclose Nicolo' Rossini: Nothing to disclose Identifying Key Predictors of Mortality and Liver D ecompensation in Hepatocellular Carcinoma Patients Treated with Yttr ium-90 Radioembolization M. Arabi, H. Alghamdi, *A. A. F. Almesned*, O. Alan azi, M. Alghamdi, M. Bukhaytan, M. Alkhalaf, M. Almaimoni, N. Alagraf y; Riyadh/SA ([email protected]) Purpose or Learning Objective: This study aimed to identify the predictors of mortality and liver decompensation in patients with HCC treated with Y-90 radioembolization.

Methods

or Background: A retrospective analysis of 140 patients with HCC who underwent Y-90 radioembolization was conducted. Kaplan‒Meier and multivariate Cox regression analyses were performed to identify the significant predictors of mortality.

Results

or Findings: The cohort comprised 69.3% males with a mean age of 71.3 ±11.9 years. Most patients (73.6%) had Child-P ugh class A cirrhosis and 34.3% had BCLC stage B disease. Among the 140 patie nts, 57.1% died after treatment and liver decompensation was recorded in 39.2%. The median survival was significantly longer in those without liver decompensation (3.2 vs 0.7 years, p<0.001). Multivariate analysis revealed that male sex (adjusted odds ratio [aOR] 5.889, p=0.009), cirrhosis (aOR 6. 82, p=0.047), and international normalized ratio (INR) (aOR 316.664, p=0.013) were independent predictors of liver decompensation. Cox regression analysis revealed several significant predictors of mortality. Ascites (HR 2. 012, 95% CI, 1.122–3.61; p=0.019), portal vein invasion (HR 1.695, 95% CI, 1 .057–2.718; p=0.029), and diabetes mellitus (HR 1.823, 95% CI, 1.017–3.265; p =0.044) were associated with increased mortality risk. Conversely, non-mult ifocal HCC (HR 0.593, 95% CI, 0.369–0.955; p=0.031), treatment of the liver l obe other than the right lobe (HR, 0.482; 95% CI 0.236–0.986, p=0.046), and age ≥60 years (HR 0.288, 95% CI, 0.139–0.597; p=0.001) were associated with a reduced risk of mortality.

Conclusion

This study identified the key predictors of mortali ty in patients with HCC undergoing Y-90 radioembolization, potenti ally improving patient selection and management strategies.

Limitations

While this study provides valuable insights, severa l limitations should be acknowledged. The retrospective nature of the study introduces potential biases in patient selection and data coll ection. The lack of post- infusion dosimetry limits the precision of dose-res ponse analyses. Funding for this study: The study was not supported by funding. Ethics committee - additional information: The study was approved by the institutional review board, and the need for inform ed consent was waived. This study was conducted in accordance with the 2010 gui delines of the Declaration of Helsinki. Author Disclosures: Omar Alanazi: Nothing to disclose Muath Almaimoni: Nothing to disclose Mohammed Bukhaytan: Nothing to disclose Mohammed Alkhalaf: Nothing to disclose Meshari Alghamdi: Nothing to disclose Hamdan Alghamdi: Nothing to disclose Abdulaziz Abdullah F Almesned: Nothing to disclose Nawaf Alagrafy: Nothing to disclose Mohammad Arabi: Nothing to disclose Application of cross-modality image registration sy stem for localising intraoperative colorectal cancer liver metastases d uring ablation *X. Wu*; Hangzhou/CN ([email protected]) Purpose or Learning Objective: Localisation of target tumours under CT guidance can be challenging due to insufficient sof t tissue resolution and metal artifacts. This study aims to validate the accuracy of the automatic image registration system (AIRS) in localising target les ions throughout the CT-guided CRLM ablation procedure, thereby exploring a novel guidance method for interventional procedures.

Methods

or Background: This retrospective, single-center study included patients with CRLM who underwent CT-guided liver ab lation between January 2021 and August 2023. Three experienced physicians collectively annotated the visibility and lesion centre positions of CRLMs on both the preprocedural contrast-enhanced MRI and intraoperative CT image s eries, which served as the ground truth. The AIRS and two junior physician s delineated the lesion centre positions in the same CT sequences. The loca lisation errors of the AIRS and junior physicians were analysed using the non-p arametric Kruskal–Wallis test for one-way analysis.

Results

or Findings: One hundred and twenty consecutive patients with 22 4 CRLMs treated across 128 sessions were enrolled. Th ere were 128 pairs of MR-pCT (pre-procedural CT) multi-modal registration s and 1,008 pairs of pCT- iCT (intra-procedural) mono-modal registrations. AI RS demonstrated superior localisation error than the physician group in loca lising lesions suboptimal visible on pCT (5.94±2.61 mm vs 8.04±5.32 mm, p=0.006), lesions excellently Friday Abstract-based Programme 149 visible on iCT (5.14±2.65 mm vs 6.15±3.84 mm, p=0.01) and lesions suboptimal visible on iCT (6.13±2.80 mm vs 8.94±4.60 mm, p＜0.001).

Conclusion

Compared with less experienced physicians, an AIRS can quickly and accurately locate target lesions in CT-guided c olorectal cancer liver metastasis ablation procedures, especially for lesi ons with poor visibility, thus paving the way for a new navigation method in color ectal cancer liver metastasis ablations.

Limitations

This was a retrospective investigation with a relat ively small sample size, which may have restricted the generali sability of our findings. Funding for this study: None Ethics committee - additional information: This single-centre retrospective study was approved by the local ethics committee an d was exempted from informed consent. Author Disclosures: Xia Wu: Nothing to disclose Deep Learning-Based Reconstruction and Superresolut ion for MR-guided Thermoablation *M. T. Winkelmann*¹, J. Kuebler¹, S. Gassenmaier¹, D. Nickel², K. Nikolaou¹, S. Afat¹, R. Hoffmann¹; ¹Tuebingen/DE, ²Erlangen/DE ([email protected]) Purpose or Learning Objective: This study explores the impact of deep learning-enhanced image generation for T1-weighted volume-interpolated breath-hold examinations (DL-VIBE) on image quality and procedural parameters during MR-guided thermoablation of liver malignancies, compared to standard VIBE images (SD-VIBE).

Methods

or Background: 34 consecutive patients (mean age: 65.4 ± 11.5 years, women: n=13) with liver malignancies underwe nt MR-guided microwave ablation using a 1.5 T MR scanner. Intraprocedural VIBE sequences (SD- VIBE) were used to monitor needle position and asse ss the ablation zone. The raw T1-weighted VIBE data were retrospectively proc essed with a deep learning algorithm (DL-VIBE) to reduce noise and im prove sharpness. Two interventional radiologists independently evaluated the image sets in a blinded manner, comparing DL-VIBE with unprocessed SD-VIBE images. Criteria assessed included diagnostic confidence, image qual ity, noise, artifacts, and sharpness. Interrater agreement was analyzed, and n oise maps were created to evaluate signal-to-noise ratio improvements.

Results

or Findings: DL-VIBE significantly improved overall image qualit y, reduced noise and artifacts, and enhanced the sharp ness of liver contours and portal vein branches compared to SD-VIBE (P<0.001). Additionally, DL-VIBE improved imaging of the interventional path, needle tip detectability, and diagnostic confidence in needle positioning and the ablation zone (P<0.001), with high interrater agreement (κ = 0.86). Quantitative noise maps demonstrated a higher signal-to-noise ratio, and th e reconstruction process took approximately 4 seconds, reducing breath-hold time by 2 seconds.

Conclusion

DL-VIBE significantly enhances image quality and di agnostic confidence during MR-guided thermal ablation proced ures, offering time savings and potential improvements in patient outco mes.

Limitations

- Small number of patients. - Sequences were retros pectively processed, not used during actual interventions. - Image quality assessments were retrospective and may differ from real-time ev aluations. - Time savings are theoretical due to the retrospective study desi gn, with uncertainty about feasibility during actual interventions. Funding for this study: No funding was received for this study Ethics committee - additional information: This retrospective study was approved by the institutional review board (Eberhar d Karls University of Tübingen, project number: 055/2017BO2) Author Disclosures: Konstantin Nikolaou: Nothing to disclose Moritz T. Winkelmann: Nothing to disclose Saif Afat: Nothing to disclose Rüdiger Hoffmann: Nothing to disclose Dominik Nickel: Nothing to disclose Jens Kuebler: Nothing to disclose Sebastian Gassenmaier: Nothing to disclose Machine learning using MR imaging radiomics and cli nical features can predict the response of large hepatocellular carcin oma to transarterial radioembolization O. Sarioğlu, A. Canturk, *R. C. Yarol*, H. Gulmez, E. Derebe k, A. Gülcü; Izmir/TR ([email protected]) Purpose or Learning Objective: To evaluate the potential of machine learning-based models for predicting the response o f large hepatocellular carcinoma to transarterial radioembolization

Methods

or Background: A total of 49 patients (38 responder and 11 non- responder) were included in the study. Laboratory r esults and clinical conditions were collected. Treatment response was a ssessed according to mRECIST criteria from the 3-month follow-up MR exam inations. Complete or partial response was categorized as the responder g roup, while stable or progressive disease was classified as the non-respo nder group. Radiomics features were extracted from contrast-enhanced T1-w eighted images (CE-T1) and T2-weighted images (T2WI). 141 radiomics featur es were obtained from each lesion. Classification learning models were us ed to create prediction models for TARE response. 5-fold cross-validation t echnique was utilized to identify the prediction rates of treatment response .

Results

or Findings: Number of radiomics features demonstrated statistic ally significant differences between the groups are 9 an d 12 on T2W and CE-T1 images, respectively. The model based on radiomics features obtained from CE-T1 images demonstrated an accuracy rate of %79.6 to predict response with an AUC of 0.92. The sensitivity and specificit y rates were %79 and %100, respectively. The accuracy and AUC rates of the mod el using radiomics features extracted from T2W images were %79.6 and 0 .77, respectively. Sensitivity and specificity rates of the model were %80 and %67, respectively. When only clinical and laboratory parameters were u sed, the model showed an accuracy rate of %77.6 and an AUC of 0.65. The sens itivity and specificity values of the clinical and laboratory model were %7 9 and %50, respectively. Another model using both clinical and CE-T1 radiomi cs features showed an accuracy rate of %73.5

Conclusion

Machine learning-based radiomics models based on MR I can predict the response of large hepatocellular carcin oma to transarterial radioembolization

Limitations

None Funding for this study: None Ethics committee - additional information: Ethics committee approval obtained Author Disclosures: Aytaç Gülcü: Nothing to disclose Ali Canturk: Nothing to disclose Raif Can Yarol: Nothing to disclose Orkun Sarioğlu: Nothing to disclose Hakan Gulmez: Nothing to disclose Erkan Derebek: Nothing to disclose 09:30-11:00 Research Stage 3 Research Presentation Session: Cardiac RPS 1303 The evolving impact of artificial intelligence (AI) in cardiac imaging Moderator A. Isaak; Bonn/DE Formulation of a predictive model for total cardiac volume (TCV) estimation: Optimizing donor-recipient size matchin g and outcomes *S. Gowda*, V. Raj, R. Kothari; Bengaluru/IN ([email protected]) Purpose or Learning Objective: Accurate donor heart size measurement is crucial for successful heart transplantation (HT). Traditional weight-based donor-to-recipient (D-R) size matching in paediatri c HT has poor correlation with cardiac size and significantly restricts the d onor pool. The aim of the study is to develop a novel predictive model to accuratel y calculate Total Cardiac Volume (TCV) tailored to the Indian population, aim ing to expand the donor pool and reduce size mismatches.

Methods

or Background: This multi-centre study incorporated paediatric and young adults (ages 0-30) with normal CT chest angio grams. TCV was predicted using common variables such as weight, he ight, gender and cardiac width on chest radiograph (CXR) with CT derived TCV (3D segmentation) as the gold standard. Three predictive models were ana lysed, and subjects were split into training and testing data. Model A- weig ht only Model B- weight, height, gender and age Model C- Model B plus horizo ntal cardiac width from CXR.

Results

or Findings: Model C showed highest accuracy in predicting TCV with an R² of 0.94 for training data and 0.91 for t esting data, with mean absolute percentage error (MAPE) of 3%. Model A was weakest with an R² of 0.82 for training data, 0.68 for testing data, and a MAPE of 6.3%.

Conclusion

TCV can be accurately predicted using readily avail able donor metrics. The proposed D-R TCV matching model can si gnificantly expand the donor pool and improve size matching in paediatric heart transplantation in India. Friday Abstract-based Programme 150

Limitations

Single centre study, which may also have an in-buil t case selection bias. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Owing to the retrospective nature of the study, ethical committee approval was waived off by the institutional ethics committee. Author Disclosures: Richa Kothari: Nothing to disclose Vimal Raj: Nothing to disclose Suraj Gowda: Nothing to disclose Prospective Comparison of Automated vs. Human-Guide d Cardiac MRI Planning *C. G. Glessgen*¹, L. A. Crowe¹, J. Wetzl², M. Schm idt², S. S. Yoon², J-P. Vallee¹, J-F. Deux¹; ¹Geneva/CH, ²Erlangen/DE Purpose or Learning Objective: Cardiac MRI (CMR) is demanding due to the number of planning steps and parameters requiring c ontinuous monitoring. The high mistake risk can impact procedure quality, sca n times, and data homogeneity. The impact of an AI-based automated CM R planning software on procedure errors and scan times compared to huma n-guided examinations is evaluated.

Methods

or Background: Consecutive patients undergoing non-stress CMR were prospectively enrolled into two acquisition mo des: manual or automated utilizing prototype software (Siemens Healthineers, Erlangen, Germany). Patients with pacemakers or targeted indications we re excluded. All underwent the same CMR protocol with contrast administration, in breath-hold (BH) or free breathing (FB). Supervising radiologists recorded p rocedure errors (plane prescription, forgotten views, incorrect propagatio n of a cardiac plane, field-of- view mismanagement). Scan times and Dead Phase (non -acquisition portion) were computed from scanner logs. Most data were non -normally distributed and compared using nonparametric tests.

Results

or Findings: Eighty-two patients (mean age, 51.6 years; 56 male) were included. Forty-four patients underwent automa ted CMR and 38 manual CMR. The rate of procedure errors per CMR was lower (p=0.01) in automated (0.45) than in manual (1.13). The ratio of error-fr ee examinations was higher (p=0.03) in automated (31/44; 70.5%) than in manual (17/38; 44.7%). Automated studies were shorter than manual studies in FB (30.3 vs. 36.5 minutes, p<.001) but had similar durations in BH (4 2.0 vs. 43.5 minutes, p=0.42). Dead Phase was lower in automated studies for both FB and BH strategies (p<.001).

Conclusion

AI-based automation performed cardiac MRI studies a t a clinical level with fewer planning errors and improved effic iency compared to human planning.

Limitations

No reproductibility analysis of plane adjustments b etween radiologists was performed. Radiologists could not realistically be blinded to the study arm as automatic acquisitions were perfor med with almost no visible human interaction. Funding for this study: None Ethics committee - additional information: All patients gave informed consent Author Disclosures: Lindsey A. Crowe: Nothing to disclose Jean-Paul Vallee: Nothing to disclose Carl Guillaume Glessgen: Nothing to disclose Seung Su Yoon: Employee: Siemens Healthineers Michaela Schmidt: Employee: Siemens Healthineers Jean-François Deux: Nothing to disclose Jens Wetzl: Employee: Siemens Healthineers Predicting Mortality After Transcatheter Aortic Val ve Replacement Using AI- Based Fully Automated Left Atrioventricular Cou pling Index *E. Zsarnóczay*¹, A. Varga-Szemes², U. J. Schoepf², S. Rapaka³, N. Fink², M. Vecsey-Nagy², P. Sharma³, P. Maurovich-Horvat¹, T. S. Emrich²; ¹Budapest/HU, ²Charleston, SC/US, ³Princeton, NJ/US Purpose or Learning Objective: To determine whether artificial intelligence (AI)–based fully automated assessment of left atrio ventricular coupling index (LACI) can provide incremental value above other tr aditional risk factors for predicting mortality among patients with severe aor tic stenosis (AS) undergoing coronary CT angiography (CCTA) before tr anscatheter aortic valve replacement (TAVR).

Methods

or Background: This retrospective study evaluated patients with severe AS who underwent CCTA examination before TAV R between 2014 and 2019. An AI-prototype software fully automatically calculated left atrial (LA) and left ventricular (LV) end-diastolic volumes and LAC I was defined as the ratio between them. Clinical parameters, the Society of T horacic Surgeons Predicted Risk of Mortality (STS-PROM) risk score, and all-cause mortality after TAVR were recorded. Uni- and multivariate Cox proportional hazard

Methods

were used to identify the predictors of mor tality in models adjusting for relevant significant parameters, STS-PROM score , and patients with preserved LV ejection fraction (EF).

Results

or Findings: A total of 656 patients (77 years [IQR, 71-84 years ]; 387 [59.0%] male) were included. The all-cause mortalit y rate was 21.6% over a median follow-up time of 24 (10–40) months. When ad justing for clinical confounders, LACI≥43.7% was found to independently predict mortality (adjusted HR, 1.52, [95CI: 1.03,2.22]; p=0.032). Af ter adjusting for the STS- PROM score in a separate model, LACI ≥43.7% remained an independent prognostic parameter (adjusted HR, 1.47, [95CI: 1.0 3,2.08]; p=0.031). In a sub- analysis of patients with preserved LVEF, LACI rema ined a significant predictor (adjusted HR, 1.72 [95CI: 1.02,2.89]; p=0.042).

Conclusion

AI-based fully automated assessment of LACI can be used independently to predict mortality in patients unde rgoing TAVR, including those with preserved LVEF.

Limitations

This study was performed in a single-center and sin gle-vendor setting, using an AI-powered software prototype, se lection bias may exist because only patients with available outcomes data were included. Funding for this study: Not applicable. Ethics committee - additional information: Not applicable. Author Disclosures: Emese Zsarnóczay: Nothing to disclose Puneet Sharma: Employee: Siemens Healthineers Pál Maurovich-Horvat: Nothing to disclose Milán Vecsey-Nagy: Nothing to disclose Uwe Joseph Schoepf: Research/Grant Support: Bayer, Bracco, Elucid Bioimaging, Guerbet, HeartFlow Saikiran Rapaka: Employee: Siemens Healthineers Tilman Stephan Emrich: Consultant: Siemens Medical Solutions USA Inc Nicola Fink: Nothing to disclose Akos Varga-Szemes: Research/Grant Support: Siemens Deep Learning Denoising Algorithm for Improved Asse ssment of Coronary Arteries in Transcatheter Aortic Valve Imp lantation CT Imaging *L. R. M. Lanzafame*¹, T. D'Angelo¹, A. Othman², C. Booz³; ¹Messina/IT, ²Mainz/DE, ³Frankfurt/DE ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the impact of a deep learning-based denoising (DLD) technique on im age quality and diagnostic accuracy for the assessment of coronary arteries in pre-procedural transcatheter aortic valve implantation (TAVI) CT p lanning.

Methods

or Background: A retrospective analysis was conducted on 200 patients with severe aortic stenosis who underwent CT scans for TAVI planning between October 2022 and April 2024. Conve ntional images were reconstructed, and denoised images were generated u sing DLD model.

Objective

image quality was assessed by measuring t he mean Hounsfield unit (HU) and standard deviation (SD) in the aortic root , coronary arteries, and subcutaneous fat to calculate noise, signal-to-nois e ratio (SNR), and contrast- to-noise ratio (CNR). Two independent readers subje ctively evaluated sharpness, noise, vascular contrast, and overall im age quality using a 5-point Likert scale. Diagnostic performance was compared b etween original and denoised images by assessing accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive val ue (NPV), using invasive coronary angiography as the reference standard.

Results

or Findings: Denoised images demonstrated significantly improved SNR (37.5 ± 12.8 vs. 12.3 ± 4.1) and CNR (45.3 ± 15.4 vs. 14.7 ± 4.4), along with reduced noise levels (16.9 ± 7.9 vs. 47.9 ± 11.6 HU) (all p<0.001). Subjective evaluations also favored denoised images in terms of sharpness, noise reduction, contrast, and overall quality (all p<0.001). DLD reconstructions revealed higher diagnostic performance, showing a o n a per-segment basis sensitivity of 95.9%, specificity of 94.3%, PPV of 86.5%, NPV of 98.4%, and accuracy of 94.8%.

Conclusion

The DLD algorithm significantly improves image qual ity and diagnostic accuracy in pre-TAVI CT imaging for coro nary artery evaluation.

Limitations

The retrospective design prevented evaluation of im age quality at reduced radiation doses. Furthermore, the results a re specific to our acquisition protocol. Funding for this study: This research did not receive external funding. Ethics committee - additional information: The study was approved by the Ethics Committee of Johannes Gutenberg University o f Mainz (Ref. Nr. 2022- 16477_1) Author Disclosures: Christian Booz: Speaker: Siemens Healthineers Ludovica Rosa Maria Lanzafame: Nothing to disclose Tommaso D'Angelo: Speaker: Philips Speaker: Bracco Ahmed Othman: Nothing to disclose Friday Abstract-based Programme 151 A recommendation: test-retest reliability of radiom ic features in myocardial T1 and T2 mapping *M. Manzke*¹, F. C. Laqua², B. Böttcher¹, A-C. Klem enz¹, M-A. Weber¹, B. Baeßler², F. G. Meinel¹; ¹Rostock/DE, ²Würzburg/ DE ([email protected]) Purpose or Learning Objective: To investigate the reproducibility of radiomic features in myocardial native T1 and T2 mapping.

Methods

or Background: Cardiac MRI T1 maps from 50 healthy volunteers (29 women and 21 men, mean age 39.4 ± 13.7 years) u nderwent two identical cardiac MRI examinations at 1.5T. The protocol incl uded native T1 and T2 mapping in both short-axis and long-axis orientatio n. For T1 mapping, we investigated standard (1.9 x 1.9 mm) and high (1.4 x 1.4 mm) spatial resolution. After manual segmentation of the left v entricular myocardium, 100 radiomic features from seven feature classes were e xtracted and analyzed. Test–retest repeatability of radiomic features was assessed using the intraclass correlation coefficient (ICC) and classi fied as poor (ICC 0.90).

Results

or Findings: For T1 maps acquired in short-axis orientation at standard resolution, repeatability was excellent fo r 6 features, good for 29 features, moderate for 19 features and poor for 46 features. We identified 15 features from 6 classes which showed good to excell ent reproducibility for T1 mapping in all resolutions and all orientations. Fo r short-axis T2 maps, repeatability was excellent for 6 features, good fo r 25 features, moderate for 23 features and poor for 46 features. 12 features from 5 classes were found to have good to excellent repeatability in T2 mapping independent of slice orientation.

Conclusion

We have identified a subset of radiomic features wi th good to excellent repeatability independent of slice orient ation and spatial resolution. We recommend using these features for further radio mics research in myocardial T1 and T2 mapping.

Limitations

This study was limited to healthy volunteers. The r eproducibility of radiomic features in patients with diffuse or focal myocardial disease cannot be directly concluded. Funding for this study: The study was in part funded by the Federal Ministr y of Education and Research (BMBF) through the Networ k University Medicine „NUM 2.0“ (grant number 01KX2121). Ethics committee - additional information: This study was approved by the institutional review board and written informed con sent was obtained from all volunteers prior to enrollment. Author Disclosures: Fabian Christopher Laqua: Nothing to disclose Benjamin Böttcher: Nothing to disclose Mathias Manzke: Nothing to disclose Bettina Baeßler: Author: This study has been suppor ted by the Deutsche Forschungsgemeinschaft (DFG, German Research Founda tion) within the Priority Programme SPP 2177 Radiomics (BA 6438/4–2) and by the Federal Ministry of Education and Research (BMBF; “SWAG” pr oject). BB is founder and CEO of Lernrad GmbH and has received speaker fe es by Bayer Vital GmbH. Ann-Christin Klemenz: Nothing to disclose Felix G. Meinel: Author: Unrelated to this work, Dr . Meinel has received institutional research support from GE Healthcare a nd speaker’s honoraria from GE Healthcare, Circle Cardiovascular Imaging a nd Bayer Vital. Marc-André Weber: Nothing to disclose Reproducibility of an AI-assisted plane positioning tool for cardiac MRI *B. Böttcher*¹, K. K. Deyerberg¹, A-C. Klemenz¹, L- M. Watzke¹, M. Gorodezky², M. Manzke¹, M-A. Weber¹, F. G. Meinel¹; ¹Rostock/DE , ²Munich/DE ([email protected]) Purpose or Learning Objective: Plane positioning in cardiac magnetic resonance imaging (cMRI) is crucial for diagnostic image quality and comparability of cardiac functional parameters in f ollow-up exams. Manual planning is influenced by user’s training making it susceptible for inter-reader variability and errors. This prospective cohort stu dy aims to investigate the reproducibility of an artificial intelligence-based planning approach against state-of-the-art manual plane prescription.

Methods

or Background: 25 healthy participants (mean age 41.5, range: 23- 65 years, mean BMI 25.2 kg/m²) underwent two identi cal cMRI exams on a 1.5T scanner (Signa Artist, GE HealthCare). Short a xis, 2-, 3- and 4-chamber planes (FOV: 34x34cm2, matrix size: 200x224, slice thickness: 8mm) were acquired using an AI-based planning tool (TeslaFlow prototype, GE HealthCare) and manual planning. Short axis left ve ntricular volumetric analysis (end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV) and ejection fraction (EF)) were perfor med using an established post-processing software (cvi42, Circle Cardiovascu lar Imaging). The Wilcoxon matched-pairs signed rank test with a significance level of p≤0.05 was used to compare the first to the second exam for both the m anual and automated planning.

Results

or Findings: Volumetric parameters calculated on manual and AI- assisted planned images showed following median of differences between both scans: EDV -5.0ml (p=0.220), -2.8ml (p=0.474); ESV 0.1ml (p=0.560), 2.0ml (p=0.367); SV -3.6ml (p=0.096), -4.2ml (p=0.043) an d EF -1.2% (p=0.329), - 2.8% (p=0.045), respectively. The only statisticall y significant differences were observed in SV and EF for AI-based planning, though the deviation is not clinically relevant.

Conclusion

AI-based planning for cMRI showed high reproducibil ity without clinically relevant variability between follow-up s cans. This novel technique can simplify and accelerate cMRI maintaining high diagn ostic quality.

Limitations

This study was conducted on a cohort of healthy ind ividuals at a single MRI scanner, provided by a single vendor. Funding for this study: None. Ethics committee - additional information: The study was designed as a prospective, single-center cohort study and approve d by the responsible institutional review board of the Medical Universit y Center of Rostock. Author Disclosures: Benjamin Böttcher: Nothing to disclose Margarita Gorodezky: Employee: GE HealthCare Mathias Manzke: Nothing to disclose Ann-Christin Klemenz: Nothing to disclose Felix G. Meinel: Nothing to disclose Marc-André Weber: Nothing to disclose Lena-Maria Watzke: Nothing to disclose Karolin Kristina Deyerberg: Nothing to disclose Accelerated Deep Learning-Based Function Assessment in Cardiovascular Magnetic Resonance *F. Fanelli*, D. De Santis, L. Pugliese, G. G. Bona , C. Santangeli, T. Polidori, G. Tremamunno, D. Caruso, A. Laghi; Rome/IT ([email protected]) Purpose or Learning Objective: Cardiovascular magnetic resonance (CMR) is the reference standard for the assessment of car diac function, achieved through conventional balanced steady-state free pre cession (bSSFP) cine sequences, which represent a considerable part of t he CMR exams, contributing to patient discomfort. The aim of our study was to evaluate diagnostic accuracy and image quality of deep-learn ing(DL)cine sequences for LV and RV parameters compared to bSSFP cine sequenc es in CMR.

Methods

or Background: From January to April 2024, patients with clinicall y indicated CMR were prospectively included. LV and R V were segmented from short-axis bSSFP and DL cine sequences. LV and RV e nd-diastolic volume, end-systolic volume, stroke volume, ejection fracti on, and LV end-diastolic mass were calculated. The acquisition time of both sequences was registered.

Results

were compared with paired-samples t-test or Wilcoxon signed-rank test. Agreement between DL cine and bSSFP was asses sed using Bland- Altman plots. Image quality was graded by two reade rs based on blood-to- myocardium contrast, endocardial edge definition, a nd motion artifacts, using a 5-point Likert scale (1= insufficient quality; 5= e xcellent quality).

Results

or Findings: Sixty-two patients were included (mean age: 47±17 years, 41 men). No significant differences between DL cine and bSSFP were found for all LV and RV parameters (P≥ .176). DL cine was significantly faster (1.35 ±.55 m vs 2.83 ± .79 m; P< .001). The agreement between DL cine and bSSFP was strong, with near-zero bias and good limi ts of agreement. Overall image quality was comparable (median: 5, IQR: 4-5; P= .330), while endocardial edge definition of DL cine (median: 4, IQR: 4-5) was lower than bSSFP (median: 5, IQR: 4-5; P= .002).

Conclusion

DL cine allows fast and accurate quantification of LV and RV parameters and comparable image quality with conven tional bSSFP.

Limitations

Not applicable Funding for this study: No funding was provided for this study. Ethics committee - additional information: This study has been approved by local Ethics committee . Author Disclosures: Luca Pugliese: Nothing to disclose Damiano Caruso: Nothing to disclose Curzio Santangeli: Nothing to disclose Federica Fanelli: Nothing to disclose Domenico De Santis: Nothing to disclose Giuseppe Tremamunno: Nothing to disclose Tiziano Polidori: Nothing to disclose Andrea Laghi: Nothing to disclose Giovanna Grazia Bona: Nothing to disclose Friday Abstract-based Programme 152 Super-Resolution Deep Learning Reconstruction to Im prove the Accuracy of CT Fractional Flow Reserve: Comparison to Model-based Iterative Reconstruction *N. Tomizawa*, Y. Nozaki, R. Fan, Y. Kawaguchi,, K. Takamura, F. Shinichiro, K. Kumamaru, T. Minamino, S. Aoki; Bunkyo-Ku/JP ([email protected]) Purpose or Learning Objective: The purpose of this study was to compare the diagnostic performance of CT fractional flow re serve (CT-FFR) using model-based iterative reconstruction (MBIR) and sup er-resolution deep learning reconstruction (SR-DLR) to detect function ally significant stenosis as assessed by invasive FFR.

Methods

or Background: This single-center retrospective study included 79 patients (mean age, 70 years ± 11 [SD]; 57 men) who underwent coronary CT angiography showing intermediate stenosis (30% ‒70% stenosis) and subsequent invasive FFR between February 2022 and M arch 2024. Vessels with heavy calcification were not excluded from the analysis. Computational fluid dynamics was used to calculate the CT-FFR usi ng MBIR and SR-DLR images. Per-vessel diagnostic performance to detect FFR ≤0.80 in coronary angiography was compared by analyzing receiver oper ating characteristic (ROC) curves.

Results

or Findings: Of the 98 vessels evaluated, 46 vessels (47%) had functionally significant stenosis. The median (inte rquartile range) calcium score was 462 (134–932). CT-FFR values calculated using b oth MBIR (mean difference: −0.088; 95% CI: −0.129, −0.048; p <0.001) a n d SR-DLR (mean difference: −0.026; 95% CI: −0.050, −0.002; p = 0.03) we re un dere s tima te d compared to invasive FFR. The area under the ROC cu rve to diagnose functionally significant stenosis was higher for SR -DLR (0.88; 95% CI: 0.80, 0.95) than for MBIR (0.76; 95% CI: 0.67, 0.86; p = 0.003). CT-FFR calculated using SR-DLR had improved diagnostic accuracy (88% vs. 70%, p <0.001) and specificity (87% vs. 63%, p <0.001) over MBIR, but had similar sensitivity (89% vs. 78%, p = 0.06).

Conclusion

SR-DLR images improved the diagnostic performance o f CT-FFR over MBIR images in detecting functionally signific ant stenosis as assessed by invasive FFR.

Limitations

This study is retrospective and used a single CT ve ndor. Multi- vendor multi-center study is necessary to confirm t he findings. Funding for this study: None Ethics committee - additional information: Approved by the Ethics Committee of Juntendo University on May 2, 2024 (No . E23-0040-H02) Author Disclosures: Fujimoto Shinichiro: Nothing to disclose Yuko Kawaguchi,: Nothing to disclose Shigeki Aoki: Nothing to disclose Kazuhisa Takamura: Nothing to disclose Tohru Minamino: Nothing to disclose Yui Nozaki: Nothing to disclose Nobuo Tomizawa: Nothing to disclose Kanako Kumamaru: Nothing to disclose Ruiheng Fan: Nothing to disclose Evaluating the Feasibility of a Customised GPT-4 Mo del for Extracting CAD-RADS Classification from Coronary CT Angiograph y Reports *V. Vingiani*, B. Proner, N. Cortellini, R. Vallett a, T. Gorgatti, A. Posteraro, V. Corato, M. Bonatti; Bolzano/IT ([email protected]) Purpose or Learning Objective: This study assessed the feasibility of a customized GPT-4 model in categorising cardiac radi ological reports using the Coronary Artery Disease Reporting and Data System ( CAD-RADS) classification.

Methods

or Background: A customised GPT-4 model was developed using the CAD-RADS 2.0-2022 guidelines, provided as a PDF , and fine-tuned on 30 clinical scenarios. The model was tested on 118 ano nymized Coronary CT Angiography (CCTA) reports. Data included patient m etrics and report details (e.g., length, conclusions). The reports were also reviewed by a radiologist with 9 years of experience, who categorised them accordi ng to the CAD-RADS classification. The time required for manual assess ment was recorded. The GPT-4 model's performance was evaluated using Cohen 's kappa for agreement and the Wilcoxon test for comparing proce ssing times between the model and the radiologist.

Results

or Findings: The mean patient age was 59.6 years (±10.9), with 4 1% women. Reports were authored by 10 radiologists, wi th 88% in Italian and 12% in German. The median report length was 1,456 chara cters, with conclusions in 76% of reports. The GPT-4 model showed substanti al agreement with the radiologist, achieving a Cohen's kappa of 0.79 (95% CI: 0.68 - 0.89). It significantly reduced processing time, averaging 16 seconds per report compared to 57 seconds for the radiologist (P < 0.0 001).

Conclusion

These findings suggest that a customized GPT-4 mode l is a promising tool for autonomously categorising radiol ogical findings using the CAD-RADS classification when the original report do es not include it, thereby offering a time-efficient alternative. Implementing such a system could assist clinicians and cardiologists in consistently interp reting reports by providing CAD-RADS classifications when they are not explicit ly reported.

Limitations

The GPT model was fine-tuned using only 30 scenario s. A larger dataset could be used to improve the model's perfor mance. Funding for this study: None Ethics committee - additional information: This study was conducted in accordance with the principles of the Declaration o f Helsinki Author Disclosures: Andrea Posteraro: Nothing to disclose Matteo Bonatti: Nothing to disclose Valentina Corato: Nothing to disclose Tommaso Gorgatti: Nothing to disclose Bernardo Proner: Nothing to disclose Riccardo Valletta: Nothing to disclose Nino Cortellini: Nothing to disclose Vincenzo Vingiani: Nothing to disclose DLR-based Motion Correction of Coronary CTA: Prelim inary Evaluation F. Tatsugami¹, A. Streiff², T. Higaki¹, A. Labani², W. Fukumoto¹, S. El Ghannudi², K. Haioun³, K. Awai¹, *M. Ohana*²; ¹Hiroshima/JP, ²Strasbourg/FR, ³Tokyo/JP ([email protected]) Purpose or Learning Objective: DLR-based Motion Correction (MC-DLR) for Coronary CTA has the potential to reduce/eliminate kinetic artifacts in CCTA more effectively than traditional algorithms, but i ts clinical impact is still unknown. We aim to evaluate the effect of MC-DLR on coronary luminal and stenosis assessment in a varied CCTA cohort.

Methods

or Background: Sixty CCTA (20 with HR75) with various degrees of stenosis (50% CAD-RADS 1 & 2, 50% CAD-RADS 3 & 4) were retrospectively selected f rom 2 tertiary centers. All scans were acquired on 4th/5th-gen wide-area de tector CT within 1 heartbeat. Best phase for each included CCTA was re constructed without and with MC-DLR, using Super Resolution DLR with 1024 m atrix-size. MC-DLR subdivides the data required for volume reconstruct ion into smaller time sections to estimate coronary artery motion. Four r adiologists with varying levels of expertise independently and randomly revi ewed all 120 datasets to: (A) grade the luminal/wall image quality using a 3- level scale, for the 9 coronary artery segments, and (B) assess CAD-RADS. Statistical analysis used descriptive and Bayesian approaches.

Results

or Findings: For each reader and for the pooled analysis, overal l luminal/wall image quality score was significantly better with MC-DLR than without (p<0.05). Per segment, the positive effect was more consistent on the RCA (improvement in 62% of cases) than on the LAD ( 44%) and the Cx (21%). Effect of MC-DLR was non-existent in cases with abs ent/minimal coronary kinetic artifacts. Non-significant changes in intra /inter-reader variability were noted in CAD-RADS 3/4.

Conclusion

MC-DLR significantly enhances coronary artery lumin al and wall image quality in cases with moderate or severe kine tic artifacts, suggesting a potential clinical role in refining stenosis assess ment when above >50%.

Limitations

Quantitative analysis with attenuation profile curv es was not performed. Funding for this study: No funding was received for this study. Ethics committee - additional information: IRB from Strasbourg University Hospital Author Disclosures: Wataru Fukumoto: Nothing to disclose Aissam Labani: Nothing to disclose Soraya El Ghannudi: Nothing to disclose Fuminari Tatsugami: Nothing to disclose Toru Higaki: Nothing to disclose Kazuo Awai: Nothing to disclose Amandine Streiff: Nothing to disclose Mickaël Ohana: Consultant: Boehringer Ingelheim Con sultant: Canon Medical Systems Europe Karim Haioun: Employee: Canon Medical Systems Japan Performance of AI-based automated coronary artery c alcium density quantification on CT Y. J. Suh¹, C. Kim², W-S. Yoo¹, J. Y. Kim³, S. Chan g¹, C. H. Park⁴, *S. J. Hong*⁵, D. H. Yang¹, H. S. Yong¹; ¹Seoul/KR, ²Ansan/KR, ³ Daegu/KR, ⁴Cheonan/KR, ⁵Guri/KR Purpose or Learning Objective: Coronary artery calcium (CAC) density on electrocardiogram (ECG)-gated CT has been suggested as an inverse prognostic marker for prediction of future adverse cardiovascular events. We aimed to evaluate the performance of artificial int elligence (AI)-based automated CAC density quantification on ECG-gated c alcium scoring CT Friday Abstract-based Programme 153 (CSCT) and non-ECG-gated low-dose chest CT (LDCT), using multi- institutional datasets.

Methods

or Background: A total of 1,540 pairs of CSCT-LDCT scans from a multicenter database were retrospectively included. AI-based automated CAC quantification was conducted on the CSCT and LDCT. For cases with CAC score>0, mean and peak CAC density was calculated f rom the labeled CAC. For peak CAC density factors, a value of 1 to 4 was assigned based on the measured peak density attenuation (1: 130-199HU; 2: 200-299HU; 3: 300- 399HU; 4: >400HU). The reliability and agreement of the mean CAC density and peak CAC density categories obtained from the a utomated scoring were analyzed compared to manual measurement, using the intraclass correlation coefficient (ICC), Bland-Altman analysis, and weigh ted kappa (κ) statistics, respectively.

Results

or Findings: A total of 808 CSCT scans and 579 LDCT scans were positive for CAC. Automated mean density measuremen t demonstrated excellent ICCs on CSCT and LDCT (0.988 [95% CI, 0.9 86-0.989] vs. 0.956 [95% CI, 0.948-0.962]). Mean bias with 95% limits o f agreement for the mean density was 0.7 ± 22.4 on CSCT and 0.6 ± 27.4 on LDCT. In terms of peak density category, automated measurement on CSCT and LDCT exhibited excellent reliability with manual measurement (weig hted κ 0.980 [95% CI, 0.969-0.991) and 0.964 [95% CI, 0.945-0.982]).

Conclusion

AI-based automated CAC quantification can provide a ccurate and reliable measurement of CAC density on CSCT and LDCT across multi- institutional datasets.

Limitations

Prognostic value of AI-based CAC density should be further investigated. Funding for this study: The Researcher Supporting Program funded by Korean Society of Cardiovascular Imaging (KOSCI) an d the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT)(No. 2021R1A2C4002195) Ethics committee - additional information: Approval numbers: KC22RIDI0156, 2021-12-027, 2022GR0064, 2021AS0371, 2022-01-001, 2021-0303, 2021-12-029, and 4-2021-1589 Author Disclosures: Dong Hyun Yang: Nothing to disclose Young Joo Suh: Nothing to disclose Suyon Chang: Nothing to disclose Chan Ho Park: Nothing to disclose Jin Young Kim: Nothing to disclose Su Jin Hong: Nothing to disclose Cherry Kim: Nothing to disclose Hwan Seok Yong: Nothing to disclose Won-Seok Yoo: Nothing to disclose 09:30-11:00 Research Stage 4 Research Presentation Session: Neuro RPS 1311 Decoding the mind: sculpting neuroimaging with technology Moderator A. Krainik; Grenoble/FR ([email protected]) Author Disclosures: Alexandre Krainik: Advisory Board: Geodaisics Clinical Evaluation of 3D Motion-Correction via Sco ut Accelerated Motion Estimation and Reduction (SAMER) framework versus C onventional T1- Weighted MRI at 1.5 T in Brain Imaging *L. Leukert*, A. Kronfeld, R. Paul, M. A. Brockmann , S. Altmann, A. Othman; Mainz/DE ([email protected]) Purpose or Learning Objective: To evaluate the presence of motion artifacts in 1.5 T T1-weighted MRI scans using 3D motion corr ection via the Scout Accelerated Motion Estimation and Reduction (SAMER) framework versus conventional image reconstruction.

Methods

or Background: MRI long scan times often cause motion artifacts, reducing image quality. SAMER uses an ultrafast pre -scan and repeated acquisition of a minimal number of additional k-spa ce encoding lines, enabling feasible computation times. A preliminary study (14 volunteers) assessed SAMER’s effect on induced motion at 3T. The main st udy (82 patients) compared conventional resonstruction (Non-Moco) and SAMER (SAMER Moco) motion correction using 3D T1-weighted imagin g at 1.5T. Radiologists evaluated images with a 5-point Likert scale.

Results

or Findings: In the preliminary study, SAMER Moco showed significant improvements over Non-Moco across all i maging parameters (p < 0.001), with 52.4% and 66.7% of cases rated as exce llent or good for artifact freedom and image quality, compared to 21.4% for No n-Moco. The main study underlined these findings. SAMER Moco demonstrated superior image quality and outperformed Non-Moco, particularly in diagnost ic confidence and overall image quality (p < 0.0001). Diagnostic confidence w as rated excellent or good in 93.8% of SAMER Moco cases versus 72.0% for Non-M oco. Similarly, 84.6% of SAMER Moco cases had excellent or good image qua lity (56.8% for Non- Moco). Odds ratios favoured SAMER Moco (5.444 and 5 .807, respectively, p < 0.0001). Multi-reader agreement was excellent acros s all parameters.

Conclusion

The use of SAMER in T1-weighted imaging is feasible in clinical practice and significantly enhances the reliability of 1.5 T brain MRI by successfully mitigating motion artifacts.

Limitations

This study's limitations include its single-centre design, reliance on a single 3D MR sequence, and inclusion of both c ontrast-enhanced and non-contrast scans. SAMER may also be affected by p atient-induced k-space gaps. Funding for this study: This research received no funding from any public, commercial, or not-for-profit sources. The authors declare that they have no competing interests that are relevant for the conte nt of this article. Ethics committee - additional information: This single-center prospective study was approved by our institution's local ethic s committee, and written informed consent was obtained (approval number 2021 -15811). Our study was conducted in accordance with the Declaration of Hel sinki and its amendments. Author Disclosures: Roman Paul: Nothing to disclose Laura Leukert: Nothing to disclose Andrea Kronfeld: Nothing to disclose Marc A Brockmann: Nothing to disclose Ahmed Othman: Nothing to disclose Sebastian Altmann: Nothing to disclose Comparison of Photon-Counting CT and Conventional C T for Determining Rotational Orientation of Directional D BS Electrodes: A Phantom Study *D. Fedders*¹, A. Hellerbach², M. Eichner², C. Pank nin³, S. Faby³, J. Wirths², V. Visser-Vandewalle², H. Treuer², S. Hunsche²; ¹Ch emnitz/DE, ²Cologne/DE, ³Forchheim/DE ([email protected]) Purpose or Learning Objective: Accurate determination of the rotational orientation of directional deep brain stimulation ( DBS) electrodes is crucial for optimizing therapeutic outcomes in functional neuro surgery. Conventional CT methods, relying on artifact analysis, face precisi on limitations, especially at certain angles. Photon-counting detector CT (PCD-CT ), with its superior resolution, offers a potential alternative. This st udy compares the efficacy of PCD-CT against conventional CT-based artifact analy sis in determining DBS electrode orientation.

Methods

or Background: A phantom study was conducted using directional leads from Boston Scientific, Medtronic, and Abbott embedded in cylindrical phantoms. The phantoms were scanned with PCD-CT for direct orientation detection and conventional CT for stripe artifact a nalysis. Scans covered varying polar angles to assess accuracy and consist ency. Key metrics included orientation accuracy and dependency on lead positio n relative to the CT gantry.

Results

or Findings: PCD-CT demonstrated high accuracy across all tested angles, independent of lead alignment. In contrast, conventional CT showed reduced precision, particularly at extreme angles w here artifact detection was unreliable. PCD-CT enabled consistent, precise asse ssments of segmented contacts, enhancing postoperative DBS programming.

Conclusion

PCD-CT offers a robust solution for determining the rotational orientation of DBS electrodes, overcoming limitatio ns of conventional artifact- based methods. This supports more accurate electrod e positioning and programming, potentially improving functional neuro surgery outcomes.

Limitations

The phantom-based design may not replicate clinical complexity, limiting generalizability. The study only evaluated specific directional DBS leads, so results may not apply to other types. Add itionally, PCD-CT’s limited availability could hinder immediate clinical applic ation. Funding for this study: None beside scanning time at the research facility from Siemens Ethics committee - additional information: Phantom study Friday Abstract-based Programme 154 Author Disclosures: Harald Treuer: Nothing to disclose Jochen Wirths: Nothing to disclose Veerle Visser-Vandewalle: Nothing to disclose Dieter Fedders: Nothing to disclose Christoph Panknin: Other: Computed Tomography, Siem ens Healthineers AG, Forchheim/DE Sebastian Faby: Other: Computed Tomography, Siemens Healthineers AGC Alexandra Hellerbach: Nothing to disclose Markus Eichner: Nothing to disclose Stefan Hunsche: Nothing to disclose A Multimodal MRI-Based Machine Learning Framework f or Classifying Cognitive Impairment in Cerebral Small Vessel Disea se *G. Lin*, W. Chen, M. Chen, J. Ji; Lishui/CN Purpose or Learning Objective: This study aims to propose a multimodal magnetic resonance imaging (MRI)-based machine lear ning framework to effectively classify mild cognitive impairment (MCI ) and no cognitive impairment (NCI) in patients with cerebral small ve ssel disease (CSVD).

Methods

or Background: We enrolled 223 patients with CSVD, categorized into NCI (n = 121) and MCI (n = 102) groups based o n neurocognitive assessments. Multimodal MRI data, including T1-weig hted, resting-state functional MRI, and diffusion tensor images, were c ollected. Image preprocessing, feature extraction, and feature sele ction methods were applied to obtain MRI features from the three modalities. T he AutoGluon platform was utilized for model development, and traditional mac hine learning algorithms were applied for comparison. The models were valida ted using a validation cohort of 97 patients with CSVD, and their performa nce was assessed via receiver operating characteristic curve (ROC) analy sis.

Results

or Findings: The AutoGluon model to distinguish MCI from NCI based on multimodal MRI features demonstrated a hig h area under the ROC curve (AUC), accuracy, sensitivity, specificity, an d F1-score in the testing set (0.894, 85.65%, 84.31%, 86.78%, and 84.31%, respect ively) and validation cohort (0.846, 79.38%, 81.82%, 77.36%, and 78.26%, respectively). Other models built using traditional machine learning alg orithms had AUCs of 0.661– 0.732, and their prediction accuracies were signifi cantly lower than that of the AutoGluon model (P < 0.001).

Conclusion

Our study provides a multimodal MRI-based machine l earning framework, utilizing the AutoGluon platform, that o utperforms traditional algorithms in classifying MCI and NCI in patients w ith CSVD, offering a promising tool for the early prediction of MCI in C SVD.

Limitations

As a retrospective study, it is susceptible to sele ction bias, which may limit its generalizability. Funding for this study: This study is supported by Zhejiang Public Welfare Research Program (LGF20H220002, LGF19H180010), and Zhejiang Provincial Healthcare Program (2024KY562) Ethics committee - additional information: This study was approved by the Ethics Committee of the Fifth Affiliated Hospital o f Wenzhou Medical University (approval number: 2024-266) Author Disclosures: Minjiang Chen: Nothing to disclose Jiansong Ji: Nothing to disclose Weiyue Chen: Nothing to disclose Guihan Lin: Nothing to disclose Detection of intracranial hemorrhage using ultralow -dose brain computed tomography with deep learning reconstruction versus conventional-dose computed tomography *C. Otgonbaatar*¹, H. Kim², P-H. Jeon², S. H. Jeon² , S. Cha², J-K. Ryu¹, H. Shim¹, S. M. Ko², J. Kim²; ¹Seoul/KR, ²Wonju-si/ KR ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the diagnostic performance, image quality, and radiation dose amon g ultralow-dose protocol with deep learning reconstruction (DLR), ultralow-d ose computed tomography (CT) with iterative reconstruction (IR), and conven tional-dose protocols for detecting intracranial hemorrhage.

Methods

or Background: This retrospective study enrolled 93 patients. All patients underwent follow-up noncontrast CT with ul tralow-dose setting after initial conventional-dose CT within 5 days. A conve ntional-dose CT was obtained using 123–188 mA and IR. Ultralow-dose CT was obtained using 50 mA with IR and DLR. Qualitative assessments and qua ntitative assessments (image noise, differentiation between gray and whit e matter, and artifact) were conducted. The diagnostic performance for detecting intracranial hemorrhage using ultralow-dose CT with IR and ultralow-dose CT with DLR was assessed.

Results

or Findings: An approximately 84.0% reduction in median volume C T dose index was found in the ultralow-dose CT protoc ol (5.6 mGy) compared with conventional-dose CT (35.02 mGy; IQR: 33.09–37 .36). Ultralow-dose CT with DLR significantly (p < 0.001) improved image n oise, SNR, and CNR compared with ultralow-dose CT with IR and conventi onal-dose CT. Ultralow- dose CT with DLR resulted in higher sensitivity (99 .3% vs. 98.6%) and specificity (97.5% vs. 97.5%) for detecting intracr anial hemorrhage than ultralow-dose CT with IR.

Conclusion

Ultralow-dose CT with DLR is an acceptable techniqu e that provides higher image quality and diagnostic perfor mance with a reduction in radiation dose of approximately 87.7% compared with conventional-dose CT.

Limitations

We did not investigate the effect of a tube current of <50 mA on the diagnostic performance of intracranial hemorrha ge and image quality. Further validation is required to investigate a low er effective dose of <0.21 mSv. Additionally, all results were limited to one scanner, and acquisition parameters may require adjustment for different CT vendors. Funding for this study: None Ethics committee - additional information: No Author Disclosures: Sung Min Ko: Nothing to disclose Pil-Hyun Jeon: Nothing to disclose Jinwoo Kim: Nothing to disclose Sungjin Cha: Nothing to disclose Jae-Kyun Ryu: Nothing to disclose Hyunjung Kim: Nothing to disclose Sang Hyeon Jeon: Nothing to disclose Hackjoon Shim: Nothing to disclose Chuluunbaatar Otgonbaatar: Nothing to disclose Correlation of diffusion tensor imaging findings in cerebral sensorimotor regions with neurophysiological deficits in patient s after spinal cord injury A. Zimny¹, *W. N. Machaj*¹, P. Podgórski¹, W. Fortu na¹, J. Huber², B. Bobek-Billewicz³, P. Tabakow¹; ¹Wrocław/PL, ²Poz nań/PL, ³Gliwice/PL Purpose or Learning Objective: The aim of the study is to examine the correlation between DTI findings, clinical motor an d sensory deficits, and motor evoked potential (MEP) parameters. This will provid e insights into the potential of DTI metrics as biomarkers for predicting functio nal recovery and guide therapeutic interventions in patients with chronic spinal cord injury (SCI).

Methods

or Background: A total of 29 patients with SCI (both paraplegic p- SCI and tetraplegic t-SCI), matched by sex and age to 29 healthy controls, were neurologically and neurophysiologically evalua ted, including MEPs recorded from upper and lower limb muscles. Diffusi on tensor imaging (DTI) was performed using a 3 Tesla MRI scanner and proce ssed using Human Motor Area (HMAT) and Sensorimotor Area Tract (SMAT T) templates.

Results

or Findings: No significant DTI differences were found between p - SCI and t-SCI or p-SCI and healthy controls. Howeve r, patients with t-SCI had lower fractional anisotropy (FA) in primary motor ( M1) and sensorimotor (S1) tracts, pre-supplementary motor area (pre-SMA) trac ts, M1 and S1 cortices, and left pre-SMA cortex compared to controls. In t- SCI patients, higher motor scores correlated with increased FA in ventral prem otor area (PMv) tracts and cortices, and higher sensory scores with higher FA in S1 tracts. MEP amplitudes from rectus femoris also positively corr elated with FA in motor tracts, M1, PMd, PMv, and SMA cortices.

Conclusion

DTI findings reveal distant degeneration in the sen sorimotor cortex and supraspinal tracts in chronic SCI, which correlates with clinical motor and sensory scores, as well as MEP parameters from rectus femoris muscles in t-SCI patients. DTI metrics can serve as potential biomarkers to predict motor and sensory recovery in patients with SCI and to guide and track therapeutic interventions.

Limitations

Cross-sectional design and the small sample size. Funding for this study: Grant NCBiR ERA-NET-NEURON/13/2018, Wroclaw Medical University grant SB. Ethics committee - additional information: The study was performed in accordance with the Declaration of Helsinki and was approved by the Bioethics Committee of the Wroclaw Medical University. Author Disclosures: Juliusz Huber: Author: Nothing Anna Zimny: Author: Nothing Paweł Tabakow: Author: Nothing Weronika Natalia Machaj: Author: Nothing Przemysław Podgórski: Author: Nothing Barbara Bobek-Billewicz: Author: Nothing Wojciech Fortuna: Author: nothing Friday Abstract-based Programme 155 MR Neurography at 3T and 7T for the assessment of p roximal nerve damage in polyneuropathies *J. M. E. Jende*¹, C. Mooshage¹, K. Zhang¹, T. Plat t¹, C. Neelsen¹, M. Bendszus¹, H-P. Schlemmer¹, M. Ladd¹, F. Kurz²; ¹Heidelberg/DE, ²Geneva/CH Purpose or Learning Objective: Disorders of the peripheral nervous system such as polyneuropathies pose a huge challenge to t he global healthcare system. The exact pathophysiology underlying most p olyneuropathies remains poorly understood. Previous studies on 3T MR neurog raphy (MRN) have found that the maximum of fascicular nerve damage in vari ous polyneuropathies is located at the level of the sciatic nerve although clinical symptoms usually occur further distally.

Methods

or Background: To understand the clinical impact and physiological

Background

of fascicular sciatic nerve lesions, 10 patients with distal symmetric polyneuropathy and 10 healthy controls matched for age and BMI underwent T2-weighted, high resolution MRN of the right thigh at 3T and 7T. At 7T, the maximum fascicular diameter and the average number of nerve fascicles per slice were measured. At 3T, additional diffusion-we ighted and T2-relaxometry sequences were acquired and the sciatic nerve’s fra ctional anisotropy (FA) and T2 relaxation times (T2R) were calculated.

Results

or Findings: At 7T, the sciatic nerve’s maximum fascicular diame ter in patients with polyneuropathies was larger compar ed to controls (1.41mm±0.16 vs. 1.03mm±0.07; p=0.049mm). In patients with polyneuropathies, the fascicular diameter was negat ively correlated with the number of nerve fascicles (r=-0.72;p=0.018) and the FA (r=-0.78;p=0.017mm). Positive correlations were found between the fascic ular diameter and T2R (r=0.76;p=0.019).

Conclusion

The results indicate that polyneuropathies cause a fusion of nerve fascicles that results in a reduced number of nerve fascicles and a larger fascicular diameter that is associated with a struc tural and functional decline represented by a decrease in nerve FA and an increa se in T2R. This study is the first to show that the fascicular diameter of t he sciatic nerve is directly related to changes in FA, T2R and clinical neuropat hy status. The causes of fascicular fusion remain to be determined.

Limitations

Cohort Size Funding for this study: Else Kröner Fresenius Foundation (EKFS) Ethics committee - additional information: This study was appoved by the local ethics committee of Heidelberg University Hos pital. Author Disclosures: Tanja Platt: Nothing to disclose Ke Zhang: Nothing to disclose Christian Neelsen: Nothing to disclose Johann Malte Enno Jende: Nothing to disclose Mark Ladd: Nothing to disclose Felix Kurz: Nothing to disclose Martin Bendszus: Nothing to disclose Christoph Mooshage: Nothing to disclose Heinz-Peter Schlemmer: Nothing to disclose Deep-learning-reconstructed 3D MR neurography of ex traforaminal cranial and spinal nerves F. Ensle¹, *F. Zecca*², B. J. Kerber¹, M. Lohezic¹, J. Kroschke¹, K. Pawlus¹, R. Guggenberger³; ¹Zurich/CH, ²Cagliari/IT, ³Winter thur/CH ([email protected]) Purpose or Learning Objective: To assess and compare DESS and post- contrast STIR sequences in deep-learning(DL)-recons tructed 3D MR neurography of the extraforaminal cranial and spina l nerves.

Methods

or Background: Eighteen consecutive exams of 18 patients with unclear cephalgia undergoing head-and-neck MRI at 1 .5T were retrospectively included (mean age: 51 ± 14 years, 11 female). 3D DESS and post-contrast 3D STIR sequences were reconstructed with a prototype DL algorithm. Two blinded readers qualitatively evaluated visualizati on of the inferior alveolar (IAN), lingual (LN), facial (FN), hypoglossal (HN), greater occipital (GON), lesser occipital (LON) and greater auricular (GAN) nerves, as well as overall image quality, vascular suppression and artifacts. Apparent signal-to-noise ratio (aSNR) and contrast-to-noise ratios (aCNR) we re measured. Qualitative ratings were compared between sequences using Wilco xon signed-rank test, quantitative analysis with paired sample Student’s t-testing.

Results

or Findings: DESS demonstrated significantly improved visualizat ion of the LON and GAN and proximal GON (p < 0.015). Po st-contrast STIR showed significantly enhanced visualization of the LN, HN and distal IAN (p 0.08). With regard to overall image quality and artifacts, there was also no significant difference between sequences. Post-contrast STIR ac hieved superior vascular suppression, reaching statistical signifi-cance for one reader (p = 0.039). Quantitatively, there was no significant difference between sequences (p > 0.05).

Conclusion

Our findings suggest that 3D DESS generally provide s improved visualization of spinal nerves (GON, LON, GAN), whi le post-contrast 3D STIR facilitates enhanced delineation of extraforaminal cranial nerves (IAN, LN, HN). 3D DESS and post-contrast 3D STIR could each add va lue to head-neck MRN protocols, depending on the main clinical area of i nterest.

Limitations

Retrospective study. Limited sample size. Funding for this study: Not applicable. Ethics committee - additional information: This study was approved by the institutional review board. Author Disclosures: Maelene Lohezic: Employee: GE HealthCare Fabio Zecca: Nothing to disclose Falko Ensle: Nothing to disclose Roman Guggenberger: Nothing to disclose Karolina Pawlus: Nothing to disclose Bjarne Jonas Kerber: Nothing to disclose Jonas Kroschke: Nothing to disclose Enhancing Imaging Efficiency in Advanced Diffusion Imaging Using Denoising and Post-Processing Techniques *V. Sedlák*, K. Vambersky, A. Kavková, K. Sichova, D. Netuka, T. Belsan, M. Majovsky; Prague/CZ Purpose or Learning Objective: The objective of this study is to demonstrate how the application of advanced denoising technique s, such as MP-PCA and P2S, along with post-processing methods for enhanci ng angular resolution, can significantly reduce imaging times in advanced diffusion imaging. This approach aims to preserve or improve image quality while reducing the acquisition time burden in clinical and research se ttings.

Methods

or Background: We acquired two sets of advanced diffusion MRI data from patients with glial brain tumors: one usi ng a full-length acquisition and the other using a fast acquisition protocol. Th e full-length dataset was processed directly, while the fast protocol data wa s enhanced with denoising algorithms (e.g. MP-PCA, P2S) and angular super-res olution reconstruction techniques. We evaluated data quality by assessing signal-to-noise ratio (SNR), angular resolution, and diagnostic accuracy for predicting glioma grade and IDH mutation status.

Results

or Findings: Data from 100 patients with glial brain tumors were processed and analyzed. The shortened protocol data , when enhanced by MP- PCA and P2S denoising, provided a significant impro vement in SNR, aligning closely with or even superseding the quality of the full-length acquisition. Angular super-resolution reconstruction techniques further enhanced the angular resolution without extending scan time. The diagnostic accuracy for predicting glioma grade and IDH mutation status was comparable between the full-length and processed shortened datasets, demon strating that these post- processing methods can preserve clinical diagnostic value while reducing scan duration

Conclusion

Denoising techniques and angular resolution enhance ment significantly improve the quality of shortened diff usion MRI acquisitions, maintaining diagnostic accuracy for glioma grading and IDH status while reducing scan time.

Limitations

Results are based on 100 glioma patients and requir e validation in broader populations. The computational demands m ay limit immediate clinical application, and performance may vary with MRI hardware and sequence parameters. Funding for this study: This study was supported by the Grant Agency of Charles University, grant number GAUK 222623 Ethics committee - additional information: Approved by the Ethics committee of the Military. University Hospital Prag ue Author Disclosures: Martin Majovsky: Nothing to disclose Vojtěch Sedlák: Nothing to disclose Anna Kavková: Nothing to disclose David Netuka: Nothing to disclose Kamil Vambersky: Nothing to disclose Kristyna Sichova: Nothing to disclose Tomas Belsan: Nothing to disclose DTI-ALPS Mapping: A Novel Method that Can Comprehen sively Reflect the DTI-ALPS pattern X. Fan, *G. Cheng*, X. Zhang, N. Zhang; Shenzhen/CN Purpose or Learning Objective: Our objectives are to depict the whole white matter DTI-ALPS changes pattern in cognitive impair ment and provide an intuitive and comprehensive method to assess the ac tivity of the glymphatic system.

Methods

or Background: The glymphatic system is increasingly recognized as a critical factor in the pathogenesis of dementi a. We creatively propose a novel method to analyze the whole white matter diff usion tensor image analysis along the perivascular space (DTI-ALPS). W e included 304 Friday Abstract-based Programme 156 participants from the Shenzhen Multimodal Aging Res earch (STAR) Cohort recruited in Peking University Shenzhen Hospital, i ncluding 182 cognitively unimpaired (CU) participants, 93 participants with mild cognitive impairment, and 29 patients with dementia. All participants und erwent 3.0T MRI scans with DTI sequences. We calculated the DTI-ALPS values us ing the conventional

Method

and depicted the whole white matter DTI-ALPS mapping with the novel

Method

we developed. We performed the analysis of v ariance (ANOVA) to find the differences in ALPS values among three groups a nd used post-hoc tests to find intergroup differences in the conventional met hod and DTI-ALPS mapping, respectively.

Results

or Findings: We revealed the whole brain ALPS pattern using the DTI-ALPS mapping. Over 13 out of 30 regions showed significant differences (p < 0.05) in the inter-group analysis, which provi ded additional information beyond the significant differences based on the con ventional ROI-based ALPS.

Conclusion

In conclusion, the DTI-ALPS mapping provides a robu st, intuitive, and comprehensive way to evaluate the changing patt ern of the glymphatic system, overcoming the limitations introduced by co nventional ROI-based DTI- ALPS calculating methods.

Limitations

Future studies should integrate additional statisti cs such as minimum, maximum, and standard deviation into ALPS- mapping analyses for a comprehensive understanding of the DTI-ALPS mapping . Funding for this study: This study is principally supported by the Shenzhen Science and Technology Program (KCXFZ 2021102016340 8012). Ethics committee - additional information: the Ethics Committee of Peking University Hospital Author Disclosures: Xiang Fan: Nothing to disclose Xiqian Zhang: Nothing to disclose Guanxun Cheng: Nothing to disclose Na Zhang: Nothing to disclose HDD-Net: Hippocampus Dual Decoder Network for autom ated segmentation of hippocampus from computed tomograph ic scans *S. J. Ahn*, W. J. Son, J. Y. Lee, H. Lee; Seoul/KR ([email protected]) Purpose or Learning Objective: Changes of brain hippocampal volumes are closely associated with the development of Alzheime r’s disease. In this work, we develop a new deep learning (DL) network model f or volumetric hippocampal segmentation from computed tomography ( CT) head images, a task that has been challenged due to the modality’s limited brain contrast.

Methods

or Background: HDD-Net: The proposed network model is characterized by four major elements – 1) an encode r , 2) two parallel decoders (namely, seg-decoder and edge-decoder), 3) a feature cross module (FCM) fusing features from the two decoders and 4) a cross loss computing differences between outputs. Datasets and preproces sing: 150 pairs of MRI- CT volumetric head images collected at Gangnam Seve rance Hospital were used for model training (N=120) and internal valida tion (N=30), while 47 pairs were selected from Seoul St. Mary’s Hospital databa se for external validation. Ground-truth hippocampal labels were generated from T1-weighted MR images using FreeSurfer, and then were processed by a Gaussian high-pass filter leading to reference edge maps. Each pair of MR-CT images were coregistered using SPM12. Training and Evaluation: The DL model was trained with a cost function combining segmentation, edge, and cross losses. Its performance was evaluated by calculating Dice coeff icient and intersection- over-union (IOU). The performance our model was com pared with that of conventional U-Net model.

Results

or Findings: Dice and IOU of our model is higher than those of U -net for internal validation set (DICE : 0.840 vs. 0.822 , IOU: 0.726 vs. 0.699) and external validation (DICE : 0.784 vs. 0.751, IOU: 0 .650 vs. 0.613)

Conclusion

Results suggest feasibility of DL-based automated h ippocampal segmentation from CT scans and its improved perform ance via edge decoding.

Limitations

The applicability of the model could be enhanced by training it on patients with dementia Funding for this study: None Ethics committee - additional information: Gangnam severance hospital IRB Author Disclosures: Sung Jun Ahn: Nothing to disclose Hyunyeol Lee: Nothing to disclose Ji Young Lee: Nothing to disclose Won Jun Son: Nothing to disclose Initial experience with 60kVp craniocervical CT ang iography: achieving 0.2 millisievert while maintaining diagnostic perfo rmance via artificial intelligence iterative reconstruction Y. Han¹, L. Peng², T. Meng², Q. Sun¹, G. Zhang², *T . Wang*², X. Wang¹; ¹Jinan/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: To describe the initial experience and evaluate the clinical feasibility of 60kVp cranioce rvical CT angiography (CTA) with artificial intelligence iterative reconstructi on (AIIR).

Methods

or Background: Sixty consecutive patients scheduled for craniocervical CTA were prospectively enrolled and underwent a 60kVp (350 ref. mAs) and a followed 120kVp (120 ref. mAs) cran iocervical CTA, in a 5- minute interval, using two separate contrast medium (CM) injections. The 120kVp scans were reconstructed with hybrid iterati ve reconstruction (HIR), while the 60kVp scans were reconstructed with HIR a nd AIIR. Two radiologists diagnosed the stenosis, the intracranial aneurysm ( IA), and the vascular anatomic variant (VAV) in consensus using a 5-point confidence scale (1=definitely absent, 5=definitely present) on a pe r-patient basis, which was used for a receiver operating characteristic analys is, and evaluated the vessel visibility (1=blur, 5=clear). Image noise on the co mmon carotid artery (CCA) was measured. The diagnostic performance and image quality of 60kVp scans were evaluated, using 120kVp scans as the reference standard.

Results

or Findings: The mean effective dose and CM dosage was 0.18 ± 0.04 mSv and 30.58 ± 5.15 ml, respectively, for 60k Vp acquisition, corresponding to an 83.02% and 38.46% reduction as compared to 120kVp acquisition. Under 60kVp acquisition, AIIR outperfo rmed HIR in diagnosing all three abnormal manifestations, showing higher AUC ( stenosis: 0.98 vs 0.57; IA: 0.92 vs 0.62; VAV: 0.92 vs 0.79; all p<0.05). N o significant difference in vessel visibility was found between 60kVp AIIR and reference images (4.32±0.97 vs 4.42±0.72, p=0.412), while AIIR images showed lower image noise (10.14±4.64 HU vs 14.68±5.23 HU; p<0.05).

Conclusion

The 60kVp craniocervical CTA with AIIR has the pote ntial for profound dose reduction without compromising the im age quality and the diagnostic performance.

Limitations

N/A Funding for this study: N/A Ethics committee - additional information: This study was approved by the local IRB Author Disclosures: Qizhong Sun: Nothing to disclose Tiantian Wang: Employee: at United Imaging Healthca re Yicheng Han: Nothing to disclose Guozhi Zhang: Employee: at United Imaging Healthcar e Liying Peng: Nothing to disclose Ximing Wang: Nothing to disclose Ting Meng: Nothing to disclose 12:30-13:30 Research Stage 1 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 1405 Artificial intelligence in musculoskeletal imaging Moderator S. Gitto; Milan/IT Author Disclosures: Salvatore Gitto: Other: Abiogen Pharma, Biolive Alp inion, Bracco Imaging Development and validation of deep learning model f or screening low bone mineral density using chest radiographs: A mul ticentre, multinational study *J. Song*, M. Kim, G. Lee, J. Jeong, S. J. Bae, J-M . Koh, N. Kim; Seoul/KR ([email protected]) Purpose or Learning Objective: This study aimed to develop and validate a deep learning model for screening of patients with low bone mineral density (BMD) using chest radiographs (CXRs).

Methods

or Background: We retrospectively collected CXR data paired with DXA results from patients aged 50 and above from fi ve different resources. Each patient's BMD was classified using a T-score t hreshold of -1.0, with scores of -1.0 or above defined as ‘normal’ and tho se below as ‘low BMD’. Of Friday Abstract-based Programme 157 the 57,589 CXRs from Hospital A, 55,600 were utiliz ed for training, and 1,989 were used for internal validation. For external val idation, 3338, 938, and 295 CXRs were collected from B, C hospitals and D platf orm, respectively, representing diverse patient demographics and clini cal backgrounds. A deep learning model was developed to perform binary clas sification of patients' BMD as either normal or low, based on their CXRs.

Results

or Findings: In the A dataset, the model yielded an AUC of 0.95 and demonstrated sensitivity of 0.97, specificity of 0. 65, and F1 score of 0.90. In the datasets B, C, and D, the model achieved AUCs o f 0.91, 0.89, and 0.82. The model’s sensitivity was 0.87, 0.88, and 0.64; s pecificity was 0.77, 0.71, and 0.85; and F1 score was 0.80, 0.88, and 0.76, re spectively.

Conclusion

The proposed low BMD screening system demonstrated performance exceeding an AUC of 0.8 in all external datasets, highlighting the robustness of the system. Notably, the system showe d promising performance even on the D dataset, which comprised individuals of completely different racial backgrounds. This suggests the potential to promptly identify patients with low BMD from CXRs, the most widely used imagin g modality globally.

Limitations

First, this is a retrospective study. Second, a sub stantial proportion of the dataset comprises a single nation al population. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Ethics approval was obtained from the Ethics Committee of Asan Medical Center (N o. 2019-1226), the Public institutional Bioethics Committee (No.2024-0256-001 ) and the Ethics Committee of Korea VHS Medical Center (No.2022-10-0 03-001). Author Disclosures: Jeongmin Song: Employee: Promedius Inc. Namkug Kim: Shareholder: Promedius Minjee Kim: Employee: Promedius Inc. Jung-Min Koh: Advisory Board: Promedius Sung Jin Bae: Nothing to disclose Gaeun Lee: Employee: Promedius Inc. Jinhoon Jeong: Employee: Promedius Inc. Evaluating the Impact of Artificial Intelligence on Fracture Detection: A Multinational Randomized Crossover Study on diagn ostic thinking efficacy *B. J. Van Der Zwart*¹, H. C. Ruitenbeek¹, M. Boese n², M. W. Brejnebol², G. Gunes³, K-G. A. Hermann⁴, K. Ziegeler⁴, E. Oei¹, J. J. Visser¹; ¹Rotterdam/NL, ²Copenhagen/DK, ³Ortaca/TR, ⁴Berlin/DE ([email protected]) Purpose or Learning Objective: To assess the impact of AI assistance on fracture detection on conventional radiographs by c onducting a multi-country, multicenter randomised crossover study.

Methods

or Background: Radiography data from 1,500 consecutive adult cases with suspected posttraumatic fractures were g athered along with relevant clinical information, with 500 cases from each of three European sites. All cases were read by senior and junior radiologis ts and orthopedic surgeons both without and with AI assistance in two sessions separated by at least four weeks. A reference standard was established by expe rt radiologists with the help of clinical data and follow-up imaging. The me an change in diagnostic accuracy was measured both per case (sensitivity an d specificity) and per fracture (sensitivity).

Results

or Findings: Sensitivity at the case-level increased with the AI assistance for all reader groups with +0.074 for se nior radiologists, +0.181 for junior radiologists, +0.095 for both senior and jun ior orthopedic surgeons. The specificity was negatively impacted with the AI ass istance for the radiologists with changes of -0.010 for senior and -0.027 for ju nior radiologists, respectively. The specificity increased for orthopa edic surgeons with +0.016 and +0.024 for senior and junior surgeons, respecti vely. The changes in sensitivity per fracture with the AI assistance wer e +0.089 and +0.168 for senior and junior radiologists and +0.082 and 0.111 for senior and junior surgeons, respectively.

Conclusion

Our study demonstrates that AI assistance enhances fracture detection on conventional radiographs, yielding imp roved patient-wise sensitivity across participating centers. The chang es in specificity was positive for orthopedic surgeons, but negative for radiologi sts.

Limitations

The study was conducted in a simulated setting, pot entially impacting reader performance. Additionally, the Haw thorne effect may have influenced reader behaviour, as participants were a ware they were being observed. Funding for this study: This project has received funding from the European Union’s Horizon 2020 Research and Innovation Progra mme under grant agreement no. 954221. The results presented in this work reflect only the views of the authors. The Commission is not respons ible for any use that may be made of the information it contains. Ethics committee - additional information: This study was approved by the institutional review board of the Erasmus MC, Rotte rdam (Study ID: MEC- 2021-0430) and the need to obtain informed consent was waived by the institutional review boards of Charité Universitäts medizin–Berlin (no. EA4/079/22) and the Danish Patient Safety Authority . Author Disclosures: Huibert C. Ruitenbeek: Nothing to disclose Edwin Oei: Nothing to disclose Mikael Boesen: Advisory Board: Radiobotics Katharina Ziegeler: Nothing to disclose Gözde Gunes: Employee: Radiobotics Bastiaan Johannes Van Der Zwart: Nothing to disclos e Mathias Willadsen Brejnebol: Nothing to disclose Kay-Geert A. Hermann: Nothing to disclose Jacob Johannes Visser: Nothing to disclose Post implementation validation - False Positives an d Negatives in AI Fracture Detection in Clinical Workflow: A Deep Div e *R. Sivanandan*¹, J. Vardal²; ¹Sandvika/NO, ²Dramme n/NO ([email protected]) Purpose or Learning Objective: This study evaluates post-implementation monitoring of an AI fracture detection algorithm, f ocusing on false positives and negatives, their clinical implications, and mitigat ion strategies. It also assesses the algorithm's impact on workflow and patient outc omes and further possibility of research studies.

Methods

or Background: The algorithm was implemented sequentially across 5 hospitals in Vestre Viken Health Trust, Norway, w ith a follow-up validation conducted at the primary hospital using 1284 cases. AI results were negative (60.8%), positive (37.9%) and doubtful (1.3%). Our new workflow after AI implementation allowed AI-negative patients to be d ischarged, while AI-positive cases were referred to clinicians. Radiographers in itially reviewed AI results and guided patients flow, while radiologists report ing all cases. Orthopedic residents and consultants subsequently examined the se images with AI results.

Results

or Findings: Validation revealed 86% true negatives, 2% false negatives, 7% true positives, 4% false positives, a nd 4% doubtful cases. These

Results

were consistent with preliminary external v alidation. Patients with AI- positive had false positives (5.6%) were primarily attributed to old fractures, skin folds and heterotopic calcifications, with min imal clinical significance. Among the patient sent home with AI-negative result , false negatives (2.3%) mainly included minimal knee effusion in adults, be nign bone lesions, and small avulsion fractures requiring conservative tre atment. Only one patient with an avulsion fracture and drop finger was erroneousl y discharged and recalled for clinical re-evaluation.

Conclusion

The validation results aligned with pre-implementat ion findings, with false positives and negatives having minimal c linical impact. This study highlights the potential for further research that is planned to evaluate the necessity of radiologist reporting for AI-positive cases, given that clinicians already review these images.

Limitations

Few patients’ clinical results were not analyzed du e to restricted access to patient journal. Few patients status afte r imaging were missing. Funding for this study: No funding Ethics committee - additional information: Information was collected as per the Ethics Author Disclosures: Jonas Vardal: Nothing to disclose Ramprabananth Sivanandan: Nothing to disclose Assessing the Generalisability of a Paediatric Wris t Fracture Detection AI Model Using a Novel Dataset *C. Pauling*, O. Arthurs, B. Kanber, S. C. Shelmerd ine, E. Allan, E. Ashworth; London/UK ([email protected]) Purpose or Learning Objective: The purpose of this study was to assess the generalisability of an artificial intelligence (AI) model, trained on open-source data, for the detection of fractures and other abno rmalities in paediatric wrist radiographs using a novel, external, and multi-cent ric dataset.

Methods

or Background: A novel retrospective case dataset was curated from two paediatric trauma centres in London, Engla nd. The dataset comprises 865 images with a mean patient age of 10.4 ± 3.5 [standard deviation] years. Ground truth annotations for the external test data set were established by consensus opinion of at least two paediatric radiol ogists. To imitate real-world prospective data, no pre-processing was applied to the external data and only invalid scans were excluded. A YOLOv7-X model was t rained on GRAZPEDWRI-DX, an open-source paediatric wrist trau ma radiograph dataset. After achieving an optimal performance on the test split of data, the model was used to perform inference on the novel ex ternal data and the performance metrics were compared.

Results

or Findings: The sensitivity of the model for the detection of f ractures was 89.0% on the test split of the open-source data . When evaluating on the novel external data, the sensitivity decreased by 3 2.6%. The reduction in the performance of the model across all detection class es was less severe, with a Friday Abstract-based Programme 158 change to mean Average Position (mAP) of -0.139mAP@ 0.5 (-0.067 mAP@[0.5:0.95]).

Conclusion

The model failed to adequately generalise to an ext ernal dataset evidenced by a notable decline in fracture detectio n sensitivity. It is of critical importance to ensure that AI models intended for us e in a prospective clinical setting are externally validated. Additionally, dat a quality and pre-processing procedures can significantly impact model performan ce.

Limitations

The open-source training dataset contains annotatio ns for additional pathologies which are not included in th e external test dataset. Funding for this study: CP is funded by the Great Ormond Street Hospital Children’s Charity (GOSHCC) (Award Number: VS0618). OJA is funded by an NIHR Career Development Fellowship (NIHR-CDF-2017-1 0-037). SCS is funded by an NIHR Advanced Fellowship Award (NIHR-3 01322). Ethics committee - additional information: Ethical approval was provided by the National Health Service (NHS) Health Research A uthority (HRA) (IRAS ID: 274278, REC reference 22/PR/0334) Author Disclosures: Emily Ashworth: Nothing to disclose Cato Pauling: Nothing to disclose Susan Cheng Shelmerdine: Nothing to disclose Owen Arthurs: Nothing to disclose Baris Kanber: Nothing to disclose Emma Allan: Nothing to disclose AI in radiological imaging of soft-tissue and bone tumours: a systematic review evaluating against CLAIM and FUTURE-AI guide lines D. J. Spaanderman¹, M. Marzetti², X. Wan¹, A. Scars brook², E. Oei¹, D. Grünhagen¹, *S. Klein*¹, M. P. A. Starmans¹; ¹Ro tterdam/NL, ²Leeds/UK ([email protected]) Purpose or Learning Objective: Soft-tissue and bone tumours (STBT) are rare, diagnostically challenging lesions with varia ble clinical behaviours and treatment approaches. This systematic review aims t o provide an overview of Artificial Intelligence (AI) methods using radiolog ical imaging for diagnosis and prognosis of STBT, highlighting challenges in clini cal translation, and evaluating study alignment with the Checklist for A I in Medical Imaging (CLAIM) and the FUTURE-AI international consensus g uidelines for trustworthy and deployable AI to promote clinical t ranslation of AI methods.

Methods

or Background: The systematic review identified literature from several bibliographic databases, covering papers pu blished before 17/07/2024. Original research published in peer-reviewed journa ls, focused on radiology- based AI for diagnosis or prognosis of primary STBT was included. Exclusion criteria were animal, cadaveric, or laboratory stud ies, and non-English papers. Abstracts were screened by two of three independent reviewers to determine eligibility. Included papers were assessed against the two guidelines by one of three independent reviewers. (PROSPERO Registration : CRD42023467970)

Results

or Findings: The search identified 15,015 abstracts, and 325 art icles were included for evaluation. Studies performed mod erately on CLAIM, averaging a score of 28∙9±7∙5 out of 53, but poorly on FUTURE-AI, averaging 5∙1±2∙1 out of 30.

Conclusion

Imaging-AI tools for STBT remain at the proof-of-co ncept stage, indicating significant room for improvement. Future efforts by AI developers should focus on design (define unmet clinical need, intended clinical setting and integration), development (build on previous wo rk, training with data reflecting real-world usage, explainability), evalu ation (addressing biases, evaluating using best practices), and data reproduc ibility and availability. Following these recommendations could improve clini cal translation of AI methods.

Limitations

Limitations include single-reviewer scoring due to the high volumn of literature included and assessment against as-of -yet unpublished, FUTURE- AI guidelines. However, FUTURE-AI were developed by a large group of international medical AI experts. Funding for this study: Hanarth Fonds, ICAI Lab, NIHR, EuCanImage Ethics committee - additional information: Not applicable to the systematic review. Author Disclosures: Stefan Klein: Grant Recipient: Payment to instituti on. Stefan Klein is scientific director of the ICAI lab “Trustworthy AI for MRI”, a public-private research program partially funded by General Electric Health care. Edwin Oei: Grant Recipient: Payment to institution. Edwin Oei co-leads a project embedded in the ICAI lab “Trustworthy AI fo r MRI”, a public-private research program partially funded by General Electr ic Healthcare. Andrew Scarsbrook: Nothing to disclose Douwe Jan Spaanderman: Nothing to disclose Matthew Marzetti: Nothing to disclose Martijn Pieter Anton Starmans: Nothing to disclose Dirk Grünhagen: Nothing to disclose Xinyi Wan: Nothing to disclose Multi-Center External Validation of an Automated Me thod Segmenting and Differentiating Atypical Lipomatous Tumors from Lipomas Using Radiomics and Deep-Learning on MRI D. J. Spaanderman¹, S. Hakkesteegt¹, D. Hanff¹, C. Messiou², L. Nardo³, D. Grünhagen¹, C. Verhoef¹, M. P. A. Starmans¹, *S. Klein*¹; ¹Rotterdam/NL, ²London/UK, ³Sacramento, CA/US ([email protected]) Purpose or Learning Objective: Differentiating between lipomas and atypical lipomatous tumors (ALTs) on imaging is challenging, often requiring biopsies. This study aimed to externally and prospectively va lidate a radiomics model to distinguish between lipomas and ALTs using MRI acro ss three large, multi- center cohorts. Additionally, the model was extende d with automatic and minimally interactive segmentation methods to impro ve clinical applicability.

Methods

or Background: Three cohorts were analyzed: two for external validation (US data from 2008–2018 and UK data from 2011–2017), and one for prospective validation (Netherlands, 2020–2021) . Patient data, including MDM2 amplification status and MRI scans, were colle cted. An automatic segmentation method was developed for T1-weighted M RI scans, with interactive segmentation applied in case of poor qu ality. Radiomics model performance was compared with that of two radiologi sts.

Results

or Findings: The cohorts included 150 (54% ALT), 208 (37% ALT), and 86 patients (28% ALT) from the US, UK, and Neth erlands, respectively. Automatic segmentation succeeded in 78% of cases, w hile 22% required interactive segmentation, with only 3% needing manu al adjustments. External validation yielded AUCs of 0.74 (95% CI: 0.66, 0.82 ) (US) and 0.86 (0.80, 0.92) (UK), and prospective validation achieved an AUC of 0.89 (0.83, 0.96) (Netherlands). The radiomics model performed simila rly to radiologists in all cohorts.

Conclusion

The radiomics model, combined with automated and mi nimally interactive segmentation methods, effectively diffe rentiated between lipomas and ALTs, matching the performance of expert radiol ogists and potentially reducing the need for biopsies.

Limitations

First, the segmentation workflow was performed by a single clinician, hence the potential impact of different users on radiomics performance was not assessed. Second, MDM2 amplific ation status, determined by core needle biopsy or resected specim ens, may include false negatives, affecting the accuracy of the ground tru th. Funding for this study: This research was supported by an unrestricted gran t of Stichting Hanarth Fonds, The Netherlands. MPAS a nd SK acknowledge funding from the research project EuCanImage (Europ ean Union's Horizon 2020 research and innovation programme under grant agreement Nr. 95210). This study was supported by the National Institute for Health Research (NIHR) Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, London, and by Th e Royal Marsden Cancer Charity. The work was also supported by the In Vivo Translational Imaging Shared Resources with funds from NCI P30CA093373. T he views expressed are those of the author(s) and not necessarily thos e of the NIHR or the Department of Health and Social Care. Ethics committee - additional information: The study protocol was approved by the local medical ethics review committee (MEC-2 020-0175), and performed in accordance with national and international legis lation. Informed consent was required and obtained exclusively from participants in the prospective study cohort. For the training and external validation co horts, approval by the local medical ethics review committee and the waiver of i nformed consent were previously reported. Author Disclosures: Stefan Klein: Nothing to disclose Christina Messiou: Nothing to disclose Douwe Jan Spaanderman: Nothing to disclose Martijn Pieter Anton Starmans: Nothing to disclose Lorenzo Nardo: Nothing to disclose Dirk Grünhagen: Nothing to disclose Stefanie Hakkesteegt: Nothing to disclose Cornelis Verhoef: Nothing to disclose David Hanff: Nothing to disclose Radiopsy, quantitative wb-mri adc and fat fraction sequences for discrimination of smoldering multiple myeloma and m ultiple myeloma: a prospective observational study *G. Feliciani*¹, A. Rossi¹, C. Cerchione¹, E. Loi¹, E. Antognoni¹, A. Cattabriga², M. Marchesini¹, D. Barone¹, A. Sarnelli¹; ¹Meldola/ IT, ²Bologna/IT ([email protected]) Purpose or Learning Objective: To distinguish between Multiple Myeloma and High-Risk Smouldering Myeloma at staging using image-based biomarkers obtained from Whole Body-MRI (WB-MRI) Ap parent Diffusion Coefficient (ADC) and Fat Fraction (FF) sequences.

Methods

or Background: From January 2021 to March 2024, we enrolled consecutive myeloma patients at staging into an pro spective trial and divided them into Smouldering Multiple Myeloma (SMM) and Mu ltiple Myeloma (MM). Friday Abstract-based Programme 159 All patients underwent WB-MRI. We use the term "Rad iopsy" to indicate the quantification and modelling of image characteristi cs nearby the biopsy site to predict patient status. A radiologist placed a cyli ndrical VOI nearby the biopsy site and 5 more identical VOIs on distant sites suc h the pelvis bone and on D11 and L5 vertebrae. LASSO was used to select the most predictive features and build logistic regression models, which were th en validated using the test set. ROC curves were used as metrics for models’ pe rformance assessment.

Results

or Findings: The study included 102 patients (46 males, mean age 63 ± 12 [SD]) with 60 diagnosed with MM and 42 with SMM. 144 quantitative features were extracted from the VOI at the biopsy site WB-MRI ADC and FF sequences for each patient. Radiopsy model showed a median AUC of 0.80 (0.75-0.90) in the training phase and a median AUC of 0.70 (0.55-0.80) in the test phase. The best predictive model had an AUC of 0.95 and 0.75 in the training and test phase, respectively. The models u sed to predict patient status at biopsy site were also predictive in distant VOIs .

Conclusion

Conclusions: Radiopsy models can distinguish betwee n MM and SMM with good performance nearby the biopsy site. R adiopsy can be used to predict disease invasion on distant sites where bio psy is not possible or not feasable

Limitations

Single center Funding for this study: None Ethics committee - additional information: protocol name: AccuMRI code: IRST 100.15 Author Disclosures: Domenico Barone: Nothing to disclose Claudio Cerchione: Nothing to disclose Anna Sarnelli: Nothing to disclose Eleonora Antognoni: Nothing to disclose Arrigo Cattabriga: Nothing to disclose Alice Rossi: Nothing to disclose Emiliano Loi: Nothing to disclose Matteo Marchesini: Nothing to disclose Giacomo Feliciani: Nothing to disclose Feasibility of generating sagittal radiographs from coronal images using deep learning in adolescent idiopathic scoliosis *M. E. Pellegrino*¹, T. Bassani¹, A. Cina², F. Galb usera², A. Cazzato¹, D. Albano¹, L. M. Sconfienza¹; ¹Milan/IT, ²Zurich/C H ([email protected]) Purpose or Learning Objective: Minimizing radiation exposure is crucial in clinical monitoring of adolescent idiopathic scolio sis (AIS). Generative adversarial networks (GANs) have gained prominence in medical imaging due to their ability to learn complex patterns and gene rate high-quality synthetic images by transforming one type of image into anoth er. This study explores GANs to generate synthetic sagittal radiographs fro m coronal views in AIS patients.

Methods

or Background: A retrospective dataset of 3,935 AIS patients with mild-to-moderate scoliosis (Cobb angle <45°) was an alyzed. The subjects underwent radiographic spine and pelvis examination using the EOS system, which acquires coronal and sagittal images simultan eously. The dataset was split into training (85%, n=3,356) and validation ( 15%, n=579). A pix2pix-based GAN model was trained to generate sagittal images f rom coronal views, targeting real sagittal views. To evaluate accuracy , 100 subjects from the validation set were randomly selected for manual me asurement of lumbar lordosis (LL), sacral slope (SS), pelvic incidence (PI), and sagittal vertical axis (SVA) by two radiologists in both synthetic and rea l images.

Results

or Findings: Of the 100 synthetic images, 69 were deemed assessable. Intraclass correlation coefficient rang ed from 0.93 to 0.99 for measurements in real images and from 0.83 to 0.88 f or synthetic images. Correlations between parameters in real and synthet ic images (mean values between raters) were 0.52 (LL), 0.17 (SS), 0.18 (PI ), 0.74 (SVA). Errors in parameters showed minimal correlation with Cobb ang le. The mean±SD absolute errors were 7±7° (LL), 9±7° (SS), 9±8° (PI), 1.1±0.8 cm (SVA).

Conclusion

While the model generates sagittal images consisten t with

Reference

images, their quality is not sufficient f or clinical parameter assessment, except for promising results in SVA, wh ich describes lateral plumb line alignment.

Limitations

The quality of sagittal images is insufficient for assessing clinical parameters, except for SVA. Funding for this study: Italian Ministry of Health. Ethics committee - additional information: Not applicable. Author Disclosures: Andrea Cazzato: Nothing to disclose Andrea Cina: Nothing to disclose Luca Maria Sconfienza: Nothing to disclose Fabio Galbusera: Nothing to disclose Maria Elena Pellegrino: Nothing to disclose Tito Bassani: Nothing to disclose Domenico Albano: Nothing to disclose 12:30-13:30 Research Stage 2 Research Presentation Session: Abdominal and Gastrointestinal RPS 1401 Insights into gastroesophageal disease Moderator G. Bagnacci; Siena/IT ([email protected]) Predicting Efficacy of Neoadjuvant Therapy in Esoph ageal squamous cell carcinoma Using Spectral CT Parameters *J. Qu*, W. Li, F. Chu; Zhengzhou/CN ([email protected]) Purpose or Learning Objective: To explore the differences in spectral CT parameters among patients with esophageal squamous cell carcinoma (ESCC), categorized by different tumor regression g rade (TRG) groups.

Methods

or Background: This prospective study analyzed 31 patients with ESCC who underwent neoadjuvant therapy. All patient s received energy- spectrum CT scans within 7 days before surgery, and the following parameters were quantified: 40KeV, 70KeV, and 100KeV CT values in the arterial (AP) and venous phase (VP), iodine concentration (IC) in the normal esophageal wall, lesion and aorta, and Z Effective values. Finally, we assessed the differences in parameters among the different TRG groups (0, 1, 2, 3).

Results

or Findings: In this study, a total of 19 participants were anal yzed. 6 (31.58%) were in the TRG0 group, 3 (15.79%) in the TRG1 group, 6 (31.58%) in the TRG2 group, and 4 (21.05%) in the TRG3 group . Significant differences were observed among the four groups for arterial ph ase K40–70 keV (F=4.86, P=0.015) and K70–100 keV (F=4.78, P=0.016). However , no significant differences were found for NICD, NICratio, venous p hase K40–70 keV and venous phase K70–100 keV (P>0.05).

Conclusion

Arterial phase parameters differ significantly amon g TRG groups, indicating potential in predicting tumor regression , while NICD, NICratio, and venous phase parameters do not.

Limitations

One limitation of this study is the small sample si ze, which may affect the generalizability of the findings. Additi onally, the statistical methods employed were relatively simple, which may not full y capture the complexity of the data. Future research should involve larger sam ple sizes and develop models to further evaluate the significant paramete rs, enhancing the accuracy and reliability of the results. Funding for this study: This study has received funding by the Projects of the General Program of the National Natural Science Fou ndation of China (No.82271979), Henan Province Medical Science and T echnology Tackling Program Joint Construction Project (No. LHGJ2023009 6)，Henan Province Central Plains Talent Program (Nurturing talent Ser ies)（No.20240220） Ethics committee - additional information: The Research Ethics Committee of the Affiliated Cancer Hospital of Zhengzhou Univ ersity (Henan Cancer Hospital) has approved this study (NCT03635619). Co nsent to participate Written informed consent was obtained from all subj ects (patients) in this study. Author Disclosures: Wenshi Li: Nothing to disclose Jinrong Qu: Nothing to disclose Funing Chu: Nothing to disclose Diagnostic accuracy and reliability of CT-based Nod e-RADS for esophageal cancer *J. Leonhardi*, B. Schnarkowski, T. Denecke, H-J. M eyer; Leipzig/DE ([email protected]) Purpose or Learning Objective: Staging in patients with oesophageal cancer is of high importance for treatment decisions. Rega rding correct nodal staging, CT imaging was reported with low accuracy. Our obse rvational retrospective study tried to elucidate the potential diagnostic b enefit of the Node-RADS classification to discriminate benign from malignan t lymph nodes.

Methods

or Background: 126 patients (n= 15 females, 11.9%) with a mean age of 62.1 ± 10.4 years were included. Patients were surgically resected and lymph nodes were analyzed histopathologically. CT s cans were acquired within one month before surgery. N = 182 lymph nodes were independently scored by two radiologists, following the Node-RADS classific ation. This lymph node scoring system ranges from 1 to 5, reflecting proba bility of malignancy (1–very low to 5–very high).

Results

or Findings: 54 patients were nodal positive (42.9 %), whereas 7 2 patients were nodal negative (57.1 %). N1 stage was found in n= 28 cases Friday Abstract-based Programme 160 (22.2 %), N2 in n = 20 cases (15.9 %) and N3 stage in n = 6 cases (4.8 %). The tumors were squamous cell carcinomas (36 cases) , adenocarcinomas (88 cases) and mixed adenoneuroendocrine carcinomas (2 cases). Inter-reader agreement reached k = 0.65 (p<0.001) for the overal l Node-RADS scoring. Malignancy rates for Node-RADS-scores, ranging from score 1 - 5, were as following: 30 %, 14%, 81%, 90.1%, 86.5%. Total Node -RADS score showed statistically significant differences between N0 an d N1-3 stage (N0: 2.68 ± 1.31 versus N1-3: 3.54 ± 1.11, p<0.001). ROC curve analy sis yielded an AUC of 0.69, a threshold of 2 resulted in a sensitivity of 0.77 and a specificity of 0.55.

Conclusion

Node-RADS scores were associated with malignancy of lymph nodes and might help to improve staging in oesophag us carcinomas.

Limitations

For potential clinical translation there is need fo r external validation. Funding for this study: This study did not receive funding of any kind. Ethics committee - additional information: No. of the approval: 106-16- 14032016 Author Disclosures: Timm Denecke: Nothing to disclose Hans-Jonas Meyer: Nothing to disclose Jakob Leonhardi: Nothing to disclose Benedikt Schnarkowski: Nothing to disclose The value of deep learning image reconstruction alg orithm in improving dual-energy CT image quality of gastric cancer *T. Bei*, J. Li; Zhengzhou/CN ([email protected]) Purpose or Learning Objective: To investigate the value of deep learning image reconstruction algorithm (DLIR) in improving image quality of the dual- energy CT scans of gastric cancer by comparing with the adaptive statistical iterative reconstruction Veo (ASiR-V) algorithm.

Methods

or Background: The original DECT images of 80 patients with surgical pathology confirmed gastric cancer between February 2023 and July 2023 were retrospectively collected. The virtual mo no-energy images (VMI) and iodine images at three-phase including arterial phase, venous and delayed phase, were reconstructed using ASiR-V50% and DLIR of three strengths (DLIR‑L/M/H). Objective evaluation indicators for assessi ng image quality included noise (SD), signal-to-noise ratio of gastr ic cancer lesions (SNR lesion) and muscle signal-to-noise ratio (SNR muscle), cont rast-to-noise ratio of lesions (CNR lesion), and normalized iodine concent ration (NIC). Subjective evaluation indicators included image noise and imag e sharpness. One-way analysis of variance and Kruskal-Wallis test were p erformed to compare the differences of image quality among different recons truction images.

Results

or Findings: Compared with ASiR-V50%, DLIR-M and DLIR-H significantly reduced the SD value of the images (P <0.05). The SNR lesion and SNR muscle of DLIR-M and DLIR-H were significantly higher than those of ASiR-V50% (P<0.05). The CNR lesion of DLIR-H were s ignificantly higher than those of ASiR-V50% and DLIR-L (P0.05). Subjective scor ing results showed that the DLIR‑H images displayed the lowest noise, and the highes t image sharpness, significantly higher than ASIR-V50% (P<0 .05).

Conclusion

Compared with ASiR-V50%, DLIR significantly reduced the noise of DECT images of gastric cancer and improved the i mage quality. Among the three deep learning reconstruction algorithms , DLI R-H had the best noise reduction performance, image noise and image sharpn ess.

Limitations

Not applicable. Funding for this study: National Natural Science Foundation of China (82202146) Ethics committee - additional information: This is a retrospective study Author Disclosures: Jing Li: Nothing to disclose Tianxia Bei: Nothing to disclose Photon-Counting CT: Can Virtual Non-Contrast Match True Non-Contrast Imaging? *G. Marras*, F. Pisu, A. Palmisano, A. Esposito; Mi lan/IT ([email protected]) Purpose or Learning Objective: Virtual Non-Contrast (VNC) has emerged as a potential alternative to True Non-Contrast (TNC) imaging, offering dose reduction while maintaining good tissue contrast. P hoton-Counting CT (PCCT) for improved spectral separation should improve mat erial decomposition providing accurate VNC. The aim of the present stud y is to evaluate the diagnostic reliability of VNC compared to TNC in PC CT images.

Methods

or Background: This retrospective study included consecutive patients submitted to Contrast-Enhanced CT examinat ion acquired with PCCT Naeotom Alpha (Iopromide or Iopamidol 370, 60 ml) f rom June to September 2024 for clinical indications with the availability of TNC. VNC were obtained from arterial (VNCa) and venous phases (VNCv) (Qr40 , Q3). Using PACS tool, TNC, VNCa and VNCv were aligned and for the extract ion of Hounsfield Unit, Regions of interest (ROI) were drawn in right liver lobe, spleen, kidney, pancreas, aorta, omental and subcutaneous fat, para vertebral muscle, vertebral cortical and spongious bone. Co-registrat ion and minor adjustments were made to account for breathing variations. Atte nuation values were then compared using Tukey’s HSD or Dunn’s test.

Results

or Findings: A total of 38 studies have been analyzed. Significa nt differences in mean HU values between TNC and both VNC were observed for most tissues (p<.01), except the spleen. Highest at tenuation differences were found for subcutaneous/omental fat and bone cortica l and spongious with HU difference higher than 15HU in more than 50% of cas es for fat and 100% for bone. No significant HU differences were found betw een VNCa and VNCv, except in the spleen.

Conclusion

VNC obtained from PCCT provide different attenuatio n values compared to TNC. Caution should be exercised in rou tine clinical practice.

Limitations

Higher sample size is required to confirm the prese nt data. Funding for this study: None Ethics committee - additional information: Aimomics Author Disclosures: Antonio Esposito: Nothing to disclose Anna Palmisano: Nothing to disclose Gloria Marras: Nothing to disclose Francesco Pisu: Nothing to disclose Sarcopenia effects on surgical outcomes and surviva l after gastrectomy (SESGa study) V. Sbacco, F. Puccetti, D. Palumbo, *G. J. Ortu*, A . Campisi, E. Mazza, F. De Cobelli, U. Elmore, R. Rosati; Milan/IT ([email protected]) Purpose or Learning Objective: Sarcopenia, defined as a progressive and generalized loss of muscle mass, strength, and func tion, is linked to increased morbidity and mortality in several nonmalignant con ditions. Recently, it has emerged as an important factor in oncology, correla ting with higher chemotherapy toxicity. However, its impact on survi val and surgical outcomes in cancer patients remains unclear. This study eval uates the effects of sarcopenia on the response to neoadjuvant chemother apy (NACT), overall survival, and postoperative outcomes in gastric can cer patients.

Methods

or Background: A retrospective analysis was conducted on a prospective database of patients treated between Se ptember 2017 and December 2023. All patients received NACT with the FLOT regimen followed by total or subtotal gastrectomy (open/laparoscopic ). Body composition parameters, including skeletal muscle area (SM), vi sceral (VAT), and subcutaneous adipose tissue (SAT), were measured fr om CT images before and after NACT using a semi-automated segmentation software (Sliceomatic 5.0). Sarcopenia was defined by the skeletal muscle index (SMI) adjusted for gender.

Results

or Findings: The study included 101 patients, predominantly male (60.4%), with distal adenocarcinoma (71.3%). Post-N ACT sarcopenia was present in 65 patients (64.4%), while visceral obes ity was found in 33 (32.7%). NACT-induced sarcopenia had no significant effect o n postoperative outcomes, except for a higher incidence of wound co mplications (p=0.012). Longitudinal changes in body composition showed a c orrelation between sarcopenia and a poorer tumor regression grade (TRG ).

Conclusion

There is a strong association between NACT administ ration and the onset of sarcopenia (p = 0,001) , which, howeve r, had minimal impact on survival and clinical outcomes in our center. This underscores the importance of completing the full neoadjuvant FLOT regimen.

Limitations

Although our goal is to expand it, our cohort of pa tients is currently quite limited. Further exploration of sar copenia in gastric cancer is warranted. Funding for this study: Unfunded study. Ethics committee - additional information: 28/Int/2015 Author Disclosures: Francesco Puccetti: Nothing to disclose Diego Palumbo: Nothing to disclose Gabriele Jacopo Ortu: Nothing to disclose Elena Mazza: Nothing to disclose Valentina Sbacco: Nothing to disclose Antonino Campisi: Nothing to disclose Riccardo Rosati: Nothing to disclose Francesco De Cobelli: Nothing to disclose Ugo Elmore: Nothing to disclose Friday Abstract-based Programme 161 12:30-13:30 Research Stage 3 Research Presentation Session: Cardiac RPS 1403 Valves and pulmonary hypertension: the role of cardiac CT and MRI Moderator R. P. J. Budde; Rotterdam/NL ([email protected]) Author Disclosures: Ricardo P.J. Budde: Advisory Board: Bayer, payment to Erasmus MC; Board Member: ESCR; Research Grant/Support: Siemens, Hear tflow, Bayer, Bracco all payements to Erasmus MC; Speaker: Bayer Mitral valve annulus assessment: a comparison betwe en 3D-TOE, CCT and surgical ring *E. Moliterno*, L. Giarletta, A. Pasquini, M. Masse tti, L. Natale, R. Marano; Rome/IT ([email protected]) Purpose or Learning Objective: To compare the size assessment of the mitral annulus (MA) obtained by preoperative CCT wi th the intraoperative 3D- TOE in patients undergoing surgical mitral valve re pair (MVr) and assess their agreement with the MA measured intraoperatively by the surgeon.

Methods

or Background: 55 patients with severe primary mitral regurgitatio n and candidates to MVr by the Carpentier technique w ith annuloplasty underwent pre-surgery CCT to exclude coronary arter y disease (CAD). We compared CCT-derived MA sizing on short-axis view a ccording to the D-shape MV segmentation model obtaining inter-trigonal dist ance, septal-to-lateral and inter-commissural distances, with those obtained by 3D-reconstruction of intra- operative TOE and the surgical implanted ring sizes , using intraclass correlation.

Results

or Findings: Good agreement resulted between the inter-trigonal distance measured by CCT and the surgical ring (ICC 0.89 [CI 0.330-0.985; p < 0.05]) and between inter-commissural distance obt ained by 3D-TOE and surgical ring (ICC 0.81 [CI 0.458-0.936; p < 0.05]) , while the inter-trigonal distance measured by 3D-TOE showed a moderate agree ment (ICC 0.63 [CI 0.056-0.852; p < 0.05]). Excellent agreement result ed between the inter- trigonal distance assessed by intraoperative 3D-TOE and CCT (ICC 0.95 [CI 0.755-0.989; p<0.05]).

Conclusion

Our study shows the good accuracy of the pre-proced ural CCT- based MA size assessment in comparison with the int raoperative 3D-TOE and the conventional surgical sizing, proposing CCT as a complementary non- invasive imaging technique in predicting eligibilit y and surgical ring sizing in patients candidate to MVr.

Limitations

The major study limitations are the relatively smal l sample size and the monocentric design. Funding for this study: None Ethics committee - additional information: No additional information Author Disclosures: Massimo Massetti: Nothing to disclose Eleonora Moliterno: Nothing to disclose Luigi Natale: Nothing to disclose Annalisa Pasquini: Nothing to disclose Riccardo Marano: Nothing to disclose Lorenzo Giarletta: Nothing to disclose CT-based planning of transcatheter pulmonary valve implantation in patients operated for Tetralogy of Fallot and Doubl e Outlet Right Ventricle *P. Marakhouskay*, P. Chernoglaz, K. Marakhouski, K . Drozdovski; Minsk/BY ([email protected]) Purpose or Learning Objective: The aim of the study was to quantify the

Method

of planning transcatheter pulmonary valve im plantation (TPVI) in patients who have undergone surgery for congenital heart defects(CHD) by measuring in both ECG and non-ECG gated CTA sizes o f pulmonary artery(PA).

Methods

or Background: 22 patients with CHD as Tetralogy of Fallot and Double outlet Right Ventricle(DORV), operated in in fancy, were included into study and divided into two groups: study group, in which direct intravascular balloon sizing of PA was performed, and control gro up in which TPVI was rejected for excessive artery size without performi ng angiography. CTA of all cases were retrospectively analyzed and 4 sizing zo nes were measured: RVOT, supravalvular zone, middle-segment of and bif urcation zone of PA. After that calculations to find appropriate patient -specific sizing methods by comparing parameters with an actual stent size were made (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org;2022) .

Results

or Findings: Analysis showed no significant correlation between diastolic RVOT sizing with any other measuring poin ts or an actual stent size (P=0,0004). Only supravalvular zone was accepted as an independent residual in regression analysis (F-ratio 22, P=0,0004). Afte r implementing a regression equation to calculate appropriate size of PA stent on the control group, all the stents exceeded 30 mm (TPVI exclusion point). Mann- Whitney U-Test showed that the significant difference between an actual s tent size and calculated size for study group (U 91,00,P = 0,5678) and no signifi cant difference between calculated sizes of two groups (U 0,50 P = 0,0004).

Conclusion

Our study showed that any cardiac CT-based preTPVI landing zone sizing in diastolic phase can be based solely on measuring supravalvular zone of PA (r= 0,7992, F-ratio 22, P=0,0004), espec ially as an exclusion criteria (U 0,50 P = 0,0004).

Limitations

None Funding for this study: No funding Ethics committee - additional information: Approved by the Ethics Committee of the NACPS No. 11 22.03.24. Author Disclosures: Kirill Marakhouski: Nothing to disclose Palina Marakhouskay: Nothing to disclose Konstantin Drozdovski: Nothing to disclose Pavel Chernoglaz: Nothing to disclose Lung Perfusion Changes Following Pulmonary Valve Re placement in Repaired Tetralogy of Fallot Patients: A Time-Resol ved MR Angiography Study A. F. F. Tekin, *B. Tütüncüoğlu*, T. Banaz, Ö. Altun, Y. C. Kartal, S. Ozkök; Istanbul/TR ([email protected]) Purpose or Learning Objective: Tetralogy of Fallot (TOF) is a congenital heart disease surgically repaired in early life.Gui delines recommend pulmonary valve replacement (PVR) for these patients. However , the effects of PVR on lung perfusion are not yet fully understood.In our study, we aimed to evaluate the changes in lung perfusion following PVR in pati ents with surgically corrected TOF.

Methods

or Background: The study included 10 patients with surgically corrected TOF (M/F: 6/4, mean age: 16 years, median : 11-19 years).Lung perfusion was assessed using 4D time-resolved magne tic resonance angiography (MRA) before and 1 year after pulmonary valve replacement (PVR) .Gadolinium-based contrast was administered a nd images were analyzed with Philips Intellispace workstation.For measurement, ROI 's were manually placed in the main pulmonary artery before bifurcation, the aortic arch, upper, middle, and lower lobes of the right a nd left lungs.Changes in both ventricular parameters, pulmonary insufficiency rat io, and perfusion were assessed before and after PVR. Statistical analyses were performed using SPSS .Data were reported as medians and the Wilcoxo n-sum rank test was used.

Results

or Findings: Following PVR, a reduction in pulmonary regurgitati on fraction and right ventricular end-diastolic volume was observed, while no changes were seen in right ventricular ejection fra ction and left ventricular volume. After PVR, a difference was noted in right and left lung perfusion but no differences were observed between lung zones.

Conclusion

Lung perfusion assessment is a non-invasive, radiat ion-free imaging technique for the diagnosis and monitoring of various respiratory diseases.Improved pulmonary artery flow volumes, re duced perfusion heterogeneity, and enhanced perfusion dynamics foll owing PVR reflect a positive impact on pulmonary hemodynamics.MRA-perfu sion imaging can be a valuable tool for evaluating perfusion changes in t his patient population.

Limitations

The planning of retrospective studies using a large r number of patient groups is necessary. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was approved by Istanbul Medeniyet University Göztepe Training and Research Hospital Clinical Trials Ethics Commitee Author Disclosures: Serçin Ozkök: Nothing to disclose Berk Tütüncüoğlu: Nothing to disclose Ali Fuat Fuat Tekin: Nothing to disclose Yiğit Can Kartal: Nothing to disclose Tuba Banaz: Nothing to disclose Ömer Altun: Nothing to disclose Friday Abstract-based Programme 162 Role of CT-derived Extracellular Volume Fraction fo r Predicting Response to Percutaneous Aortic Valve Replacement *C. Gnasso*, D. Vignale, A. Palmisano, S. Barbieri, E. Agricola, A. Esposito; Milan/IT ([email protected]) Purpose or Learning Objective: Aortic stenosis (AS) causes myocardial microstructural modifications: the development of i nterstitial fibrosis, in particular, negatively impacts patients’ prognosis. Echocardiography, the gold standard for AS phenotypization, can’t investigate microstructural changes. CT -routinely performed for pre-procedural TAVI planni ng- can be implemented with delayed scans to assess myocardial fibrosis as ECV quantification. The prognostic role of ECV has already been demonstrate d. The study aims to investigate the capacity of ECV to improve risk str atification on top of echocardiographic classification (low-flow low-grad ient AS -LF-LG- vs high- gradient AS -HG-).

Methods

or Background: Prospective, single-center study (IRCCS OSR Milan), enrolling consecutive patients undergoing C T for TAVI planning (Oct2020-Mar2023); clinical, echocardiographic, and laboratoristic data were collected. Patients were categorized as LF-LG or HG . The CT protocol included a low-dose delayed scan (5 mins after iodi xanol 320 administration, 85-110 mL based on patient’s BMI). ECV was calculat ed with the formula: (1- hematocrit)x(ΔHUmyocardium/ΔHUblood), with ΔHU being the HU-difference in pre and post-contrast scans. The composite endpo int (death; heart failure hospitalization) was collected after a 1-year follo w-up.

Results

or Findings: The final cohort consisted of 415 patients (82 year s [78- 85]); 87 reached the endpoint. At Cox multivariable analysis including clinical, echo, and CT data, ECV (using the cut-off of 29% de rived from Youden index) resulted an independent prognosticator of the endpo int, along with sex, hypercholesterolemia, diabetes and transvalvular gr adient, with HR of 1.828 (95%CI1.151-2.903, p=0.0106), significantly higher in the LF-LG group (31.9(27.6-35.6) vs 28 (25.4;31), p<0.001). At Kapl an-Meier analysis, adding ECV to echo classification led to an improvement in risk stratification: HG-AS patients with higher ECV had a worse prognosis even compared to LF-LG-AS with low ECV (log-rank<.0001).

Conclusion

CT-ECV improves risk stratification in AS patients on top of echocardiographic evaluation.

Limitations

Lack of MR as reference standard Funding for this study: None Ethics committee - additional information: All patients signed informed consent. protocol number: CTMyoC 112/INT/2019 Author Disclosures: Davide Vignale: Nothing to disclose Antonio Esposito: Nothing to disclose Anna Palmisano: Nothing to disclose Simone Barbieri: Nothing to disclose Eustachio Agricola: Nothing to disclose Chiara Gnasso: Nothing to disclose Relationship between myocardial strain and extracel lular volume: Exploratory study in patients with severe aortic st enosis undergoing photon-counting detector CT *C. Lisi*¹, V. Mergen², L. J. Moser², K. Klambauer² , H. Alkadhi², M. Eberhard²; ¹Milan/IT, ²Zurich/CH ([email protected]) Purpose or Learning Objective: Diffuse myocardial fibrosis and altered deformation are relevant prognostic factors in aort ic stenosis (AS) patients. The aim of this exploratory study was to investigat e the relationship between myocardial strain, and myocardial extracellular vol ume (ECV) in patients with severe AS with photon-counting detector (PCD)-CT.

Methods

or Background: We retrospectively included 77 patients with severe AS undergoing PCD-CT imaging for transcatheter aort ic valve replacement (TAVR) planning between January 2022 and May 2024 w ith a protocol including a non-contrast cardiac scan, an ECG-gated helical coronary CT angiography (CCTA), and a cardiac late enhancement scan. Myocardial strain was assessed with feature tracking from CCTA and EC V was calculated from spectral cardiac late enhancement scans.

Results

or Findings: Patients with cardiac amyloidosis (n=4) exhibited significantly higher median mid-myocardial ECV (48. 2% versus 25.5%, p=0.048) but no significant differences in strain v alues (p>0.05). Patients with prior myocardial infarction (n=6) had reduced media n global longitudinal strain values (-9.1% versus -21.7%, p0.05). Significant correlatio ns were identified between global longitudinal, circumferential, and radial st rains, and CT-derived left ventricular ejection fraction (EF) (all, p<0.001). Patients with low-flow, low- gradient AS and reduced EF exhibited lower median g lobal longitudinal strain (GLS) values compared with those with high-gradient AS (-15.2% versus -25.8, p<0.001). In these patients, the baso-apical mid-my ocardial ECV gradient correlated with GLS values (R=0.33, p<0.05).

Conclusion

In patients undergoing PCD-CT for TAVR planning, EC V and GLS may enable to detect patients with cardiac amyl oidosis and patients with reduced myocardial contractility after myocardial i nfarction. Patients with low- flow, low-gradient AS and reduced EF showed lower m edian GLS correlating with basal LV fibrosis.

Limitations

Retrospective design, monocentric design, small sam ple number Funding for this study: No funding Ethics committee - additional information: No additional information Author Disclosures: Costanza Lisi: Nothing to disclose Victor Mergen: Nothing to disclose Matthias Eberhard: Nothing to disclose Lukas Jakob Moser: Nothing to disclose Konstantin Klambauer: Nothing to disclose Hatem Alkadhi: Nothing to disclose Cardiac magnetic resonance 4d flow in mitral annulo plasty: impact on left ventricular flow dynamics and functional correlatio ns with different types of devices *G. C. Pambianchi*, G. Cundari, L. Marchitelli, C. Catalano, N. Galea; Rome/IT ([email protected]) Purpose or Learning Objective: To evaluate functional and fluidodynamic modifications after MAP with different types of ann uloplasty prosthetics using CMR 4D-flow techniques and left atrial (LA) strain feature-tracking

Methods

or Background: We enrolled 12 patients treated with MAP (7 semi- flexible and 5 flexible incomplete rings), and had them undergo CMR at least 12 months after the procedure. The protocol include d cineMR and 4D-Flow sequences with a whole heart coverage. We quantifie d TotalFlowVolume (TFV), PeakVelocity (Vmax), and WallShearStress (WS S) at LV inflow. We analyzed LA volumes and strain. MAP were compared w ith 30 age- and sex- matched healthy controls (10 studied with 4D-flow)

Results

or Findings: MAP patients had lower EF (49.3±4.2% vs62.34±6.3%; p<0.021), reduced TFV (52.3±7.8ml vs69.6±9.7ml; p=0.039), and increased Vmax (159.7±21.3cm/s vs125.1±35.3 cm/s; p=0.002) compared to controls; the WSS resulted comparable between the groups (0.23±0. 11 vs0.21±0.44; p=0.354). As for the flow patterns analysis, intrav entricular and beneath the mitral valve flow, there was similar vortex formati on among the two groups. MAP patients had decreased Reservoir (20.6±20.1% vs 22.9±2.5%; p=0.033), lower Conduit (9.1±3.48% vs12.7±1.8%; p = 0.005), and increased Booster Pump strain (12.4±1.8% vs8.9±2.3%; p=0.001) compared to controls. The time between the surgery and CMR was inversely correlate d with TVF (r: -0.95; p=0.04) but didn't affect the WSS.

Conclusion

Mitral annuloplasty with leaflets preservation did not considerably alter intraventricular flow patterns compared to he althy controls; while the prosthetic ring causes slight stenosis, WSS isn't s ignificantly increased. Atrial function was preserved in MAP but still reduced if compared to healthy controls. Detailed evaluation of hemodynamic change s post-mitral repair and potential impact on the choice of annuloplasty devi ces and techniques to optimize long-term outcomes.

Limitations

The study had a numerically limited sample, patient s didn't have pre-annuloplasty CMR,. We didn't analyze TKE and vi scous energy loss. Funding for this study: Not applicable Ethics committee - additional information: The study was submitted to local ethics committee Author Disclosures: Giacomo Carlo Pambianchi: Nothing to disclose Giulia Cundari: Nothing to disclose Livia Marchitelli: Nothing to disclose Nicola Galea: Nothing to disclose Carlo Catalano: Nothing to disclose The non-invasive right heart catheter: Hemodynamic classification of pulmonary hypertension using 4D flow MRI *G. Reiter*, G. Kovacs, C. Reiter, H. Olschewski, M . Fuchsjäger, U. Reiter; Graz/AT ([email protected]) Purpose or Learning Objective: Mean pulmonary arterial pressure (mPAP), pulmonary arterial wedge pressure (PAWP), and pulmo nary vascular resistance (PVR) are the hemodynamic parameters mea sured during right heart catheterization (RHC) for diagnosis and class ification of pulmonary hypertension (PH). This study aimed to assess the a ccuracy of 4D flow MRI in predicting these parameters non-invasively.

Methods

or Background: 103 patients with known or suspected PH (PH/non- PH, 77/26) prospectively underwent both RHC and who le-heart 4D flow MRI at 3T. From 4D flow data, the duration of vortical blo od flow along the main pulmonary artery (t_vortex), the left atrial accele ration factor (acc) and the cardiac output (CO_MR) were determined to derive es timates for mPAP via mPAP_MR = 16 + 0.63·t_vortex, PAWP via PAWP_MR = −6.2 + 10.1·a c c , a n d Friday Abstract-based Programme 163 PVR via PVR_MR = (mPAP_MR – PAWP_MR)/CO_MR. Relatio nships between invasive and 4D flow MRI-derived parameters were analyzed by correlation analysis and t-test. The performance of 4D flow MRI-derived parameters to predict the hemodynamic classificatio n of PH was assessed by receiver operating characteristic curve analysis.

Results

or Findings: The area under the curve (AUC) for predicting PH (mPAP > 20 mmHg) using mPAP_MR was 0.96. In PH pati ents, mPAP and mPAP_MR correlated strongly (r = 0.94) and showed n o significant bias (0.6 ± 4.5 mmHg, p=0.24). Within PH patients, the AUCs for predicting post-capillary PH (PAWP > 15 mmHg) from PAWP_MR and PVR > 2 WU fro m PVR_MR were 1.00 and 0.95, respectively. PAWP and PAWP_MR correlated strongly (r = 0.94) and demonstrated no significant bias (0.4 ± 1.6 mmHg, p=0.05). PVR and PVR_MR correlated strongly (r = 0.85) but demon strated a significant bias (0.8 ± 2.2 WU, p<0.01).

Conclusion

4D flow MRI allows accurate non-invasive diagnosis and hemodynamic classification of PH.

Limitations

Single-center study Funding for this study: None Ethics committee - additional information: Medical University of Graz, Austria Author Disclosures: Horst Olschewski: Nothing to disclose Ursula Reiter: Nothing to disclose Gert Reiter: Employee: Research and Development Sie mens Healthineers Michael Fuchsjäger: Nothing to disclose Clemens Reiter: Nothing to disclose Gabor Kovacs: Nothing to disclose Feasibility of pulmonary arterial pulse wave veloci ty assessment from 4D flow MRI *C. Reiter*, G. Reiter, G. Kovacs, D. Scherr, A. Sc hmidt, H. Olschewski, M. Fuchsjäger, U. Reiter; Graz/AT ([email protected]) Purpose or Learning Objective: To assess the feasibility of measuring pulmonary arterial pulse wave velocity (PWV) – a po tential prognostic marker in pulmonary hypertension (PH) – from 4D flow MRI.

Methods

or Background: Thirty-one healthy subjects (15 females; age, 60 ± 10 years) and 10 patients with PH (6 females; age, 66 ± 11 years; mean pulmonary arterial pressure, 46 ± 11 mmHg) were pro spectively recruited for 4D flow MRI at 3T. PWV was calculated using the tra nsit-time approach (cvi42). Centerline segmentation of the pulmonary a rtery was performed twice, from the main pulmonary artery once to the left and once to the right pulmonary artery; PWV was measured in the main pulm onary artery as well as the entire segmented vessels. The pulmonary artery cross-section area was obtained from a multiplanar-reconstructed plane thr ough the center of the main pulmonary artery. Results were analyzed using t-tes ts and correlation analysis.

Results

or Findings: In healthy subjects, the main pulmonary arterial PW V was 2.4 ± 0.2 m/s with no significant difference between the main-to-right (2.7 ± 0.3 m/s) and main-to-left pulmonary PWV (2.7 ± 0.3 m/s, p=0.583). Moreover, no significant sex differences were obser ved (p=0.430). In PH subjects the PWV was higher than in controls in the main (6.1 ± 1.8 m/s, p<0.001), the main-to-right (7.7 ± 3.0 m/s, p<0.001) and the main-to-left pulmonary artery (7.1 ± 2.5 m/s, p<0.001). Main, main-to-right and main-to-left pulmonary artery PWV correlated significantly with the average pulmonary artery cross-section area (r = 0.78, 0.61 and 0.59, respectively; p<0.001 in all cases).

Conclusion

Pulmonary arterial PWV assessment from 4D flow MRI is feasible. Normal ranges align with values reported from 2D flow measurements, show small variations in healthy subj ects, and differ compared PWV in patients with PH.

Limitations

Small sample size. Funding for this study: OeNB Anniversary Fund 17934 Ethics committee - additional information: Approval was obtained from the Ethics committee of the Medical University of Graz, Austria. Author Disclosures: Horst Olschewski: Nothing to disclose Ursula Reiter: Nothing to disclose Gert Reiter: Employee: Research & Developement Siem ens Healthineers Michael Fuchsjäger: Nothing to disclose Daniel Scherr: Nothing to disclose Clemens Reiter: Nothing to disclose Albrecht Schmidt: Nothing to disclose Gabor Kovacs: Nothing to disclose 12:30-13:30 Research Stage 4 Research Presentation Session: Neuro RPS 1411 Charting the brain's next frontier: glymphatic system imaging Moderator E. T. Tali; Ankara/TR ([email protected]) Association of Sleep Quality and Memory Performance With Glymphatic Function in Medical Interns and Residents *H. Wu*; Chongqing/CN ([email protected]) Purpose or Learning Objective: Poor sleep quality is prevalent among medical interns and residents, often characterized by insufficient sleep .Our

Objective

is to examine the associations between gl ymphatic function, sleep quality, and neuropsychological performance within this population.

Methods

or Background: 30 medical Interns and 15 residents on 2-week Internal Medicine and Radiology rotations were incl uded . Sleep profile was accessed using questionnaires and polysomnography. Wechsler Memory Scale (WMS) was used for evaluation of the memory p erformance. The MRI for each subject was scheduled at least 72 hours after the completion of the night shift. Diffusion tensor imaging analysis along the perivascular space (DTI- ALPS) index was used to evaluate glymphatic functio n.

Results

or Findings: Multivariate linear regression model for interns gr oup determined that intern year (unstandardized β=-0.0045 [SE 0.0001]; p < 0.001), and the N3 sleep duration, index (unstandar dized β=0.0071 [SE=0.0004]; p = 0.0013) were independently associa ted with DTI-ALPS. Increased DTI-ALPS was linked to enhanced memory sc ores (unstandardized β = −0.77 [SE = 0.23]; p = 0.014). Interns group had higher percentages of N1 and N2 sleep and a modestly higher amount of REM sl eep than residents group. The higher percentages of N1, N2, and REM sl eep were offset by a lower percentage of N3 sleep in interns. There was no significant difference of memory scales between these two groups.

Conclusion

Interns tend to have poorer sleep structure compare d to residents, likely due to their more frequent extend ed overnight shifts, whereas residents rarely or never work such extended shifts . This study identified associations between DTI-ALPS and memory performanc e, suggesting the potential of DTI-ALPS as a biomarker for evaluating the neuropsychological status of young doctors.

Limitations

Our study does not include longitudinal data on tem poral changes in DTI-ALPS between night shifts for individual par ticipants. Funding for this study: None Ethics committee - additional information: The authors report no disclosures relevant to the Abstract. Author Disclosures: Hao Wu: Nothing to disclose Small vessel disease and glymphatic changes in hosp italized and non- hospitalized COVID-19 patients: A study on peak ske letonised mean diffusivity and DTI-ALPS B. Genç, *A. Özçağlayan*, M. S. Buruk, L. Incesu, K. Aslan; Samsun/TR ([email protected]) Purpose or Learning Objective: COVID-19 has also been associated with the development of dementia, cortical atrophy, and cogn itive impairments such as brain fog, which are challenging or often impossibl e to detect using conventional MRI. Peak width of skeletonized mean d iffusivity (PSMD), a recently developed quantitative marker, has been pr oposed as a sensitive biomarker for small vessel disease. The aim of this study is to investigate changes in PSMD, associated with small vessel disea se, and DTI-ALPS parameters, related to the glymphatic system, in CO VID-19 patients.

Methods

or Background: Clinical, demographic data, and MRI images were obtained from the "neuroCOVID MRI dWIand fMRI with reversal learning" dataset available on OpenNeuro (https://openneuro.o rg/datasets/ds005364/ versions/1.0.0 . After denoising and eddy current c orrections, DTI-ALPS and PSMD index measurements were performed similarly to the previous literature. The values of the hospitalized and non-hospitalized COVID groups were compared separately with the control group. Friday Abstract-based Programme 164

Results

or Findings: The PSMD index was significantly higher in the hospitalized COVID-positive group (326 x10 ⁻⁶ mm²/s) compared to the control group (298 x10⁻⁶ mm²/s, p=0.028), while there was no significant di fference between the non-hospitalized COVID-positive group ( 301 x10⁻⁶ mm²/s) and the control group (p=0.953). no significant differences were observed in left or right DTI-ALPS values between the hospitalized, non-hospi talized COVID-positive groups, and the control group.

Conclusion

Our study shows that hospitalized COVID-19 patients exhibit an increase in PSMD index, indicative of small vessel disease, while no such risk exists in non-hospitalized COVID-19 patients. Addit ionally, contrary to initial concerns, COVID-19 does not appear to cause glympha tic dysfunction.

Limitations

The use of the DTI-ALPS method to measure glymphati c function remains controversial. Perivascular spaces account for only 1% of cerebral tissue, making it difficult for this method to diff erentiate diffusion within perivascular spaces from diffusion along other axes . Funding for this study: No funding was received for this study. Ethics committee - additional information: We did not apply for ethics approval as the data utilized in this study were ob tained from the OpenNeuro database. Author Disclosures: Mehmet Seyfi Buruk: Nothing to disclose Bariş Genç: Nothing to disclose Lütfi Incesu: Nothing to disclose Kerim Aslan: Nothing to disclose Ali Özçağlayan: Nothing to disclose Changes in cerebrospinal and vitreous fluid density after iodine contrast administration in acute ischemic stroke patients: p otential role of the glymphatic system? V. Khasminsky, J. Naftali, *G. Danieli*, E. Uriel; Tel Aviv/IL ([email protected]) Purpose or Learning Objective: The aim of our study was to measure the changes in density of the ventricular CSF and vitre ous fluid of the eyes in AIS patients who underwent emergent diagnostic CT work- up and endovascular thrombectomy. Our assumption was that iodine contra st medium leaked into infarcted brain parenchyma would be cleared by the glymphatic system into CSF and would be detected there by CT.

Methods

or Background: A cohort of 119 subjects with the diagnosis of AIS who underwent head NCCT, CT-angiography, CT-perfusi on, endovascular thrombectomy, and another follow-up head NCCT in up to 36 hours was selected. The density of CSF in Hounsfield units (H U) was measured in the frontal horns of lateral ventricles, in the third v entricle and in the vitreous body of both eyes on the admission head NCCT and compare d to the same measurements on the follow-up head NCCT.

Results

or Findings: Small but statistically significant increases in HU were observed across various locations on the follow-up CT. The average HU in the frontal and third ventricles combined rose from 3.8 9±1.49 to 5.01±2.24, a difference of 1.12±2.14 (P < 0.001). The average HU in both eyeballs rose from 5.93±2.75 to 7.14±4.18, a difference of 1.2±4.57 (P = 0.003). In the subgroup with ASPECT score below 8 we found a diffe rence of 1.68±2.14 (P < 0.001) in the ventricles and 3.16±5.29 (P < 0.011) in the vitreous.

Conclusion

Our findings can be explained in the framework of g lymphatic system theory by the clearence of leaked iodine via the bulk flow of CSF in the perivenular space to the lateral ventricles. The de nsity increase in the vitreous fluid may be due to secondarily increased permeabil ity of blood-retinal barrier or connections between intra-cranial and intra-ocul ar glymphatic systems.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Uid: Rmc-0825-23 Author Disclosures: Guy Danieli: Nothing to disclose Vadim Khasminsky: Nothing to disclose Jonathan Naftali: Nothing to disclose Eitan Uriel: Nothing to disclose Diffusion tensor imaging along the perivascular spa ce (DTI-ALPS) for the glymphatic system evaluation in idiopathic and GBA- associated Parkinson’s disease L. De Carolis, *R. Pascuzzo*, D. Aquino, S. A. Dell a Seta, R. Eleopra, F. Doniselli, R. Cilia; Milan/IT ([email protected]) Purpose or Learning Objective: Glymphatic system dysfunction has been associated with neurodegeneration in Parkinson's di sease (PD). However, its role in PD patients carrying GBA mutations (GBA-PD) , a key genetic risk factor for PD, remains unclear. This study aimed to compar e glymphatic function in idiopathic PD (iPD) and GBA-PD using Diffusion Tens or Imaging Along the Perivascular Spaces (DTI-ALPS) and investigate corr elations with motor and non-motor symptoms.

Methods

or Background: Brain DTI sequences (32 directions, b=1000 s/mm2) acquired with a 3T MRI scanner were retrospe ctively collected from 40 PD (20 iPD, 20 GBA-PD) and 40 age- and sex-matched patients with essential tremor (ET, serving as controls). DTI-ALPS indeces were computed for both hemispheres using Taoka’s method. Severity of motor and non-motor symptoms of PD patients were assessed by Hoehn and Yahr (H&Y) and Unified Parkinson’s Disease Rating scales (UPDRS). DTI-ALPS values were compared between groups of patients using Mann-Whit ney test and were correlated with clinical variables using Spearman’s (rho) correlation coefficient.

Results

or Findings: DTI-ALPS values were similar between iPD and GBA- PD patients. However, DTI-ALPS values in the hemisp here contralateral to the less affected side (DTI-ALPS-ctrl) were significant ly higher (p<0.001) in PD patients with H&Y stage <2 (1.42±0.07) compared to others (PD with H&Y stage ≥2: 1.26±0.18; ET: 1.27±0.19). In GBA-PD patients, DTI-ALPS-ctrl values were positively correlated with visual memor y (rho=0.726, p<0.001) but negatively correlated with sleep disturbances (rho= -0.576, p=0.008), while in iPD they were negatively correlated with age at ons et (rho=-0.591, p=0.006), worse response to dopaminergic medication (rho=-0.5 88, p=0.007), and total UPDRS score (rho=-0.591, p=0.006).

Conclusion

Glymphatic dysfunction is linked to disease severit y in PD and may play distinct roles in the pathology of iPD and GBA-PD.

Limitations

The study limitations are the relatively small samp le size and the absence of healthy controls. Funding for this study: Funding was provided by the Italian Ministry of Hea lth Ethics committee - additional information: The study is retrospective Author Disclosures: Sara Adriana Della Seta: Nothing to disclose Domenico Aquino: Nothing to disclose Fabio Doniselli: Nothing to disclose Roberto Eleopra: Nothing to disclose Roberto Cilia: Nothing to disclose Lanfranco De Carolis: Nothing to disclose Riccardo Pascuzzo: Nothing to disclose Diffusion-Tensor MRI Study of the relationship betw een glymphatic system asymmetry and onset lateralization in Parkin son's disease *Z. Li*; Nanjing/CN ([email protected]) Purpose or Learning Objective: Parkinson's disease (PD) is commonly characterized by asymmetric motor symptoms. Previou s studies have implicated abnormal α-synuclein aggregation as a key factor in PD pathogenesis and suggested that glymphatic system d ysfunction may be responsible for the disease progression. However, t he relationship between glymphatic system asymmetry and the side of motor s ymptom onset in PD remains unclear.

Methods

or Background: 27 left-onset PD (LPD) patients, 36 right-onset PD (RPD) patients, and 49 age- and sex-matched healthy controls (HCs) were included in this study. The bilateral hemispheric A LPS indices were used to evaluate glymphatic function. Asymmetry of the glym phatic system was assessed by the asymmetry index (AI). Partial corre lation analysis was conducted to examine the relationship between impai red glymphatic system function and motor deficits.

Results

or Findings: Compared to HCs, RPD patients exhibited a significa nt reduction in the left ALPS index, while both left a nd right ALPS indices were significantly reduced in LPD patients. In both LPD patients and HCs, the right ALPS index was lower than the left, suggesting a na tural leftward asymmetry. However, this asymmetry was diminished in RPD patie nts, as indicated by a reduced AI. Moreover, in RPD patients, the Unified Parkinson's Disease Rating Scale Part III score showed a negative correlation with the left ALPS index, and with AI.

Conclusion

This study demonstrated PD patients with different onset sides have different patterns of glymphatic system functi on. The glymphatic system asymmetry may provide new insights into the mechani sm underlying the lateralized onset of PD.

Limitations

Firstly, being a retrospective cohort study conduct ed at a single center, it involved a relatively small sample size. Secondly, further research is needed to investigate the potential therapeutic imp lications of targeting the glymphatic system in PD Funding for this study: National Natural Science Foundation of China (81671258) Nanjing Medical University-Qilu Clinical Research Fund Project (2024KF0254) Ethics committee - additional information: This study was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University, and the requirement for written informed consent wa s achieved. IRB number: No. 2014-SRFA-097. Author Disclosures: Zihan Li: Nothing to disclose Friday Abstract-based Programme 165 Glymphatic System Evaluation in Essential Tremor an d Parkinson’s Disease: an MRI Study Using Diffusion Tensor Imagin g Analysis Along Peri-vascular Spaces (DTI – ALPS) *A. Innocenzi*, M. Cella, G. Saltarelli, P. Badini, F. Pistoia, A. Catalucci, F. Bruno, E. Di Cesare, A. Splendiani; L'Aquila/IT Purpose or Learning Objective: This study aims to assess the presence of glymphatic system alterations in patients with Esse ntial tremor (ET) and Parkinson’s disease (PD) through the analysis of di ffusion tensors along perivascular spaces (DTI-ALPS index).

Methods

or Background: We retrospectively evaluated 35 patients (19 PD, 16 TE, 28 males, mean age 67 years) who were eligib le for Vim thalamotomy using MRgFUS (2018-2022) and 17 healthy controls (1 3 female, mean age 39 years). All MR images were obtained using a 3.0 Tes la MRI and a 32-channel head coil. DTI images were analyzed using open-sour ce software. The ALPS indices were compared between the ET and PD patient populations, respectively. Both these populations were compared to healthy controls.

Results

or Findings: PD and ET groups were matched for disease duration, age, gender and cognitive score. The ALPS index in PD patients showed a mean of 1.33, with no statistically significant differences compared to ET patients (1.31, p-value = 0.31), both reduced compa red to healthy controls (1.62; p-value < 0.001). We observed a more intense tremor in ET patients rather than in PD patients (FTM 34,14 VS 28,64; p-v alue = 0.041). There was a significant correlation between the ALPS index and tremor intensity in ET patients (R = -0.76; p-value < 0.001). We did not f ind statistically significant correlations between ALPS –index and MoCA score in PD and TE groups.

Conclusion

There were no differences observed in the MRI indic es of the glymphatic system between ET and PD patients, and b oth are reduced with respect to healthy subjects. Our results suggest th e need for further studies to better define the role of the ALPS index as a marke r of disease progression and to evaluate the possibility of neuro-degenerati ve pathophysiology in ET.

Limitations

Small sample size. Funding for this study: This research received no external funding. Ethics committee - additional information: Our study received approval from the Internal Review Board of the University of L'Aq uila (protocol code 21 January 2020 n. 01/2020), and all participating pat ients provided signed informed consent to take part in the study. Author Disclosures: Alessandra Splendiani: Nothing to disclose Gaspare Saltarelli: Nothing to disclose Antonio Innocenzi: Nothing to disclose Ernesto Di Cesare: Nothing to disclose Francesca Pistoia: Nothing to disclose Marco Cella: Nothing to disclose Alessia Catalucci: Nothing to disclose Federico Bruno: Nothing to disclose Pierfrancesco Badini: Nothing to disclose Glymphatic system evaluation in idiopathic intracra nial hypertension- can ALPS index solve all issues of the disease? *R. Dahiya*, S. Rohilla; Rohtak/IN Purpose or Learning Objective: idiopathic intracranial hypertension (IIH) is a common headache disorder in young and middle age fe males. A few theories postulate impaired cerebral glymphatic clearance in IIH, however there is a paucity of methods to quantify glymphatic activity in human brains. The purpose of this study was to use diffusion-tensor i maging to evaluate the glymphatic clearance of IIH patients and how it may relate to various clinical parameters.

Methods

or Background: This observational cross-sectional study was conducted on 25 patients clinically diagnosed with IIH. DTI was used to separately evaluate the diffusivity in lateral asso ciation and projection fibers, with the degree of diffusivity used as a surrogate for glymphatic function (diffusion tensor image analysis along the perivasc ular space). Glymphatic clearance was correlated with several clinical metr ics, including lumbar puncture opening pressure and Frisen papilledema gr ade and combined conduit score for transverse sinus stenosis.

Results

or Findings: At an ALPS Index cut off of 1.64, ROC curve analysi s showed sensitivity of 100%, specificity of 84.6%, P PV of 83.3%, NPV of 100%, DA of 91.3% and PLR of 6.5 with AUC of 0.973 demons trating excellent diagnostic performance in differentiating severe fr om mild IIH. Also ALPS index showed significant negative correlation with BMI(P= 0.012), CSF pressure(p=0.032), papilledema(p=0.001) indicating increasing ALPS index(i.e.worsened glymphatic clearance) with incre asing severity of disease. CSF pressure and papilledema showed positive correl ation(p=0.008).

Conclusion

Our results show that patients with IIH possess imp aired glymphatic clearance, which is directly related to the extent of clinical severity and ALPS Index can be used for clinical diagnosis a s a non invasive investigation.

Limitations

Since we did not record response to treatment on mo st recent follow up and its correlation with ALPS, we could a ssess whether ALPS can be used to guide treatment for IIH Funding for this study: Not provided Ethics committee - additional information: The ethics committee notification can be found under the number BREC/22/231. Author Disclosures: Seema Rohilla: Nothing to disclose Ravi Dahiya: Nothing to disclose Evaluation of the glymphatic system in patients wit h intracranial tumours with blood-brain barrier disruption *L-P. Schmill*, S. Seehafer, S. Aludin, S. Peters, O. Jansen, N. Larsen; Kiel/DE ([email protected]) Purpose or Learning Objective: After decades of doubt, it has finally been proven that the human brain has its own glial-lymph atic system. Its role in various diseases has been studied and discussed. Al though its exact function is still unclear, perivascular spaces and fluid tra nsport by glial cells are thought to be the main components, and drainage appears to occur via the perivenous spaces and meningeal lymphatic vessels. These have been delineated by MRI for the first time in recent years, but there is st ill a lack of established methods to assess their functionality. The aim of this stud y was to investigate whether intracranial tumours have increased drainage via ad jacent meningeal lymphatic vessels.

Methods

or Background: MRI scans were performed on patients with glioblastomas and cerebral metastases of different primary tumours. The protocol included a fluid-attenuated inversion reco very sequence and a contrast-enhanced T1-weighted black blood sequence, which can be used to delineate meningeal lymphatic vessels. This study l ooked specifically at the contrast enhancement of these vessels and compared the signal intensities of lymphatics near the intracranial mass with those co ntralateral to it.

Results

or Findings: Results from the first 5 out of 30 patients show increased visual contrast enhancement in the mening eal lymphatics adjacent to the tumour compared to the contralateral side. T hese results indicate an increased lymphatic drainage of intracranial masses in addition to their venous drainage.

Conclusion

The current aim is to gain new insights into the gl ymphatic system and its relationship with the pathogenesis o f brain diseases. This study has already provided evidence for increased glympha tic drainage in intracranial tumours. However, the exact drainage a reas and which sequences that can be used for optimal detection of lymphatic vessels are still unclear.

Limitations

Not applicable Funding for this study: No funding was provided for this study. Ethics committee - additional information: Ethic approval was obtained from the IRB of the Medical Faculty of the Christian-Alb rechts-University Kiel. Author Disclosures: Sönke Peters: Nothing to disclose Lars-Patrick Schmill: Nothing to disclose Schekeb Aludin: Nothing to disclose Svea Seehafer: Nothing to disclose Olav Jansen: Nothing to disclose Naomi Larsen: Nothing to disclose Friday Abstract-based Programme 166 14:00-15:30 Research Stage 1 Research Presentation Session: Emergency Imaging RPS 1517 Novel approaches in emergency imaging strategies Moderator A. Ugarte; San Sebastian/ES ([email protected]) Risk factors and prognostic significance of inferio r vena cava volume defined by initial polytrauma CT-imaging *H-J. Meyer*, V. Sotikova, T. Denecke, M. Struck; L eipzig/DE ([email protected]) Purpose or Learning Objective: The role of the inferior vena cava (IVC) volume measurement in trauma patients is not yet fu lly understood. The aim of the present study was to identify associations betw een the IVC volume and red blood cell (RBC) transfusion and massive transfusio n (≥10 RBC) within 24 hours after admission, as well as 24-hour and 30-da y mortality in trauma patients.

Methods

or Background: A retrospective analysis was conducted on all consecutive trauma patients who required emergency tracheal intubation and mechanical ventilation before initial whole-body CT imaging at a level-1 trauma center over a 12-year period. The IVC volume was de termined in the initial trauma CT scan.

Results

or Findings: A total of 438 patients (75.3% male) with a median age of 50 years, and a median injury severity score (IS S) of 26 points were included in the analysis. Median IVC volume was 36. 25 cm3, and RBC transfusion and massive transfusion were performed in 197 and 90 patients, respectively. The 24-hour and 30-day mortality rate s were 7.3% and 23.3%, respectively. IVC volume was found to be independen tly associated with the necessity of RBC transfusion and 24-hour mortality (HR 0.98, 95% CI 0.96– 0.99, p =0.01 and HR 0.96, 95% CI 0.93–0.99, p =0.0 25, respectively), while associations with massive transfusion and 30-day mo rtality were not statistically significant in multivariable analyses .

Conclusion

The initial IVC volume may serve as a predictor of the general need for RBC transfusion, although it does not reac h the prognostic threshold for massive transfusion. The association with 24-ho ur mortality rather than 30- day mortality suggests the possibility of its diagn ostic efficacy in short-term outcomes.

Limitations

First, it is a single center retrospective study. S econd, there might be bias induced by infusion therapy before the CT. Funding for this study: None Ethics committee - additional information: Ethics committee at the Medical Faculty of Leipzig University, Leipzig, Germany (IR B00001750, project ID 441/15ek, September 14, 2020) Author Disclosures: Timm Denecke: Nothing to disclose Manuel Struck: Nothing to disclose Hans-Jonas Meyer: Nothing to disclose Veronika Sotikova: Nothing to disclose Unmasking the bleeding: post-traumatic CT-scan angi ography of pelvic ring with and without pelvic belt *L. J. Pavan*, N. Ouamrane, G. Paesani, P-A. Ranc, K. Desalos, T. Vivarrat-Perrin, M-E. Amoretti, N. Amoretti; Nic e/FR ([email protected]) Purpose or Learning Objective: Patients who suffered a high energy trauma with pelvic ring fracture usually come to whole-bod y CT-room with a pelvic binder that improves outcome but may mask an active bleeding during CT contrast injection. Aim of study was to evaluate th e importance of removing pelvic binder in the early imaging of pelvic ring f racture.

Methods

or Background: All consecutive post-traumatic whole-body CT- scans performed in our emergency department from Ja nuary 2022 to December 2023 were reviewed. Patients with pelvic r ing fracture and a 3- phase CT evaluation (non-contrast, late arterial an d portal venous) were included. CT-scan were directly performed without p elvic binder for hemodynamically stable patients. For hemodynamicall y unstable patients a first acquisition was performed with a tighten pelv ic binder, and if no bleeding was found a second 2-phase contrast injection was r ealized after loosening the pelvic binder to reveal any possible hidden bleedin g.

Results

or Findings: Out of 847 whole-body CT scan performed in the considered period, a total of 149 patients (87 men, 62 women, mean age 43,6±17 years) with a pelvic ring fracture (78 Tile-A, 42 Tile-B, 29 Tile-C) were included. Seven patients were hemodynamically unsta ble, requiring a first CT examination with tighten pelvic binder and a second acquisition with loosen pelvic binder. Of these, 2 patients (2/7, 29%) show ed a pelvic active bleeding only with loosen pelvic binder. Of the 142 hemodynamically stable patients who dire ctly underwent CT without binder, 10 presented an active bleeding.

Conclusion

Pelvic binders are a useful tool but may interfere with CT imaging, masking an active pelvic bleed. An acquisi tion with loosen pelvic binder should always be performed in order to avoid false negative examination.

Limitations

The retrospective design of the study. Funding for this study: None Ethics committee - additional information: None since observational and retrospective Author Disclosures: Luca Jacopo Pavan: Nothing to disclose Kevin Desalos: Nothing to disclose Nadine Ouamrane: Nothing to disclose Thomas Vivarrat-Perrin: Nothing to disclose Gaelle Paesani: Nothing to disclose Nicolas Amoretti: Nothing to disclose Paul-Alexis Ranc: Nothing to disclose Marie-Eve Amoretti: Nothing to disclose Single-pass split-bolus abdominal computed tomograp hy (CT) versus conventional biphasic CT in abdominal trauma patien ts *S. Gautam*, R. Gupta, A. Sharma; New Delhi/IN ([email protected]) Purpose or Learning Objective: To compare the image quality in single-pass split-bolus abdominal computed tomography (CT) and conventional biphasic CT in abdominal trauma patients.

Methods

or Background: 66 consecutive patients of abdominal trauma referred for CT were randomized into two groups: th e study group (n = 33) scanned using the split-bolus technique; and the co ntrol group (n = 33) scanned using the conventional biphasic technique. CT image quality was analyzed subjectively by two observers based on a 5 -point Likert scale. The images were also analyzed quantitatively for attenu ation values achieved by region of interest (ROI) placements in major arteri es, veins and solid organs. In addition, radiation dose in terms of Dose Length Pr oduct (DLP) was compared in the two groups.

Results

or Findings: The image quality in both groups ranged from good t o excellent in most cases. There was no statistically significant difference in subjective image quality in both the groups as asse ssed by Likert score. Attenuation values in solid organs and major venous structures were significantly higher in the split-bolus group (p <0 .001). Arterial attenuation values were significantly higher in the control gro up (p <0.001) but diagnostic levels were achieved in all patients. There was a r eduction of 31.1% in DLP in the split-bolus group.

Conclusion

Split-bolus technique offers comparable image quali ty and higher solid organ and venous enhancement than conventiona l biphasic protocol at a reduced radiation dose.

Limitations

The sample size was limited, excluding pediatric pa tients. The split-bolus protocol used a fixed 120 ml contrast d ose, unlike the weight-based regime in dual-phase CT. Most patients were young m ales, so the average contrast in the dual-phase group was similar. Altho ugh whole-body CT is increasingly used in trauma centers, we applied bot h protocols only for abdominal scans. Funding for this study: None Ethics committee - additional information: Abstract is approved by institutional ethics committee Author Disclosures: Shubham Gautam: Author: 1st author Anuradha Sharma: Author: 2nd author Rohini Gupta: Author: Corresponding author Contrast Timing Pulmonary CT-Angiography: fixed tri gger delay in the ascending aorta vs. pulmonary trunk *G. G. De Almeida*, O. Krzystek, J. Heimer, T. Niem ann, A. Euler; Baden/CH Purpose or Learning Objective: CT angiography (CTA) is the gold standard for diagnosing pulmonary artery embolism (PE). Opti mal scan timing is crucial for homogeneous enhancement of the pulmonary arteri es (PA). Clinical differentiation between PE and aortic dissection is not always clear. This study tested whether bolus tracking in the ascending aort a provides sufficient Friday Abstract-based Programme 167 diagnostic enhancement of the PA, whilst improving enhancement in the thoracic aorta in patients with suspected PE, with the objective of allowing additional diagnostic information within the same e xam.

Methods

or Background: Retrospective image analysis of 200 patients scanned for PE between 03.2024 and 07.2024 was cond ucted. Patients were imaged using a third-generation dual-source CT with application of 70 mL of iodinated contrast medium and bolus tracking trigge ring in either the pulmonary trunk (A) or the ascending aorta (B). A fixed trigg er delay of 7 seconds and automatic tube voltage selection were applied. CT a ttenuation and contrast-to- noise ratio (CNR) were measured at the pulmonary tr unk, main pulmonary arteries, segmental superior lobe arteries, ascendi ng, and descending aorta. A mixed-effects model and post-hoc tests were applied .

Results

or Findings: No significant difference in CNR was found when comparing both techniques for the pulmonary tree (a ll p>.05). Mean CNR at the pulmonary trunk, main pulmonary artery, and seg mental pulmonary arteries were 13.8, 12.9, 12.5 for group A and 13.9, 13.6, 1 3.6 for group B, respectively. In the ascending and descending aorta , CNR was significantly higher in group B when compared to group A (13.1 vs . 8.8 and 11.8 vs. 6.9; both p<.001).

Conclusion

Bolus tracking in the ascending aorta showed simila r contrast enhancement of the pulmonary tree compared to trigg ering in the pulmonary trunk while improving enhancement in the thoracic a orta, allowing simultaneous evaluation of both regions.

Limitations

Single center, rectrospective study. Funding for this study: No dedicated funding Ethics committee - additional information: Rectrospective study with anonymised patient data. Author Disclosures: Jakob Heimer: Nothing to disclose Oezlem Krzystek: Nothing to disclose Tilo Niemann: Nothing to disclose Gonçalo Garcia De Almeida: Nothing to disclose Andre Euler: Nothing to disclose Contrast-Enhanced Ultrasound for Post-Traumatic Spl een Lesion Assessment: A Radiation-Free Diagnostic Alternative *N. Finardi*, F. Cicchetti, E. Xhepa, C. Lanza, A. M. Ierardi, G. Carrafiello; Milan/IT Purpose or Learning Objective: Spleen injuries are common in abdominal trauma. While FAST and contrast-enhanced CT are com monly used to detect parenchymal or vascular spleen lesions, contrast-en hanced ultrasound (CEUS) is emerging as a radiation-sparing alternative. CEU S can detect lesions within minutes using microbubble agents, providing a fast and safe method for monitoring damage. This study aims to develop a CEU S protocol for monitoring spleen trauma cases managed non-operativ ely.

Methods

or Background: A prospective, single-center observational cohort study was conducted at the Radiology Department of Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico in Milan. The study involves 28 patients who presented to the emergency department with abdo minal trauma and were diagnosed with parenchymal or vascular spleen injur ies managed non- operatively. Initial assessments included contrast- enhanced CT and subsequent follow-up evaluations using contrast-enh anced ultrasound (CEUS) at 48 hours, 12-15 days, 30 days, and 60 days post- trauma to monitor lesion progression and healing.

Results

or Findings: The primary endpoint was to evaluate the diagnostic accuracy of CEUS in detecting complications followi ng non-operative management of splenic trauma. Among 28 patients, 25 (89.3%) underwent embolization, with CEUS at 48 hours revealing splen ic infarcts in 10 (35.7%). No increased free fluid, vascular lesions, or hemat oma progression cases were observed in 22 patients (78.6%).

Conclusion

CEUS proved to be a reliable, radiation-free tool f or early detection of complications in non-operatively manag ed splenic trauma. Its high diagnostic accuracy, particularly at 48 hours, supp orts its use as a valuable alternative to CT in follow-up care.

Limitations

Small population and short follow-up. Funding for this study: No funding was received for this study. Ethics committee - additional information: Pending approval by the ethics committee. Author Disclosures: Anna Maria Ierardi: Nothing to disclose Niccolò Finardi: Nothing to disclose Francesco Cicchetti: Nothing to disclose Edon Xhepa: Nothing to disclose Carolina Lanza: Nothing to disclose Gianpaolo Carrafiello: Nothing to disclose Multiplanar reconstruction of MDCT-images of the mi dface for CT- exophthalmometry: performance and reproducibility b etween different post-processing programs *U. G. Müller-Lisse*, K. Donij, S. G. Priglinger, S . Otto, C. R. Hintschich; Munich/DE ([email protected]) Purpose or Learning Objective: Measuring ocular protrusion (OP) on MDCT- images is particularly useful in cranio-facial trau ma. We step-wisely performed multi-planar reconstruction of MDCT-images (MPR) fo r CT-exophthalmometry and investigated if results differ between physicia ns and between different MPR-programs.

Methods

or Background: One dentist and one radiologist independently reformatted MDCT-images from primary multidetector- row-CT reconstructions and measured inter-frontozygomatic base-length and OP, applying two commercial and one freely available post-processing programs for MPR, respectively, in fifteen consecutive patients with cranio-facial trauma (five female, age 24-88 years), with ethics-committee app roval. Ease of patient-and- exam selection, 3D-reconstruction, alignment of orb its, fine adjustment, and measurements were rated 1-10 for each MPR-program. Wilcoxon-matched- pairs-signed-ranks-tests and two-tailed Student-T-t ests for paired data compared MPR-steps and distance-measurements, respe ctively (significance- level, p<0.05).

Results

or Findings: All MPR-programs allowed physician-generated MPR- reformatting and measurements of base-length and OP , although with different levels of ease (range, 5-10). Results of distance m easurements varied by 0.2- 0.4 mm, correlated highly (Pearson-r=0.9411-0.9956) and did not differ significantly between different MPR-programs and di fferent observers with few exceptions.

Conclusion

CT-exophthalmometry results appear highly reproduci ble and stable between different MPR-programs when differen t physicians independently reformatted CT-images and measured ba se-length and ocular protrusion.

Limitations

The study is limited to three exemplary post-proces sing programs, two independent observers with different training background and a small number of consecutive patients. However, it d emonstrates that the principles of observer-performed multiplanr reconst ruction of MDCT-images and measurements relating to CT-exophthalmometry tr ansfer between different post-processing programs and Physicians with differ ent subspecialty training. Funding for this study: No funding has been obtained for this study. Ethics committee - additional information: Ethics committee of the Medical Faculty, LMU Ludwig-Maximilians-Universität München , Vote No. 20-633 KB Author Disclosures: Kathleen Donij: Nothing to disclose Siegfried Georg Priglinger: Nothing to disclose Christoph Rudolf Hintschich: Nothing to disclose Ullrich G. Müller-Lisse: Nothing to disclose Sven Otto: Nothing to disclose Analyzing the Prevalence of Injury and Violence in Transgender Females Using Radiology Reports *R. Chopra*, K. Patel, B. Rosner, O-P. Hamnvik, B. Khurana; Boston, MA/US ([email protected]) Purpose or Learning Objective: Given the high risk of violence with significant underreporting among transgender and ge nder diverse patients, this study aims to investigate the prevalence and dispar ities in injuries and potential violence between transgender female and cisgender f emale patients by analyzing radiology reports.

Methods

or Background: We utilized our hospital's Research Patient Data Registry to identify 263 transgender female patient s and 525 age, race, and ethnicity-matched cisgender women. Adjusted inciden ce rate ratios (aIRR) and Odds ratios were calculated to compare imaging and injury patterns. Two radiologists blinded to the study's purpose assesse d the likelihood of intimate partner violence (IPV) based on radiology reports. EMRs were reviewed for violence documentation in all patients with radiolo gically evident injuries.

Results

or Findings: In our cohort, 25.4% (67/263) of cases sustained 14 1 injuries, compared to 14.7% (77/525) of controls wh o sustained 98 injuries. Injury rates were higher in cases (aIRR: 3.3 [2.5-4 .3] P<0.0001), particularly for cranial (7.8 [2.1-29.1] P<0.0001), facial (36.4 [8. 6-153.8] P<0.0001), and thoracic injuries (4.9 [1.4-17] P=0.01), with 78.9% of facial fractures (15/19) involving the midface. The percentage of imaging st udies in emergency departments was significantly higher in the cases t han in the controls (OR = 5.3 [3.3, 8.3]) (P<0.0001). Radiologists suspected IPV in 12 cases and 1 control, with 75% of cases confirming violence and 50% reporting IPV. A higher number of cases with radiologically evident injuries reported experiencing IPV (OR 6.5; [2.7-15.9]; P<0.0001) com pared to controls. Friday Abstract-based Programme 168

Conclusion

Transgender females experience significantly higher injury rates, particularly to the head, face, and chest, with fre quent presentations to emergency departments, indicating an elevated risk of violence and gaps in preventive care. By recognizing these patterns, rad iologists can help identify at-risk patients and facilitate timely IPV screenin g and support.

Limitations

Retrospective single-institution, self-reporting by patient. Funding for this study: National Institute of Biomedical Imaging and Engineering (NIBIB), National Institute of Health Ethics committee - additional information: Approved by Mass General Brigham IRB Author Disclosures: Krishna Patel: Nothing to disclose Bharti Khurana: Nothing to disclose Bernard Rosner: Nothing to disclose Rohan Chopra: Nothing to disclose Ole-Petter Hamnvik: Nothing to disclose Abbreviated MRI in traumatic injury of spine *C. Loberg*, L. Küsters, A. Gisevius, D. Roggenland , E. Yilmaz, T. Schildhauer, C. Kruppa, M. Aach; Bochum/DE ([email protected]) Purpose or Learning Objective: Traumatic ligamentous injury is common in trauma of spine and can be overlooked easily. Patie nts outcome is based on fast diagnosis and surgery. MRI is the standard of care in evaluation of ligamentous injury. We invested if an abbreviated p rotocol (AP) consisting only of one T2wSTIR acquisition is suitable for detectio n of ligamentous injury.

Methods

or Background: A cohort of 100 patients with underlying spinal trauma who underwent MRI were selected. Full MRI pr otocol (FP) comprised T1w, T2w, T2wSTIR sequences at 1.5T. For abbreviate d MRI protocol (AP) we chose a rapid protocol that ensures the acquisition of maximum of contrast and maximum spatial-resolution images based on T2wSTIR acquisition. Two radiologist with 5 years and 16 years of experience reviewed the images to characterize ligaments, fracture, spinal instabilit y and traumatic disc herniation.

Results

or Findings: MRI acquisition for FP was 23 minutes versus 4.12 f or the AP. Average time to read the single T2wSTIR and complete FP was 3.4 minutes versus 14.3 minutes. 42 ligamentous injurie s were detected. Specificity and positive predictive value (PPV) of AP versus FDP were equivalent (96.5% to 94.2% and 23.9% v 23.2%).

Conclusion

An MRI acquisition of 4.20 minutes and expert radio logist reading time of 3.40 minutes are sufficient to confirm liga mentous injury of spine. With a reading time < 4 minutes diagnostic accuracy was equivalent to that of the FDP.

Limitations

This was a retrospective single center study. Funding for this study: There was no funding Ethics committee - additional information: Ethic Comittee Ruhr University Hospital Bochum 178/ 24 Author Disclosures: Emre Yilmaz: Nothing to disclose Astrid Gisevius: Nothing to disclose Thomas Schildhauer: Nothing to disclose Mirko Aach: Nothing to disclose Leonie Küsters: Nothing to disclose Christina Loberg: Nothing to disclose Daniela Roggenland: Nothing to disclose Christiane Kruppa: Nothing to disclose Diagnostic accuracy and time efficiency of a novel deep learning algorithm for the assessment of intracranial hemorr hage *C. Booz*¹, T. Vogl¹, V. Koch¹, L. D. Gruenewald¹, A-I. Nica¹, T. D'Angelo², M. Dimitrova¹, G. M. Bucolo¹, I. Yel¹; ¹Frankfurt/D E, ²Messina/IT Purpose or Learning Objective: To evaluate diagnostic accuracy and time efficiency of a deep learning-based pipeline using a Dense U-net architecture for the assessment of intracranial hemorrhage (ICH) in unenhanced head CT scans.

Methods

or Background: This retrospective study included 1004 CT scans of 1004 patients (mean age, 71 ± 11 years; 496 men and 508 women) who had undergone an unenhanced head CT scan for the assess ment of ICH. All CT scans were analyzed by the algorithm and a board-ce rtified radiologist independently for the presence of ICH. In case of I CH presence, ICH had to be defined as intraparenchymal hemorrhage (IPH), intra ventricular hemorrhage (IVH), subarachnoid hemorrhage (SAH), subdural hemo rrhage (SDH) and epidural hemorrhage (EDH). Additionally, the time u ntil first temporary diagnosis of ICH was measured. Three experienced bo ard-certified radiologists analyzed the CT scans in consensus reading sessions to establish the standard of reference for hemorrhage presence and c lassification.

Results

or Findings: The reference standard revealed a total of 1108 dif ferent ICH presences (IPH, n=344; IVH, n=52; SAH, n=326; S DH, n=356; EDH, n=30). The algorithm showed a high diagnostic accur acy for the assessment of ICH with a sensitivity of 92%, specificity of 95% a nd an accuracy of 93%. Concerning the most frequently present different IC H types in this study, the sensitivity was 92%, 93% and 93% (IPH, SAH and SDH, respectively), and the specificity was 95%, 96% and 95% (IPH, SAH and SDH, respectively). Regarding analysis time, the algorithm was signific antly faster compared to the temporary report of the assigned radiologist (16 ± 3 s vs 273 ± 11 s, p < 0.001).

Conclusion

A novel deep learning algorithm provides high diagn ostic accuracy combined with time efficiency for the iden tification and classification of ICH in unenhanced CT scans.

Limitations

Single-center retrospective study Funding for this study: No funding was received. Ethics committee - additional information: The local IRB approved this study. Author Disclosures: Christian Booz: Speaker: Siemens Healthineers Leon D. Gruenewald: Nothing to disclose Ibrahim Yel: Speaker: Siemens Healthineers Mirela Dimitrova: Nothing to disclose Thomas Vogl: Nothing to disclose Vitali Koch: Nothing to disclose Andreea-Ioana Nica: Nothing to disclose Tommaso D'Angelo: Speaker: Bracco Speaker: Philips Giuseppe Mauro Bucolo: Nothing to disclose Beyond ASL: SWI as the Superior Diagnostic Tool for Status Epilepticus When Conventional MRI Falls Short *S. K. Kondapavuluri*, S. K. Patan, R. P. Yadav; Vi jayawada/IN Purpose or Learning Objective: This study evaluates the diagnostic utility of susceptibility-weighted imaging (SWI) in status epi lepticus (SE), focusing on its role in identifying cerebral perfusion and oxygenat ion changes, particularly when conventional MRI sequences (diffusion, T2, FLA IR) appear normal. SWI’s superior resolution for detecting subtle veno us and metabolic changes makes it a better alternative to arterial spin labelling (ASL) in certain clinical settings. The study aims to demonstrate SWI’s value as both a practical and superior tool in perfusion imaging.

Methods

or Background: This observational study involved 50 patients with confirmed SE who underwent MRI within 6 hours of se izure onset. The imaging protocol included diffusion-weighted imaging (DWI), fluid-attenuated inversion recovery (FLAIR), SWI, and ASL. SWI patterns of ven ous oxygenation were compared with ASL perfusion maps to assess concorda nce in detecting hyper- and hypo-perfused regions, particularly in cases wh ere conventional MRI sequences were unremarkable.

Results

or Findings: SWI revealed distinct patterns of cerebral venous changes, categorised into four groups: Group 1 (20 patients) exhibited generalised hyperperfusion on ASL with globally dim inished cortical veins on SWI, indicating global hyperoxygenation. Group 2 (1 3 patients) showed focal hyperperfusion with focally diminished cortical vei ns, reflecting focal hyperoxygenation. Group 3 (10 patients) displayed f ocal hyperperfusion with focally prominent cortical veins due to focal deoxy genation. Group 4 (7 patients) demonstrated generalised hyperperfusion w ith globally prominent veins, indicating global deoxygenation.

Conclusion

SWI is a valuable tool for detecting oxygenation ch anges in SE, even when conventional MRI sequences appear normal. Its superior resolution and availability make it a better alternative for d etecting subtle venous and metabolic changes, supporting its broader adoption as a complementary or standalone tool in epilepsy management.

Limitations

Not applicable Funding for this study: No funding was provided for this study. Ethics committee - additional information: Ethics approval for this observational study was obtained from the relevant ethics committee. Informed consent was secured from all participants, ensuring compliance with ethical standards for research involving human subjects. Th e study adhered to the principles outlined in the Declaration of Helsinki. Author Disclosures: Sushen Kumar Kondapavuluri: Nothing to disclose Sharuq Khan Patan: Nothing to disclose Ratan Pal Yadav: Nothing to disclose Prognostic Indicators of Conservative Treatment Fai lure in Adhesive Small Bowel Obstruction: Insights from CT Imaging *A. Ammirabile*, E. Desiato, A. M. A. Lucia, S. Giu dici, M. Francone, D. Del Fabbro, E. Lanza; Milan/IT ([email protected]) Purpose or Learning Objective: This study aimed to identify the CT imaging features associated with the failure of conservativ e management using oral water-soluble contrast medium in patients presentin g to the Emergency Room with Adhesive Small Bowel Obstruction (ASBO). Friday Abstract-based Programme 169

Methods

or Background: This retrospective single-center study included all consecutive patients admitted to the ER from Februa ry 2019 to February 2023 with ASBO, who underwent contrast-enhanced CT at di agnosis and received conservative treatment. The assessed CT findings we re type and location of the transition zone, ASBO severity, presence of fat notch sign, beak sign, small bowel feces sign, peritoneal free fluid, and pneuma tosis intestinalis. Univariable and multivariable logistic regression a nalyses were performed to evaluate the association between these radiological parameters and treatment outcomes.

Results

or Findings: Among the 106 patients included (median age 74.5 years), conservative management succeeded in 59 cas es (55.7%), while 47 patients (44.3%) required surgery after initial non -operative treatment failure. Failure was more common in patients with previous A SBO episodes (p = 0.03), female gender (p = 0.04), and was associated with a longer hospital stay (p < 0.001). At multivariable analysis, the fat notch si gn (OR = 2.95; p = 0.04) and the beak sign (OR = 3.42; p = 0.04) were significan tly associated with conservative treatment failure.

Conclusion

Two CT features - the fat notch sign and the beak s ign - were significantly correlated with the failure of conser vative management in ASBO. These findings highlight the importance of an early identification of patients who may benefit from undelayed surgical interventio n.

Limitations

The limitations of the study are the retrospective, single-center design and the evaluation of a limited number of CT signs and laboratory values. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is retrospective. Author Disclosures: Elena Desiato: Nothing to disclose Angela Ammirabile: Nothing to disclose Ada Maria Antonella Lucia: Nothing to disclose Ezio Lanza: Nothing to disclose Daniele Del Fabbro: Nothing to disclose Marco Francone: Nothing to disclose Simone Giudici: Nothing to disclose 14:00-15:30 Research Stage 2 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 1505 Generative AI in radiology Moderator C. Blüthgen; Zurich/CH Precision, Non-classifications, and Misclassificati ons of General and Medical Large Language Models in Liver Lesions Clas sification using LI- RADS from Unstructured Radiology Reports J. Lu¹, F. F-Y. Tang¹, J. Ng¹, C. Chan², H. M. Chen g¹, P. L. H. Yu¹, W. K. W. Seto¹, *W. H. K. Chiu*¹; ¹Hong Kong/HK, ²H ampshire/UK ([email protected]) Purpose or Learning Objective: Large Language Models (LLM) are powerful tools for data extraction and summarization. Howeve r, scant evidence exists as to whether a medial-specific LLM is necessary to pe rform radiology tasks. This study evaluates the performance between a general a nd a medical LLM in extracting and categorizing liver lesions from radi ology reports according to the LI-RADS.

Methods

or Background: A total of 273 anonymized unstructured Computed Tomography (CT) reports, written by 115 radiologist s from 5 institutions containing 599 liver observations were retrospectiv ely collected. These reports were fed into GPT-4 and MedLM to assign LI-RADS cat egories for each observation using zero-shot prompts (GPT4sp and Med LMsp) and instructions post-prompt engineering (GPT4pe and MedLMpe). Groun d truths and quality of the CT reports were derived by 2 board-certified radiologists.

Results

or Findings: At lesion level, the accuracies for correctly class ifying malignant lesions (LR-4/5/M) were 0.584, 0.634, 0.6 68, and 0.84 for GPT-4sp, MedLMsp, GPT-4pe, and MedLMpe respectively with Med LM outperforming GPT-4 using both simple prompts (p=0.023) and promp t engineering (p<0.001). At patient level, the accuracies were 0. 762, 0.744, 0.791, and 0.883, respectively, with prompt engineering outperforming simple prompts in MedLM (p < 0.001). Prompt engineering improved performanc e by reducing non- classification in both MedLM (11.5% vs 33.7%, p<0.0 01) and GPT-4 (29.4% vs 38.4% p<0.001). The quality of the CT reports of th e 31 misclassified/non- classified patients on MedLMpe were considered aver age (median LIKERT score 3/5) with a Fleiss’ ĸ value 0.563 (95%CI 0.356 - 0.770).

Conclusion

While general LLM exhibits potential in text-based medical tasks, our findings suggest that medical LLM yields superi or performance.

Limitations

Limitations include a small sample size, lack of pr ompt engineering exploration, and only one of general an d medical LLM used. Funding for this study: None Ethics committee - additional information: The ethics committee notification can be found under Ref: KC/KE-23-0083/ER-3. Author Disclosures: Philip L. H. Yu: Nothing to disclose Ho Ming Cheng: Nothing to disclose Wan Hang Keith Chiu: Nothing to disclose Chelsea Chan: Nothing to disclose Justin Ng: Nothing to disclose Wai Kay Walter Seto: Nothing to disclose Fanny Fong-Yi Tang: Nothing to disclose Jianliang Lu: Nothing to disclose Evaluating the Performance of LLaMA 3.1 in Classify ing Mammography Reports Based on BIRADS Scores *A. Kumar*, V. K. Venugopal; New Delhi/IN ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the performance of the LLaMA 3.1 large language model ( LLM) in classifying mammography reports based on the Breast Imaging-Rep orting and Data System (BIRADS) classification without fine-tuning the model.

Methods

or Background: A total of 930 mammography reports, covering a range of BIRADS classifications from 0 to 6, were p rocessed using the LLaMA 3.1 open-source LLM (8B version). The model was pro mpted using a five-shot prompting technique. The classification accuracy of the algorithm was analyzed, and errors in classification were recorde d. Among the 930 reports, 8 instances of errors were identified, with 4 cases w here BIRADS 2 was incorrectly classified as BIRADS 4 by the model.

Results

or Findings: The LLaMA 3.1 model demonstrated a classification accuracy of 921 correct classifications out of 930 reports, yielding an overall accuracy rate of 98.99%. Despite the model's strong performance, errors were present, particularly in the misclassification of l ower BI-RADS scores, with some benign reports (BIRADS 2) being classified at a higher risk level (BIRADS 4)

Conclusion

LLaMA 3.1, even without fine-tuning, shows signific ant potential for accurately classifying mammography reports base d on BIRADS scoring. This indicates that large language models could ser ve as valuable tools in medical imaging analysis, offering high accuracy wi th minimal adjustments.

Limitations

The study is limited by the occurrence of misclassi fication in a small number of cases, particularly in distinguishi ng between benign and higher-risk categories. Further studies with larger datasets and fine-tuning may be needed to improve reliability. Funding for this study: The study didn't receive any funding Ethics committee - additional information: Anonymized data was used Author Disclosures: Amit Kumar: Nothing to disclose Vasantha Kumar Venugopal: Nothing to disclose Evaluating local open-source large language models for data extraction from unstructured reports on mechanical thrombectom y in patients with ischemic stroke *A. Meddeb*¹, A. Othman², N. F. Grauhan², M. Scheel ³, J. Nawabi³; ¹Reims/FR, ²Mainz/DE, ³Berlin/DE ([email protected]) Purpose or Learning Objective: To assess the effectiveness of open-source Large Language Models (LLMs) in extracting clinical data from unstructured mechanical thrombectomy reports in patients with is chemic stroke caused by a vessel occlusion.

Methods

or Background: We deployed local open-source LLMs to extract data points from free-text procedural reports in pa tients who underwent mechanical thrombectomy between September 2020 and June 2023 in our institution. The external dataset was obtained from a second university hospital and comprised consecutive cases treated between Sep tember 2023 and March 2024. Ground truth labeling was facilitated b y a human-in-the-loop (HITL) approach, with time metrics recorded for bot h automated and manual data extractions. We tested three models—Mixtral, Q wen, and BioMistral— assessing their performance on precision, recall, a nd F1 score across 15 clinical categories such as National Institute of H ealth Stroke Scale (NIHSS) scores, occluded vessels, and medication details.

Results

or Findings: The study included 1000 consecutive reports from ou r primary institution and 50 reports from a secondary institution. Mixtral showed the highest precision, achieving 0.99 for first ser ies time extraction and 0.69 for Friday Abstract-based Programme 170 occluded vessel identification within the internal dataset. In the external dataset, precision ranged from 1.00 for NIHSS score s to 0.70 for occluded vessels. The HITL approach yielded an average time savings of 65.6% per case, with variations from 45.95% to 79.56%.

Conclusion

LLMs showed high performance in automated clinical data extraction from medical reports. Incorporating HITL annotations enhances precision and also ensures the reliability of the e xtracted data. This methodology presents a scalable privacy-preserving option that can significantly support clinical documentation and re search endeavors.

Limitations

Variability in the quality and consistency of the i nput data, such as differences in terminology, formatting, or detai l level in the reports, can affect the performance of the models Funding for this study: None Ethics committee - additional information: This retrospective study was approved by the ethics committee of the Charité Uni versity Hospital in Berlin (No. EA4/062/20).The requirement for informed conse nt was waived due to the retrospective design of the study. Author Disclosures: Nils F. Grauhan: Nothing to disclose Aymen Meddeb: Research/Grant Support: Berlin Instit ute of Health Jawed Nawabi: Nothing to disclose Michael Scheel: Nothing to disclose Ahmed Othman: Nothing to disclose Large language models in healthcare: DRAGON perform ance benchmark for clinical NLP *J. S. Bosma*¹, K. Dercksen¹, M. De Rooij¹, F. Ciom pi¹, A. Hering¹, J. Geerdink², H. E. Huisman¹; ¹Nijmegen/NL, ²Almelo /NL ([email protected]) Purpose or Learning Objective: Artificial Intelligence (AI) requires large-scale annotated datasets to train clinical algorithms to perform at an expert level. Natural Language Processing (NLP) shows great poten tial to annotate large volumes of data from clinical routine and facilitat e the training of these algorithms. This study aims to introduce a benchmar k for clinical NLP algorithms, including Large Language Models (LLMs), to assess the ability of algorithms to extract information from medical repo rts.

Methods

or Background: The DRAGON (Diagnostic Report Analysis: General Optimization of NLP) challenge has three ob jectives. First, it provides a unique and publicly available cloud-based benchma rk for clinical NLP that spans 28 clinically relevant tasks. 28,824 annotate d medical reports from five Dutch care centers from multiple imaging modalities (MRI, CT, X-ray, histopathology) and conditions spanning the entire body (lungs, pancreas, prostate, skin, etc.) are used. The tasks are desig ned to facilitate automated dataset curation and include predicting diagnoses, extracting lesion sizes, identifying protected health information, and more. Second, we release foundational LLMs pretrained using four million cli nical reports from a sixth Dutch care center. Third, we investigate three pret raining strategies across five architectures by evaluating LLMs using the DRAGON b enchmark.

Results

or Findings: Results showed the superiority of domain-specific pretraining (benchmark score of 0.770, 95% CI 0.755 -0.785) and mixed- domain pretraining (0.756, 95% CI 0.739-0.773), com pared to general-domain pretraining (0.734, 95% CI 0.717-0.752, p<0.005). T he best model achieved excellent or good performance for 18/28 tasks and p oor or moderate performance for 10/28 tasks.

Conclusion

The DRAGON benchmark showed that NLP is ready to fa cilitate data curation in some settings, enabling high-quali ty, low-cost, and large-scale annotation, and uncovered where innovations are nee ded to improve clinical NLP.

Limitations

Half of the tasks were sourced from a single academ ic tertiary care center (14/28, 50%). Funding for this study: Funding was provided by Health~Holland (LSHM20103), European Union HORIZON-HLTH-2022: COMF ORT (101079894), European Union HORIZON-2020: ProCAncer -I project (952159), European Union HORIZON-2020: PANCAIM project (10101 6851), and NWO- VIDI grant (number 18388). The collaboration projec t is co-funded by PPP Allowance awarded by Health~Holland, Top Sector Lif e Sciences \& Health, to stimulate public-private partnerships. Views and op inions expressed are however those of the author(s) only and do not nece ssarily reflect those of the European Union or European Health and Digital Execu tive Agency (HADEA). Neither the European Union nor the granting authori ty can be held responsible for them. Ethics committee - additional information: Retrospective use of anonymous patient data was approved by institutional or regio nal review boards at each contributing center (identifiers: CMO 2016-3045; IR Bd22-159; A21-0349 2; A20-0777), and was conducted in accordance with the principles of the Declaration of Helsinki. Informed consent was waive d. Author Disclosures: Joeran Sander Bosma: Nothing to disclose Henkjan En Huisman: Nothing to disclose Francesco Ciompi: Nothing to disclose Alessa Hering: Nothing to disclose Koen Dercksen: Nothing to disclose Maarten De Rooij: Nothing to disclose Jeroen Geerdink: Nothing to disclose Implementing Local Large Language Models and using Clinical Data Warehouse for Clinical Summarization and Decision S upport *M. Segeroth*, M. Bach, J. Wasserthal, J. Cyriac, M . Pradella, H-C. Breit, B. Stieltjes, E. M. Merkle, S. Yang; Basel/CH ([email protected]) Purpose or Learning Objective: Recent advances in Large Language Models (LLMs) have improved medical text summarization and decision support but raised data privacy concerns. We aim to integrate l ocal LLMs into clinical workflows for testing with real-world patient data.

Methods

or Background: Within our institutional healthcare network, a clin ical data warehouse (CDWH) serves as a central hub for q uerying all patient records and parameters, while ensuring data privacy . Exemplary parameters like temporal evolution of chemotherapies, dates an d outcomes of resections, findings from previous imaging examinations, etc. w ere extracted for oncology patients. The collected data were fed via a prompt into local LLMs. We utilized privateGPT and Ollama as the primary platform, allo wing integration of clinical treatment guidelines. Regarding LLMs we tested Llam a3-70B and the German- language SauerkrautLM Mixtral 8X7B Instruct which b oth ran on a Nvidia A100 GPU with 80 GB memory. A set of anonymized data was processed with cloud- based ChatGPT-4 and Claude-3 for comparison.

Results

or Findings: Using the privateGPT platform both tested LLMs ran on a single GPU with maximally 65 GB of memory usage. Both LLMs created text summaries within 15 seconds and provided decision s upport in under 5 seconds per request. For all brain cancer cases the local LLMs provided a correct and reasonable summary of medical history. In decision-making for a prostate tumor board, the decision accuracy amounte d to 7 out of 10 test cases. For anonymized data, accuracy between the lo cal LLMs and both ChatGPT-4 and Claude-3 was 8 out of 10 test cases.

Conclusion

Integration of local LLMs into clinical workflow or research task is possible. Local LLMs were able to summarize medical history or clinical data for tumor boards, preserving local data privacy pol icies.

Limitations

Only two local LLMs were evaluated on sophisticated datasets. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Bram Stieltjes: Nothing to disclose Jacob Wasserthal: Nothing to disclose Michael Bach: Nothing to disclose Hanns-Christian Breit: Nothing to disclose Maurice Pradella: Nothing to disclose Joshy Cyriac: Nothing to disclose Martin Segeroth: Nothing to disclose Elmar M. Merkle: Nothing to disclose Shan Yang: Nothing to disclose Training and Evaluation of Sentence Transformer Mod el for Retrieval Augmented Generation on Radiology Reports *K. Arzideh*, H. Schäfer, A. Idrissi-Yaghir, C. S. Schmidt, J. Haubold, R. Hosch, F. Nensa; Essen/DE ([email protected]) Purpose or Learning Objective: In many medical settings physicians often have to sift through unstructured documents to find important information. This manual process is time-consuming and can lead to mi ssed details. Retrieval Augmented Generation (RAG) can help physicians to q uickly locate relevant information. By using Sentence Transformer models f ine-tuned for retrieval tasks, similarity search between input query and do cument passages can be performed to find relevant context. However, most p ublicly available models are not specifically fine-tuned for the radiology d omain and are therefore very limited in finding clinically relevant information.

Methods

or Background: Document chunks from 400,000 German clinical notes including radiology reports and doctoral note s, were provided as input to the SauerkrautLM-SOLAR-Instruct Large Language Mode l. The model was prompted to generate clinically related questions a nd answers based on these chunks. The model generated 11 million clinically r elated question-answer pairs to fine-tune a multilingual-e5-large model. F or evaluation, 1,717 question- answer pairs were generated from 215 radiology repo rts. A radiologist filtered out unrelated or incorrect pairs for a realistic ev aluation. The fine-tuned model was then integrated into a RAG system, and its answ ers were compared to those from a non-fine-tuned model using the same da taset. Friday Abstract-based Programme 171

Results

or Findings: Fine-tuning the model resulted in improved performa nce metrics. The BLEURT score increased from 0.551 to 0 .563, indicating enhanced alignment with human judgment. Similarly, the BERTScore F1 rose from 0.750 to 0.756.

Conclusion

By using LLM to generate synthetic questions out of real world documents and fine-tuning sentence transformer mode ls on these question and document pairs, information retrieval performan ce can improve as indicated by automated evaluation metrics.

Limitations

The evaluation was only carried out for documents i n German. Fine-tuning on documents written in other languages and from other hospital sites could lead to a broader applicability. Funding for this study: None Ethics committee - additional information: This study was approved by the Ethics Committee of the Medical Faculty of the Univ ersity of Duisburg-Essen (approval number 23-11557-BO). Due to the study's r etrospective nature, the requirement of written informed consent was waived by the Ethics Committee of the Medical Faculty of the University of Duisbur g-Essen. All methods were carried out in accordance with relevant guidelines and regulations. Author Disclosures: Johannes Haubold: Nothing to disclose Henning Schäfer: Nothing to disclose Cynthia Sabrina Schmidt: Nothing to disclose Kamyar Arzideh: Nothing to disclose René Hosch: Nothing to disclose Ahmad Idrissi-Yaghir: Nothing to disclose Felix Nensa: Nothing to disclose Large Language Models for Simplified Interventional Radiology Reports: A Comparative Analysis *E. Can*¹, W. Uller¹, K. Vogt¹, F. Busch², N. Bayer l³, A. Kader², M. R. Makowski², K. K. Bressem², L. C. Adams²; ¹Fre iburg/DE, ²Munich/DE, ³Erlangen/DE ([email protected]) Purpose or Learning Objective: To quantitatively and qualitatively evaluate and compare the performance of leading large langua ge models (LLMs), including proprietary models (GPT-4, GPT-3.5 Turbo, Claude-3-Opus, and Gemini Ultra) and open-source models (Mistral-7b an d Mistral-8x7b), in simplifying 109 interventional radiology reports.

Methods

or Background: Qualitative performance was assessed using a five- point Likert scale for accuracy, completeness, clar ity, clinical relevance, naturalness, error rates, including trust-breaking and post-therapy misconduct errors. Quantitative readability was assessed using Flesch Reading Ease (FRE), Flesch-Kincaid Grade Level (FKGL), SMOG Inde x, and Dale-Chall Readability Score (DCRS). Paired t-tests and Bonfer roni-corrected p-values were used for analysis.

Results

or Findings: Qualitative evaluation showed no significant differ ences between GPT-4 and Claude-3-Opus for any metrics (al l Bonferroni-corrected p- values: p=1), while they outperformed other models across five qualitative metrics (p < 0.001). GPT-4 had the fewest content a nd trust-breaking errors, with Claude-3-Opus second. All models exhibited som e trust-breaking and post-therapy misconduct errors, with GPT-4-Turbo an d GPT-3.5-Turbo with few-shot prompting showing the lowest error rates, and Mistral-7B and Mistral- 8x7B the highest. Quantitatively, GPT-4 surpassed C laude-3-Opus in readability metrics (all p < 0.001), with a median FRE score of 69.01 (IQR: 64.88-73.14) versus 59.74 (IQR: 55.47-64.01) for Cl aude-3-Opus. GPT-4 also outperformed GPT-3.5-Turbo and Gemini Ultra (both p < 0.001). Inter-rater reliability was strong (κ = 0.77-0.84).

Conclusion

GPT-4 and Claude-3-Opus demonstrated superior perfo rmance in generating simplified IR reports, but the presen ce of errors across all models, including trust-breaking errors, highlights the need for further refinement and validation before clinical implement ation.

Limitations

This study was based on predefined metrics, which, while comprehensive, may not capture all aspects of patie nt understanding and engagement. Future research should include real-wor ld data, a broader range of medical documents, and consider patient feedback to more accurately assess the clinical utility of these models. Funding for this study: This study did not receive any specific funding fro m public, commercial, or not-for-profit sectors. Ethics committee - additional information: Since the reports did not include any real patient data, institutional review board a pproval was not required. This ensures that the study adhered to ethical standards by avoiding the use of real patient information and thereby eliminating the nee d for formal ethical approval processes typically required for studies involving human subjects. Author Disclosures: Keno K. Bressem: Nothing to disclose Elif Can: Nothing to disclose Marcus R. Makowski: Nothing to disclose Lisa C. Adams: Nothing to disclose Katharina Vogt: Nothing to disclose Nadine Bayerl: Nothing to disclose Avan Kader: Nothing to disclose Felix Busch: Nothing to disclose Wibke Uller: Nothing to disclose Automated Radiology Controlling - Using Large Langu age Models for Prediction of Radiological Services based on Radiol ogical Reports *K. Arzideh*, A. Idrissi-Yaghir, H. Schäfer, K. A. Borys, J. Haubold, F. Nensa, R. Hosch; Essen/DE ([email protected]) Purpose or Learning Objective: In hospitals worldwide, controlling of radiological services is a manual process. In Germa ny, the so-called “Gebührenordnung für Ärzte” (GOÄ) regulates the bil ling of private medical or dental services, i.e. services outside the public h ealth insurance scheme. GOÄ numbers can be used to indicate which private relat ed clinical interventions were performed during treatment. These numbers are documented by going through radiological reports and picking out releva nt information, which is time- consuming and error-prone.

Methods

or Background: In order to automate billing of radiological servic es, a Large Language Model (LLM) was fine-tuned to gene rate GOÄ digits out of radiology reports. In total, 1,000,000 radiology re ports and GOÄ digit pairs were split into 80 % training and 20 % test dataset . Training was performed on a Phi-3-small-8k-instruct model. For evaluation, th e test dataset was compared against the numbers generated by the model.

Results

or Findings: The fine-tuned LLM achieved an accuracy of 75 % calculated for the generation of GOÄ numbers. These generated GOÄ codes were identical to the ground truth. 83 % of the pre dicted codes were present in the ground truth, but may not have been a complete match.

Conclusion

LLM are capable of automatically extracting relevan t controlling codings based on radiology reports only. Therefore, LLMs could be used as an enhanced method for the automation of controlling t asks in radiology.

Limitations

The LLM needs human feedback and manual correction in order to achieve human-like results. The radiology report s used in this study were also written in German language. The use of dataset s in other languages and from other hospitals could enable broader generaliz ability. Funding for this study: None Ethics committee - additional information: This study adhered to all guidelines defined by the approving institutional r eview board of the investigating hospital. The Institutional Review Bo ard waived written informed consent due to the study's retrospective nature. Co mplete anonymization of all data was performed before inclusion in the study. Author Disclosures: Katarzyna Anna Borys: Nothing to disclose Johannes Haubold: Nothing to disclose Henning Schäfer: Nothing to disclose Kamyar Arzideh: Nothing to disclose René Hosch: Nothing to disclose Ahmad Idrissi-Yaghir: Nothing to disclose Felix Nensa: Nothing to disclose Insights and Challenges in Implementing Vision Tran sformers for Thorax Radiography *S. Hyska*, A. Wollek, T. Lasser, M. Ingrisch, B. O . T. Sabel; Munich/DE Purpose or Learning Objective: This study aimed to evaluate the performance of a Vision Transformer (ViT)-based AI model, trained on publicly available chest radiography datasets, when applied to real-world data from our clinic. The model's performance in detecting pleura l effusion, pneumothorax, cardiomegaly, and consolidation was examined, along with potential confounders.

Methods

or Background: The AI model, pre-trained on ImageNet and fine- tuned on >700,000 public chest X-rays (CXR), was te sted on an internal dataset of 113 CXR, including 23 pneumothorax, 29 c ardiomegaly, 31 consolidation, 52 pleural effusion cases, and 29 no rmal CXR. The model’s performance was assessed through ROC-curves, AUC, Y ouden Coefficient, and sensitivity/specificity metrics. Logistic regre ssion, odds ratios, and Fisher's test were used to analyse confounding factors. Friday Abstract-based Programme 172

Results

or Findings: The model correctly identified all normal CXRs. For pleural effusion, sensitivity was 96.2% and specifi city 98.4%, indicating strong performance. For pneumothorax, sensitivity was only 26.1% with 96.7% specificity. Pneumothorax size and presence of thor acic tubes were significant confounders. Cardiomegaly was detected with 55.2% s ensitivity and 96.4% specificity, whereas concomitant pleural effusions, obscuring the heart contours, act as a potential confounder. Consolidat ion was detected with 45.2% sensitivity and 91.5% specificity, and higher density consolidations were more easily identified.

Conclusion

This study emphasizes the challenges AI models face when integrated into clinical practice, demonstrating th e importance of carefully and clinically assessing model performance on real-worl d-data, especially in the context of confounding factors. While our ViT model showed strong performance for pleural effusion and normal finding s, its detection of pneumothorax, cardiomegaly, and consolidation was l imited. Known confounders, e. g. pneumothorax size and presence o f thoracic tubes, were confirmed, and new ones, such as pleural effusion i n cardiomegaly and density of consolidations, were identified.

Limitations

The exploratory nature and limited number of CXR we re key limitations. Funding for this study: This work was funded in part by the German federal ministry of health’s program for digital innovation s for the improvement of patient- centered care in healthcare [grant agreeme nt no. 2520DAT920]. Ethics committee - additional information: Approval by an ethics committee is present. Author Disclosures: Bastian Oliver Theodor Sabel: Nothing to disclose Sardi Hyska: Nothing to disclose Alessandro Wollek: Nothing to disclose Michael Ingrisch: Nothing to disclose Tobias Lasser: Nothing to disclose 14:00-15:30 Research Stage 3 Research Presentation Session: Paediatric RPS 1512 Body imaging in children: from head to toe Moderator L. B. Laborie; Bergen/NO ([email protected]) Author Disclosures: Lene Bjerke Laborie: Other: Involved in development of AI-algorithms for paediatric hip radiographs with software company Vi siana (DK). No financial benefits. Automatic Identification and Classification of Pedi atric Glomerulonephritis on Ultrasound Images Based on De ep Learning and Radiomics *J. Kou*, Y. Tang; Chongqing/CN ([email protected]) Purpose or Learning Objective: Glomerulonephritis (GN) includes a diverse range of kidney diseases that often exhibit subclin ical manifestations in children. While renal biopsy is the gold standard, its invasiveness, susceptibility to sampling errors, and time requirements impede ra pid diagnosis. This study aimed to create a noninvasive diagnostic model for childhood GN by integrating deep learning and radiomics techniques using renal ultrasound images.

Methods

or Background: A total of 469 renal ultrasound images were selected from children undergoing ultrasound-guided biopsy and split into training and validation sets at an 8:2 ratio to tra in a U-Net model for kidney segmentation. Radiomic features were extracted from the segmented regions and categorized by GN types: IgA nephropathy (127 c ases), minimal change disease (83 cases), and Henoch-Schönlein purpura ne phritis (103 cases). These categories were also split into training and validation sets at an 8:2 ratio. ANOVA was used for feature selection in the trainin g set, followed by LASSO regression for dimensionality reduction, yielding 3 7 features. A random forest algorithm was then used to develop a GN classificat ion model, which was evaluated using the validation set.

Results

or Findings: The segmentation model demonstrated excellent performance, achieving 95.19% accuracy on the valid ation set. Thirty-seven selected features were used to build a strong class ification model, which showed high accuracy and predictive power across GN categories, with AUC values between 0.91 and 0.98.

Conclusion

The combination of deep learning and radiomics usin g renal ultrasound images shows great potential for classif ying childhood GN subtypes, offering a noninvasive method to enhance diagnostic efficiency and patient outcomes.

Limitations

Firstly, the relatively limited data sources may ha ve introduced some regional bias to our findings. Furthermore, we focused only on the pathological subtypes of three GN, which inevitably limited the scope of the model. Funding for this study: 0 Ethics committee - additional information: the Ethics Committee of Children's Hospital at Chongqing Medical University . Author Disclosures: Jun Kou: Nothing to disclose Yi Tang: Nothing to disclose Biomarkers of Primary Sclerosing Cholangitis detect ed with delayed gadolinium-enhanced Magnetic Resonance Imaging in p ediatric patients *F. Maccioni*, V. Cardinale, E. Damato, L. Busato, S. Veraldi, A. Valenti, C. Catalano; Rome/IT ([email protected]) Purpose or Learning Objective: Primary sclerosing cholangitis (PSC) is a severe liver disease frequently associated with inf lammatory bowel disease (IBD) with a late diagnosis, mostly based on biliar y changes at MRCP. To identify specific biomarkers for PSC using delayed phases of gadolinium enhancement to detect ductal fibrosis, such as inte stinal fibrosis in Crohn's disease.

Methods

or Background: A prospective study based on the association of MRCP and gadolinium-enhanced MRI, including delayed (7 minutes) phases, was performed in 3 groups of pediatric patients, on e with PSC and IBD, one with IBD only, and one of controls. Three radiologi sts blindly and independently analyzed: a) intra and extrahepatic bile ducts stri ctures (IHBDs, EHBDs) at MRCP; b) gallbladder volume; c) gallbladder wall ga dolinium-enhancement; d) IHBDs-EHBDs gadolinium-enhancement.

Results

or Findings: We included 39 patients, 12 with PSC and IBD (31%) , 16 with IBD only (41%) and 11 controls (28%). At MR CP, IHBDs strictures were detected in 82% PSC-IBD patients (p<0.001). De layed enhancement of gallbladder wall was observed in 100% of PSC-IBD pa tients (100% sensitivity, 90% specificity, (p<0.001); delayed enhancement of the extrahepatic biliary ducts in 89% (89% sensitivity, 100% specificity, (p <0.001); and delayed enhancement of the intrahepatic biliary duct in 55, 6% (56% sensitivity, 100% specificity).

Conclusion

Delayed gadolinium-enhancement of the gallbladder w all and extrahepatic bile duct showed remarkable sensitivit y and specificity for PSC. These biomarkers may potentially increase MRI diagn ostic accuracy in high- risk IBD patients.

Limitations

The main limitation is the small number of patients . Funding for this study: No funding Ethics committee - additional information: The study was approved by the ethics commitee of our hospital. Author Disclosures: Vincenzo Cardinale: Nothing to disclose Alessandra Valenti: Nothing to disclose Silvio Veraldi: Nothing to disclose Elio Damato: Nothing to disclose Carlo Catalano: Nothing to disclose Ludovica Busato: Nothing to disclose Francesca Maccioni: Nothing to disclose A nomogram model based on Combi-Elastography for pr eoperative differential diagnosis of biliary atresia *J. Chen*, F. Xu, Y. Gao, M. Yu, Y. Tang; Chongqing /CN ([email protected]) Purpose or Learning Objective: This study constructs a nomogram prediction model based on combi-elastography indexes and labor atory indicators for the differential diagnosis of biliary atresia (BA) and other cholestatic liver diseases (non-BA), with a view to recognizing BA at an early stage and carrying out treatment in time.

Methods

or Background: A total of 111 children aged < 180 d with cholestat ic hepatitis are included in the study, 75 in the BA g roup and 36 in the non-BA group. Conventional ultrasound, combi-elastography, and laboratory tests are performed on each patient before pathologic biopsy. The variables are selected through logistics regression to construct a nomogram model, and the effectiveness of the model is evaluated.

Results

or Findings: Multifactorial logistic regression analysis shows t hat gamma-glutamyl transferase (GGT) , total bilirubin (TBIL) , and liver fibrosis- related F index (FI) of the combi-elastography inde x could be used as independent predictors to differentiate BA from oth er causes of cholestasis. A nomogram model of these three indexes is constructe d which shows better Friday Abstract-based Programme 173 performance, with an area under the operating chara cteristic curve (AUC) of 0.887 (p < 0.001), sensitivity of 83.3%, and specif icity of 81.3%. The internal validation of the model is performed using 1,000 bo otstrap resamples and Decision curve analysis indicates that this model h ad a better diagnostic efficacy and accuracy.

Conclusion

The nomogram model based on combi-elastography inde xes and laboratory indicators has certain value in differen tiating BA from other cholestatic liver diseases.

Limitations

First, our sample size was small and it was a singl e-center study.Futher, we can expand the sample size to veri fy the validity of the nomogram model and provide more valuable insights f or clinicians. Second, additional indicators can be included in the future , such as MMP-7 to improve the diagnostic efficacy of the nomogram model. Funding for this study: Chongging Municipal Science and Health Joint Medical Research Project(2024MSXM050) Ethics committee - additional information: This study was approved by the Institutional Review Board of the Children's Hospit al of Chongqing Medical University (ethical approval number 2024-216), and all examinations and surveys were conducted after obtaining consent from the parents. Author Disclosures: Fenglin Xu: Nothing to disclose Yi Tang: Nothing to disclose Jingyu Chen: Nothing to disclose Yang Gao: Nothing to disclose Mingzhu Yu: Nothing to disclose Ultrasound spleen stiffness as a marker of portal v ein anastomotic stenosis following paediatric liver transplantation : correlation with transhepatic portal venography *D. C. Missud*, S. Le Cam, I. Mannes, M. Duché, S. Franchi-Abella; Le Kremlin-Bicêtre/FR ([email protected]) Purpose or Learning Objective: Spleen stiffness measurement (SSM) is known as a biomarker of significant portal hyperten sion in chronic liver disease, but has been poorly studied in the context of liver transplantation (LT). SSM may be particularly interesting to evaluate por tal vein anastomotic stenosis, a common and serious complication of LT.

Methods

or Background: Retrospective study including all portal vein stenoses assessed with transhepatic portal venograp hy (TPV) among paediatric liver recipients between 2015 and 2024. Diagnostic performance of SSM for the diagnosis of portal vein stenosis prior to TPV was evaluated.

Results

or Findings: 36 children who underwent 58 TPV (stenosis group) were included, with a median age at TPV of 3.0 year s. Findings were compared to those of 58 randomly selected paediatri c liver recipients without portal vein complication (control group). In the st enosis group, there were 17 mild, 20 moderate, and 21 severe stenoses. SSM was significantly increased in the moderate and severe groups vs no or mild ste nosis groups (p < 0.005). The Area Under the Curve was 0.96 for significant v s no to mild stenosis. A SSM cut-off of 33.7 kPa led to a sensitivity of 0.9 76 and a specificity of 0.840 for the diagnosis of significant stenosis.

Conclusion

SSM correlates very well with transhepatic portal v enography when there is a suspicion of significant portal vei n anastomotic stenosis following LT in paediatric patients. Further studie s may demonstrate that SSM is a good biomarker of portal hypertension followin g LT in paediatric patients, regardless of the aetiology.

Limitations

This is a retrospective study which doesn't have th e strength of a prospective clilnical trial. This is a series of on ly 58 TPVs, but it is nonetheless one the largest series ever published regarding pae diatric liver recipients with portal vein complications. Funding for this study: No external funding Ethics committee - additional information: This is a usual care clinical retrospective study, under review by the local ethi cs committee. Author Disclosures: Inès Mannes: Nothing to disclose David Charles Missud: Nothing to disclose Mathieu Duché: Nothing to disclose Solène Le Cam: Nothing to disclose Stéphanie Franchi-Abella: Nothing to disclose Evaluation of the Validity of Image-Defined Risk Fa ctors (IDRFs) in Abdominal Neuroblastoma *Z. Can Beyoğlu*, N. G. Akyel, E. Arslantaş, T. Banaz, M. Söyleyici, E. Ayaz, S. Akpınar Tekgündüz; Istanbul/TR ([email protected]) Purpose or Learning Objective: The presence of image-defined risk factors (IDRFs) in neuroblastoma plays a crucial role in cl inical decision-making, particularly when choosing between primary tumor re section and neoadjuvant chemotherapy. This study aims to evaluate how the p resence of IDRFs influences the surgical outcomes of patients with a bdominal neuroblastoma, specifically focusing on complete tumor resection a nd the likelihood of recurrence within one year.

Methods

or Background: This retrospective study involved 60 patients diagnosed with abdominal neuroblastoma. Of these, 3 0 patients had tumors with IDRFs present, while the remaining 30 patients had no IDRFs identified. The recurrence rates and surgical outcomes were ana lyzed over a one-year follow-up period. Factors such as tumor residuals a nd overall recurrence rates were examined to assess the impact of IDRFs on surg ical success and prognosis.

Results

or Findings: A total of 60 patients were included in the analysi s. Among the 30 patients with IDRFs, 9 patients showed no signs of recurrence within one year, while 12 patients experienced tumo r recurrence, and 9 had residual tumor tissue post-surgery. In the group of 30 patients without IDRFs, 22 patients did not experience recurrence, 5 patien ts had recurrences, and 3 had residual tumor tissue after surgery. These find ings suggest that the presence of IDRFs significantly impacts the likelih ood of recurrence and surgical success.

Conclusion

The study found that the presence of IDRFs in abdom inal neuroblastoma patients is an important predictor of recurrence. Patients with IDRFs had a lower non-recurrence rate (30%) compare d to those without IDRFs (73%). Therefore, IDRF presence can be a key factor in determining surgical approach and prognosis.

Limitations

The study is limited by the small sample size and v ariability in imaging techniques used during follow-up, which cou ld affect the consistency of the results. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was approved by the Ethics Committee of Başakşehir Çam and Sakura City Hospital, Istanbul/Turkey Author Disclosures: Merve Söyleyici: Nothing to disclose Sibel Akpınar Tekgündüz: Nothing to disclose Esra Arslantaş: Nothing to disclose Nazli Gülsüm Akyel: Nothing to disclose Tuba Banaz: Nothing to disclose Zeki Can Beyoğlu: Nothing to disclose Ercan Ayaz: Nothing to disclose Validation of a New Scoring System for Residual Tum or Assessment After Surgery in Pediatric Neuroblastoma: Prelimina ry Results *J. F. Schäfer*¹, J. Spogis¹, J. Fuchs¹, B. Hero², T. Simon², A. Eggert³, M. Müller⁴, S. Warmann³, B. Timmermann⁵; ¹Tübingen/DE, ²Cologne/DE, ³Berlin/DE, ⁴Heidelberg/DE, ⁵Essen/DE ([email protected]) Purpose or Learning Objective: The SIOPEN HR-NBL2 protocol for the treatment of high-risk neuroblastoma (NB) is a mult inational trial with radiation randomization for patients with macroscopic residua l tumors after induction chemotherapy and tumor resection. However, precise definitions of residual tumor on cross-sectional imaging are not available yet. This study aims to validate a newly developed scoring system for asses sing residual tumors proposed by the German Neuroblastoma registry.

Methods

or Background: Patients treated according to the GPOH NB guidelines and irradiated at the West German Proton Therapy Center Essen were retrospectively included if pre-/post-operativ e and current MRI were available at the time of radiotherapy. The score is based on MRI (tumor size and diffusion restriction), mIBG uptake, and the su rgical report, assigning a lesion score from 1 to 3 for each point in time. An onymized imaging data were uploaded to a browser-based imaging platform (mRay, Germany) for multi- reader, multi-institutional evaluation.

Results

or Findings: A total of 15 patients (mean age 5.1y; SD 2.1y) wit h 45 MRIs were assessed by two experienced readers (R1/R 2). The mean preoperative tumor volume was 100 ml (range 3–344 m l). Image-defined risk factors were identified in all patients except one. Based on the surgical reports, complete macroscopic resection was achieved in 9 pa tients, while imaging analysis by R1 and R2 confirmed complete resection in 4 cases. On a lesion- based analysis, surgery identified 6 residual lesio ns in 6 patients, while R1 and R2 identified 25 lesions in 11 patients (mean size: 13 mm, range: 4–32 mm). The inter-reader agreement for MRI scoring was exce llent (ICC 0.87; 95% CI: 072–0.94). Friday Abstract-based Programme 174

Conclusion

Regarding MRI findings, the newly developed scoring system for residual tumors in pediatric neuroblastoma is feasi ble. Further validation through multi-institutional, platform-based reading s is planned.

Limitations

Preliminary data Funding for this study: No Funding. Ethics committee - additional information: University of Essen, Germany Author Disclosures: Angelika Eggert: Nothing to disclose Beate Timmermann: Nothing to disclose Thortsen Simon: Nothing to disclose Jörg Fuchs: Nothing to disclose Jürgen F Schäfer: Nothing to disclose Barbara Hero: Nothing to disclose Steven Warmann: Nothing to disclose Michael Müller: CEO: providing the research platfor m Jakob Spogis: Nothing to disclose Imaging predictors of rupture in pediatric solid tu mors *G. G. Koodaly*, V. Smriti, A. D. Baheti, S. Kulkar ni, N. Shetty, K. B. Gala, M. Ramadwar, S. Quereshi, G. Chinnaswamy; Navi Mumb ai/IN ([email protected]) Purpose or Learning Objective: Tumor rupture poses as a life-threatening complication in pediatric solid tumors, and require s a high-risk protocol management. This study aims to identify the potenti al imaging predictors for tumor rupture.

Methods

or Background: The clinical data of children with pediatric solid tumor rupture at our institution from January 2021 to June 2024 were reviewed retrospectively.

Results

or Findings: Total of 22 cases, which comprised of 14 hepatoblastoma, 6 Wilms’ tumor, 1 neuroblastoma, an d 1 Ewing sarcoma were analyzed. Patients were aged between 2 to 10 years (median 4.5 years). 17 patients were treatment naïve and 5 on chemotherapy . Patients commonly presented with abdominal pain, distension, nausea, vomiting, and signs of shock due to significant drop in hemoglobin. Imagin g via ultrasound and CT scans revealed tumors, with a notable correlation b etween tumor rupture and factors such as high PRETEXT scores, tumor sizes ov er 10 cm, a greater percentage of necrotic component, intratumoral blee d and hyperdense ascitic fluid (>25 Hounsfield units). Among the hepatoblast oma cases, 71% had PRETEXT III and above. At diagnosis, 64% of the pat ients had hemoglobin levels ≤ 8 g/L, with 18% at ≤ 6 g/L. Seven patients required angioembolization, 2 underwent surgery and rest were treated conservat ively. Unfortunately, one patient developed tumor lysis syndrome and 3 succum bed to tumor rupture.

Conclusion

This study identifies the imaging predictors of tum or rupture in pediatric solid tumors and associated risk factors, such as chemotherapy, larger size of tumor and other high-risk factors.

Limitations

Retrospective study. Its prevalence in cohort is no t known. Funding for this study: No funding was received for this study. Ethics committee - additional information: Retrospective study Author Disclosures: Genesis Giddo Koodaly: Nothing to disclose Suyash Kulkarni: Nothing to disclose Sajid Quereshi: Nothing to disclose Mukta Ramadwar: Nothing to disclose Nitin Shetty: Nothing to disclose Kunal Bharat Gala: Nothing to disclose Akshay Dwarka Baheti: Nothing to disclose Vasundhara Smriti: Nothing to disclose Girish Chinnaswamy: Nothing to disclose The Risk of Pediatric and Adolescent Hematologic Ma lignancies Associated with Medical Imaging (RIC) *R. Smith-Bindman*¹, S. Albers², M. Kwan³, W. Bolch ⁴, E. Bowles⁵, C. Stewart¹, R. Greenlee⁶, J. Pole⁷, D. L. Miglioretti²; ¹San Francisco, CA/US, ²Davis, CA/US, ³Oakland, CA/US, ⁴Gainesville, FL/US, ⁵Seattle, WA/US, ⁶Marshfield, WI/US, ⁷Brisbane/AU ([email protected]) Purpose or Learning Objective: Risks of hematologic malignancies associated with medical imaging ionizing radiation exposure have not been evaluated in the U.S. or Canada.

Methods

or Background: This retrospective cohort study followed 3,724,622 children born at one of 6 U.S. healthcare systems o r in Ontario, Canada from 1/1/1996 to 4/30/2016 from birth until the earliest of a cancer diagnosis, death, emigration from Ontario, 6 months after disenrollme nt from healthcare system, age 21, or 12/31/2017. Active bone marrow radiation doses from medical imaging examinations were estimated. Hazards ratios (HR) and relative risks (RR) of hematologic malignancies associated with cu mulative radiation exposure were estimated.

Results

or Findings: A total of 2,961 hematologic malignancies were diagnosed during 35,735,719 person-years of follow up, including lymphoid malignancies (n=2,349, 79.3% of malignancies), myel oid or acute leukemia (myeloid, n=460, 15.5%); and histiocytic and dendri tic cell malignancies (H&D, n=129, 5.1%). Malignancy risk increased with cumula tive dose (p<0.0001); e.g., risk was 1.70 times higher among children wit h a cumulative dose of 15 to < 20 mGy vs. <1mGy (95%CI=1.27-2.28). The risk of m alignancy was 3.7 times higher (95%CI=2.82-4.71) for children with a cumulative exposure of 100 mGy vs. no exposure and was significantly elevated for cancer subtypes. RRs decreased with increasing time since exposure and i ncrease with age at exposure and attained age. We estimate 27 excess he matologic malignancies by age 21 per 10,000 children with a cumulative exp osure of 30 mGy or higher vs. <1 mGy, equivalent to average dose of approxima tely 2 head CTs.

Conclusion

Children and adolescents who undergo radiation-base d medical imaging are at a small, but significant increased r isk of hematologic malignancy.

Limitations

While reverse causation is a potential limitation, analyses of clinical indications confirmed symptoms related to hematologic malignancy were rare in included studies. Funding for this study: US National Institutes of Health, National Cancer Institute R01CA185687, R50CA211115 Ethics committee - additional information: The requirement for individual informed consent was waived for the study Author Disclosures: Erin Bowles: Nothing to disclose Marilyn Kwan: Nothing to disclose Jason Pole: Nothing to disclose Rebecca Smith-Bindman: Nothing to disclose Carly Stewart: Nothing to disclose Diana L Miglioretti: Nothing to disclose Wesley Bolch: Nothing to disclose Susan Albers: Nothing to disclose Robert Greenlee: Nothing to disclose Ultra-low dose CT for non-accidental injury A. A. Mohammed, M. F. Mcentee, A. England, N. Moore , E. K. Mahon, M. Maher, *R. Young*; Cork/IE Purpose or Learning Objective: The aim of this study is to compare two whole-body CT protocols for SPA and to assess wheth er the ULD protocol can provide diagnostic images of sufficient quality com pared to the standard dose (STD).

Methods

or Background: In this cross-sectional study, two sets of images o f a newborn whole-body anthropomorphic phantom were a cquired using different protocols, one with STD and the other wit h ULD protocol. The effective dose (ED) of both protocols was calculate d using the Monte Carlo dose simulation approach. The image quality arising from both protocols was then assessed at the ECR 2024 Congress using a four -section questionnaire. The questionnaire included demographic information, a comparison of the visualization of different bony anatomical structur es, and confidence in diagnosis using either protocol. The Wilcoxon signe d-rank test was used to evaluate the significant differences between STD an d ULD image quality scores. VGC analyser was used for image quality rat ing and comparison.

Results

or Findings: 46 participants were included in this study. For al l body parts, STD showed significantly higher image qualit y than ULD (AUCVGC=0.75). 76% of the participants were confide nt to use the STD protocol for SPA diagnosis, whereas, 41% were confi dent to use the ULD protocol. The percentage effective dose difference between protocols was 93.5%(STD=0.56 mSv vs. ULD=0.04mSv) and most of the participants underestimated the dose reduction.

Conclusion

This study successfully compared the STD and ULD wh ole-body CT in phantom and shows ULD CT is a promising techn ique which may compete with digital radiography for SPA diagnosis.

Limitations

The limitations of the study are: 1-the use of phan tom involves no patient movement experienced, and no pathologies. 2 -sampling bias was due to the observers as only people attending the ECR 2 024 could participate. 3- only compares two whole-body CT protocols. Funding for this study: Taif University Ethics committee - additional information: The ethical approval provided by the University College Cork Author Disclosures: Mark F. Mcentee: Nothing to disclose Niamh Moore: Nothing to disclose Rena Young: Nothing to disclose Ahmed Abdulahad Mohammed: Nothing to disclose Andrew England: Nothing to disclose Eimear Kate Mahon: Nothing to disclose Michael Maher: Nothing to disclose Friday Abstract-based Programme 175 Deep Learning-based Detection of Pediatric Bone Tum ors Using X-ray Imaging S. Consalvo, A. Curto Vilalta, *A. W. Marka*, S. Br eden, B. Schlossmacher, C. Eisfeld, D. Rückert, R. Von Eisenhart-Rothe, F. Hinterwimmer; Munich/DE ([email protected]) Purpose or Learning Objective: To address the challenge in musculoskeletal radiology of early detection of bone tumours in chi ldren with x-ray imaging. While machine learning (ML) has shown proficiency i n differentiating tumour entities, a critical gap remains in initial tumour detection, particularly for non- oncology-trained professionals and general practiti oners. Paediatric tumours are often incidentally discovered, underscoring the need for more sophisticated tools.

Methods

or Background: This retrospective study utilized X-ray data from a diverse cohort of paediatric patients from our loca l musculoskeletal tumour database. The dataset comprised 817 images (567 pat hological and 250 healthy) from 511 patients, including ten benign, i ntermediate and malignant tumour entities. We employed the ResNet18 architect ure for classification, supported by cross-validation techniques and excess ive data augmentation strategies. Our methodology focused on enhancing th e ML system's ability to generalize across various clinical scenarios.

Results

or Findings: The ML model demonstrated high performance with an accuracy of 96.39%, a sensitivity of 96.0%, and a s pecificity of 96.0% into “tumour” and “no tumour”. The variance in cross-val idation splits was 0.05, 0.10, and 0.13, respectively, indicating stable res ults across different test sets. These metrics reflect the model's reliability and p otential effectiveness in clinical settings.

Conclusion

Current ML applications in orthopaedic oncology are progressing yet remain insufficiently performant for widespread clinical use. However, our findings underscore the potential of ML tools in ai ding both young professionals and general practitioners. Additionally, future adv ancements should focus on multimodal approaches that incorporate not only X-r ay data but also MRI and, crucially, clinical data. Integrating these diverse data sources will enhance the performance and applicability of ML in diagnosing a nd managing paediatric bone tumours, offering a more holistic and effectiv e approach to patient care.

Limitations

90% Monocentric Data. Funding for this study: Nemetschek Innovation Foundation and Bavarian Ministry of Science and the Arts Ethics committee - additional information: Klinikum rechts der Isar, Technical University of Munich. Author Disclosures: Carolin Eisfeld: Nothing to disclose Anna Curto Vilalta: Nothing to disclose Sebastian Breden: Nothing to disclose Alexander Wolfgang Marka: Nothing to disclose Florian Hinterwimmer: Nothing to disclose Sarah Consalvo: Nothing to disclose Daniel Rückert: Nothing to disclose Rüdiger Von Eisenhart-Rothe: Nothing to disclose Benjamin Schlossmacher: Nothing to disclose 14:00-15:30 Research Stage 4 Research Presentation Session: Neuro RPS 1511 Insights into brain tumours: from visible to invisible and back again Moderator C. Eraslan; Izmir/TR ([email protected]) Improved Brain Tumor visualization with 3T Stack-of -Stars Echo Unbalanced T1 Relaxation-Enhanced Steady-State MRI – A Two Center Clinical Study *A. Toth*¹, R. Edelman², J. A. Chetta¹, J. Joyce¹, M. V. Spampinato¹, R. Zi³, K. T. Block³, A. Varga-Szemes¹; ¹Charleston, SC/US, ²Evanston, IL/US, ³New York, NY/US ([email protected]) Purpose or Learning Objective: The novel stack-of-stars echo unbalanced T1 relaxation-enhanced steady-state (SOS echo-uT1RE SS) sequence aims to provide improved motion robustness and enhanced dar k blood contrast, and to improve the visualization of small metastases and l ow enhancing lesions. This study compared the image quality and diagnostic uti lity of SOS echo-uT1RESS with the widely used magnetization-prepared rapid a cquisition gradient-echo (MPRAGE) sequence in brain tumor imaging.

Methods

or Background: This two-center prospective study involved 25 adults with known brain tumors (n= 5 intra-axial pr imary brain tumors; n= 11 intra-axial brain metastases; n= 9 extra-axial brai n tumors). Each participant underwent 3T contrast enhanced MRI of the brain wit h both standard MPRAGE and prototype SOS echo-uT1RESS sequences. Co ntrast-to-noise ratio (CNR) and tumor-to-brain contrast were quanti tatively analyzed. Image quality, lesion conspicuity, and image artifacts we re scored on a 4-point Likert scale. Diagnostic performance and assessment of the vascular and dural involvement were compared side-by-side by 2 readers .

Results

or Findings: There was no significant difference in CNR between MPRAGE and SOS echo-uT1RESS (27.0 ± 19.2 vs. 26.5 ± 14.9, respectively; p = 0.84). SOS echo-uT1RESS demonstrated a 1.6-fold improvement in tumor- to-brain contrast compared with MPRAGE (0.7 ± 0.4 v s. 0.4 ± 0.3, respectively; p < 0.001). Image quality and artifacts were simila r for both sequences, while SOS echo-uT1RESS showed improved lesion conspicuity , diagnostic performance and enhanced detection of vascular and dural invasion.

Conclusion

SOS echo-uT1RESS showed promising results for post- contrast evaluation of brain tumors on 3T MRI. This techniqu e enhanced lesions visibility, achieving approximately 1.6-fold improv ement in tumor-to-brain contrast compared to MPRAGE. It offered superior di agnostic performance and improved detection of vascular and dural involvemen t.

Limitations

The limitations of the study are the relatively sma ll patient cohort and that quantitative measurements were performed b y a single observer. Funding for this study: Funding was provided by NIH HHS United States (1R01CA263091 and 1R21CA273280). Ethics committee - additional information: The study was approved by the Institutional Review Board (Pro00128013)). Author Disclosures: Kai Tobias Block: Nothing to disclose Robert Edelman: Nothing to disclose Ruoxun Zi: Nothing to disclose Adrienn Toth: Nothing to disclose Jennifer Joyce: Nothing to disclose Justin A Chetta: Nothing to disclose Akos Varga-Szemes: Nothing to disclose M. Vittoria Spampinato: Nothing to disclose Multiparametric MRI‑based radiomics with interpretable machine learning for predicting progesterone receptor expression in meningioma: A multicenter study *G. Lin*, W. Chen, J. Ji; Lishui/CN Purpose or Learning Objective: This study aimed to develop and validate an interpretable machine learning-based prediction mod el for preoperatively predicting progesterone receptor (PR) expression in meningioma patients using multiparametric magnetic resonance imaging (M RI).

Methods

or Background: The study retrospectively enrolled 739 patients wit h pathologically confirmed meningioma from three medi cal centers, dividing them into four cohorts: training (n = 294), internal tes t (n = 126), external test 1 (n = 217), and external test 2 (n = 102). Radiomics char acteristics were derived from T2-weighted and contrast-enhanced T1-weighted MRI images, followed by feature selection. A machine learning-based comb ined model was developed by incorporating radiomics scores (rad-sc ores) from the optimal radiomics model along with clinical predictors. The Shapley additive explanation (SHAP) method was employed to visually represent the process of making predictions. The prognostic value of the mod el was evaluated using Kaplan-Meier survival analysis.

Results

or Findings: Among the 739 patients, 299 (40.5%) had negative PR expression confirmed by pathology. Twelve radiomics features derived from multiparametric MRI were selected to build the radi omics model. Tumor location and enhancement pattern were identified as key clinical predictors and were combined with rad-scores to create a combined model utilizing the extreme gradient boosting (XGBoost) algorithm. The combined model demonstrated strong accuracy and robustness, with a rea under the curve values of 0.907, 0.827, 0.846, and 0.807 across tra ining, internal test, external test 1, and external test 2 cohorts, respectively. The survival analysis indicated that the combined model was able to effectively cat egorize patients based on recurrence outcomes.

Conclusion

The XGBoost combined model, utilizing multiparametr ic MRI, shows promise for predicting PR expression in menin gioma patients. The SHAP visualization enhances the model’s clinical ap plicability.

Limitations

As a retrospective study, it is susceptible to info rmation selection bias. Funding for this study: This work was supported by the Key Project of Joint Construction by Provincial and Ministerial Authorit ies (Grant No.WKJ-ZJ-2452 to Minjiang Chen), Medical and Health General Proje ct of Zhejiang Province (Grant No. 2023KY425 to Guihan Lin, Grant No. 2024K Y562 to Shuiwei Xia), Friday Abstract-based Programme 176 and Medical and Health Youth Innovation Project of Zhejiang Province (Grant No. 2023RC115 to Weiyue Chen). Ethics committee - additional information: All procedures performed in studies involving human participants were in accord ance with the ethical standards of the institutional and/or national rese arch committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the Instituti onal Review Board and Human Ethics Committee of the Fifth Affiliated Hosp ital of Wenzhou Medical University (2024-336), the Sixth Affiliated Hospita l of Wenzhou Medical University, and the Third Affiliated Hospital of We nzhou Medical University, with the requirement for patient informed consent b eing waived due to its retrospective nature. All patients’ information was anonymized prior to the analysis. Author Disclosures: Jiansong Ji: Nothing to disclose Weiyue Chen: Nothing to disclose Guihan Lin: Nothing to disclose Histogram Analysis in Predicting the Intracranial M eningioma Grading Based on Amide Proton Transfer-Weighted Imaging *Y. H. Lee*, B-H. Kim, M. Kim, S-D. Kim; Ansan-Si/K R ([email protected]) Purpose or Learning Objective: To determine whether amide proton transfer- weighted (APTW) histogram analysis is useful for pr edicting the grade of meningioma

Methods

or Background: We retrospectively enrolled a total of 48 patients (M:F=16:32, mean age: 60.0 years; grade 1:grade 2/3 =36:12) with pathologically proven intracranial meningioma who u nderwent mDIXON 3D- APT sequence of the fast spin echo method in additi on to conventional 3T MR protocols prior to surgical resection. From the rep resentative APTW images of the tumor registered with gadolinium-enhanced T1 im ages, the following parameters of each histogram were obtained: every 5 intervals from 5th to 95th percentile, mean, median, maximum, minimum, standar d deviation, kurtosis and skewness. The diagnostic performance of each AP TW histogram parameter for differentiating grade 2/3 from grade 1 intracranial meningioma were evaluated by drawing the receiver operating ch aracteristic (ROC) curves and calculating the cut-off values

Results

or Findings: Among all histogram parameters, only maximum, standard deviation and 75th,80th,85th,90th, and 95t h percentiles for APTW signal in grade 2/3 were significantly higher than those of grade 1 (p<0.05). According to the ROC curve comparison analysis, the areas under the curves of 75th,80th,90th,95th percentile, maximum and stan dard deviation to discriminate grade 2/3 from grade 1 were 0.706,0.71 3,0.722,0.734, 0.738, 0.722, and 0.718, respectively.

Conclusion

Histogram analysis of APTW imaging can be used in c linical practice for grading of intracranial meningiomas

Limitations

1.ROI-dependency, 2.small number of participants Funding for this study: None Ethics committee - additional information: None Author Disclosures: Young Hen Lee: Nothing to disclose Baek-Hyun Kim: Nothing to disclose Sang-Dae Kim: Nothing to disclose Myungji Kim: Nothing to disclose Prognostic utility of intratumoral susceptibility s ignals in adult diffuse gliomas: a radiopathological study *J. I. Tudela Martínez*, V. Vázquez Sáez; Murcia/ES Purpose or Learning Objective: Intratumoral susceptibility signals (ITSS) are promising radiological markers for assessing diffus e gliomas. This study evaluates the relationship between ITSS grading and key radiological and histopathological prognostic factors in adult diffu se gliomas.

Methods

or Background: Between January 1st, 2022, and April 30th, 2024, we selected 99 patients diagnosed with adult diffus e glioma who met the following criteria: age over 18 years, MRI scans al lowing ITSS quantification and confirmed pathological diagnosis with available molecular testing. Radiological variables included tumor volume, subve ntricular zone involvement and relative cerebral blood volume (rCBV) on MRI pe rfusion. Histopathological features examined were WHO-2021 grade, Ki-67 index, mitotic count, necrosis, microvascular proliferation, and key prog nostic mutations (IDH, p53, ATRX, and CDKN2A/B). Spearman’s correlation and chi -square tests were used for quantitative and qualitative variables, re spectively. Multiple logistic regression models were developed to predict WHO tum or grade, categorized as low (1-2) or high (3-4), based on ITSS grade, tu mor volume, and rCBV.

Results

or Findings: ITSS grades 0-1 were more common in oligodendrogliomas and astrocytomas, while grades 2 -3 were linked to glioblastomas (p<0,001). ITSS grade positively corr elated with rCBV, tumor volume, WHO grade, mitotic count, and Ki-67 index ( p<0,001). Higher ITSS grades also showed increased necrosis and microvasc ular proliferation (p<0,001). IDH mutations and 1p/19q co-deletions we re more prevalent in grades 0-1 (p<0,001 and p=0,001, respectively), whi le CDKN2A/B alterations correlated with grades 2-3 (p=0,02). Regression mod els showed AUCs of 0,937 and 0,960 for ITSS combined with rCBV and tum or volume, respectively (p=0,000).

Conclusion

ITSS represent valuable biomarkers for assesing dif fuse gliomas, offering diagnostic and prognostic insights that ca n guide clinical decision- making. Additionally, combining ITSS with MRI-rCBV and tumor volume enhances predictive capacity of these radiological parameters.

Limitations

ITSS grading remains semi-quantitative; further stu dies should focus on fully quantifying ITSS data. Funding for this study: None Ethics committee - additional information: We consulted with the ethics and research committee regarding the need for approval for the study. They confirmed that, due to its observational nature, su ch approval is not required. Author Disclosures: Victoria Vázquez Sáez: Nothing to disclose Jose Ignacio Tudela Martínez: Nothing to disclose Relationship between Whole-tumor MRI-based Fractal Analysis and Molecular Features in IDH-wildtype Glioblastoma *B. Zhang*, J. Zhou; Lanzhou/CN Purpose or Learning Objective: Molecular mechanisms and specific genes involved in the growth of Glioblastoma (GBM) are im portant factor in deciding the treatment strategy. In this study, we aimed to non-invasively explore the relationship between whole-tumor MRI-based fractal features and molecular features of GBM.

Methods

or Background: The clinical and imaging data of 104 patients with IDH-wildtype GBM at our hospital between November 2 018 and June 2024 were retrospectively analyzed. The molecular featur es of GBM were collected by molecular sequencing and immunohistochemical met hod, including MGMT promoter methylation, 1p/19q-codeleted, TERT promot er mutation, Ki67, and P53 status. The volume of interest of whole tumor w as manually segmented slice-by-slice using ITK-SNAP software. Fractal fea tures of whole-tumor in contrast-enhanced T1-weighted imaging were extracte d using Image J software. As many as 24 fractal features (fractal d imensions and lacunarity) were generated within each volume of interest. Corr elation analyses were performed using Spearman correlation analysis. Logi stic regression was used to build prediction models.

Results

or Findings: The L1 and L4 were positively correlated with 1p/19 q- codeleted (correlation coefcient: 0.213 and 0.212). The L5 was positively correlated with TERT promoter mutation (correlation coefcient: 0.326). The L1, L4, and L9 were negatively correlated with TERT pro moter mutation (correlation coefcient: -0.251, -0.310, -0.196, res pectively). The L1, L3, and L4 were positively correlated with Ki-67 proliferation index (correlation coefcient: 0.226, 0.200, 0.241, respectively). MGMT promoter m ethylation and P53 had no correlation with fractal features. The AUC of fr actal features predicting 1p/19q-codeleted was 0.677 and predicting TERT prom oter mutation was 0.755.

Conclusion

The fractal features were correlated with 1p/19q-co deleted, TERT promoter mutation, and Ki67 status in IDH-wildtype GBM. Fractal features can be used as non-invasive quantitative parameters to predict the molecular features of GBM.

Limitations

Not Funding for this study: This study was supported by the National Natural Science Foundation of China (grant no. 82071872 and 82371914), the Science and Technology Program of Gansu Province (grant no. 21YF5FA123 and 21JR11RA105), and the China International Medical F oundation (grant no. Z- 2014-07-2101). Ethics committee - additional information: This study was approved by the Medical Ethics Committee of the Second Hospital of Lanzhou University (approval number : 2020A-070) and informed consent was waived. Author Disclosures: Junlin Zhou: Nothing to disclose Bin Zhang: Nothing to disclose T1 Curves in the evaluation of radionecrosis or dis ease recurrence *C. Monopoli*, A. Romano, G. De Rosa, A. Romano, G. Moltoni, A. M. Ascolese, G. Capriotti, A. Bozzao; Rome/IT ([email protected]) Purpose or Learning Objective: Radiation treatment of brain metastases creates diagnostic doubts in the differential diagn osis between disease progression or radionecrosis induced by radiosurger y. MRI with the administration of contrast medium doesn't offer the possibility of reliably distinguishing the two pathological entities. The a im of our study is to verify the presence of radionecrosis or disease recurrence thr ough the evaluation of T1 enhancement curves. Friday Abstract-based Programme 177

Methods

or Background: 40 brain metastases undergone to radiosurgery were evaluated (32 from lung, 4 from breast, 2 from melanoma and 2 colorectal). All patients underwent MRI examination with dynamic T1 acquisitions with contrast medium. For each lesions the T1 enhancement curve was extracted by positioning a region of interest c orresponding to the solid component, excluding the necrotic areas. All patien ts underwent a PET-DOPA study and the result of the examination was used as the gold standard to distinguish radionecrosis from disease progression. The PET investigations identifies three stages of the disease; radionecros is (rSUV1.9)

Results

or Findings: 4 types of T1 enhancement curves have been identifi ed (A-D). Curve A showed constant growth over time; cu rve B showed faster growth in its initial portion and constant growth o ver time; curve C showed rapid initial growth and a final plateau; Curve D s howed rapid growth and rapid final washout. Of the 40 lesions, 13 showed uptake compatible with radionecrosis, 12 with a mixed picture and 15 with disease progression. Curves A and B corresponded to radionecrosis or mix ed in 90% of cases, curves C and D corresponded to a progression of the disease in 95% of cases.

Conclusion

T1 enhancement curves allows to distinguish a condi tion of radionecrosis from a progression of the disease.

Limitations

Small enrolled population Funding for this study: Not funding received Ethics committee - additional information: Not applicable. Author Disclosures: Cristiana Monopoli: Nothing to disclose Anna Maria Ascolese: Nothing to disclose Andrea Romano: Nothing to disclose Gabriella Capriotti: Nothing to disclose Alessandro Bozzao: Nothing to disclose Allegra Romano: Nothing to disclose Giulia Moltoni: Nothing to disclose Giulia De Rosa: Nothing to disclose PET/MRI in brain primary and secondary tumors treat ed with radiochemotherapy: a radiomic-based analysis of bra in Perfusion MRI and 11C-Methionine PET images acquired by a integra ted hybrid system *E. Masiello*, M. Barbera, F. Fallanca, S. Paola, A . Castellano, A. Falini, N. E. Anzalone; Milan/IT ([email protected]) Purpose or Learning Objective: Perfusion-weighted MRI (PWI) and 11C- methionine PET (MET-PET) provide valuable hemodynam ic and metabolic insights for assessing brain tumors. This study aim ed to investigate the diagnostic role of PWI and MET-PET, using radiomic analysis, in distinguishing progression (PD), pseudoprogression (PsP), and radi onecrosis (RN) in patients with brain tumors treated with radiotherapy (RT) or radiochemotherapy.

Methods

or Background: Patients with primary and secondary brain neoplasms who developed post-treatment lesions of a t least 1 cm within the radiation field were retrospectively enrolled. All patients underwent simultaneous PET/MRI examinations according to a st andardized protocol. Radiomics features were extracted from the 3D-segme ntation of parametric maps, including relative cerebral blood volume (rCB V) from DSC, plasma volume (Vp) and vascular permeability (Ktrans) from DCE, relative cerebral blood flow (CBF) based on pseudo-Continuous Arteria l Spin Labeling (pCASL), and Standardized Uptake Value (SUV) from MET-PET. F or each lesion, imaging data were compared with outcomes based on R ANO criteria or histological examination.

Results

or Findings: A semi-automatic 3D segmentation of 52 lesions (23 PD and 29 RN) was performed using PMod software (v. 3. 7) to extract 263 radiomic features. After feature selection, CBF Max imum gray level and SUV Maximum gray level were the most correlated (p < 0. 001). In terms of accuracy, the highest area under the curve (AUC) fo r PET features was SUV Entropy, while Vp Mean showed the highest AUC for P WI features. SUV Entropy achieved the highest sensitivity for detect ing PD (95.65%), while rCBV Entropy demonstrated the highest specificity (79.31 %). PET and PWI parameters exhibited similar overall accuracy, rang ing from 71.15% to 76.92%.

Conclusion

Radiomics features from MET-PET and PWI demonstrate strong potential in accurately distinguishing PD and PsP f rom RN. Combining both modalities using radiomics enhances overall diagnos tic accuracy.

Limitations

Small and heterogeneous population Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is retrospective. Author Disclosures: Andrea Falini: Nothing to disclose Nicoletta Emanuela Anzalone: Nothing to disclose Edoardo Masiello: Nothing to disclose Scifo Paola: Nothing to disclose Antonella Castellano: Nothing to disclose Maurizio Barbera: Nothing to disclose Federico Fallanca: Nothing to disclose Multiparametric MRI-based clinical radiomics model for predicting TERTp genotype and overall survival in oligodendrogliomas *J. Zhao*¹, X. Ke¹, T. Gan¹, W. Hu¹, C. Xue², S. Li ³, Q. Zhou¹, J. Zhou¹; ¹Lanzhou/CN, ²Qingdao/CN, ³Chengdu/CN ([email protected]) Purpose or Learning Objective: To test the hypothesis that combining features from multiparametric MRI with clinically r elevant prognostic risk factors provides a more accurate prediction of TERTp genoty pe and overall survival(OS)in patients with oligodendrogliomas(OGS ).

Methods

or Background: Preoperative multiparametric MRI sequences (T1WI, T2WI, and CE-T1-3D) from 135 patients with O GS (grades 2 and 3) were collected and randomly divided into training ( n = 95) and validation (n = 40) sets. Radiomics features were extracted, and th e least absolute shrinkage and selection operator regression was used to selec t the most relevant features. Clinical relevant features identified thr ough univariate and multivariate logistic regression analyses were incorporated to e stablish a clinical radiomics model. This model was used to develop a predictive TERTp genotype nomogram. Kaplan–Meier curves were used to assess O S differences between TERTp groups; the log-rank test determined significance.

Results

or Findings: The T2WI-based clinical radiomics model demonstrate d superior performance in predicting the TERTp genoty pe, with mean area under the receiver operating characteristic curve (AUC) v alues of 0.90 (95% CI: 0.88, 0.92; P = 0.0002) in the training set and 0.83 (95% CI: 0.82, 0.85; P < 0.0001) in the validation set. The one year, two year and t hree year survival probability prediction models achieved AUC values of 0.85, 0.80 , and 0.79, respectively.

Conclusion

The multiparametric MRI-based clinical radiomics mo del provides the most accurate prediction of the TERTp genotype. Combined with clinical relevant prognostic risk factors, the prognostic mo del offers precise prediction of OS in patients with OGS.

Limitations

The limitation of this study was three-dimensional tumor segmentation was manually performed, future researc h should explore automated and efficient segmentation methods to red uce workload and simplify clinical application. Funding for this study: Funding were provided by National Natural Science Foundation of China (82071872, 82371914), Science a nd Technology Program Funding Project of Gansu Province ( 21JR7RA404). Ethics committee - additional information: The ethics committee notification can be found under the number 2021A-348. Author Disclosures: Shenglin Li: Nothing to disclose Wanjun Hu: Nothing to disclose Jun Zhao: Nothing to disclose Junlin Zhou: Nothing to disclose Caiqiang Xue: Nothing to disclose Qing Zhou: Nothing to disclose Xiaoai Ke: Nothing to disclose Tiejun Gan: Nothing to disclose Intraoperative and postoperative MRI detection of i schemia in brain tumor surgery: a retrospective study on predictive factors and ischemic evolution *M. R. López De La Torre Carretero*, C. Mbongo, P. Corral Alonso, D. A. Zambrano, Á. R. Cabrera Abud, C. D. Solano, J . M. Rodríguez Ortega, M. Macías de la Corte Hidalgo, M. Calvo Imirizaldu; Pamplona/ES ([email protected]) Purpose or Learning Objective: Intraoperative magnetic resonance imaging (iMRI) is an increasingly valuable tool in neurosur gical oncology, particularly for guiding tumour resection by real-time detection of residual tumour and margins, but also acute ischemic complications usin g diffusion weighted imaging (DWI). However, some studies suggest iMRI m ay underestimate ischemia compared to early postoperative MRI (epMRI ), highlighting the importance of thorough postoperative evaluation. Ou r study aimed to compare iMRI´s ability to detect ischemia against epMRI and late postoperative MRI (lpMRI), and to identify predictive factors for pos toperative ischemia in patients undergoing brain tumour resection.

Methods

or Background: This retrospective study included 106 patients undergoing brain tumor resection at our centre. iMR I, epMRI (5-7 days post- surgery), and lpMRI (30 days post-surgery) were per formed. Two radiologists analysed imaging to detect intraoperative ischemia (IOI), early postoperative ischemia (EPI), and late postoperative ischemia (LP I) using (DWI) and quantifying ischaemic volume on apparent diffusion coefficient (ADC) maps. Variables such as age, sex and tumour histology wer e recorded. Statistical analysis was conducted (StataNow 18.5) using McNema r tests, Pearson’s chi- squared test, and multivariate logistic regression.

Results

or Findings: McNemar test revealed that iMRI tends to underestimate ischaemia compared to epMRI (p= 0.003 9). Only tumour type was a significant predictor of EPI (p=0.041), with glioblastoma patients having lower probabilities of EPI compared to other tumour types (p=0.041). EPI and Friday Abstract-based Programme 178 LPI were significantly associated (p <0.001), indic ating ischaemia detected in epMRI didn’t progress in lpMRI.

Conclusion

Our results suggest iMRI may underestimate the isch emia compared to epMRI. Glioblastoma was associated with a lower risk of ischemia, highlighting the importance of personaliz ed surgical approach. The strong relationship between EPI and LPI emphasizes the need for close follow- up to monitor late complications.

Limitations

Retrospective study. Clinical outcomes could be useful (further studies) Funding for this study: No funding was provided for this study Ethics committee - additional information: Not applicable Author Disclosures: Carmen Mbongo: Nothing to disclose José Marlon Rodríguez Ortega: Nothing to disclose Paula Corral Alonso: Nothing to disclose Marta Macías de la Corte Hidalgo: Nothing to disclo se Manuel Rafael López De La Torre Carretero: Nothing to disclose Marta Calvo Imirizaldu: Nothing to disclose Daniel Alfonso Zambrano: Nothing to disclose Álvaro Rafael Cabrera Abud: Nothing to disclose Carlos Delgado Solano: Nothing to disclose Supratentorial Changes in Patients with Cerebellopo ntine Angle Tumors: A Comprehensive Morphologic Analysis *A. Çolakoğlu*, B. Genç, K. Aslan, L. Incesu; Samsun/TR ([email protected]) Purpose or Learning Objective: Neuroplasticity that develops in intracranial tumors may guide the management of post-surgical or radiotherapy treatment. To our knowledge, there is no study investigating m orphometric changes in the brain of patients with cerebellopontine angle tumor s. Our aim in this study is to investigate supratentorial morphometric changes in patients with cerebellopontine angle tumors.

Methods

or Background: The study included 29 patients with cerebellopontine angle tumors who had not yet recei ved any treatment, and 53 age- and sex-matched healthy controls. Voxel-based morphometry and surface-based morphometry analyses were performed u sing CAT12, running under SPM12, to examine gray matter volume changes and cortical thickness changes in these patients. A general linear model w as used for statistical analysis, and a p-value<0.05 with family-wise error (FWE) correction was considered statistically significant.

Results

or Findings: Our VBM results showed an increase in gray matter volume in the thalamus, ventral diencephalon, cingu late gyrus, precuneus, cuneus, superior parietal lobe, and parahippocampal gyrus in these patients (p<0.05 FWE). Our SBM results revealed an increase in cortical thickness in the right superior parietal and paracentral gyri in patients with cerebellopontine angle tumors (p<0.05 FWE).

Conclusion

Our study is the first to demonstrate an increase i n gray matter volume and cortical thickness in the supratentorial region of patients with cerebellopontine angle tumors. These findings may b e associated with neuroplastic changes in these patients.

Limitations

The limitations of the study are its single-center design, its retrospective nature, and the absence of neurocogni tive tests. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is approved by clinical studies ethics committee of Ondokuz Mayıs Universit y. The reference number is 2024090641. Author Disclosures: Bariş Genç: Nothing to disclose Lütfi Incesu: Nothing to disclose Arda Çolakoğlu: Nothing to disclose Kerim Aslan: Nothing to disclose 16:00-17:30 Research Stage 1 Research Presentation Session: Chest RPS 1604 Lung cancer screening and nodule assessment Moderator D. Shaham; Jerusalem/IL ([email protected]) Author Disclosures: Dorith Shaham: Consultant: AstraZeneca Automated triage to a virtual lung nodule clinic us ing Veye Lung Nodule volumetry could provide a cost saving compared to s tandard care *G. Dixon*¹, V. Dattani¹, M. Jordan², J. Rodrigues¹ ; ¹Bath/UK, ²Coventry/UK ([email protected]) Purpose or Learning Objective: Pulmonary nodules are frequently identified on non-screening CT chest scans. The management of pulmonary nodules in the UK is governed by the British Thoracic Society (BTS) Guidance 2015. Artificial intelligence can provide automated nodul e identification and volumetry. The clinical and cost effectiveness of i ncorporating automated nodule detection into patient pathways are currentl y unclear.

Methods

or Background: All CT scans including the thorax over one week in a single UK centre during March 2022 were identifie d and their original reports reviewed. CT scans were categorised as to whether t he original report gave a BTS recommendation or not. Currently, patients with out a BTS recommendation are assumed to undergo a full lung M DT discussion to determine the recommendation. A virtual nodule clin ic was modelled whereby all patients with >1 nodule of >80mm3 identified us ing Veye Lung Nodules who did not receive a BTS recommendation in the origina l CT report were referred to a virtual nodule clinic. The cost of this approa ch was compared to current practice.

Results

or Findings: 80/357 scans had > 1 pulmonary nodule >80mm3 identified using Veye Lung Nodules. 51/80 patients did not receive a BTS Guideline based recommendation in the original radi ologist report. A potential cost saving of £9.48 per case to review these cases in a virtual lung nodule clinic was calculated. This represents a cost savin g of £483.48 during the study period (one week) and a potential annual cost savin g of between £25,141 and £36,854.

Conclusion

The study identified a potential cost saving of aut omated triage of lung nodules into a virtual lung nodule clinic.

Limitations

Further work is ongoing to model longitudinal costs of this cohort of patients. Funding for this study: No specific funding was used for this study Ethics committee - additional information: N/A Author Disclosures: Giles Dixon: Speaker: Aidence Vruti Dattani: Nothing to disclose Mary Jordan: Nothing to disclose Jonathan Rodrigues: Other: Aidence Speaker: Aidence SOLACE - CT acquisition protocols in lung cancer sc reening across Europe *M. F. G. Konrad*¹, E. Nischwitz¹, V. Palm¹, O. Von Stackelberg¹, A. Baca-Stera², K. Błasińska², M. Adamek³, J. Chorostowska-Wynimko², H-U. Kauczor¹; ¹Heidelberg/DE, ²Warsaw/PL, ³Gdansk/ PL ([email protected]) Purpose or Learning Objective: Exploration of the current status of institutional and technical factors of CT acquisiti on protocols applied in Europe that may affect the radiation exposure of screening participants in lung cancer screening procedures.

Methods

or Background: To achieve an overview of the applied CT acquisition protocols in lung cancer screening acro ss Europe, data were acquired from leading investigators responsible for the definition of CT acquisition protocols in the screening centres of t he SOLACE consortium. Data regarding institutional and technical factors of CT acquisition protocols were collected through a baseline survey. All data embod y the current status between June and October 2024.

Results

or Findings: Survey responses of 16 screening centres from 10 European countries (Croatia, Czechia, Estonia, Fran ce, Greece, Hungary, Ireland, Italy, Poland and Spain) were received. In relation to institutional factors, the CT acquisition protocols are establish ed and modified by personnel Friday Abstract-based Programme 179 of multiple professional roles (radiologists, radio graphers, medical physicists, manufacturer personnel). In eight of 16 (50 %) inst itutions the protocol is established by a multiprofessional team. CT protoco ls are mostly modifiable (88 %). Regarding technical factors, the number of detector rows in the z- direction ranges from 16 to 128. In 13 of 16 centre s (81 %) automatic exposure control is applied. The applied reconstructed slice thickness and increment lies in the range of 0.625 - 1.25 mm.

Conclusion

Lung cancer screening in Europe within the SOLACE c onsortium is widely implemented reaching the technical limits of currently operated devices. The definition of CT acquisition protocols as a team effort and the possible modification of protocols are subjects of imaginable improvements.

Limitations

The brief survey data stems only from institutions with research focus. Funding for this study: This project is co-funded under the EU4Health Programme 2021–2027 under grant agreement no. 10110 1187 Ethics committee - additional information: Not a patient study Author Disclosures: Joanna Chorostowska-Wynimko: Nothing to disclose Alicja Baca-Stera: Nothing to disclose Katarzyna Błasińska: Nothing to disclose Hans-Ulrich Kauczor: Nothing to disclose Oyunbileg Von Stackelberg: Nothing to disclose Emily Nischwitz: Nothing to disclose Viktoria Palm: Nothing to disclose Mathis Franz Georg Konrad: Other: This contribution has been prepared on behalf of the SOLACE consortium Mariusz Adamek: Nothing to disclose Variations in participant positioning, scan directi on and scanogram angle influence organ-specific radiations doses in routin e low-dose chest CT for lung cancer screening *L. D'Hondt*¹, C. Haentjens¹, A. Snoeckx², K. Bache r¹; ¹Ghent/BE, ²Antwerp/BE ([email protected]) Purpose or Learning Objective: In lung cancer screening (LCS), non-ideal positioning, changes in scan direction or scanogram angles are likely to occur due to the high turnover of participants and large volume of scans performed. Additionally, since we are dealing with healthy ind ividuals, careful management of radiation dose is crucial. This study aims to sy stematically simulate how participant positioning and scanning parameters aff ect organ-specific radiation doses.

Methods

or Background: Using the Alderson Rando phantom, we performed CT scans under varying conditions on two scanners ( GE Revolution, Siemens SOMATOM Definition Flash) to establish automatic tu be current modulation (ATCM) variations. ImpactMC Monte Carlo software si mulated low-dose scans for 32 patient-specific voxel models, calculating p ercentage dose differences to lungs, heart, thyroid, liver, and female breasts. D eviations included lateral/vertical mispositioning (2cm increments), s can direction changes (craniocaudally, caudocranially), and varying scano gram angles (anteroposterior, posteroanterior, lateral, combine d).

Results

or Findings: Vertical deviations caused linear increase in organ doses when positioned closer to the tube during sca nogram, with the GE scanner showing a 12-24% increase per 2cm deviation s. For the Siemens scanner this linear increase was less pronounced (3 -9% increase per 2cm deviation). Lateral mispositioning increased organ doses by <10% for both scanners. Caudocranial scanning only increased thyr oid doses by 15.8% for GE, while Siemens in fact showed a 19.7% decrease i n thyroid dose and significant increases in lung (18.7%), heart (28.7% ), liver (32.6%), and breast (27.9%) doses. For both scanners, the highest doses occurred with a posteroanterior scanogram. Siemens showed 72% dose reduction with dual scanogram, while GE showed minimally 34% reduction with an anteroposterior scanogram.

Conclusion

Non-ideal positioning and varying CT parameters can significantly affect organ doses, potentially under estimating the anticipated organ-specific doses and related radiation-induced cancer risk in LCS.

Limitations

Limitations are use of a phantom. Funding for this study: Funding was provided by the FWO “Kom op tegen Kanker”-project for lung cancer screening research in Belgium. (Project number: G0B1922N). Ethics committee - additional information: Not applicable Author Disclosures: Klaus Bacher: Nothing to disclose Louise D'Hondt: Nothing to disclose Annemiek Snoeckx: Nothing to disclose Claudia Haentjens: Nothing to disclose Pilot National Lung Cancer Screening in Poland – su mmary of the northern macroregion results - continuation of the Gdansk experience *J. Bidzinska*, J. Rymarowicz, K. Kobyłecka, K. Dzi adziuszko, E. Szurowska, W. Rzyman; Gdansk/PL ([email protected]) Purpose or Learning Objective: The aim of he study was to evaluate the

Results

of the Pilot National Lung Cancer Screening program in the northern macroregion of Poland.

Methods

or Background: After primary prevention, lung cancer screening is the strategy to reduce lung cancer-related mortalit y. A total of 3068 asymptomatic volunteers between 50 and 74 years of age with a smoking history of at least 20 pack years underwent screeni ng with the use of low-dose computed tomography of the chest. The radiological nodule management protocol was designed for a 2-year follow-up. It wa s created specifically for the national program and implemented nationwide. In the northern macroregion, LDCT was performed in 4 radiological centers. All r adiologists were trained. The program included an anti-smoking intervention.

Results

or Findings: At the time of submitting this abstract, 59 (1.92%) participants of the Program were diagnosed with lun g cancer. In this group, 43 participants had resectable lung cancer, which is 7 3% of cases, including 22 women and 21 men. Few participants undergo diagnost ic workups. In addition, we assessed (or the analysis is ongoing) the lung c ancer screening adherence rate, number and type of lung nodules, presence and severity of emphysema, CAC, incidental findings, and effectiveness of smok ing cessation intervention.

Conclusion

Lung cancer screening with low-dose computed tomogr aphy of the chest is an effective tool for reducing lung ca ncer-related mortality. LDCT is an effective tool for the identification of comorbi dities. It should be implemented as a National Lung Cancer Screening Strategy.

Limitations

Quality control in one of the sites revealed that t he radiologist should be re-trained periodically. Funding for this study: The program was financed by the Ministry of Health. The lung cancer prevention program was implemented in accordance with the competition regulations No. POWR.05.01.00-IP.00-010 /19 Operational Program Knowledge Education and Development 2014-20 20 Priority axis V Support for the health area Measure 5.1 Prevention programs. Ethics committee - additional information: The study was approved by the Institutional Review Board of the Medical Universit y of Gdańsk (NKBBN/72/2020). During realisation of the program we asked all aprticipants for the consent to use of data in the future analys es and studies. Only 8 participanst did not consent. Author Disclosures: Katarzyna Dziadziuszko: Nothing to disclose Katarzyna Kobyłecka: Nothing to disclose Julia Rymarowicz: Nothing to disclose Witold Rzyman: Nothing to disclose Edyta Szurowska: Nothing to disclose Joanna Bidzinska: Nothing to disclose Implications of 2D single slice vs. 3D whole-chest body composition phenotyping on outcome prediction: insights from th e National Lung Screening Trial *J. Jahn*, F. B. Pallasch, M. Jung, M. Reisert, F. Bamberg, J. Weiß; Freiburg/DE ([email protected]) Purpose or Learning Objective: Body composition (BC) is linked to outcomes in cardiovascular (CV) disease and cancer. BC is ty pically estimated from a 3rd lumbar slice, but the role of standardized chest he ights and 3D volumes is unclear. Here, we used a fully automated deep learn ing network 1) to investigate the correlation between 2D single-slice areas and 3D whole-chest volumes and 2) to explore their association with mo rtality in heavy smokers.

Methods

or Background: Using baseline data from the National Lung Screening Trial, BC was estimated on chest CT as sk eletal muscle (SM), intramuscular (IMAT), and subcutaneous adipose tiss ue (SAT). Correlations between 2D thoracic vertebra slices and 3D chest vo lumes were explored. Associations between BC measures and all-cause mort ality were assessed for both approaches, secondarily for CV and lung cancer mortality. Kaplan-Meier curves (categories 85%) and Cox regr ession models adjusted for demographics and CV risk factors were used.

Results

or Findings: Among 23,361 individuals (mean age 61.4±5, 41.6% female), 1,616 (6.9%) all-cause deaths occurred ove r a median follow-up of 6.5 years. The highest correlation between 2D and 3D vo lumes was at T4 vertebra. For 3D volumes, lower SM, lower SAT, and higher IMAT were associated with higher mortality in Kaplan-Meier cu rves (p<0.01). Associations between these groups remained robust after multivar iable adjustment (adjusted hazard ratio (aHR): 1.51, 95% CI 1.33-1.7 1, p<0.001; aHR: 1.21, 95% CI: 1.04-1.40, p<0.05; aHR: 1.68, 95% CI: 1.36- 1.78, p<0.001, respectively). Largely similar patterns were found for 2D measures, CV, and lung cancer mortality. Friday Abstract-based Programme 180

Conclusion

BC measures independently predict mortality in heav y smokers beyond traditional clinical risk factors. 2D measur es at T4 and 3D BC showed similar results, offering interchangeable use for i dentifying high-risk individuals in lung cancer screening, potentially enhancing per sonalized prevention.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: All National Lung Screening Trial participants provided written informed consent for the original trial and secondary use of the data was approved by the local IRB. Author Disclosures: Marco Reisert: Nothing to disclose Jakob Weiß: Nothing to disclose Matthias Jung: Nothing to disclose Fabian Bamberg: Speaker: Bayer Healthcare Research/ Grant Support: Siemens Healthineers Consultant: Bayer Healthcare S peaker: Siemens Healthineers Research/Grant Support: Bayer Healthca re Fabian Bernhard Pallasch: Nothing to disclose Johannes Jahn: Nothing to disclose Photon-Counting Detector CT Provides Superior Subso lid Nodule Characterization Compared to Same-Day Energy-Integr ating Detector CT *J. Wang*, L. Song, W. Song; Beijing/CN ([email protected]) Purpose or Learning Objective: To investigate the performance of photon- counting detector (PCD) CT compared to conventional energy-integrating detector (EID) CT in identifying subsolid nodule (S SN) characteristics.

Methods

or Background: Participants with SSNs who underwent same-day EID CT and PCD CT between October 2023 and April 20 24 were prospectively included. The 1.0 mm EID CT images and, subsequentl y, 1.0, 0.4, and 0.2 mm PCD CT images were reviewed to assess image noise a nd subjective image quality on a 5-point Likert scale. SSN characterist ics, including lobulation, spiculation, pleural retraction, air cavities, intr a-nodular vessel sign, internal vascular changes, and heterogeneous solid component s, were evaluated.

Results

or Findings: Forty-eight participants (mean age: 56 ± 11 years; 16 males) with 89 SSNs were included. PCD CT significa ntly reduced radiation dose when using matched scans (1.79 ± 0.39 vs. 2.17 ± 0.57 mSv, P < 0.001). Compared to 1.0 mm EID CT, 1.0 mm PCD CT images exh ibited significantly lower objective image noise and higher subjective i mage quality (all P < 0.001). Compared to EID CT, PCD CT demonstrated enhanced vi sualization of subtle characteristics, except for lobulation, with a 0.4 mm section thickness offering a favourable balance between ultra-high resolution an d perceived image quality for radiologists.

Conclusion

PCD CT facilitated radiation dose reduction and out performed conventional EID CT in terms of image quality and v isualization of SSN characteristics.

Limitations

First, the primary focus was on SSNs with various i maging characteristics suggestive of malignancy; consequen tly, SSNs < 6 mm were not included in the study. Second, quantitative SSN analysis was not performed because the study aimed to explore the qu alitative value of PCD CT in assessing SSN characteristics. Third, evaluation of nodules on a mediastinal window was not performed in this study. Fourth, not all SSNs yielded histopathologic results. Funding for this study: This work has received funding by the National Natural Science Foundation of China (NSFC No. 82171 934) and the National High Level Hospital Clinical Research Funding (2022 -PUMCH-B-069). Ethics committee - additional information: The Institutional Review Board of our hospital approved this prospective study [I-23P J1459]. Author Disclosures: Wei Song: Nothing to disclose Lan Song: Nothing to disclose Jinhua Wang: Nothing to disclose Diagnostic Outcomes of Korean Lung-RADS Category 2b Nodules in Lung Cancer Screening: A Single Tertiary Hospital E xperience *Y. Kim*, T. Ha, S. You, J. Sun; Suwon/KR ([email protected]) Purpose or Learning Objective: To evaluate the clinical outcomes of Korean Lung-RADS category 2b nodules in lung cancer screen ing and to compare the diagnostic performance between Korean Lung-RADS and Lung-RADS.

Methods

or Background: This retrospective study included 2,908 participant s who underwent low-dose chest CT scans at a single t ertiary hospital between September 2019 and January 2024, as part of the Kor ean national lung cancer screening program. The modified Korean Lung-RADS ad ded category 2b for nodules sized as category 3 or 4 but likely benign, aiming to reduce false positives. All CT scans were interpreted using the Korean Lung-RADS categorization (version 1.0 until 2021, version 1.1 from 2022, and version 2022 from 2024). To assess interobserver agreement, thre e radiologists independently reassessed each 2b nodule, indicating whether they would classify it as category 2b. Diagnostic performance comparisons were conducted using McNemar and permutation tests, whil e interobserver agreement was analyzed using multirater Fleiss κ statistics.

Results

or Findings: 1,270 participants with confirmed diagnoses were fi nally included (mean age, 61.5 ± 5.2 years; 1,243 men). Lung cancer was identified in 28 participants (2.2%). Category 2b nodules were found in 24 participants (1.9%), with one diagnosed as lung cancer (4.2%) an d the others benign (95.8%). Interobserver agreement was fair with a Fl eiss kappa value of 0.36 (95% CI 0.17-0.79). In terms of diagnostic performa nce, Korean Lung-RADS showed significantly higher specificity (92.5%, 1,1 49 of 1,242) compared to Lung-RADS (90.7%, 1,126 of 1,242; p < 0.001), while sensitivity was comparable between Korean Lung-RADS (89.36%, 25 of 28) and Lung-RADS (92.8%, 26 of 28; p = 1).

Conclusion

Korean Lung-RADS category 2b helped reduce the fals e-positive rate in lung cancer screening. However, significant variability was observed among radiologists in classifying category 2b nodul es.

Limitations

Retrospective study with small sample size. Funding for this study: No Ethics committee - additional information: This retrospective study was approved by the institutional review board, which w aived the requirement for written informed consent (AJOUIRB-DB-2024-399). Author Disclosures: Youna Kim: Nothing to disclose Joosung Sun: Nothing to disclose Seulgi You: Nothing to disclose Taeyang Ha: Nothing to disclose Safety of CT-guided core needle biopsy in patients with interstitial lung abnormalities (ILAs) *M. Balbi*¹, S. Capelli², A. Caroli², N. C. Culasso ¹, R. Senkeev¹, D. Morbidelli¹, G. Reboli¹, L. Righi³, A. Veltri¹; ¹Turin/IT, ²Rani ca/IT, ³Orbassano/IT ([email protected]) Purpose or Learning Objective: To evaluate the safety of CT-guided core needle biopsy (CNB) in patients with interstitial l ung abnormalities (ILAs).

Methods

or Background: Consecutive CT-guided pulmonary biopsies performed at the San Luigi Gonzaga Hospital (Orbass ano, Italy) from February 2010 to December 2023 (n=3251) were retrospectively reviewed to identify patients with ILAs who underwent CNB for the assess ment of a pulmonary lesion (n=73, case group). A control group of 73 CN B patients with no evidence of ILAs, matched for age, sex, emphysema severity, and lesion depth and dimensions, was selected to compare the complicatio n rates. Logistic regression was performed to identify risk factors f or complications within the case group, as well as to compare complication rate s between the case and control groups.

Results

or Findings: Complications occurred in 21/73 cases (29%) and 24/ 73 controls (33%), including 2 major complications in cases (3%) and 3 in controls (4%). Considering the overall occurrence of complic ations, there were no statistically significant differences between cases and controls, as assessed by Fisher’s exact test (p=0.72) and conditional logist ic regression adjusted for matched variables (p=0.72). Among patient, procedur al, and lesion-related data, three variables were found to be risk factors for complications in the case group based on univariate, multivariate, and stepwi se logistic regression with bidirectional elimination, guided by the Akaike inf ormation criterion (AIC): longer time of the needle within the lung (odds rat io (OR), 1.17; 95% confidence interval (CI), 0.72–13.02; p = 0.126), n eedle traversal of ILAs (OR, 7.04; 95% CI, 2.07–26.28; p = 0.002), and multiple pleural passes (OR, 8.06; 95% CI, 1.26–70.46; p = 0.035).

Conclusion

CT-guided CNB in patients with ILAs showed safety c omparable to patients without ILAs. However, crossing ILAs du ring the procedure may increase the complication risk.

Limitations

-Retrospective, single-center study -Limited number of cases -Lack of inter-reader agreement Funding for this study: The study did not receive any funding. Ethics committee - additional information: The institutional review board (Comitato Etico Territoriale Interaziendale of the AOU Città della Salute e della Scienza di Torino) approved this retrospective obse rvational case-control study and waived the requirement for written informed con sent. Author Disclosures: Serena Capelli: Nothing to disclose Luisella Righi: Nothing to disclose Andrea Veltri: Nothing to disclose Rouslan Senkeev: Nothing to disclose Noemi Cristina Culasso: Nothing to disclose Davide Morbidelli: Nothing to disclose Anna Caroli: Nothing to disclose Maurizio Balbi: Nothing to disclose Giulia Reboli: Nothing to disclose Friday Abstract-based Programme 181 Deep Learning-Based Image Domain Reconstruction Enh ances Image Quality and Pulmonary Nodule Detection in Ultralow- Dose CT with Adaptive Statistical Iterative Reconstruction-V *K. Ye*¹, B. Pan², J. Li³, Z. Pan², H. Yuan¹, N-J. Gong¹; ¹Beijing/CN, ²Shagnhai/CN, ³Fujian/CN ([email protected]) Purpose or Learning Objective: To evaluate the image quality and lung nodule detectability of ultralow-dose CT (ULDCT) wi th adaptive statistical iterative reconstruction-V (ASiR-V) post-processed using a deep learning image reconstruction (DLIR)-based image domain comp ared to low-dose CT (LDCT) and ULDCT without DLIR.

Methods

or Background: A total of 210 patients undergoing lung cancer screening underwent LDCT (mean ± SD, 0.81 ± 0.28 mSv) and ULDCT (0.17 ± 0.03 mSv) scans. ULDCT images were reconstructed wi th ASiR-V (ULDCT- ASiR-V) and post-processed using DLIR (ULDCT-DLIR). The quality of the three CT images was analyzed. Three radiologists de tected and measured pulmonary nodules on all CT images, with LDCT resul ts serving as references. Nodule conspicuity was assessed using a five-point Likert scale, followed by further statistical analyses.

Results

or Findings: A total of 463 nodules were detected using LDCT. Th e image noise of ULDCT-DLIR decreased by 60% compared to that of ULDCT- ASiR-V and was lower than that of LDCT (p <0.001). The subjective image quality scores for ULDCT-DLIR (4.4[4,1, 4.6]) were also higher than those for ULDCT-ASiR-V (3.6[3.1, 3.9]) (p <0.001). The overal l nodule detection rates for ULDCT-ASiR-V and ULDCT-DLIR were 82.1% (380/463 ) and 87.0% (403/463), respectively (p 1 mm was 2.9% (ULDCT-ASiR-V vs. LDCT) an d 0.5% (ULDCT- DLIR vs. LDCT) (p =0.009). Scores of nodule imaging sharpness on ULDCT- DLIR (4.0 ± 0.68) were significantly higher than those on ULDCT-ASiR-V (3.2 ± 0.50) (p <0.001).

Conclusion

DLIR-based image domain improves image quality, nod ule detection rate, nodule imaging sharpness, and nodul e measurement accuracy of ASiR-V on ULDCT.

Limitations

Only ASiR-V algorithm was evaluated, and more diffe rent IR algorithms should be included in the future. Funding for this study: With improved image quality and a higher nodule detection rate, deep learning-based image domain re construction is beneficial in facilitating the use of ultralow-dose CT in lung cancer screening. Ethics committee - additional information: Peking university third hospital review board Author Disclosures: Huishu Yuan: Nothing to disclose Kai Ye: Nothing to disclose Boyang Pan: Nothing to disclose Jie Li: Nothing to disclose Zhenglin Pan: Nothing to disclose Nan-Jie Gong: Nothing to disclose 16:00-17:30 Research Stage 2 Research Presentation Session: Musculoskeletal RPS 1610 Imaging of the peripheral joints: shoulder to foot Moderator A. Serfaty; Cabo Frio, Rio de Janeiro/BR ([email protected]) Relationship Between Coraco-Glenoid Ligament Varian ts and SLAP Lesions: Insights from a 3-T Arthro-MRI Study *M. Curti*, A. Cozzi, V. Chianca, S. Rizzo, F. Del Grande; Lugano/CH ([email protected]) Purpose or Learning Objective: The insertion site of the coraco-glenoid ligament (CGL) of the shoulder seems to be involved in superior labrum anterior and posterior (SLAP) lesions. Anatomical a nd arthro-magnetic resonance imaging (MRA) studies have reported five CGL variants. This study aimed to evaluate inter-reader reliability in the i dentification of CGL presence and variants on 3-T MRA, also investigating potenti al associations between CGL variants and SLAP lesions.

Methods

or Background: In this retrospective cohort study, three board- certified musculoskeletal radiologists (20, 10, and 5 years of experience) evaluated 1136 consecutive MRA examinations, 577 (5 0.2%) of patients with SLAP lesions. Inter-reader reliability in assessing CGL presence and variants was evaluated with Fleiss’ κ. After 1:1 patient matching according to age and sex, the association between the presence of SLAP l esions and CGL variants was evaluated with binary logistic regression, calc ulating odds ratios (ORs) and their 95% confidence intervals (CI).

Results

or Findings: “Almost perfect” reliability was found for the iden tification of CGL variants (κ 0.832, 95% CI 0.813–0.851), the most frequent bein g type I (393/1136 patients, 34.6%), and type II (320/1136 p atients, 28.2%). 1:1 matching resulted in two groups of 427 patients wit h and without SLAP lesions (256 males in each group; median age 52 years, IQR 44–59 years). Among these 854 matched patients, compared to the type I CGL variant, all other CGL variants had high ORs for the presence of SLAP lesi ons, ranging from OR 4.3 (95% CI 2.5–7.5) of type III to OR 39.9 (95% CI 24. 5–65.0) for type II.

Conclusion

3-T MRA grants high reliability in identifying CGL presence and variants. In an age and sex matched comparison, the type II CGL variant was strongly associated with the presence of SLAP lesio ns.

Limitations

Single-center study Funding for this study: None Ethics committee - additional information: Local etichs commitee Author Disclosures: Andrea Cozzi: Nothing to disclose Vito Chianca: Nothing to disclose Stefania Rizzo: Nothing to disclose Filippo Del Grande: Nothing to disclose Marco Curti: Nothing to disclose CT Findings and the Time-course of Myositis Ossific ans in Short Rotator Muscles *A. Fujii*, M. Katsumata, C. Sato, S. Tsukahara, T. Wada, M. Yamamoto, H. Kondo, H. Oba, A. Yamamoto; Tokyo, Japan/JP Purpose or Learning Objective: Non-traumatic myositis ossificans (MO) in the pelvic region in patients with neurological dis orders are classically reported on radiographs. However, CT findings have been rare ly reported and the time- course has not been proven. The purpose is to inves tigate the patients’ background, CT findings and the time-course of MO i nvolving the short external rotator muscles (SERMs) of the hip.

Methods

or Background: Of patients hospitalized for acute cerebrovascular disease between 2001 and 2023 in our hospital, 1,46 9 CT scans were performed for 436 patients. Swelling, calcification and ossification of SERMs, evaluated for each hip by two radiologists, were ju dged as calcification when the average density of the lesion was above 100HU, and as ossification when bone marrow fat was confirmed. Medical charts were reviewed for their clinical course and neurological findings.

Results

or Findings: SERMs swelling was found in 22 cases, 29 hips (age 61.0±3.0, 9 males, 13 females). Follow-up CT scans were examined in 6 (6/436, 1.4%) (aged 35–72, mean 54.7±12.7, 2 male, 4 female). All showed ipsilateral limb paralysis due to cerebral hemorrha ge. Swelling was detected in 8 hips of 6 patients, 4 unilaterally (right:left, 3 :1) and 2 bilaterally. Follow-up CT revealed calcification in all swelling lesions in S ERMs . The post-onset time- course in 6 hips was 13–36 days for swelling, 23–53 days for calcification, and 49–360 days for ossification. In 2 hips, calcificat ion disappeared on follow-up CT.

Conclusion

CT imaging illustrated the drastic changes in the m aturation process of SERMs-MO. Understanding the time-course of SERMs-MO is essential in the management, and frequency of follo w-up CT for distinguishing neoplasms and infectious conditions.

Limitations

This was a retrospective study of a few patients. T he follow-up durations varied between patients because examinati ons of the lesions were not always performed. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by Teikyo University Medical Research Ethics Committee (23-030). Author Disclosures: Hiroshi Oba: Nothing to disclose Satoshi Tsukahara: Nothing to disclose Megumi Katsumata: Nothing to disclose Akio Fujii: Nothing to disclose Asako Yamamoto: Nothing to disclose Chiaki Sato: Nothing to disclose Masayoshi Yamamoto: Nothing to disclose Hiroshi Kondo: Nothing to disclose Takeshi Wada: Nothing to disclose Friday Abstract-based Programme 182 Predicting the Surgical Requirement for Bone Substi tutes in Distal Radius Fractures: Evaluation of Dual-Energy-Derived Metrics L. D. Grünewald, V. Koch, *S. Mahmoudi*, S. Martin, J-E. Scholtz, I. Yel, C. Booz, T. Vogl; Frankfurt/DE ([email protected]) Purpose or Learning Objective: Distal radius fractures (DRF) are commonly diagnosed in emergency departments. Depending on fr acture characteristics and patient factors, bone substitutes may be needed , a decision usually made during surgery. However, preoperative preparations, such as obtaining informed consent, are necessary. This study aimed t o evaluate metrics from routine CT scans as surrogates for bone texture to predict bone substitute use in surgery.

Methods

or Background: Distal radius scans of patients who underwent dual- energy CT (DECT) between 01/2016 and 08/2021 were r etrospectively analyzed. Cortical HU, trabecular HU, cortical thic kness, and DECT-based bone mineral density (BMD) were measured. Patient r ecords and follow-up images were reviewed to determine bone substitute u se during surgery. Receiver-operating characteristic (ROC) analysis id entified AUC values for BMD, HU values, and cortical thickness, while logis tic regression assessed their association with bone substitute use.

Results

or Findings: A total of 263 patients (median age, 52 years; 132 women; 192 fractures) were included. ROC analysis s howed a higher AUC for DECT-derived BMD compared to cortical HU, trabecula r HU, and cortical thickness (0.87 vs. 0.62, 0.52, and 0.60, respectiv ely; P < .001). Logistic regression confirmed a significant association betw een lower DECT-derived BMD and the use of bone substitutes (Odds Ratio, 0. 94; P = .02), while cortical HU, trabecular HU, and cortical thickness were not significantly associated (P > .05).

Conclusion

Routine CT scans of the distal radius can predict t he use of bone substitutes in DRF surgical management, facilitatin g preoperative planning. DECT-derived BMD offers superior predictive perform ance compared to cortical HU, trabecular HU, and cortical thickness.

Limitations

Preselection bias, as patients received Xrays befor e CT. Funding for this study: No funding was received for this study Ethics committee - additional information: Consent waived due to the retrospective nature of the study Author Disclosures: Simon Martin: Nothing to disclose Christian Booz: Nothing to disclose Ibrahim Yel: Nothing to disclose Thomas Vogl: Nothing to disclose Jan-Erik Scholtz: Nothing to disclose Vitali Koch: Nothing to disclose Scherwin Mahmoudi: Nothing to disclose Leon David Grünewald: Nothing to disclose Evaluation of findings in recalcitrant tennis elbow , to rule out SMILE lesion (Symptomatic Micro-Instability In Lateral El bow) - a retrospective study: *S. Rajan*, J. S. Chatha, H. Mahajan; New Delhi/IN ([email protected]) Purpose or Learning Objective: Purpose: Recalcitrant lateral epicondylitis, often presents diagnostic challenges, particularly in ruling out symptomatic micro-instability in the lateral elbow (SMILE lesio n). This study aims to evaluate the radiological findings in a retrospective cohort , focusing on defining MRI findings with arthroscopically confirmed SMILE lesi ons.

Methods

or Background: Methods: A total of 313 consecutive elbow MRIs referred for tennis elbow of which 7 arthroscopical ly proven cases of SMILE lesions were identified as test cases, and key imag ing findings were documented as a checklist. The remaining 306 cases were randomized and anonymized and analyzed by two experienced radiolog ists (9 and 22 years of experience). On MR, key findings included lateral u lnar collateral ligament (LUCL) tear, partial tear of the extensor carpi rad ialis brevis (ECRB), widened ulnotrochlear joint or radiocapitellar posterior sh ift . Statistical significance of findings was assessed using Chi-square tests for ca tegorical variables and Fisher's exact test for small sample sizes.

Results

or Findings: Results: Of the 306 evaluated cases, 122 (25.8%) showed partial LUCL tears, 12 (3.9%) had complete L UCL tears, 268 cases (87.5%). had deep surface partial tears of ECRB. St atistical analysis revealed a significant association between the presence of L UCL tears (partial and complete) and abnormal radiocapitellar alignment (p < 0.01). Abnormal ulnotrochlear alignment was significantly associate d with the presence of ECRB tears (p < 0.05). Intra-articular abnormalitie s included posterior radial synovial plica in 22 cases (7.18%), synovitis in 3 cases (0.98%), and chondral lesions in the radial head in 39.78% of cases, with minimal occurrence of lateral capitellar lesions (0.65%).

Conclusion

Conclusions: The study underscores a significant as sociation between MRI findings such as LUCL tears, ECRB injur ies, and joint misalignments.

Limitations

Arthroscopic proof was not available in the cases. MR arthrography could possibly have helped. Funding for this study: None Ethics committee - additional information: Retrospective study using HIPAA guidelines Author Disclosures: Jagneet Singh Chatha: Nothing to disclose Sriram Rajan: Nothing to disclose Harsh Mahajan: Nothing to disclose Detecting and differentiating bone marrow edema in the knee using dual- energy CT with water-calcium and water-hydroxyapati te material decomposition W. Xiong¹, J. Han², G. Zhang², *T. Wang*², M. Cheng ¹, L. Hu¹, W. Li¹, Z. Ding¹, X. He¹; ¹Nanchang/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: To evaluate the diagnostic performance of dual-energy CT (DECT) with water-calcium and water- hydroxyapatite (HAP)

Material

decomposition for quantitatively detecting and differentiating bone marrow edema (BME) in knee.

Methods

or Background: This retrospective study included 26 patients who underwent DECT and magnetic resonance imaging (MRI) on the same day for the knee. The DECT images were post-processed to ge nerate water-calcium and water-HAP material decomposed images, on which two readers independently measured the water mass density of ed ema zone and contralateral normal bone marrow by drawing ellipse regions of interest. The MRI was served as the reference for presence and ex tent of BME. The edema extent was graded as mild or severe by visual asses sment. The diagnostic performance of DECT was evaluated using the receive r operating characteristic (ROC) analysis.

Results

or Findings: Thirty-two bones were confirmed with BME at MRI, caused by fractures (n=16), arthritis (n=10) and br uises (n=6). The mean water mass density of severe BME, mild BME and normal bon e marrow was 1073.9 ± 15.2, 1052.9 ± 13.0, and 1012.5 ± 22.4 mg/cm3 (p <0.001), respectively, on water-calcium images; and 1012.9 ± 15.6, 979.7 ± 23.8, and 959.1 ± 12.8 mg/cm3 (p <0.001), respectively, on water-HAP image s. In detecting the presence of BME, the area under the ROC curve (AUC) of water-calcium and water-HAP images was 0.976 and 0.885 (p = 0.001), r espectively, and the accuracy was 89.1% and 79.7%. In differentiating se vere and mild BME, the AUC of the two decomposed images was 0.836 and 0.89 6 (p = 0.318), respectively, with accuracy of 70.3% and 87.5%.

Conclusion

Dual-energy CT with water-calcium decomposition sho wed superior diagnostic performance for quantitatively detecting BME in knee, while water-HAP decomposition showed better performance i n differentiating severe and mild BME.

Limitations

Not applicable. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study was a retrospective study. Author Disclosures: Wenfeng Li: Nothing to disclose Zhichao Ding: Nothing to disclose Tiantian Wang: Employee: Central Research Institute , United Imaging Healthcare Xi He: Nothing to disclose Guozhi Zhang: Nothing to disclose Wei Xiong: Nothing to disclose Leiming Hu: Nothing to disclose Jintao Han: Nothing to disclose Maochu Cheng: Nothing to disclose Synthetic Multiplanar Reformation Generates Quantit atively Comparable Knee MR Images to Standard-Of-Care Acquisitions T. C. Arnold, L. Wang, L. Yao, A. Shankaranarayanan , *S. Pasumarthi Venkata*; Menlo Park, CA/US ([email protected]) Purpose or Learning Objective: Knee MRI protocols acquire multiple acquisition planes for three main 2D sequences: T1w , T2w, and PD. However, protocol sequence number is limited by time constra ints, meaning some contrast and orientation combinations are not avail able. Here, we develop a deep-learning-based method to synthesize 2D sequenc es using complementary contrast and acquisition plane inform ation from two input images, and quantitatively compare synthesized and ground truth acquisitions

Methods

or Background: We retrospectively analyzed data from 50 knee MRI participants. Each contained three 2D sequences: ax ial T2 fat-sat, coronal T2 fat-sat, and coronal PD. The coronal T2 fat-sat ser ved as our target sequence and the remaining sequences were our network inputs . All images were coregistered to the target image. Patients were div ided into training (N=34), Friday Abstract-based Programme 183 validation (N=8), and testing (N=8) sets. Quantitat ive metrics (PSNR, SSIM, and NMI) measured the similarity between the synthe sized coronal T2 and acquired images.

Results

or Findings: The algorithm demonstrated excellent quantitative performance when comparing acquired and synthesized coronal T2 fat-sat images across all three metrics (SSIM: 0.86±0.13, P SNR: 29.98±5.93 dB, and NMI: 1.33±0.10). We also compared the synthesized i mages to a coronal reformat of the acquired axial T2 fat-sat (SSIM: 0. 68±0.22, PSNR: 23.33±5.21 dB, and NMI: 1.21±0.10) to illustrate the significant benefit of our algorithm over simple image reformation (paired t-test, SSIM: p<0.005, PSNR & NMI: p<0.001). Relative to simple reformats, the synthes ized image showed a 26% SSIM increase, 29% PSNR increase, and 10% NMI incre ase. Qualitatively, pathological presentation in the synthesized image was similar to acquired sequences.

Conclusion

This study presents a deep-learning-based method to synthesize high quality multiplanar reformations in knee MRI. Synthesized images exhibited strong quantitative similarity to acquire d images.

Limitations

In future work, these methods should undergo qualit ative review by radiologists. Funding for this study: This study was funded by Subtle Medical. Ethics committee - additional information: All data was retrospective and anonymized and therefore not considered human subje cts research. Author Disclosures: Srivathsa Pasumarthi Venkata: Employee: Subtle Medi cal Inc. Lanhong Yao: Employee: Subtle Medical Long Wang: Employee: Subtle Medical Thomas Campbell Arnold: Employee: Subtle Medical Ajit Shankaranarayanan: Employee: Subtle Medical A three-dimensional magnetic resonance imaging-base d scoring system to evaluate continuity, thickness and shape of the injured anterior cruciate ligament *N. Giannotti*¹, A. Liu², H. Gauffin², S. Filbay³, J. Kvist², M. Englund⁴; ¹Sydney/AU, ²Linköping/SE, ³Melbourne/AU, ⁴Lund/SE ([email protected]) Purpose or Learning Objective: Anterior cruciate ligament (ACL) injuries are common, and ACL reconstruction (ACLR) is considered vital for restoring knee stability. However, recent evidence evaluating non- surgical approaches to the injured ACL challenges these notions. To date, no s tudies have used three- dimensional magnetic resonance imaging (3D-MRI) to systematically assess morphological changes of the injured ACL at differe nt time-points in non- reconstructed patients. The aim of this study was t o develop and investigate the reliability of a semi-quantitative scoring syst em to assess morphological changes of the injured ACL using 3D-MRI.

Methods

or Background: Baseline and 4 follow-ups within 2 years 3T 3D-MRI scans were acquired with an isotropic proton-densit y fat-saturated sequence. Scans were censored from further analysis after ACL R. 3D-MRI data were reconstructed on oblique coronal parallel to ACL, s agittal, and axial planes. Unblinded for time sequence, two readers independen tly scored the MRI data from all available time-points. The intra-rater rel iability was measured six months after the initial evaluation. Weighted kappa (κw) was used to calculate inter- and intra-reader reliability.

Results

or Findings: 129 patients (46% female; 25 SD [7] years) with non - surgically treated ACL injury from the NACOX study were included. The scoring system developed assessed the following ACL 3D-MRI features: overall structure, fiber continuity, thickness, and shape. The injury location was scored at baseline. 2,430 gradings were collected i n total. The averaged inter- reader reliability (κw) was 0.738, 0.673, 0.710 and 0.539 for ligament s tructure, fiber continuity, thickness and shape, respectively . The intra-rater reliability (κw) ranged from 0.409 to 0.873.

Conclusion

We introduce a novel 3D-MRI-based semi-quantitative scoring system to enable assessment of the acutely injured ACL and its morphological changes over time with acceptable inter-reader reli ability. Future validation vs findings from knee arthroscopy will be required.

Limitations

No limitations were identified. Funding for this study: The NACOX study received funding from the Swedish Research Council, Medical Faculty of Linköping Univ ersity, Swedish National Centre for Research in Sports, Medical Research Cou ncil of Southeast Sweden (FORSS), ALF Region Östergötland Ethics committee - additional information: Ethical approval for the study was granted by the Swedish Ethical Review Authority , Dnr: 2016/44/31. Author Disclosures: Håkan Gauffin: Nothing to disclose Nicola Giannotti: Nothing to disclose Joanna Kvist: Nothing to disclose Stephanie Filbay: Nothing to disclose Martin Englund: Nothing to disclose Angie Liu: Nothing to disclose The posterior cruciate ligament angle in the settin g of deficient anterior cruciate ligament deficient knees: the effect of ge nder, age, time from injury and tibial slope *M. V. Bausano*¹, F. Di Maria², R. D'Ambrosi¹, L. M . Sconfienza¹, S. Fusco¹, E. Abermann³, C. Fink³; ¹Milan/IT, ²Catania/IT, ³In nsbruck/AT ([email protected]) Purpose or Learning Objective: The aim was to assess the posterior cruciate ligament (PCL) angle in anterior cruciate ligament deficient knees and correlate it with anatomic and demographic factors.

Methods

or Background: Patients included were initially noted to have an ACL tear clinically as confirmed by MRI.For each pa tient were evaluated:PCL angle (PCLA), medial tibial slope (MTS), lateral ti bial slope (LTS), medial anterior tibial translation (MATT) and lateral ante rior tibial translation (LATT). Age, sex, and time from injury to MRI were manually recorded.Age groups were predefined dichotomizing age at its mean value while for time interval from injury to MRI the cut-off of 90 days was selec ted to differentiate between chronic and acute lesions. Differences by groups we re assessed with t-test or a Wilcoxon-Mann Whitney test, according to score di stribution. Spearman rank correlations were also estimated to explore correla tion among collected variables.

Results

or Findings: A total of 193 patients were included in the study of which 91 females and 102 males with a mean age of 3 0.27±12.54.The mean time from injury to MRI was 14.18±55.77 days.On ove rall population, mean PCLA resulted 128.72±10.33°, mean MTS 3.57±2.33, mean LTS 6.07±3.52, mean MATT and LATT were respectively 4.76±2.02 and 7.01 ±2.48 mm. In 190 cases PCLA angle was≥to 105° and only in 3 was inferior.PCLA negatively correlated with medial and lateral anterior tibial translation(p<0.05).Female showed a higher PCLA compared to male (130.55±10.23 vs 127.08±10.20; p=0.019).Patients with chronic ACL injury showed a lower value of PCLA compared to patients with acute injury(p=0.032) and a superior grade of LATT(p=0.015).

Conclusion

In the setting of ACL lesions, PCLA has normal valu e in acute injury and decrease over the time. PCLA is negative ly correlated with anterior tibial translation and female have higher PCLA comp ared to male.

Limitations

First, the diagnosis was made exclusively by MRI an alysis, second the data excluded partial ACL injuries. Funding for this study: N/A Ethics committee - additional information: After Institutional Review Board approval (ACL-L2104), two of the authors reviewed t he MR images of all patients with a clinical diagnosis of acute ACL inj ury. The present study was conducted following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [3]. All procedu res were conducted in accordance with the standards highlighted in the 19 64 Helsinki Declaration and its later amendments. Author Disclosures: Elisabeth Abermann: Nothing to disclose Fabrizio Di Maria: Nothing to disclose Stefano Fusco: Nothing to disclose Luca Maria Sconfienza: Nothing to disclose Riccardo D'Ambrosi: Nothing to disclose Maria Vittoria Bausano: Nothing to disclose Christian Fink: Nothing to disclose Diminished Meniscal Height as an Indicator of Synov itis on Magnetic Resonance Imaging of the Knee S. N. Yılmazer Zorlu, *N. A. Ahmady*, K. B. Karaca, Z. Akkaya; Ankara/TR ([email protected]) Purpose or Learning Objective: To investigate the relationship between meniscal height and adjacent synovitis on magnetic resonance images (MRI) of the knee, based on the hypothesis that, in contr ast to simple effusion, hypertrophic synovial tissue will cause compressive distortion on neighboring meniscus.

Methods

or Background: Contrast-enhanced knee MRIs from patients (≥18years), acquired during 2014-2024, at a single un iversity hospital were included in this cross-sectional, retrospective stu dy. Patients with history of knee surgery, acute trauma, local and systemic mali gnancy were excluded. On post-contrast fat-suppressed(fs)-T1-weighted images , 2 independent observers assessed the presence of synovitis (synovial enhanc ement ≥2mm) adjacent to the medial meniscus body-posterior horn (MMB, MMPH) and lateral meniscus body-posterior horn (LMB, LMPH). Blinded to the syn ovitis status, meniscal heights were measured on coronal and sagittal fs-fl uid sensitive images where menisci were intact and showed < grade 2 degenerati on. The relationship between meniscal height and synovitis was evaluated using linear regression models adjusted for age, gender and body mass index (BMI). Intra-class correlations (ICC) were used to test the reproducib ility of meniscal height measurements across 10 randomly selected patients.

Results

or Findings: A total of 129 patients (85 women, mean age=46.6±15years, mean BMI=24.6±5.1 kg/m2) (119 patients on 1.5T; 10 patients on 3.0T systems; ranges of slice thickness =3.5-6mm, gap=0.3-1mm, Friday Abstract-based Programme 184 FOV=12-16 cm) were included. Mean heights of MMB, M MPH, LMB, LMPH were 5.5±1mm, 5.3±1.1mm, 5.9±0.9mm, 5.6+0.8mm, respectively. MMB, MMPH and LMPH were significantly thinner adjacent t o synovitis regions(β=- 0.9, [95%CI= -1.34, -0.44], p<0.001; β=-1.26, [95%CI=-1.79, -0.73], p<0.001; β=-0.7, [95%CI= -1.12, -0.29], p=0.001, respectively ). ICC ranged between 0.96 - 0.99 (p<0.001), indicating excellent inter-r ater agreement.

Conclusion

Diminished meniscal height adjacent to high-signal areas on fluid-sensitive images may help distinguish effusio n from synovitis on clinical knee MRIs.

Limitations

The BMI values were not available in 25 patients. Funding for this study: None Ethics committee - additional information: Ethics Committee Approval Number: I03-251-24 Author Disclosures: Kazım Burak Karaca: Nothing to disclose Sezer Nil Yılmazer Zorlu: Nothing to disclose Nesar Ahmad Ahmady: Nothing to disclose Zehra Akkaya: Nothing to disclose The precision of a novel method for automated CT-ba sed radio- stereometric analysis in evaluating tibial implant migration *M. Acke*¹, B. Keelson¹, L. H. W. Engseth², G. Van Gompel¹, F-D. Ohrn³, J. De Mey¹, A. Schulz², S. M. Röhrl², N. Buls¹; ¹Br ussels/BE, ²Oslo/NO, ³Kristiansund/NO ([email protected]) Purpose or Learning Objective: The current gold-standard for implant migration analysis is radiostereometric analysis (R SA), which is a time- consuming and resource demanding method. We aim to evaluate the precision of a novel non-invasive method using automated CT-b ased radio-stereometric analysis (CT-RSA) for tibial implant migration anal ysis in a porcine model.

Methods

or Background: A porcine knee cadaver with a tibial implant was examined with marker-based RSA and computed tomogra phy (CT). RSA acquisitions (133kV, 6.3mAs) and CT images (120kV, 100mAs, 0.625mm, 0.5s, FOV 200×200mm) were acquired in seven differe nt positions (P1,…,P7) of the cadaveric knee. To obtain enough statistical power, the 7 positions were compared to each other, resulting in 21 double exam inations (P1_P2,…,P6_P7) for each method. Post-processing of the CT data was performed by a novel in-house built automatic image processing pipeline using SimpleITK. The reference standard was zero implant motion. Maximum-Total- Point-Motion (MTPM) was calculated for RSA to ident ify the implant point with maximum translation. Similarly, from the CT-RSA dat a we calculated Total- Translation (TT) for 6 non-invasive, virtual landma rks on the implant, including the centre-of-mass and 5 peripheral landmarks. In a ddition, we computed a virtual implant mesh from the CT data, including it s displacement. For RSA and CT-RSA, displacement differences towards the refere nce-standard were assessed using a two-sample t-test.

Results

or Findings: The precision for MTPM using marker-based RSA was 0.45mm (95%CI 0.19–0.70mm). TT calculation for virt ual landmarks using CT- RSA was more precise than RSA (p<0.001) with 0.15mm (95%CI 0.12– 0.18mm). The mesh data allowed to identify the poin t of maximum translation with a displacement of 0.16mm (95%CI 0.13–0.19mm).

Conclusion

Compared to RSA, the novel automated CT-RSA shows improved precision in analysing tibial implant migr ation for a porcine cadaver.

Limitations

This is a porcine phantom-study, which limits gener alizability and

Results

may differ in humans. Funding for this study: Funding is provided by the UZBrussels, Belgium and CIRRO, Norway. Ethics committee - additional information: The study was approved by the local research committee at Oslo University Hospita l on December 13, 2021. No consent was necessary since exams were performed on a porcine cadaver phantom. Author Disclosures: Stephan Maximilian Röhrl: Nothing to disclose Manou Acke: Nothing to disclose Johan De Mey: Nothing to disclose Lars Harald William Engseth: Nothing to disclose Gert Van Gompel: Nothing to disclose Anselm Schulz: Nothing to disclose Nico Buls: Nothing to disclose Frank-David Ohrn: Nothing to disclose Benyameen Keelson: Nothing to disclose How Cone-Beam CT affects Clinical Management in Foo t Trauma *E. Mcdermott*, D. P. Moloney, S. Murphy, B. Gibney , P. J. Macmahon, E. Kavanagh; Dublin/IE ([email protected]) Purpose or Learning Objective: Foot trauma is a common presentation to the emergency department. In our institution Cone Beam Computed Tomography (CBCT) is routinely utilized in patients with suspe cted fractures with negative radiographs. Previous studies have investigated the role of CBCT in the acute trauma setting for wrist and ankle injuries, but no large study has been performed for foot trauma. Our study focuses on how CBCT can improve the diagnosis of foot fractures in a trauma setting and , with the input of orthopaedic surgery, how it can alter the patient’s treatment p lan.

Methods

or Background: Patients with foot trauma who underwent CBCT between 2019 and 2023 were reviewed. In cases of di scordance between CBCT and X-rays, images were assessed with an ortho paedic surgeon. Patients were classified into three treatment optio ns based on X-rays and CBCT: No Treatment, Immobilisation, and Surgery.

Results

or Findings: 204 trauma patients had foot X-rays and CBCT. In 10 2 cases, CBCT identified 164 additional fractures, le ading to treatment changes in 62 cases (p < 0.01). 42 changed from ‘No Treatme nt’ to ‘Immobilisation,’ 3 from ‘No Treatment’ to Surgery,’ and 3 from ‘Immobi lisation’ to ‘Surgery.’ 12 were downgraded from ‘Immobilisation’ to ‘No Treatm ent,’ and 2 from ‘Surgery’ to ‘Immobilisation’.

Conclusion

Our findings show that a large number of foot traum a goes un- diagnosed on radiographs. The involvement of orthop aedic surgery in reviewing the radiographs and CBCT to assess how tr eatment plans may change is a novel approach. We found that a signifi cant number of patients would have their type of treatment altered with the use of CBCT. Our findings show the significant role CBCT can have when integr ated into trauma care.

Limitations

Only a single centre was used for data collection. Reporting of initial radiographs by non-musculoskeletal speciali sed radiologists. Funding for this study: None. Ethics committee - additional information: Project approved by the Mater Misericordiae University Hospital's Clinical Audit and Effectiveness Committee. Author Disclosures: Peter Joseph Macmahon: Nothing to disclose Darren Patrick Moloney: Nothing to disclose Brian Gibney: Nothing to disclose Edward Mcdermott: Nothing to disclose Eoin Kavanagh: Nothing to disclose Sophie Murphy: Nothing to disclose 16:00-17:30 Research Stage 3 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 1605 Artificial intelligence: real world results and large European projects Moderator L. Martí-Bonmatí; Valencia/ES ([email protected]) Assessing the effectiveness of artificial intellige nce (AI) in prioritising CT Head interpretation: a stepped-wedge cluster-random ised trial (ACCEPT- AI) *K. Nash*¹, K. Vimalesvaran², R. Dharmadhikari³, M. Hall⁴, A. Novak¹, S. Ather¹, D. J. Lowe⁴, H. Shuaib², Accept-Ai Investigators²; ¹Oxford/UK, ²London/UK, ³Northumberland/UK, ⁴Glasgow/UK ([email protected]) Purpose or Learning Objective: To evaluate the effectiveness of artificial intelligence (AI) in prioritisation of non-contrast head computed tomography scan (NCCT). We evaluated; 1) whether there was a r eduction in report turnaround time (TAT) of prioritised NCCT, 2) the a ccuracy of the AI algorithm, and 3) the technical performance of the algorithm.

Methods

or Background: This large-scale multi-centre trial has been conducted across three emergency departments betwee n November 2023 to July 2024. Individuals above the age of 18 who pres ented to the emergency department and underwent NCCT were included. Data c ollected included demographic data, time from acquisition to report o f CT scan, referral, and discharge, and death within 28 days. Findings were categorised into prioritised Friday Abstract-based Programme 185 findings (intracranial haemorrhage, mass effect, fr acture), non-prioritised findings (atrophy and infarct), and no prioritised findings. The study was conducted in 3 stages; pre-implementation, implemen tation and post- implementation. Baseline data was collected during the pre-implementation phase, whilst the implementation phase enabled trai ning and integration of the AI tool into radiology workflow at each trust. Duri ng the post-implementation phase, AI results were visible to radiologists. Rad iology reports were coded to assess AI accuracy, with discrepant cases sent for ground truthing by two independent radiologists.

Results

or Findings: 7,500 scans from the pre-implementation phase, and 8,453 scans from the post-implementation phase have been included. The median TAT (minutes) for prioritised scans in the p re-implementation was 34 (inter-quartile range 20-55) and 35 in the post-imp lementation phase (22-59).

Conclusion

We have successfully conducted a prospective trial of AI implementation across multiple centres in the Unite d Kingdom. Initial results do not show a difference in TAT for NCCT reports after implementation of AI.

Limitations

TAT has been calculated during preliminary analysis of a larger dataset, therefore full results will be available f or presentation. Funding for this study: This study has been awarded funding from NHSx AI Award (Award reference: AI_Award02354). Ethics committee - additional information: The study protocol was approved by the Research Ethics Committee (REC) of East Midl ands (Leicester Central), in May 2023 (REC 23/EM/0108) and was conducted in a ccordance with the principles of Good Clinical Practice. Author Disclosures: Haris Shuaib: Nothing to disclose Rahul Dharmadhikari: Nothing to disclose Accept-Ai Investigators: Nothing to disclose Katrina Nash: Nothing to disclose Alex Novak: Nothing to disclose Sarim Ather: Shareholder: Share holder RAIQC Ltd Kavitha Vimalesvaran: Nothing to disclose Mark Hall: Nothing to disclose David J. Lowe: Nothing to disclose ONCOPILOT: A Promptable CT Foundation Model For Sol id Tumor Evaluation *L. Machado*¹, H. Philippe², E. Ferreres², J. Khlau t², J. Gregory¹, M. Ronot¹, D. Tordjman², P. Manceron², P. Herent²; ¹Clichy/FR, ²Paris/FR ([email protected]) Purpose or Learning Objective: Carcinogenesis leads to tumors with diverse shapes and behaviors. Although RECIST 1.1 remains t he standard for evaluation, its reliance on linear measurements and high inter-reader variability often result in misclassification. Volumetric bioma rkers, such as total tumor burden, provide more information but need automated segmentation. Traditional segmentation models struggle with compl ex lesions, have a narrow focus, and lack interactivity. Foundation models wi th transformer architecture address these issues through zero-shot learning and visual prompts (point- click, bounding box). We developed ONCOPILOT, a fou ndation model that enhances RECIST 1.1 measurements, enabling volumetr ic analysis while integrating seamlessly into radiology workflows.

Methods

or Background: ONCOPILOT was trained on 7,500 CT scans, including normal anatomy and oncological cases. Its segmentation performance was assessed against nnUnet, a state-of -the-art baseline, using the DICE coefficient. The evaluation also included comparisons with radiologists for RECIST 1.1 long-axis measurements, annotation speed, and inter-reader variability.

Results

or Findings: ONCOPILOT outperformed state-of-the-art models, achieving a mean DICE of 0.78 post-editing versus 0 .70 for the baseline. Its RECIST error (7.4%, 1.1 mm) was not significantly d ifferent to radiologists' (8.6%, 1.3 mm). ONCOPILOT-assisted measurements wer e quicker than manual (17.2 vs. 20.6 seconds, p < .05), with reduc ed inter-reader variability (1.7 vs. 2.4 mm, p < .05).

Conclusion

ONCOPILOT is among the first foundation model appli cations in radiology, acting as an interactive AI assistant fo r oncological evaluation. It improves RECIST reproducibility, facilitates access to volumetric biomarkers, and seamlessly integrates into radiology workflows, reducing inter-reader variability and measurement time. This approach off ers substantial potential to advance oncology research and enhance clinical care .

Limitations

ONCOPILOT showed reduced performance on small tumor s, particularly lung lesions, which were overrepresent ed in the test set. Future iterations should use a more balanced dataset and c over a broader range of tumor types. Funding for this study: This work was granted access to the HPC resources of IDRIS under the allocation 2024-AD011013489R2 ma de by GENCI. Ethics committee - additional information: The data used is anonymous and publically available Author Disclosures: Julien Khlaut: Employee: Raidium Jules Gregory: Nothing to disclose Pierre Manceron: Founder: Raidium Maxime Ronot: Nothing to disclose Léo Machado: Employee: Raidium Paul Herent: Founder: Raidium Elodie Ferreres: Employee: Raidium Hélène Philippe: Employee: Raidium Daniel Tordjman: Employee: Raidium Evaluating artificial intelligence for lung cancer detection on chest radiographs: multi-vendor comparison of diagnostic accuracy in a real- world UK population *A. Maiter*, P. Metherall, J. Taylor, S. Matthews, K. Hocking, E. Burton, E. Anderson, A. Swift, C. S. Johns; Sheffield/UK Purpose or Learning Objective: Automated detection of lung cancer on chest radiographs by AI could streamline diagnostic pathw ays. This study evaluated the diagnostic accuracy of commercially available s oftware from seven vendors using a large dataset of radiographs from a real-wo rld UK population.

Methods

or Background: Consecutive chest radiographs obtained at our tertiary UK centre were retrospectively identified. Chest radiographs requested from primary care for adult patients, regardless of indication, were eligible for inclusion. Software from the seven vendors evaluate d each radiograph independently. The radiologist report for each radi ograph was also interrogated. Diagnostic accuracy metrics were dete rmined by comparing the software outputs and radiologist reports against th e diagnosis of lung cancer by multidisciplinary team decision.

Results

or Findings: 5,722 chest radiographs were included from 5,592 patients (median age 59 years, 54% female, 79% whit e, 1.6% prevalence of lung cancer). The software yielded the following (m ean±SD): sensitivity 46±9%, specificity 95±4%, positive predictive value (PPV) 15±5%, negative predictive value (NPV) 99±0%, accuracy 70±4% and false positive per image rate (FPPI) 0.05±0.04. Radiologist reports yielded the following: sensitivity 66%, specificity 98%, PPV 36%, NPV 99%, accuracy 82% and FPPI 0.02.

Conclusion

All software demonstrated high specificity and NPV comparable to radiologists, and could add value by helping to exclude lung cancer. However, all software also showed lower sensitivity and PPV than radiologists, suggesting that they could increase the number of f alse negative and false positive results. While AI has potential to improve the efficiency of “straight to CT” and other diagnostic pathways for suspected lun g cancer, the risks of diagnostic errors require careful consideration.

Limitations

This study used a retrospective dataset from a sing le centre. Further testing of software performance in multi-ce ntre patient cohorts and evaluation of downstream impacts are essential prio r to clinical deployment. Funding for this study: This study was funded by the NHS South Yorkshire Integrated Care System. Ethics committee - additional information: This study received local research ethics committee approval (23/EM/0186). Th e need for dedicated patient consent was waived. Author Disclosures: Suzanne Matthews: Nothing to disclose Jonathan Taylor: Nothing to disclose Katie Hocking: Nothing to disclose Emily Burton: Nothing to disclose Andrew Swift: Nothing to disclose Ahmed Maiter: Other: Bayer Chris S Johns: Nothing to disclose Elizabeth Anderson: Nothing to disclose Peter Metherall: Nothing to disclose Evaluation of AI-assisted Diagnosis of Clinically S ignificant Prostate Cancer on MRI at Scale: Preliminary Findings from t he PI-CAI Consortium *J. J. Twilt*¹, A. Saha¹, J. S. Bosma¹, D. Yakar², M. Elschot³, J. Veltman⁴, J. Fütterer¹, H. E. Huisman¹, M. De Rooij¹; ¹Nijmeg en/NL, ²Groningen/NL, ³Trondheim/NO, ⁴Almelo/NL ([email protected]) Purpose or Learning Objective: To assess whether utilizing an internationally validated prostate artificial intelligence (AI) system enhances the accuracy of prostate MRI evaluations in diagnosing clinically s ignificant prostate cancer (csPCa; Gleason Grade ≥2), compared to non-assisted assessments, in a comprehensive international reader study.

Methods

or Background: In this retrospective study, imaging and a prostate AI system developed and benchmarked on 10,207 exami nations through an international confirmatory study (PI-CAI) were used . A total of 780 biparametric prostate MRI examinations (2015-2021) from men susp ected of csPCa, with diagnostic-sufficient image quality and no prior cs PCa findings or treatment, were included. Reference was established through hi stopathology and ≥3 Friday Abstract-based Programme 186 years of follow-up. The AI was calibrated to genera te patient-level csPCa suspicion scores (0-10) and associated lesion-detec tion maps using 420 examinations from three Dutch centers. The remainin g 360 examinations (three Dutch, one Norwegian center) were used for o utcome analysis. Sixty- one readers (53 centers, 17 countries) provided PI- RADS 3-5 annotations and patient-level csPCa suspicion scores (0-100) with a nd without AI assistance in two phases, separated by a 4-week washout period. M ulti-reader, multi-case analysis compared diagnostic outcomes at a per-pati ent level.

Results

or Findings: Preliminary analysis of 22 readers (1-13 years of prostate MRI experience) demonstrated that AI assis tance improved csPCa diagnosis, with AUROCs of 0.919 (95% CI: 0.896-0.94 2) compared to 0.870 (95% CI: 0.831-0.908) without assistance. Sensitivi ty and specificity at PI- RADS ≥3 improved from 94.1% to 96.5% and 46.6% to 49.1%, respectively.

Conclusion

AI assistance improves csPCa diagnosis and holds pr omise for improving clinical outcomes. Further research is ne eded to confirm generalizability and assess workflow efficiency.

Limitations

Preliminary findings limit the sample size and subg roup analysis considering reader expertise. Reuse of data may int roduce generalization bias, and AI was not used to guide histologic verificatio n. Funding for this study: EU Horizon 2020: ProCAncer-I (grant number 952159), Health~Holland (grant number LSHM20103). Ethics committee - additional information: Retrospective use of anonymous patient data was approved by institutional or regio nal review boards at each contributing center (identifiers: REK 2017/576; CMO 2016-3045; IRB 2018- 597; ZGT23-37), and was conducted in accordance wit h the principles of the Declaration of Helsinki. Informed consent was waive d. Author Disclosures: Jurgen Fütterer: Grant Recipient: Siemens Healthine ers Joeran Sander Bosma: Nothing to disclose Henkjan En Huisman: Grant Recipient: Siemens Health ineers Grant Recipient: Canon Anindo Saha: Speaker: Guerbet Matthijs Elschot: Nothing to disclose Derya Yakar: Nothing to disclose Jeroen Veltman: Nothing to disclose Maarten De Rooij: Nothing to disclose Jasper Jonathan Twilt: Nothing to disclose The PANORAMA study results: Pancreatic Cancer Diagn osis - Radiologists meet AI *M. Schuurmans*, N. Alves, P. Vendittelli, G. Litje ns, J. J. Hermans, H. E. Huisman; Nijmegen/NL ([email protected]) Purpose or Learning Objective: The PANORAMA study transparently evaluates radiologists and AI in detecting pancreat ic ductal adenocarcinoma (PDAC) using contrast-enhanced computed tomography (CECT).

Methods

or Background: This retrospective study includes 3338 abdominal CECTs of patients without prior history of treatmen t or positive histopathology findings of PDAC acquired between 2006 and 2021 fro m 5 centers (Netherlands, Norway, and Sweden). Of these, 2238 c ases (676 PDAC) are publicly available to develop and train AI algorith ms, and 100 and 1000 cases were sequestered for AI tuning and testing, respect ively. The test set comprises data from two external centres not presen t in the other cohorts. A subset of 400 testing cases is used for the PANORAM A reader study. Both AI and radiologists indicate PDAC likelihood and local ization of lesions. Additionally, radiologists provide a newly introduc ed PANC-RADS score to assess the urgency of expert referrals. AI is openl y developed and evaluated using common metrics through the Grand Challenge pl atform. Patient-level performance and lesion localization performance are assessed using the area under the receiver operating characteristic curve ( AUROC) and average precision (AP), respectively.

Results

or Findings: The PANORAMA reader study results will be presented for the first time at ECR 2025. Currently, 57 radio logists (39 centers,13 countries, 2-30 years of experience, median: 9 year s) participate in the study. The baseline AI algorithm (nnU-Net with cross-entro py loss) achieves 0.9776 AUROC and 0.7226 AP in the tuning set. There are cu rrently 165 registered AI challenge participants across 11 teams.

Conclusion

Transparently benchmarked AI can enable early PDAC detection at the expert-radiologist level.

Limitations

While histopathology and follow-up were considered as the

Reference

standard for the sequestered testing set, we could not guarantee this level of evidence for all cases in the public train ing set. Funding for this study: This project has received funding from the European Union’s Horizon 2020 research and innovation progra m under grant agreement No 101016851, project PANCAIM. Ethics committee - additional information: Writen ethics committee approval was obtained from every center participating in the study. Author Disclosures: Natalia Alves: Nothing to disclose Henkjan En Huisman: Nothing to disclose Megan Schuurmans: Nothing to disclose Pierpaolo Vendittelli: Nothing to disclose John J Hermans: Nothing to disclose Geert Litjens: Nothing to disclose Development and Implementation of a Web-Based Platf orm for Clinical Validation of Artificial Intelligence Models in Fiv e Cancer Types: An Initiative of the European CHAIMELEON Project *A. Galiana-Bordera*, J. Aquerreta-Escribano, P. M. Martínez Gironés, P. Lozano, G. Ribas, P. Jimenez, L. Cerda Alberich, I. Blanquer, L. Marti-Bonmati; Valencia/ES ([email protected]) Purpose or Learning Objective: This study evaluates AI-based predictive oncology models for prostate, lung, breast, and col orectal cancers through the creation of an innovative web-platform. The goal is to bridge the gap between AI development and clinical implementation, advanci ng AI adoption in healthcare as part of the European CHAIMELEON proje ct.

Methods

or Background: We developed a web-platform with a microservices architecture, including a REST API, ORTHANC PACS fo r image management, and Keycloak security. The frontend is a custom-bui lt application serving as a control panel for users. It allows clinicians to op en a patient-specific DICOM viewer alongside clinical information, AI predictio ns, and validation buttons. This interface facilitates a three-stage case revie w: standard evaluation, AI- aided assessment, and final comparison with ground truth. AI models, developed through an open challenge using anonymize d data, were integrated into the platform.

Results

or Findings: The platform demonstrates efficiency, customizabili ty, and scalability, enhancing oncological study reprod ucibility. It creates an environment where medical images, clinical data, an d AI predictions coexist, adapting to various clinical and research settings. Over 70 clinicians are using the system to evaluate more than 2000 patients, wit h 60% reviewed in less than a month. This rapid adoption highlights the pl atform's user-friendly design and potential for improving clinical workflow.

Conclusion

This platform represents a significant advancement in validating and adopting AI models in oncology. It provides a f oundation for integrating AI into clinical environments, benefiting both clinici ans and patients. The study highlights the importance of user-friendly interfac es and structured evaluation processes in bridging the gap between AI developmen t and clinical application.

Limitations

The study's retrospective nature limits long-term o utcome assessment. Further research is needed on implement ation across diverse hospital infrastructures, the learning curve for clinicians, and ethical and regulatory aspects of AI integration in clinical pr actice. Funding for this study: This particular study did not receive direct fundin g. However, it was developed as part of the CHAIMELEON project, which received funding from the European Union's Horizon 2020 research and innovation program under Grant Agreement No. 952172 . Ethics committee - additional information: Under the CHAIMELEON project, all necessary ethical committee approvals have been obtained for the use of medical images and the development of artifi cial intelligence models. These approvals cover the collection, anonymization , and utilization of patient data for research purposes. Furthermore, the web pl atform developed for evaluating these AI models has been registered and approved for use in clinical settings. This ensures compliance with dat a protection regulations and ethical standards in medical research. The project adheres to strict protocols for data handling and user access, maintaining pati ent privacy and data integrity throughout the study. Author Disclosures: Pau Lozano: Nothing to disclose Ignacio Blanquer: Nothing to disclose Luis Marti-Bonmati: Nothing to disclose Paula Jimenez: Nothing to disclose Javier Aquerreta-Escribano: Nothing to disclose Gloria Ribas: Nothing to disclose Pedro Miguel Martínez Gironés: Nothing to disclose Leonor Cerda Alberich: Nothing to disclose Adrian Galiana-Bordera: Nothing to disclose Friday Abstract-based Programme 187 Hospital patients' attitudes towards AI worldwide: Results from the COMFORT study in 74 hospitals and 43 countries *F. Busch*¹, L. Hoffmann², L. Xu², L. Zhang³, L. Sa ba⁴, M. R. Makowski¹, H. Aerts⁵, L. C. Adams¹, K. Bressem¹; ¹Munich/DE, ²Berlin/DE , ³Nanjing/CN, ⁴Cagliari/IT, ⁵Boston, MA/US ([email protected]) Purpose or Learning Objective: Too often, we see healthcare technology implementations that focus only on the clinician's point of view and undervalue the patient's perspective. Given the exponential ri se in artificial intelligence (AI) applications in healthcare, this international, mul ticentre, cross-sectional study aimed to assess hospital patients' attitudes toward s AI in healthcare worldwide.

Methods

or Background: The present COMFORT study, involving 74 network hospitals in 43 countries, employed a quantitative 26-item instrument available in 26 languages on-site between February and Novemb er 2023.

Results

or Findings: 13806 patients from Europe (41.7%, n=5764/13806), Asia (25.2%, n=3473/13806), North America (16.5%, n =2284/13806), South America (9.7%, n=1336/13806), Africa (5.3%, n=728/1 3806) and Oceania (1.6%, n=221/13806) were included. Overall, 57.6% o f respondents were positive about the use of AI in healthcare. Signifi cant differences in attitudes were observed based on demographic characteristics, health status and technological literacy. Female participants and tho se in poorer health had less positive attitudes towards the use of AI in medicin e. Conversely, higher levels of AI knowledge and frequent use of technological d evices were associated with more positive attitudes. Notably, less than ha lf of the participants expressed positive attitudes to all items related t o trust in AI, with the lowest level of trust being in the accuracy of AI in provi ding information about treatment response. Patients showed a strong prefer ence for explainable AI and clinician-led decision-making, even if this mea nt a slight compromise in accuracy.

Conclusion

This large-scale, multinational study provides a co mprehensive perspective on patient attitudes towards AI in heal thcare across six continents. The findings suggest the need for tailored AI imple mentation strategies that account for patient demographics, health status, an d preferences for explainable AI and physician oversight. All study d ata has been made publicly available to encourage replication and further rese arch.

Limitations

Non-probability sampling Funding for this study: This research is funded by the European Union (101079894). Ethics committee - additional information: Ethical approval was obtained from Charité – Universitätsmedizin Berlin (EA4/213/ 22), which served as the lead institution. Author Disclosures: Lena Hoffmann: Nothing to disclose Marcus R. Makowski: Nothing to disclose Lisa C. Adams: Nothing to disclose Hugo Aerts: Nothing to disclose Lina Xu: Nothing to disclose Longjiang Zhang: Nothing to disclose Felix Busch: Nothing to disclose Keno Bressem: Nothing to disclose Luca Saba: Nothing to disclose Implementing Artificial Intelligence with a Multi-A I Platform across 15 centres: Experience and Strategy from an Internatio nal Healthcare Organization D. Penha¹, A. Juhos¹, D. Tálos¹, L. Rosa², M. Santo s², E. Dias², R. Paroczai¹, R. Barone¹, *A. Roncacci*¹; ¹Amsterdam/NL, ²Lisboa/ PT Purpose or Learning Objective: Implementing artificial intelligence (AI) in radiology is challenging due to the variety of AI t ools and the complexity of IT infrastructure and workflows. This presentation det ails a qualitative case study examining the implementation of six AI solutions us ing a multi-AI platform across 15 centers.

Methods

or Background: A longitudinal qualitative case study was conducted in Portugal, in 15 radiology centers over two years (May to October 2024), focusing on the implementation of different AI tool s (Veye Lung Nodules, Icobrain DM, Transpara, Keros/Polaris, qXR, and ARV A) using a multi-AI platform (Incepto). Data collected included 833 day s of work observations, 86 meeting observations, and from Incepto dashboard.

Results

or Findings: The multi-country healthcare organization's AI team managed the process adhering to a standard operatin g procedure covering initiation, planning, implementation, and clinical/ operational phases. AI deployment started with Icobrain DM (11 centers) an d Veye Lung Nodules (12 centers), processing 1,699 exams and 49,278 respect ively. Transpara was deployed across three centers, resulting in 5,791 s tudies over 19 months. Keros/Polaris in two centers, with 706 studies over 18 months. ARVA and qXR were not implemented due to clinical decision. Over all, the AI implementation was successful, with over 56,000 studies processed by four tools across 15 centers. The study identified advantages of the mul ti-AI platform, including workflow, cost-effectiveness, and improved patient care. However, disadvantages such as complexity, integration chall enges, and potential vendor lock-in were also noted.

Conclusion

This case study provides valuable insights for heal thcare organizations considering AI implementation via mul ti-AI platform over stand- alone AI tools.

Limitations

The implementation of a multi-AI platform in Radiol ogy lacks a clear blueprint to follow, creating the need to def ine new procedures and metrics. These procedures suited our needs of but m ay not apply to other radiology institutions. Funding for this study: Not applicable Ethics committee - additional information: Ethics committe and data/ legal department involved and with full approval of the w hole project Author Disclosures: Daniel Tálos: Nothing to disclose Alessandro Roncacci: Nothing to disclose Luis Rosa: Nothing to disclose Arpad Juhos: Nothing to disclose Robert Paroczai: Nothing to disclose Diana Penha: Nothing to disclose Rocco Barone: Nothing to disclose Miguel Santos: Nothing to disclose Eduardo Dias: Nothing to disclose From theory to practice: Re-identification Challeng e to test imaging data anonymization effectiveness *R. Catalán Flores*¹, I. Gómez-Rico¹, P. Jimenez¹, J. Gomes Carvalho², S. Mazzetti³, R. Martínez Martínez¹, M. França², D. Regge³, L. Marti-Bonmati¹; ¹Valencia/ES, ²Porto/PT, ³Torino/IT ([email protected]) Purpose or Learning Objective: DICOM image de-identification is an effective measure to protect patient privacy and en sure compliance with the GDPR. However, no standardized methods guarantee th e irreversible anonymization of DICOM images or provide evidence o n the robustness of these procedures. Organizations such as NEMA propos e de-identification profiles for DICOM metadata, but the risk to data p rotection is assumed by the entity responsible for de-identification, as the da ngers of using these profiles cannot be accurately measured. Given this context, the Re-identification Challenge serves as a technical audit to assess the robustness of DICOM image de-identification methods. The objectives of this project are to validate the robustness of these de-identification methods a nd to gain insight into organizing a challenge within the context of a Euro pean project.

Methods

or Background: The Re-identification Challenge consisted of a single phase in which the selected 68 participants were tasked to re-identify 38 pseudonymized DICOM studies from multiple European hospitals (including Spain, Portugal and Italy), modalities and anatomic al regions. They were pseudonymized locally using the de-identification p rofiles of ChAImeleon and ProCancer-I European AI4HI projects.

Results

or Findings: Despite 74% of the participants delivered result, n one succeeded in re-identifying the studies. Based on p articipants' reports of their attempts, vulnerabilities were discovered that allo wed them to narrow the population by obtaining their geographical region.

Conclusion

This challenge is a groundbreaking initiative that may pave the way for robust evaluations of de-identification met hods and may culminate in the standardization of de-identification profiles i n the field of radiological imaging.

Limitations

This challenge involved a limited sample of DICOM s tudies to ensure GDPR compliance and protect patients’ privac y. Participant selection was controlled through specific projects to prevent unauthorized access and reduce the risk of data breaches, marking a pioneer ing effort in de- identification research. Funding for this study: This challenge is part of the ChAImeleon project, specifically of Work Package 10, titled 'Repository Sustainability'. ChAImeleon has received funding from the European Union's Hori zon 2020 research and innovation programme under grant agreement No 95217 2. Ethics committee - additional information: Ethical considerations have been paramount throughout the Challenge's development, g arnering approval from the Ethics Committee of the involved data providers institutions. Rigorous measures ensure privacy, security, and data legitim acy: patient consent forms were obtained for the medical studies; legal expert s conducted a comprehensive data protection impact assessment; an d the platform hosting the studies adheres to stringent security protocols and privacy policies. Author Disclosures: João Gomes Carvalho: Nothing to disclose Luis Marti-Bonmati: Nothing to disclose Ricard Martínez Martínez: Nothing to disclose Daniele Regge: Nothing to disclose Rocío Catalán Flores: Nothing to disclose Paula Jimenez: Nothing to disclose Manuela França: Nothing to disclose Simone Mazzetti: Nothing to disclose Ignacio Gómez-Rico: Nothing to disclose Friday Abstract-based Programme 188 Establishing an applied framework to establish the trustworthiness of an international secure data environment for AI in CT imaging (AICT Consortium) *J. Kellas*¹, S. Van Wortswinkel², E. Casany Pujol³ , R. Lee¹, E. R. Ranschaert⁴; ¹Oxford/UK, ²Borgerhout/BE, ³Barcelona/ES, ⁴Ghent/BE ([email protected]) Purpose or Learning Objective: The AICT consortium, funded by Horizon Europe (NetZeroAICT), consists of international cli nical sites across 3 continents and proposes a comprehensive trustworthi ness framework by systematically integrating ethical, legal, sustaina bility and stakeholder engagement elements throughout research and develop ment. The goal is to ensure acceptability and trust in our research, inn ovation pipeline and AI-driven radiology applications.

Methods

or Background: The NetZero AICT trustworthiness framework is based on the European Commission's Trustworthy AI m odel, including foundational elements—Lawfulness, Ethics, and Robus tness—structured with key pillars: Human Agency, Technical Robustness, Pr ivacy, Transparency, Diversity, Societal Wellbeing, and Accountability. The project adheres to GDPR compliance, local applicable privacy laws, and the EU AI Act, and integrates ethics and privacy by design, broad and deep public involvement, sustainability, innovation management, clinical val idation, and regulatory compliance.

Results

or Findings: This is an interim report (year 1 of Horizon Europe program). Here, we showcase the ‘ethics by design’ approach and an applied model of patient and public patient involvement and engagement. A public advisory group (PAG) has been formed with current m embership from 4 countries and diverse backgrounds) with member repr esentation on project leadership groups (adopting and adapting a tiered m odel implemented by large-scale UK health data infrastructure projects (including OpenSAFELY). Feedback from the Public Advisory Group and the pro ject team strongly supported the view that integrating ethics by desig n from the concept stage through to deployment has improved public confidenc e and ensured compliance with ethical and legal standards.

Conclusion

By incorporating ethics by design, legal compliance , sustainability considerations and stakeholder engagement, the (Net Zero) AICT consortium established a reliable framework for radiology AI, promoting fairness and trustworthiness among users and stakeholders.

Limitations

This framework applies to healthcare imaging AI. Funding for this study: Horizon Europe and UK Research Innovation Ethics committee - additional information: HRA number 22/HRA/2302 Author Disclosures: Regent Lee: Shareholder: AI Sentia John Kellas: Consultant: This Equals Steven Van Wortswinkel: Employee: Ziekenhuis aan de Stroom Erik R. Ranschaert: Nothing to disclose Ernest Casany Pujol: Employee: CMRAD Time impact of AI-Assisted knee MRI reading in a re al-world multi-center study within a radiology network *B. Rizk*¹, P. Cordelle², B. Dufour¹, N. Heracleous ¹, C. Thouly¹, P. Zille², F. Zanca³; ¹Sion/CH, ²Poitiers/FR, ³Leuven/BE ([email protected]) Purpose or Learning Objective: We explored the impact on reading time of KEROS, a knee-MRI multifaceted AI algorithm, across three distinct interpretation workflows.

Methods

or Background: Clinical routine data was gathered from ten differe nt centers during the daily workflow of eight radiolog ists, including four musculoskeletal subspecialists (MSKs) and four gene ral knowledge radiologists (GENs), two of whom were junior (0 to 1 year of experience in private practice). We use a standardized report rel ying on voice recognition without secretariat formatting assistance. Data col lection was performed in three phases: Phase 1, generated reports without AI assistance; Phase 2, KEROS AI diagnosis was available before their inter pretation; Phase 3, AI findings were pre-filling and auto-integrated into the structured report. Reporting time was measured from report opening to validation, using the radiology information system (RIS). The Kruskal-Wal lis test (p<0.01) assessed significant differences, and time differences were calculated using weighted means.

Results

or Findings: In Phase 1, 431 exams were read with an average reading time of 10.8±9.5 to 26.3±9.3 minutes. In Phase 2 had 429 exams, with reading times between 12.8±8.2 and 27.5±9.2 minutes. Finally, Phase 3 included 425 exams with times ranging from 9.2±6.4 to 21.6±5.6 minutes. Cases were nearly equally read by MSKs and GENs. Co mpared to Phase 1, Phase 3 showed an average time reduction of 2.1 min utes (p<0.01) (13%), primarily driven by GENs, who saved up to 3.1 minut es (17%) (p<0.01).

Conclusion

We observe the average 13.4% (p<0.01) time reductio n after implementing KEROS and pre-filled reporting (Phase 3 vs. Phase 1). Generalists, the primary users of AI-assisted knee MRI readings, see an average 17.4% (p<0.01) decrease in their reading ti me. In clinical practice, AI- assisted knee MRI reporting saves time for general radiologists, who benefit the most from AI guidance.

Limitations

No limitations Funding for this study: No funding Ethics committee - additional information: We acquired the green light by the ethical committee in Switzerland for our resear ch structure including data registry and patient general consent for research. For the specific study presented here, only timing of radiologist reportin g data was used and not patient data. Author Disclosures: Pascal Zille: Nothing to disclose Natalie Heracleous: Nothing to disclose Cyril Thouly: Nothing to disclose Philippine Cordelle: Nothing to disclose Benoît Dufour: Nothing to disclose Benoît Rizk: Nothing to disclose Federica Zanca: Nothing to disclose PARROT: A Collaborative Polyglottal Annotated Radio logy Reports Database for Open Testing of Large Language Models *B. Le Guellec*¹, K. Bressem²; ¹Lille/FR, ²Munich/D E ([email protected]) Purpose or Learning Objective: To create a database of annotated radiology reports in diverse languages on which to test Large Language Models (LLMs).

Methods

or Background: Large Language Models (LLMs) represent one of the most important advancements in artificial intel ligence in recent years. In the field of medicine, they hold the potential to tranf orm how physicians interact with and interpret medical data. However, the curre nt research and applications of LLMs in medicine are predominantly focused on English- language datasets. This narrow focus raises signifi cant concerns about the ability of LLMs to generalize across the thousands of other languages. As a

Result

of difficulties accessing high-quality data from low-resources languages, patients may be excluded from the benefits of AI-dr iven advancements in healthcare. To address this critical gap, we have l aunched the Polyglottal Annotated Reports for Open Testing (PARROT) project . PARROT seeks to gather fictional medical reports from diverse lingu istic and cultural backgrounds, manually annotated by experts for ICD- 10 codes and make them freely accessible to the global research community. PARROT is a completely open-source initiative, inviting radiologists and m edical professionals from around the world to contribute.

Results

or Findings: 2648 annotated radiology reports from 75 radiologis ts from 20 countries in 13 languages have been collect ed. Most prevalent languages were Polish (808 reports), French (480 re ports) and Italian (285 reports). Contributions from the Global South inclu ded Ivory Coast, Mexico, Madagascar, Togo, Gabon, Argentina, Algeria, Turkey and China.

Conclusion

The to the collaborative effort of 75 radiologists from 20 countries, PARROT is the largest multilingual open database of radiology reports to date.

Limitations

Annotation for this first iteration of PARROT is li mited to ICD-10 codes. Most reports originated from European countr ies and European languages. Funding for this study: No specific funding Ethics committee - additional information: None required Author Disclosures: Keno Bressem: Nothing to disclose Bastien Le Guellec: Nothing to disclose Friday Abstract-based Programme 189 16:00-17:30 Research Stage 4 Research Presentation Session: Breast RPS 1602 Interventional radiology in breast cancer care Moderator I. Allajbeu; Cambridge/UK ([email protected];) Detection and characterization of cryoimmunologic r esponse induced by ultrasound-guided cryoablation on early breast canc er: evaluation of circulating markers F. Galati, *M. Pasculli*, C. Napoletano, V. Rizzo, R. Maroncelli, F. Cicciarelli, M. Nuti, C. Catalano, F. Pediconi; Rome/IT ([email protected]) Purpose or Learning Objective: Cryoablation is a minimally-invasive procedure that uses cooling to induce necrosis of t he targeted tissue. All other minimally invasive techniques use hyperthermia, whi ch melts cell membranes and causes protein denaturation. In contrast, cryoa blation leaves tumor proteins and tumor-associated antigens intact, with the potential to stimulate an anti-tumor immune response. Thus, the purpose of th is prospective pilot study was to characterize the immune response induced by tumor cryoablation in blood samples from early breast cancer (BC) patient s.

Methods

or Background: We enrolled patients with early-stage BC, scheduled for breast surgery, not eligible for neo- adjuvant therapy, and with a cryo-feasible cancer location. Blood samples to ass ess immune response were taken before (T0) and one week after cryoablation ( T1), and before (T2) and one week after (T3) surgery. Analysis of T cell sub sets (CD3, CD8, CD4, CD137, and Tregs) and the inflammatory/damage molec ule HMGB1 were performed by flow cytometry/ELISA. Circulating cyto kines were also analyzed using the Luminex analysis.

Results

or Findings: From July 2022 to January 2023, ten patients underw ent cryoablation. Analysis of circulating markers of cr yo-immunological response revealed a progressive release of HGBM1 after cryoa blation (T0-T1, p=0.04) until surgical resection of the primary tumor (T0-T 3, p=0.02). Cryoablation followed by surgery also induced a significant decr ease in CD137 T cell subsets (total and CD4; p<0.01), and IL4 (T0-T3, p< 0.05). Finally, a significant decrease in proliferative Treg cell subsets (Ki67+T regs; p<0.05) was observed.

Conclusion

In our pilot study, cryoablation induced the releas e of HMGB1, which acts to activate the primary phases of the im mune response, and the decrease of the immunosuppressive Treg subset and t he pro-tumoral cytokine IL-4, probably released by CD4CD137 T cells.

Limitations

Although the limited number of patients, cryoablati on was a valuable method to enhance the anti-tumor response. Funding for this study: The study has received funding from the Seed Grant funding programme of the European Society of Radiol ogy (ESR) in collaboration with the European Institute for Biome dical Imaging Research (EIBIR) kindly supported by an unrestricted, non-ex clusive grant from GE Healthcare. Ethics committee - additional information: The study obtained the approval of the Institutional Review Board of “Sapienza” Uni versity of Rome (Ref.6528, approved 24.11.2021). Author Disclosures: Roberto Maroncelli: Nothing to disclose Veronica Rizzo: Nothing to disclose Francesca Galati: Nothing to disclose Marcella Pasculli: Nothing to disclose Federica Cicciarelli: Nothing to disclose Marianna Nuti: Nothing to disclose Federica Pediconi: Nothing to disclose Carlo Catalano: Nothing to disclose Chiara Napoletano: Nothing to disclose A convenient model based on mammography and magneti c resonance imaging for preoperative differentiation of breast phyllodes tumors and fibroadenomas *X. Ma*; Shanghai/CN ([email protected]) Purpose or Learning Objective: To establish a fusion model based on mammography (MG) and magnetic resonance imaging (MR I) for the preoperative differentiation of breast phyllodes tu mors (PTs) and fibroadenomas (FAs).

Methods

or Background: The clinical data, MG images, and MR images of patients with breast FAs treated in our institution from October 2019 to December 2020, as well as patients with PTs treated from January 2011 to December 2020, were retrospectively collected. Univ ariate and multivariate logistic regression analyses were conducted to sele ct independent factors and to construct a diagnostic model to differentiate PT s and FAs. The diagnostic performance of the model was evaluated using the re ceiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA).

Results

or Findings: A total of 147 patients with FAs and 138 patients w ith PTs were included in this study. The results of the multivariate logistic regression analysis showed that patient age, maximu m diameter of mass, density on MG images, lobulation on MR images, and time-intensity curve (TIC) were independent factors contributing to the differential diagnosis. Finally, the fusion model showed satisfactory discr imination (area under the curve (AUC) 0.90, 95% CI: 0.86-0.94) and calibratio n. DCA indicated good clinical benefit, as indicated by most values being within threshold probabilities.

Conclusion

MG and MRI findings help differentiate between FAs and breast PTs preoperatively. The multimodal fusion model was clinically efficacious and beneficial and thus useful for accurate clinical di agnosis and treatment.

Limitations

Our study is a retrospective and single-centre stud y, so it is necessary to validate the accuracy of the research results through multicentre studies. Funding for this study: None Ethics committee - additional information: Fudan University Shanghai Cancer Center Author Disclosures: Xiaowen Ma: Nothing to disclose Correlation of radiological and pathological tumor sizes in early-stage breast cancer based on molecular subtypes and accom panying in situ carcinoma: A retrospective multicenter study D. E. Tekcan Sanli¹, *G. Esen*²; ¹Gaziantep/TR, ²Is tanbul/TR ([email protected]) Purpose or Learning Objective: To compare the accuracy of radiological tumor sizes obtained by mammography (MMG), ultrason ography (US) and magnetic resonance imaging (MRI) with pathological sizes and to determine whether tumor size measurements change based on mol ecular subtypes and the presence of accompanying ductal carcinoma in si tu (DCIS).

Methods

or Background: A total of 559 breast cancer patients diagnosed in 11 university hospitals in Turkey between 2010 and 2022, underwent preoperative MMG, USG, and MRI, and did not receive neoadjuvant chemotherapy (NAC) were included in the study. Tumo rs were divided into histopathological (in-situ/invasive/ in-situ+invasi ve (mixed)) and molecular (Luminal A/B/HER2+/triple-) subgroups. Tumor sizes were measured on each modality retrospectively and compared with the path ologic sizes reported in the postoperative pathology reports. Comparison was per formed based on histological type (invasive/ in situ/ mixed), and m olecular subtypes.

Results

or Findings: The highest agreement in invasive tumors was obtain ed with MRI (MRI:0.831, US:0.769, MMG:0.650). In the p resence of DCIS, the agreement was strong with MRI (r:0.770), moderate w ith MMG and US (r:0.517, r:0.593, respectively). In mixed tumors, agreement was strong with MRI (r:0.817), moderate with US (r:0.656), and low with MMG (r:0.499). Based on molecular subtypes, highest correlation was obta ined with US and MRI in HER-2 (+) tumors (r:0.754, r:0.715, respectively), and with MRI in other subtypes (Luminal A-B-triple (-)) with MRI (r:0.856 -0.815-0.858; respectively). There was no statistically significant difference i n terms of other criteria.

Conclusion

This multicenter study shows that MRI is the most r eliable

Method

in preoperative determination of tumor size for both invasive and in-situ tumors and all molecular subtypes.

Limitations

In this retrospective study, the number of patients was sufficient but unbalanced when divided into subgroups. The pre sence of invasive lobular carcinoma and axillary lymph nodes were not taken i nto account separately. Funding for this study: No Ethics committee - additional information: Ethics committee approval is obtained from Acıbadem University. (No:2021-21/29) Author Disclosures: Gul Esen: Nothing to disclose Deniz Esin Tekcan Sanli: Nothing to disclose Friday Abstract-based Programme 190 Breast abnormalities identified on cross sectional imaging represent a small but significant subgroup of referrals to symp tomatic breast service *C. O'Brien*, M. R. Common, N. Hambly, N. Ní Mhuirc heartaigh, M. Bambrick, D. Duke, N. Healy, E. Downey; Dublin/IE Purpose or Learning Objective: Our retrospective study examines patients referred to a symptomatic breast centre with breast abnormality identified on cross-sectional imaging with special reference to f requency and outcomes.

Methods

or Background: All patients presenting to our symptomatic breast centre over a 29-month period (Dec. 21-May 24) were evaluated. Patients referred following detection of breast abnormality on cross-sectional imaging were identified. Results of subsequent breast imagi ng, image-guided biopsies and histopathology were analysed.

Results

or Findings: 13,336 consecutive referrals over a 29-month period were reviewed. 179 female/ 3 male patients, mean ag e 56.3 years (range 26 - 89years), were referred with breast abnormality on cross-sectional imaging. 158/182 (86.8%) were identified on CT, 9/182 (4,9%) on MRI, 14/182 (7.7%) on nuclear medicine, 1/182 (0.5%) on ultrasound. Cl inical breast examination was normal in most patients. All patients underwent breast imaging. 75/ 182 (41.2%) patients underwent image guided biopsy. His topathology results of biopsies were benign in 26/75 (34.7 %), indetermina te (B3) in 5/74 (6.8%) and malignant in 44/75 (58.6%). 2/5 patients with B3 hi stopathology underwent surgical excision yielding final benign pathology, while 3/5 opted for lesion surveillance. Overall cancer detection rate was 24. 2% (44/ 182).

Conclusion

While breast abnormalities detected on cross sectio nal imaging represent a small subgroup of referrals the cancer detection rate in this cohort is significant. With normal clinical examinations t hese patients are at risk of delays along the traditional referral pathways base d on clinical suspicion of malignancy. Direct referral of these patients to br east radiology ensures prioritisation based on lesion appearance on cross- sectional imaging.

Limitations

Only cross-sectional breast abnormalities of patien ts referred to the symptomatic breast service were reviewed. Funding for this study: Nil Ethics committee - additional information: Approval obtained from the Quality and Safety Directorate team at our institut ion. Author Disclosures: Niamh Hambly: Nothing to disclose Deirdre Duke: Nothing to disclose Marie Bambrick: Nothing to disclose Matthew R Common: Nothing to disclose Eithne Downey: Nothing to disclose Conor O'Brien: Nothing to disclose Neasa Ní Mhuircheartaigh: Nothing to disclose Nuala Healy: Nothing to disclose Mammography-based radiomic analysis in triple negat ive ductal invasive breast cancer. Comparison between BRCA and not BRCA mutated patients: Preliminary results *G. Sessa*, C. Beretta, C. Bozzola, F. Mogavero, C. Parlato, L. Nocetti, G. Ligabue, P. Torricelli, A. Pecchi; Modena/IT Purpose or Learning Objective: This study aims to evaluate the applicability of radiomics analysis to mammographic images of pat ients diagnosed with triple negative breast cancer (TNBC) in order to id entify radiomics features that can differentiate the mutational status of BRCA gen es.

Methods

or Background: This retrospective study included patients histologically diagnosed with TNBC who performed a mammographic examination between 2010 and 2021. Mammographic ima ges were reviewed and for each patient the tumor lesions were manuall y segmented in the mammographic projection where they were better dema rcable; a further elliptical ROI (region of interest) of standard siz e (100 mm2) was drawn in the most homogeneous area of the controlateral healthy gland using the analogue mammographic projection of the same date or, if not available, of the corresponding bilateral mammographic investigation closer to the time of diagnosis. Features from each ROI were extracted wi th Pyradiomics-3D.

Results

or Findings: The population included 50 patients and 51 lesions (12 BRCA+ patients and 13 lesions therein, 38 BRCAwildt ype patients and 38 lesions therein). The lesions included 37 nodules, 3 pathologic microcalcifications and 11 lesions appearing as nod ules with contextual microcalcifications. The segmentation was carried o ut on the following projections: 24 LCC, 24 RCC, 21 LMLO, 2 LML, 21 RML O, 2RML. The first preliminary analysis demonstrated the feasibility o f the radiomics study in the population examined. Based on a previous study cond ucted on DCE-MRI of the same target population, we expect to be able to identify differences in radiomic patterns which represent the fenotipic exp ression of mutations occurring on a genetic level.

Conclusion

This study demonstrated the feasability of radiomic s analysis on diagnostic mammograms of TNBC patients to build a p redictive model able to discriminate between carriers and non carriers of B RCA gene mutations.

Limitations

Small population Funding for this study: No funding received for this study Ethics committee - additional information: The research was approved by the Area Vasta Emilia Nord Est Ethical Committee (4 63/2020/OSS/AOUMO) SIRER ID 236 - EMENDAMENTO SOSTANZIALE 1.0 (prot. A OU 0008681/22 del 23/03/2022) and informed consent was obtained f rom all subjects. Author Disclosures: Giulia Sessa: Nothing to disclose Pietro Torricelli: Nothing to disclose Annarita Pecchi: Nothing to disclose Luca Nocetti: Nothing to disclose Guido Ligabue: Nothing to disclose Chiara Bozzola: Nothing to disclose Cecilia Beretta: Nothing to disclose Francesca Mogavero: Nothing to disclose Chiara Parlato: Nothing to disclose Can the new combined CB SCORE reduce the number of breast false positive biopsies? Results from a monocentric study *A. Portaluri*¹, F. M. Arico¹, X. Wang², T. Zhang², C. Sofia¹, E. Condorelli¹, F. Catanzariti¹, M. A. Marino¹; ¹Messina/IT, ²Amste rdam/NL ([email protected]) Purpose or Learning Objective: To evaluate the performance of a new combined score, CEUS-BI-RADS (CB) score, in differe ntiating between benign and malignant lesions and in reducing the number of unnecessary breast biopsies.

Methods

or Background: 331 women with a new breast lesion scheduled for US-guided biopsy (sonographically assesed as BI-RAD S 4 a-c and 5) were enrolled. For each lesion, CEUS examination was per formed before the biopsy. 2 single reader (1 high-experienced breast radiologist and 1 trainee) independently assessed all CEUS studies with qualit ative analysis, in terms of time and intensity of enhancement, enhancement patt ern and size increase after contrast administration, assigning them a CEU S score from 0 to 3. Moreover a score from 1 to 4 was assigned for each BI-RADS category sonographically assesed as BI-RADS 4a, 4b, 4c and 5 . Finally a combined score (CB score) was obtained. Descriptive statisti cs, area under the curve (AUC), receiver operating characteristic (ROC) anal ysis, sensitivity and specificity were used to investigate the diagnostic performance of the combined approach. Inter-reader reliability was mea sured using Cohen’s kappa statistics.

Results

or Findings: 294 lesions have been found. CB score showed the highest diagnostic performance compared to the CEUS score alone (average AUCs = 0.935 vs 0. 890, p <0.0001), the highest sen sitivity (87.9% vs 91.3%) and specificity (75.2% vs 80.7%) . Moderate ( κ= 0. 45) agreement was found between the two readers. Finally, CB score would ha ve obviated up to 40% of unnecessary biopsies.

Conclusion

CB score allowed to improve the diagnostic performa nce of CEUS alone for lesions assesment, reducing the rate of false positives biopsies for both experienced and unexperienced rea der.

Limitations

Deeper studies on the application of this score nee d to be carried out to improve the identification of small tumours that have minimal blood supply. Funding for this study: None Ethics committee - additional information: This is a single-center, retrospective study. Ethics committee approval was waived as all Contrast- Enhanced Ultrasound (CEUS) examinations were conduc ted as part of routine clinical care, following established protocols at o ur institution. These examinations were clinically indicated prior to bio psy, ensuring that no additional procedures were performed outside the St andard of Care (SOC). Author Disclosures: Carmelo Sofia: Nothing to disclose Elvira Condorelli: Nothing to disclose Maria Adele Marino: Nothing to disclose Francesca Catanzariti: Nothing to disclose Xin Wang: Nothing to disclose Antonio Portaluri: Nothing to disclose Tianyu Zhang: Nothing to disclose Francesco Marcello Arico: Nothing to disclose Non-surgical treatment of breast cancer: a comparis on of outcomes between Cryoablation with hormonal therapy versus C ryoablation alone and hormonal therapy alone in patients not eligible for surgery *S. E. Baldi Giorgi*, F. Di Naro, G. Migliaro, F. P ugliese, T. Amadori, S. Vidali, G. Bicchierai, V. Miele, J. Nori; Florence/IT ([email protected]) Purpose or Learning Objective: This study aims to evaluate the most effective non-surgical treatment for breast cancer in surgery-ineligible patients, comparing ultrasound-guided Cryoablation combined w ith hormonal therapy (HT) versus Cryoablation alone and hormonal therapy alone. Friday Abstract-based Programme 191

Methods

or Background: A cohort of 64 patients (mean age 83.4 years) not- suitable for surgery due to comorbidities and/or ad vanced age was enrolled, with a total of 73 biopsy-confirmed malignant breas t lesions. All the lesions were invasive ductal carcinomas (mean size 14.8 mm) , hormone-positive and HER2-negative, with no ultrasound-visible lymph nod e involvement. Patients were divided into three groups: 36 lesions were tre ated with Cryoablation and HT, 19 with Cryoablation only and 18 with HT only. A locoregional staging was performed at baseline with contrast-enhanced-mammog raphy (CEM) and ultrasound, followed by CEM and ultrasound follow-u p at 12 months post- treatment. Only patients completing the follow-up w ere included. Lesion size was compared at baseline and 12 months after-treatm ent. Fisher's exact test was used for group comparison.

Results

or Findings: Of the 73 lesions, 47 completed the 12-months follo w- up: 20 in the Cryoablation-with-HT group, 9 in the Cryoablation-only group, and 18 in the HT-only group. Tumor size reduction was g reatest in the Cryoablation-with-HT group (94%, mean reduction of 15.4 mm), followed by Cryoablation-only (82%, mean reduction of 9.7 mm), and HT-only (43%, mean reduction of 4.6 mm). Tumors with absent or low CEM -enhancement, suggesting residual-disease reduction, were most fr equent in the Cryoablation- with-HT group (80%), followed by Cryoablation-only (77.8%), and HT-only (38.9%). Fisher’s test revealed a significant diffe rence between the Cryoablation-with-HT and the HT-only groups (p<0.00 15), expressing the added value of Cryoablation.

Conclusion

Cryoablation with hormonal-therapy significantly re duces tumor size and residual disease more effectively than the rapy alone, making it a promising option for patients not-eligible for surg ery.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by Comitato Etico Regione Toscana - Pediatrico (refere nce number: 165/2024). Author Disclosures: Giulia Bicchierai: Nothing to disclose Francesca Pugliese: Nothing to disclose Jacopo Nori: Nothing to disclose Sofia Vidali: Nothing to disclose Vittorio Miele: Nothing to disclose Sofia Elisabetta Baldi Giorgi: Nothing to disclose Tommaso Amadori: Nothing to disclose Giuliano Migliaro: Nothing to disclose Federica Di Naro: Nothing to disclose Vacuum-Assisted Biopsy in The Era of Low-Risk Ducta l Carcinoma in Situ (LR-DCIS) Active Surveillance: Real World Data and Implications *H. L. Couto*¹, C. N. Valadares², B. F. De Paula Ri cardo¹, A. N. Soares¹, P. H. Toppa¹, B. A. Coelho³, C. C. Pessoa⁴, N. Sharma⁵, E. Carvalho Pessoa⁴; ¹Belo Horizonte/BR, ²São Paulo/BR, ³Montes Claros/B R, ⁴Botucatu/BR, ⁵Leeds/UK ([email protected]) Purpose or Learning Objective: Evaluate vacuum assisted biopsy (VAB) as diagnostic test of LR-DCIS in the context of real-w orld clinical practice.

Methods

or Background: Database analysis of 116 cancers [both invasive breast cancers (IC) and ductal carcinoma in situ (D CIS)] diagnosed by VAB submitted to standard surgical treatment with compl ete histological data from VAB and surgery from 04/13/2017 to 11/28/2020. The VAB results were matched to the surgical pathology, considered the g old standard. The pathological diagnoses were grouped into malignanci es requiring immediate surgical treatment [DCIS with high risk (HR-DCIS) o f progression to IC or IC] versus those eligible to active surveillance (LR-DC IS). HR-DCIS/IC were considered positive while LR-DCIS negative results. VAB sensitivity, specificity, positive predictive value (PPV), negative predictiv e value (NPV), and accuracy were obtained.

Results

or Findings: Median age was 56 (20-91); final median invasive tu mor size (T) of 6mm (0,8 – 25) and 8mm (2 – 65) for DCI S; 65.52% were US- guided (70/116) and 44.48% (46/116) stereotactic gu ided; 42.24% (49/116) were only masses, 26,72% (31/116) masses associated with calcifications and 31.03% (36/116) calcifications. Out of 116 malignan cy cases diagnosed by VAB, 15 (12.9%) resulted in LR- DCIS in the biopsy, 10 (8.6%) confirmed LR- DCIS in surgery, and 5 (4.3%) were upgraded to HR-D CIS/IC in surgery. VAB showed 95.28% sensitivity and 100% specificity. The positive predictive value (PPV) was 100%, and the negative predictive value ( NPV) was 66.67%. Of the 5 false negatives (FN) LR-DCIS upgraded in surgery: 3 were HR-DCIS and 2 IC (pT1a-bpN0-luminal).

Conclusion

VAB, based in conventional pathology and immunohistochemistry, had an elevated FNR LR-DCIS i n real world practice and, if applied, VAB LR-DCIS upgraded cases could b e treated by either hormone or radiation therapy isolated or combined c ounterbalanced by slight reduction of overtreatment.

Limitations

Retrospective data base Funding for this study: None Ethics committee - additional information: The study was approved by the Ethics Committee of Santa Casa of Belo Horizonte un der the number 25761019.8.0000.5138, and all methods were carried out in accordance with national guidelines Author Disclosures: Nisha Sharma: Nothing to disclose Bernardo F. De Paula Ricardo: Nothing to disclose Carla Carvalho Pessoa: Nothing to disclose Bertha Andrade Coelho: Nothing to disclose Carolina Nazareth Valadares: Nothing to disclose Henrique Lima Couto: Nothing to disclose Eduardo Carvalho Pessoa: Nothing to disclose Paola H. Toppa: Nothing to disclose Aleida N. Soares: Nothing to disclose Does touch imprint cytology prepared from core need le biopsy specimens in breast lesions provide an immediate di agnosis? Ş. Kökten, H. Kılın, *N. Voyvoda*; Istanbul/TR ([email protected]) Purpose or Learning Objective: The touch imprint method is used during the frozen study of sentinel lymph node samples in brea st cancer patients and provides a rapid response with high accuracy and en sures the correct management of patients during surgery. Similarly, i mprint preparations made from core biopsies can also be helpful in reaching rapid and accurate results. The aim of this study was to determine the diagnost ic value and accuracy of the imprint method.

Methods

or Background: Between January 2024 and March 2024, patients who were referred to the breast imaging center of o ur hospital due to a mass in the breast and planned for US-guided core needle bi opsy were included in the study. Touch imprint and core biopsy specimens were retrospectively evaluated at different times by the same pathologis t. Pathological findings in touch imprint evaluation were classified using the guideline.

Results

or Findings: 201 lesions of 178 patients with an average age of 48.05 (min: 16-max 82) were evaluated. Of the 201 lesions , 186 were breast lesions and 15 were axillary lymph nodes. Seven of 186 brea st lesions were excluded from the evaluation because touch imprint was defin ed as insufficient. The sensitivity of touch imprint was calculated as 87.5 0%, specificity as 89.16%, ppv: 54.66%, npv as 97.95% and accuracy as 88.94%. If we exclude the lymphoma patient, the sensitivity of imprint for ly mph node was calculated as 100%, specificity as 83.3%, ppv: 47.27%, npv as 100 % and accuracy as 85.5%. When not excluded, the sensitivity of imprin t cytology was 88.89%, specificity as 83.33%, ppv: 44.35%, npv as 98.05% a nd accuracy as 84.06%.

Conclusion

Imprint cytology prepared from core biopsies of bre ast lesions can provide highly accurate and rapid diagnosis. Th us, treatment can be started without delay.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: Local ethics committee approval (decision no: 2024/010.99/3/19) was obtained before the study. Author Disclosures: Şermin Kökten: Nothing to disclose Nuray Voyvoda: Nothing to disclose Halil Kılın: Nothing to disclose Differential efficacy of Cryoablation in breast can cer subtypes: ultrasound-guided scar biopsy evaluation year post- treatment *F. Pugliese*, F. Di Naro, G. Migliaro, S. E. Baldi Giorgi, T. Amadori, D. De Benedetto, C. Bellini, V. Miele, J. Nori; Flo rence/IT ([email protected]) Purpose or Learning Objective: To assess the effectiveness of cryoablation (CR) in different subtypes of breast tumors, a coho rt of 39 biopsy-proven B5 lesions underwent ultrasound-guided scar biopsy eva luation one year post- treatment.

Methods

or Background: From 2022-2023, the B5 lesions comprised 35 invasive ductal carcinoma (IDC), 2 IDC-associated D CIS, and 2 invasive lobular carcinomas. All patients were deemed inoper able for advanced age and comorbidities, leading to their enrollment in a n annual follow-up and ultrasound-guided scar biopsy. The study population was stratified into three subgroups: molecular subtype, dimensional cut-off, and growth index.

Results

or Findings: These lesions were hormone-responsive, with 19 classified as Luminal A and 20 as Luminal B. The di mensional cutoff ranged to 2.5 cm, with ice ball dimensions tailored to encomp ass a one-centimeter margin around the lesions. Data analysis revealed n otable differences in the efficacy of cryoablation among the various subgroup s. When considering Luminal A lesions with ki67>20% the complete ablati on rate was 84.2% and for Luminal B it was 90.0%. Conversely, tumors with Ki6 7 expression ≤20% exhibited higher complete ablation rates, with Lumi nal A reaching 100% and Luminal B at 84.6%. Additionally, lesions ≤10 mm in size exhibited a higher Friday Abstract-based Programme 192 complete ablation rate of 100% compared to lesions >10 mm, which showed an 80.0% success rate. Histologically, CR was ineff ective in achieving complete ablation in DCIS cases, presenting a rate of 0%, but other subtypes demonstrated a higher complete ablation rate at 91. 9%. However, none of these differences were statistically significant.

Conclusion

Cryoablation emerges as a promising primary treatme nt option for breast cancer is a safe and effective nonsurgic al alternative to breast- conserving surgery in select patients with unifocal IDC low grade, hormone receptor-positive, and ≤10 mm size

Limitations

No limitations were identified. Funding for this study: No funding was received for this study Ethics committee - additional information: The study is retrospective Author Disclosures: Francesca Pugliese: Nothing to disclose Diego De Benedetto: Nothing to disclose Jacopo Nori: Nothing to disclose Vittorio Miele: Nothing to disclose Sofia Elisabetta Baldi Giorgi: Nothing to disclose Tommaso Amadori: Nothing to disclose Chiara Bellini: Nothing to disclose Giuliano Migliaro: Nothing to disclose Federica Di Naro: Nothing to disclose 193 Saturday, March 1 Saturday Abstract-based Programme 194 08:00-09:00 Research Stage 1 Research Presentation Session: Genitourinary RPS 1707 Advances in imaging techniques for the genitourinary tract Moderator A. Shinagare; Boston, MA/US Synthesising and adapting multiple data sources to design environmentally sustainable and clinically effectiv e imaging pathways for visible haematuria J. B. John¹, K. O'Flynn², *S. Lambracos*³, B. Abdel qader⁴, S. Nalagatla⁴, S. Khadouri⁵, T. W. R. Briggs³, W. K. Gray³, J. Mcgrath¹; ¹Exet er/UK, ²Salford/UK, ³London/UK, ⁴Glasgow/UK, ⁵Leeds/UK ([email protected]) Purpose or Learning Objective: To design a risk-stratified imaging pathway that reduces greenhouse gas (GHG) emissions using d iagnostic performance evidence for ultrasound and computerised tomography urogram (CTU) in detecting upper tract urothelial cancer (UTUC).

Methods

or Background: An audit of 15 UK hospitals’ first-line imaging for visible haematuria (VH) and non-visible haematuria (NVH), and use of one- stop cystoscopy and imaging, was performed. Urology referral data from the IDENTIFY study (N = 10,896) were linked to national Hospital Episode Statistics data to estimate absolute numbers of pat ients receiving ultrasound or CTU first-line across England annually.

Results

or Findings: Ultrasound was the first-line imaging choice for VH and NVH in 53% and 93% of hospitals respectively; other hospitals used CTU as first-line. One-stop assessment was performed in 44 % of audited cases. An estimated 127,701 ultrasound and 77,880 CTU were pe rformed across England annually, including an estimated 20% and 2% of additional CTU for patients with VH and NVH respectively receiving ult rasound first due to persistent VH. Informed by these data, a risk-strat ified imaging pathway for patients with haematuria was developed. The pathway comprised: one-stop clinic ultrasound first-line for all, additional CT U for high-risk cases (VH + age>65 + smoking history), patient-initiated follow up direct to CTU for persistent VH if not performed initially. We estima te that this pathway could

Results

in 191,097 ultrasounds, 51,320 CTU and 73% of assessments completed with one-stop assessment, leading to an e stimated 269 tonnes CO2e reduction in net greenhouse gas emissions acro ss England for one year. Around 0.1% of patients would have UTUC missed with first-line ultrasound on this pathway.

Conclusion

Adoption of this evidence-based pathway will reduce GHG emissions whilst delivering greater risk-stratified imaging use, improving equity of healthcare access and reducing unwarranted use o f CTU.

Limitations

N/A Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Joseph B John: Nothing to disclose John Mcgrath: Nothing to disclose Sinan Khadouri: Nothing to disclose Bushra Abdelqader: Nothing to disclose Sarika Nalagatla: Nothing to disclose Simon Lambracos: Nothing to disclose Tim W R Briggs: Nothing to disclose Kieran O'Flynn: Nothing to disclose William K Gray: Nothing to disclose Study on Optimization of Image Quality of Computed Tomography Urography Using Dual-layer Detector Spectral CT Vir tual Monoenergetic Imaging Technique Combined with Contrast Split Bolu s Protocol *F. Zhao*, K. Li; Xi'an/CN ([email protected]) Purpose or Learning Objective: To investigate the optimal keV of dual- detector spectral CT virtual monoenergetic imaging (VMI) in computed tomography urography (CTU).

Methods

or Background: Hematuria patients who underwent dual-layer detector spectral CTU were collected. In the experi mental group, 25mL of contrast bolus was injected first, and then 20mL of contrast bolus was injected 15 minutes later, and cortico-excretory phase CT sc an was performed. In control group, non-contrast phase CT scan was perfo rmed first, then 100mL of contrast bolus was injected, and cortical, medullar y and excretory phases CT scans were performed later. The experimental group was reconstructed into 40keV, 50keV, 60keV, 70keV VMI images and mixed ene rgy images, and the excretory phase in control group was reconstructed into mixed energy image. The differences of CT value, standard deviation (SD ), signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), subjective sc ore of the image quality were compared.

Results

or Findings: There was no significant difference in SNR and CNR between the mixed energy images of the control grou p and the VMI 50keV images of the experimental group. In the score of i mage display effect, the VMI 50keV image of the experimental group had the highe st score. In the score of filling degree of contrast bolus, VMI 40keV images in the experimental group had the highest score.The effective dose in the exp erimental group was about 19.96% of that in the control group, and the amount of contrast bolus in the experimental group was about 45% of that in the con trol group.

Conclusion

The dual-layer detector spectral CT VMI technique u sed in CTU can improve the image quality while reducing the am ount of contrast bolus and radiation dose, and VMI 40～50keV is the best energy level for image display.

Limitations

A single-center study with a small number of cases Funding for this study: This study was funded by the hospital by RMB 10,000 Ethics committee - additional information: This study was approved by the Ethics Committee Author Disclosures: Fanhui Zhao: Nothing to disclose Kai Li: Nothing to disclose Prospective evaluation of high-resolution diffusion -weighted imaging accelerated by deep-learning reconstruction in mult iparametric MRI of the prostate *S. Ursprung*¹, J. Herrmann¹, E. Weiland², D. Nicke l², A. Lingg¹, S. Afat¹, S. Gassenmaier¹; ¹Tübingen/DE, ²Erlangen/DE Purpose or Learning Objective: Prostate MRI is a gatekeeper for more invasive investigations in prostate cancer diagnosi s. This study investigates the potential of deep-learning reconstruction of high-r esolution (HR) diffusion- weighted imaging (DWI) to improve image quality and lesion detectability.

Methods

or Background: Prospective study comparing multiparametric MRI according to PI-RADS 2.1 specifications with standa rd (DWI-Std) and DWI-HR (4-fold higher in-plane resolution) on a 3T MRI-sys tem. Two radiologists (7yr experience) compared image quality of DWI-Std/DWI-H R qualitatively and quantitatively using the Prostate Imaging Quality S coring System (PI-QUALv2), a 5-point Likert-scale assessing sharpness, noise, artefacts, overall impression and diagnostic confidence, and contrast-to-noise ra tio (CNR) of prostatic lesions.

Results

or Findings: 91 patients consented (17 exclusions for prostatect omy, 2 radiotherapy, 3 incomplete imaging); 69 patients were included. Average acquisition time for DWI-Std/DWI-HR was 04:30min/05 :33min. DWI-DL showed higher sharpness at all b-values and on ADC maps (p <0.001). This came at the cost of higher noise on b1000 and ADC (p<0.001) , resulting in comparable lesion detectability (p=0.28) and PI-RADS scoring ( p=1). Readers favoured DWI-HR in 47% and DWI-Std in 33%. The PI-QUAL sub-s core of DWI- Std/DWI-HR was similar (p=0.37); only the reduced S NR/Contrast in DWI-HR approached significance (p=0.054). PI-QUAL was opti mal in 67%/61%, acceptable in 28%/33% and inadequate in 6%/6% when including DWI- Std/DWI-HR. The CNR of PI-RADS 3-5 lesions in the P Z was significantly higher at b1000/on ADC maps from DWI-Std (median 10 .8 vs. 10.3, p=0.002 at b1000, 10.1 vs. 7.7, p=0.0002 for ADC). The CNR bet ween PZ/TZ was higher in DWI-Std (median 3.0 vs. 2.2; p=0.04).

Conclusion

We prospectively evaluated high-resolution DWI with DL- reconstruction, showing improved sharpness at sligh tly reduced CNR and maintained diagnostic performance in prostate MRI.

Limitations

This study was conducted on scanners of a single ve ndor. Confirmatory histology was available for PI-RADS 3+ lesions only. Funding for this study: This study received no funding. Ethics committee - additional information: Tubingen University Hospital IRB Author Disclosures: Judith Herrmann: Nothing to disclose Elisabeth Weiland: Employee: Siemens Healthineers Saif Afat: Speaker: Siemens Healthineers Research/G rant Support: Siemens Healthineers Dominik Nickel: Employee: Siemens Healthineers Andreas Lingg: Nothing to disclose Stephan Ursprung: Nothing to disclose Sebastian Gassenmaier: Nothing to disclose Saturday Abstract-based Programme 195 Ultra-Fast Biparametric MRI in Prostate Cancer Asse ssment: Diagnostic Performance and Image Quality Compared to Conventio nal Multiparametric MRI *A-M. Pausch*, V. Filleböck, C. Elsner, N. Rupp, D. Eberli, A. M. Hötker; Zurich/CH Purpose or Learning Objective: To compare the diagnostic performance and image quality of a deep-learning-assisted ultra-fas t biparametric MRI (bpMRI) with the conventional multiparametric MRI (mpMRI) f or the diagnosis of clinically significant prostate cancer (csPCa).

Methods

or Background: This IRB-approved prospective single-center study enrolled 123 biopsy-naïve patients undergoing conve ntional mpMRI and additionally ultra-fast bpMRI at 3T between 06/2023 -02/2024. Two radiologists (R1: 4 years and R2: 3 years of experience) indepen dently assigned PI-RADS scores (PI-RADS v2.1) and assessed image quality (m PI-QUAL score) in two blinded study readouts. Weighted Cohen’s Kappa ( κ) was calculated to evaluate inter-reader agreement. Diagnostic perform ance was analyzed using clinical data and histopathological results from cl inically indicated biopsies.

Results

or Findings: Inter-reader agreement was good for both mpMRI ( κ = 0.83) and ultra-fast bpMRI (κ = 0.87). Both readers demonstrated high sensitivity (≥94%/≥91%, R1/R2) and NPV (≥96%/≥95%) for csPCa detection using both protocols. The more experienced reader m ostly showed notably higher specificity (≥77%/≥53%), PPV (≥62%/≥45%), and diagnostic accuracy (≥82%/≥65%) compared to the less experienced reader. There was no significant difference in the diagnostic performanc e of correctly identifying csPCa between both protocols (p>0.05). The ultra-fa st bpMRI protocol had significantly better image quality ratings (p<0.001 ) and achieved an 80% reduction in scan time compared to mpMRI.

Conclusion

Deep-learning-assisted ultra-fast bpMRI protocols o ffer a promising alternative to conventional mpMRI for dia gnosing csPCa in biopsy- naïve patients with comparable inter-reader agreeme nt and diagnostic performance at superior image quality. However, rea der experience remains essential for diagnostic performance.

Limitations

The single-center design and the exclusion of some patients with a PI-RADS ≥ 3 who did not undergo biopsy may limit the general izability of our findings. Funding for this study: Holcim Stiftung zur Förderung der wissenschaftliche n Fortbildung Ethics committee - additional information: Cantonal Ethics Commission Zurich Author Disclosures: Niels Rupp: Nothing to disclose Clara Elsner: Nothing to disclose Daniel Eberli: Nothing to disclose Vivien Filleböck: Nothing to disclose Antonia-Maria Pausch: Nothing to disclose Andreas M. Hötker: Nothing to disclose Increased dose of iodine contrast media does not in crease the odds of contrast-associated acute kidney injury *F. B. Berglund*, P. Liss, R. Frithiof; Uppsala/SE ([email protected]) Purpose or Learning Objective: The aim of this study is to investigate if any of the two components of the g-I/eGFR ratio is inde pendently associated with the development of contrast-associated acute kidney injury (CA-AKI).

Methods

or Background: All patients admitted to the intensive care units (ICUs) of a tertiary hospital from January 2013 to February 2020 were retrospectively identified. Those who underwent iod ine contrast media (ICM)- enhanced CT exams were included in this nested case -control study. CA-AKI was defined and staged based on the creatinine and urine output criteria set forth by the Kidney Disease Improving Global Outcom es guidelines. The two components of the ratio, the dose of ICM (measured in grams of iodine) and renal function estimated by plasma creatinine, were analyzed separately in relation to the odds of developing CA-AKI.

Results

or Findings: Among the 214 patients included in the analysis PC- AKI occurred in 42 of the patients (19.6%). Median age was 61.5 years (IQR 40-73) and 59.3% were of male sex. Renal function at the d ay of the CT-scan differed between those developing PC-AKI (eGFR 56.9, IQR 35- 87) and those that did not (eGFR 81.6, IQR 58-96). However, the dose of IC M was not associated with PC-AKI development (OR 1.31 (IQR 0.49-3.47), p =0.827).

Conclusion

In this case-control study, renal function at the d ay of the examination but not the administered dose of iodine contrast media was associated with PC-AKI. This suggest that including injected amount of iodine contrast media as a variable to clinically predict risk of PC-AKI is futile.

Limitations

This is a single center case-control study where on ly 42 critically ill patients developed CA-AKI. This reduces the gen eralizability of the study as well as its power. Funding for this study: The study was supported with funding from ALF from Uppsala University Hospital, and the Swedish Resear ch Council (2014-02569 and 2014-07606). Funding bodies had no role in the design of the study, data collection, interpretation, or in the writing of th e manuscript. Ethics committee - additional information: This study was approved by the Swedish Ethical Review Authority (Dnr 2017/168 with amendment 2020- 00135). Declaration of Helsinki and its subsequent revisions were observed. Author Disclosures: Robert Frithiof: Nothing to disclose Per Liss: Nothing to disclose Felix Björn Berglund: Nothing to disclose Modified in-plane resolution while maintaining high image quality T2- weighted prostate MRI *S. J. Riederer*¹, E. Borisch¹, A. Froemming¹, A. K awashima², N. Takahashi¹; ¹Rochester, MN/US, ²Phoenix, AZ/US ([email protected]) Purpose or Learning Objective: To determine if an axial T2-weighted spin- echo (T2-WI) sequence with modified in-plane spatia l resolution could provide non-inferior performance and reduced acquisition ti me vs. a standard PI- RADSv2.1-compatible sequence.

Methods

or Background: Both sequences used 3 mm slice thickness, 16 cm FOV, acceleration R=1.5, TR>3000, TE 150. The PI-RA DS-compatible sequence used 400×230 in-plane sampling, 0.4 mm × 0 .7 mm resolution (0.280 mm2 pixel area). The modified sequence used 320×280 in-plane sampling, 0.5 mm × 0.57 mm resolution (0.285 mm2 pi xel area). Although the increased phase sampling of the modified sequence ( 280 vs. 230) required more repetitions, the reduced frequency sampling (3 20 vs. 400) allowed lower bandwidth and reduced averaging. The two sequences were both applied to 62 consecutive subjects identified for clinical prosta te MRI. The number of slices was patient-specific but identical for the two sequ ences. Results were blindly reviewed by three experienced radiologists. Each se ries was assessed individually for Image Quality (IQ) using a 0-3 sca le. For each subject the two series were also compared for overall preference on a five-point (-2, -1, 0, +1, +2) scale.

Results

or Findings: Scan time depended on slice count (29 to 45). Scan time range for the PI-RADS-compatible sequence was 2:56- 5:04 (median 3:54) and for the modified sequence 2:16-3:56 (median 3:01). Scan time reduction using the modified sequence was 37 to 80 sec (median 53 s ec). 51/62=82.2% of the PI-RADS-compatible series and 56/62=90.3% of the mo dified sequence were evaluated as diagnostically interpretable (IQ=2,3). Reviewer-averaged scores showed a significant preference for the modified se quence (p<0.001).

Conclusion

Although not adherent to PI-RADSv2.1 guidelines, ac quisition with essentially identical in-plane pixel area (0.2 8 mm2) allows 53 sec (23%) reduction in acquisition time and significantly imp roved reviewer preference vs. a PI-RADS-adherent sequence.

Limitations

Limited number of subjects Funding for this study: This work was supported by NIH. Ethics committee - additional information: Approved by Institutional Review Board (IRB) Author Disclosures: Adam Froemming: Nothing to disclose Stephen J. Riederer: Nothing to disclose Eric Borisch: Nothing to disclose Naoki Takahashi: Nothing to disclose Akira Kawashima: Nothing to disclose CEST Imaging vs. DWI with and without CEST Imaging: Capability for Distinguishing Malignant from Benign Prostatic Area s *T. Ueda*, H. Nagata, M. Nomura, T. Yoshikawa, D. T akenaka, Y. Ozawa, Y. Ohno; Toyoake/JP ([email protected]) Purpose or Learning Objective: 3D Chemical exchange saturation transfer (CEST) imaging is recently developed to obtain CEST information within entire tumor. The purpose of this study was to compare the capability for distinguishing malignant from benign prostatic area s among 3D CEST imaging, diffusion weighted imaging (DWI) and combined both information.

Methods

or Background: Fifty-two suspected prostatic cancer patients underwent DWI at b value as 0 and 1500 s/mm2 and 3D CEST imaging, surgical treatments and pathological examinations. According to the pathological results, 154 areas were determined as malignant prostatic areas, and 154 out of 470 areas were computationally selec ted as benign prostatic areas. On each 3D CEST imaging, magnetization trans fer ratio asymmetry (MTRasym) at 3.5 ppm map was generated. Then, 308 R OIs were placed over malignant or benign areas on each map, and MTRasym and ADC values were determined. Each index was compared between maligna nt and benign areas by Student’s t-test. ROC analysis was performed to compare diagnostic performance among MTRasym, ADC and combined discrim inators. Finally, sensitivity, specificity and accuracy were compared among all methods by McNemar’s test. Saturday Abstract-based Programme 196

Results

or Findings: MTRasym and ADC of malignant area had significant differences with those of benign area (MTRasym: p<0 .0001, ADC: p<0.0001). Area under the curves (AUC) of combined discriminat ors (AUC=0.86) was significantly better than that of MTRasym (AUC=0.81 , p=0.001) and ADC (AUC=0.76, p<0.0001). Specificity (SP) and accuracy (AC) of combined discriminators (SP: 72.1%, AC: 78.6%) were signific antly higher than those of MTRasym (SP: 60.4%, p<0.0001; AC: 73.1%, p<0.0001) and ADC (SP: 64.2%, p<0.0001; AC: 74.0%, p<0.0001).

Conclusion

3D CEST imaging is considered at least as valuable as DWI and can improve capability for differentiation of malig nant from benign prostatic areas with DWI.

Limitations

LImited study population Funding for this study: Research grants from Canon Medical Systems Ethics committee - additional information: Fujita Health University Hospital Author Disclosures: Yoshiyuki Ozawa: Research/Grant Support: Smoking Re search Foundation Research/Grant Support: Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, Sports, Sc ience and Technology Masahiko Nomura: Nothing to disclose Takahiro Ueda: Research/Grant Support: Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Daisuke Takenaka: Nothing to disclose Hiroyuki Nagata: Research/Grant Support: Grants-in- Aid for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Canon Medica l Systems Corporation Takeshi Yoshikawa: Nothing to disclose Yoshiharu Ohno: Research/Grant Support: Smoking Res earch Foundation Research/Grant Support: Canon Medical Systems Corpo ration The comparison between virtual non-contrast and tru e non-contrast imaging of adrenal masses on photon-counting CT *X. Bai*¹, G. Zhang¹, X. Zhang¹, J. Zhang¹, L. Chen ¹, Q. Peng¹, Z. Lin², H. Sun¹, Z. Jin¹; ¹Beijing/CN, ²Shanghai/CN Purpose or Learning Objective: To investigate the differences in CT attenuation and radiomics features between virtual non-contrast (VNC) and true non-contrast (TNC) of adrenal masses on photon -counting CT (PCCT).

Methods

or Background: Patients with adrenal masses who underwent unenhanced and portal-venous-phase PCCT were includ ed. Image reconstructions of portal-venous phase were perform ed, including Conventional VNC (VNCconv) and PureCalcium VNC (VNC pc) algorithms. For two dimensional (2D) measurements, we measured CT a ttenuation of adrenal mass at the largest slice on TNC, VNCconv, and VNCp c images, respectively. Semiautomatic segmentations of adrenal masses were performed to extract three-dimensional (3D) CT attenuation and radiomics features on TNC and VNC. The paired t-test, Bland–Altman plots and intr aclass correlation efficient (ICC) were used for statistical analyses.

Results

or Findings: The study consisted of 54 patients (27 female, mean age 45.3 years) with 68 adrenal lesions. CT attenua tion on VNCconv and VNCpc was higher than that on TNC (Mean differences , 2D: 8.03 HU for VNCconv and 5.76 HU for VNCpc, 3D: 8.75 HU for VNCc onv and 6.89 HU for VNCpc). The proportion of lipid-rich adenomas (TNC < 10 HU) correctly classified by VNCconv and VNCpc was 26.1% (6/23) an d 39.1% (9/23), respectively. TNC, VNCconv, and VNCpc attenuation d id not differ significantly between 2D and 3D measurements (all P ＞ 0.05). The ICCs of first-order features, shape features and texture features betwe en TNC and VNCconv were 0.625, 0.820 and 0.591, respectively.

Conclusion

The VNC algorithms of PCCT overestimated CT attenua tion. CT attenuation at the largest slice can replace 3D att enuation. VNC and TNC have excellent agreement on shape features and average a greement on first-order features and texture features.

Limitations

The limitations of the study are the study sample i s relatively small and the retrospective study may have a select ion bias. Funding for this study: Funding was provided by the National High-Level Hospital Clinical Research Funding (2022-PUMCH-B-06 9, 2022-PUMCH-A- 033 and 2022-PUMCH-A-035), CAMS Innovation Fund for Medical Sciences (2022-I2M-C&T-B-019) and Beijing Municipal Natural Science Foundation (L232133). Ethics committee - additional information: The study was approved by the Institutional Review Board (No. 23PJ1487). Author Disclosures: Jiahui Zhang: Nothing to disclose Gumuyang Zhang: Nothing to disclose Hao Sun: Nothing to disclose Xin Bai: Nothing to disclose Qianyu Peng: Nothing to disclose Zijing Lin: Nothing to disclose Li Chen: Nothing to disclose Zhengyu Jin: Nothing to disclose Xiaoxiao Zhang: Nothing to disclose 08:00-09:00 Research Stage 2 Research Presentation Session: Cardiac RPS 1703 Cardiac imaging: interactions with other organs and systemic diseases Moderator P. Krumm; Tübingen/DE Author Disclosures: Patrick Krumm: Research Grant/Support: Spimed AI, S iemens Healthcare; Speaker: Bayer Healthcare, Siemens Healthineers, Br acco Association between pericoronary fat attenuation in dex, fractional flow reserve and brain white matter hyperintensity: a ca se control study *J. Qin*, Y. Xu; Nanjing, Jiangsu/CN Purpose or Learning Objective: To explore the association between coronary computed tomography angiography (CCTA)- derived per icoronary fat attenuation index (pFAI), fractional flow reserve ( CT-FFR) and degree of white matter hyperintensities (WMH).

Methods

or Background: Clinical, CCTA and brain magnetic resonance imaging (MRI) data of 561 participants were retrosp ectively analyzed. WMH were assessed in periventricular (PVWMH) and deep ( DWMH) locations, and a total Fazekas score was calculated by summing the s cores for PVWMH and DWMH. The study cohort was classified into mild WMH group (score 0-2) and moderate-to-severe WMH group (score 3-6). Coronary artery disease (CAD) was defined as one or more coronary arteries with d iameter stenosis of ≥50%. The threshold value of CT-FFR was set as 0.80. Clin ical data, pFAI, CT-FFR and other coronary parameters were compared between two groups, and independent variables associated with moderate-to-s evere WMH were identified using multiple logistic regression analy sis.

Results

or Findings: Compared with patients with mild WMH, those with moderate-to-severe WMH showed larger volume of plaq ue (total, calcified, noncalcified palque), higher plaque burden, longer plaque length, higher Agatston Score, higher value of pFAI and higher pro portion of CT-FFR≤ 0.80, higher proportion of patients with CAD and aortic u lcers (all p<0.05). Multiple logistic regression indicated that age [odds ratio (OR), 1.028; p=0.025], CAD (OR, 5.282; p=0.004), plaque burden (OR, 4.101; p=0 .004) , noncalcified plaque burden (OR, 2.850; p<0.001) and pFAI (OR, 1. 109; p<0.001) were independently associated with moderate-to-severe WM H.

Conclusion

Besides the well-known factors including age, CAD a nd plaque burden, pFAI was also found to be associated with m oderate-to-severe WMH.

Limitations

As a cross-sectional study, the progression of WMH was not analyzed. Future research exploring the progression of CCTA- derived parameters including CT-FFR and pFAI and the progre ssion of WMH would be more valuable. Funding for this study: This work was supported by the grants from the Natural Scientific Foundation of China (Grant Nos. 82302163 for Yunfei Wang) and Young Scholars Fostering Fund fo the First Affi liated Hospital of Nanjing Medical University (Grant Nos. PY2022036 for Yunfei Wang). Ethics committee - additional information: The requirement of written informed consent was waived due to the retrospectiv e nature. Author Disclosures: Jie Qin: Nothing to disclose Yi Xu: Nothing to disclose Quantitative T1 mapping for the evaluation of the i ron overload in hereditary hemochromatosis *G. Prencipe*, I. Notarangelo, P. Mangano, L. Marin elli, L. Macarini, G. Guglielmi, M. Gravina; Foggia/IT ([email protected]) Purpose or Learning Objective: The aim of this study was to evaluate the efficacy of cardiac T1 mapping MRI sequences in ass essing myocardial iron overload in patients with Hereditary Hemochromatosi s, in comparison to the more commonly employed T2* sequences.

Methods

or Background: A total of 28 cardiac MRI scans, conducted between 2019 and 2023, were analyzed. All patients had Hereditary Hemochromatosis and elevated serum ferritin levels. The MRI scans were performed using a Philips Achieva dStream 1.5T scan ner with cardiac gating and included T2* and native T1 mapping sequences, w ithout the use of contrast agents. Quantitative T2* analysis was cond ucted by placing regions of interest (ROIs) in the interventricular septum (IVS ), and these results were compared with the corresponding native T1 mapping f indings. Saturday Abstract-based Programme 197

Results

or Findings: Out of the 28 patients, 25 had T2* values over 20 m sec, indicating no significant iron deposition. The rema ining three patients exhibited T2* values between 15 and 20 msec, suggesting mild iron overload. None of the patients demonstrated T2* values below 15 msec. In the three patients with mild iron overload, T1 mapping showed corresponding ly low values, consistent with their T2* measurements. However, 5 of the 25 p atients with normal T2* values had T1 values below the expected range (980- 1080 msec for the our scanner).

Conclusion

Native T1 values were reduced in patients with myoc ardial iron accumulation and correlated well with T2* measureme nts. T1 mapping provides the added benefit of detecting early-stage iron overload that might be missed by T2* alone, making it a valuable tool for the early diagnosis of iron deposition and for monitoring the effectiveness of chelation therapy

Limitations

The study is retrospective. Single center study. Fe w patients with low T2*. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is retrospective Author Disclosures: Luca Marinelli: Nothing to disclose Giuseppe Guglielmi: Nothing to disclose Ilenia Notarangelo: Nothing to disclose Matteo Gravina: Nothing to disclose Gianluca Prencipe: Nothing to disclose Paola Mangano: Nothing to disclose Luca Macarini: Nothing to disclose Sex-Specific Prognostic Value of Opportunistic Epic ardial Adipose Tissue Quantification on Lung Cancer Screening Ches t CT *E. Norton*, J. M. Brendel, I. Hadzic, T. Mayrhofer , I. L. Langenbach, M. C. Langenbach, M. T. Lu, H. Aerts, B. Foldyna; B oston, MA/US ([email protected]) Purpose or Learning Objective: To evaluate the prognostic value of epicardial adipose tissue (EAT) volume and density between women and men (who often present with different body fat distribu tion) eligible for lung cancer screening, a group with an unmet need for better ri sk stratification.

Methods

or Background: Using a validated deep-learning algorithm, we measured EAT volume (indexed to body-surface-area; cm³/m²) and density (HU) on baseline non-contrast, low-dose chest CTs f rom the National Lung Screening Trial. Associations with 12-year all-caus e and cardiovascular mortality were assessed using multivariable Cox reg ression models, stratified by sex, and adjusted for age, race/ethnicity, smoki ng status (current vs. former), pack-years, history of heart disease or st roke, diabetes mellitus, hypertension, educational status, and body-mass-ind ex.

Results

or Findings: Among 24,090 participants, 9,886 were women (41%; 61±5 years), and 14,204 were men (59%; 62±5 years). Women presented lower EAT volumes and higher densities than men (65 .7±23.2cm³/m² and - 77.2±5.1HU vs. 73.5±25.0cm³/m² and -78.0±5.2HU; p<0.001 for sex differences). A 10 cm³/m² EAT volume increase revea led similar prognostic values in women and men (all-cause mortality: aHR:1 .10 [95%-CI: 1.07–1.13] in women vs. aHR:1.11 [95%CI: 1.08–1.13] in men; ca rdiovascular mortality: aHR:1.13 [95%-CI: 1.06–1.20] in women vs. aHR:1.16 [95%-CI: 1.11–1.21] in men). EAT density demonstrated a stronger associati on with all-cause mortality in men than in women (aHR:1.41 [95%-CI: 1.28–1.56] in men vs. aHR:1.28 [95%-CI: 1.10–1.47] in women; p=0.010 for sex as in teraction term). The associations were similar with cardiovascular death (aHR:1.76 [95%-CI: 1.29– 2.41] in women vs. aHR:1.78 [95%-CI: 1.45–2.19] in men).

Conclusion

Opportunistic EAT volume and density quantification may improve risk stratification in women and men eligib le for lung cancer screening, with EAT density being a stronger predictor of all- cause death in men than women.

Limitations

Exclusively heavy smokers were included, precluding comparison with non-smokers and lower-risk smokers. Funding for this study: Original data collection for NLST was funded by NIH . Ethics committee - additional information: Insight IRB protocol #: 2017P002400 Author Disclosures: Jan Michael Brendel: Nothing to disclose Isabel Luisa Langenbach: Nothing to disclose Emilia Norton: Nothing to disclose Hugo Aerts: Nothing to disclose Ibrahim Hadzic: Nothing to disclose Thomas Mayrhofer: Nothing to disclose Borek Foldyna: Nothing to disclose Marcel Christian Langenbach: Nothing to disclose Michael T. Lu: Nothing to disclose Cardiovascular Magnetic Resonance Features in Cirrh otic Patients with Transjugular Intrahepatic Portosystemic Shunt *J. Arenja*, D. G. Aydemir, L. Naimi, I. Molwitz, M . Sterneck, G. Adam, P. Bannas, E. Tahir, J. Erley; Hamburg/DE Purpose or Learning Objective: A transjugular intrahepatic portosystemic shunt (TIPS) results in blood flow from the splanch nic to the venous circulation. We aimed to evaluate if a TIPS impacts cardiovascul ar magnetic resonance (CMR) features in patients with liver cirrhosis.

Methods

or Background: In this retrospective, monocentric study, 60 patien ts with liver cirrhosis received a CMR exam (3T, Ingen ia, Philips or Siemens). Left ventricular (LV) mass, volumes (indexed to the body mass index [BMI]), and ejection fraction (EF) were analyzed. Strain was me asured by feature tracking using Cvi42 (Circle Vascular Imaging). For statisti cs, mixed linear models were conducted.

Results

or Findings: 30 patients with TIPS (40% females, age 56±11 years , BMI 26.8±5.2 kg/m2, hepatic venous pressure gradien t (HVPG) pre-TIPS 27.9±6.7 and post-TIPS 11.2±4.1 mmHg) were compared to 30 matched patients without TIPS (47% females, age 53±14 years , BMI 26.1±5.1 kg/m2). Patients with TIPS showed a -14.8% attenuated LV gl obal radial strain [95% confidence interval (CI): -28.1 to -1.5%] compared to patients without TIPS (p = 0.037). LV mass, volumes, and longitudinal/circumfe rential strain were not significantly different between the groups. In pati ents with TIPS, LV end- diastolic mass (regression coefficient = 1.5 ml/m2 [95%-CI: 0.1 to 3%], p = 0.040) and LV global longitudinal strain (0.6% [95% -CI: 0.1 to 1.2%], p = 0.035) were associated with HPVG post-TIPS.

Conclusion

Cirrhotic patients with TIPS show attenuated LV rad ial strain compared to patients without TIPS, possibly reflect ing subclinical LV dysfunction due to TIPS-induced increase in preload . A higher post-TIPS HPVG is associated with increased LV end-diastolic mass and attenuated LV global longitudinal strain, indicating LV hypertrop hy and subclinical dysfunction with persistent portal hypertension.

Limitations

This is a retrospective study with a small sample s ize, warranting validation in a larger cohort. Funding for this study: Not applicable. Ethics committee - additional information: The study was approved by the local ethics committee of the medical association i n Hamburg. Author Disclosures: Gerhard Adam: Nothing to disclose Lieda Naimi: Nothing to disclose Isabel Molwitz: Nothing to disclose Martina Sterneck: Nothing to disclose Peter Bannas: Nothing to disclose Jennifer Erley: Nothing to disclose Jennis Arenja: Nothing to disclose Destina Gizem Aydemir: Nothing to disclose Enver Tahir: Nothing to disclose Arrhythmic Burden, Myocardial Markers, and Long-ter m Survival in Distinct Cardiac Magnetic Resonance Subsets of Syst emic Sclerosis *E. Moliterno*¹, L. Giarletta¹, G. Rovere¹, S. L. B osello¹, G. De Luca², A. Tonutti³, M. A. D'Agostino¹, L. Natale¹, R. Mara no¹; ¹Rome/IT, ²Milan/IT, ³Rozzano/IT ([email protected]) Purpose or Learning Objective: Cardiac involvement in Systemic Sclerosis (SSc) is widely recognized as heterogeneous and, wh en clinically evident, it is associated with a poor prognosis. Recently, 5 cardi ac magnetic resonance (CMR) phenotypes in SSc have been identified (Knigh t DS et al. European Heart Journal 2023). These phenotypes do not align with the existing clinical subgroup classifications or autoantibody statuses, yet each has a distinct 5- year prognosis. Our objective is to test the long-t erm prognostic significance of this classification system in an external cohort an d to compare ECG Holter monitor parameters, NT-proBNP and troponin T levels and 10-year survival outcomes across these groups.

Methods

or Background: CMR assessments were conducted in 3 Italian tertiary centers on 143 consecutive SSc patients wh o presented symptoms of dyspnea, palpitations or chest pain. Based on the C MR findings, patients were categorized into 5 distinct groups: dilated right h earts with right ventricular failure (RVF); biventricular failure with dilatatio n and dysfunction (BVF), normal function with average cavity (NF-AC), small cavity (NF-SC), and large cavity (NF-LC).

Results

or Findings: The distributions for NF-AC, NF-SC, NF-LC, BVF, and RVF were 46.2%, 22.4%, 14.0%, 14.0%, and 3.5%, resp ectively. Proportions of male patients and pulmonary function tests showe d statistically significant differences across the subsets. Troponin T and NT-p roBNP values were similar across all subsets. The NF-LC and RVF groups exhibi ted Left Bundle Branch Block (LBB) and ventricular ectopic beats (VEB) mor e frequently compared to other groups. There was a variation in 10-year surv ival rates across the groups, with patients in the RVF, NF-LC, and BVF ca tegories showing poorer prognosis. Saturday Abstract-based Programme 198

Conclusion

This data confirms the prognostic value of the prop osed CMR subsets in an another European SSc cohort, highligh ting that subsets with poorer prognosis are associated with a higher arrhy thmic burden.

Limitations

No Funding for this study: None Ethics committee - additional information: No additional information Author Disclosures: Silvia Laura Bosello: Nothing to disclose Giacomo De Luca: Nothing to disclose Maria Antonietta D'Agostino: Nothing to disclose Eleonora Moliterno: Nothing to disclose Giuseppe Rovere: Nothing to disclose Antonio Tonutti: Nothing to disclose Luigi Natale: Nothing to disclose Riccardo Marano: Nothing to disclose Lorenzo Giarletta: Nothing to disclose Evaluating the Relationship Between Systemic Inflam mation Index (SII), Systemic Inflammation Response Index (SIRI), and Co ronary Calcium (Ca) Scoring in Atherosclerotic Cardiovascular Dise ase (ASCVD) I. T. Rakıcı, *K. F. Kaldırımoğlu*, A. S. Mahmutoğlu; Istanbul/TR ([email protected]) Purpose or Learning Objective: Atherosclerosis is the main cause of coronary artery disease, with inflammation as a key factor. Studies link SII (SII = PlateletxNeutrophil/ Lymphocyte) and SIRI (SIRI = NeutrophilxMonocyte/ Lymphocyte) to inflamatory diseases and cardiovascu lar outcomes. This study examines the relationship between these markers, Ca scoring, and ASCVD.

Methods

or Background: Images of 460 patients who underwent coronary CTA and calcium scoring were analyzed and divided i nto five groups: normal calcium score, calcium score 1-100, calcium score > 100, calcium score of 0 with hypodense plaques, and calcium score of 0 with myocardial bridging. The SIRI and SII were calculated to assess significant differences and establish cutoff values.

Results

or Findings: In the group with a calcium score greater than 100, the SIRI value was significantly higher than in all oth er groups, indicating an inflammatory process in ASCVD. A notable difference in SIRI values was observed between calcium scores of 1-100 and above 100, suggesting a correlation between higher scores and elevated infl ammatory markers. In cases with a calcium score of 0 and myocardial brid ging, no significant differences in inflammatory markers were found, sug gesting no increase in the inflammatory process. The SII value showed no signi ficant differences (p>0.05) among the other groups, except that it was signific antly higher (p 100 and 1-100 groups compared to th e calcium score 0 group with myocardial bridging. No significant difference s in SIRI and SII values were observed in cases with a calcium score of 0 and hyp odense plaques, indicating SIRI and SII values were insufficient to detect inf lammation in these cases.

Conclusion

SIRI effectively discriminates between calcium scor e levels, whereas SII is not a reliable indicator. Future res earch is needed to determine clinically relevant cutoff values in this area.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by the University of Health Sciences Istanbul Training and Research Hospital Clinical Research Ethics Committee (Decision number: 91). Author Disclosures: Abdullah Soydan Mahmutoğlu: Nothing to disclose Ibrahim Taşkın Rakıcı: Nothing to disclose Kemal Furkan Kaldırımoğlu: Nothing to disclose Myocardial strain trajectories by cardiac magnetic resonance and its relationship with myocardial fibrosis in Duchenne m uscular dystrophy associated cardiomyopathy *H. Xu*, X. Ting; Cheng du/CN ([email protected]) Purpose or Learning Objective: This study aims to explore the relationship between Myocardial strain trajectories and myocardi al fibrosis by cardiac magnetic resonance (CMR) in Duchenne Muscular Dystr ophy (DMD).

Methods

or Background: DMD is a severe, X-linked genetic disorder characterized by progressive degeneration of skelet al muscle due to mutations in the dystrophin gene. And cardiomyopathy is becom ing the leading cause of death. Cardiac magnetic resonance (CMR) and late ga dolinium enhancement (LGE) are important tools in recognizing myocardial involvement, myocardial strain assessed by CMR feature tracking imaging (FT ) may demonstrate early functional changes. Trajectories is a new method us ed to explore the dynamic changes of disease progression. Dynamic changes of myocardial strain using trajectories and myocardial tissue characteristics has not been investigated. We obtained myocardial strain parameters of more th an 2 times by CMR FT

Method

on 110 DMD patients. Global circumferential strain data was used to construct trajectory model. Late gadolinium enhance ment progression was measured. Group-based trajectory modeling (GBTM) wa s performed to detect the trajectories of FT. Data t tests or one-way ANO VA adjusting for multiple comparisons.

Results

or Findings: Three circumferential strain FT trajectory groups w ere identified as: up-down FT trajectory group(N=14), F T trajectory group(N=32), the steady FT trajectory group (N= 64). The occurre nce rate of adverse cardiac event in the three trajectory analysis groups were significantly different(P=0.03).

Conclusion

This study is the first to analyze the myocardial s train trajectory of DMD patients. It provides further evidence of the c orrelation between FT and progression of myocardial fibrosis.

Limitations

The relatively small number of patients assigned to the 3 groups may result in insufficient statistical power. Funding for this study: National funding of china Ethics committee - additional information: Ethics committee of west china second university hospital Author Disclosures: Xu Ting: Nothing to disclose Huayan Xu: Nothing to disclose 08:00-09:00 Research Stage 3 Research Presentation Session: Neuro RPS 1711 Foetal and paediatric neuroimaging: unveiling the youthful brain Moderator D. Zlatareva; Sofia/BG ([email protected]) Even Low Levels of Prenatal Alcohol Exposure Can In duce Structural Alterations in Fetal Temporal Gyrification *P. Kienast*, M. Stuempflen, J. Tischer, A. Taymour tash, G. Langs, D. Prayer, G. Kasprian; Vienna/AT ([email protected]) Purpose or Learning Objective: Around 10-20% of women in Europe consume alcohol during pregnancy, leading to negati ve neurodevelopmental outcomes such as fetal alcohol spectrum disorders ( FASD). Although the harmful impact of alcohol on neuronal development d uring pregnancy is well- established, studies on its effects during the pren atal phase remain limited. This study evaluates prenatal changes in fetal gyri fication resulting from maternal alcohol consumption using automated surfac e measurements from fetal MRI data.

Methods

or Background: This study involved 500 fetal MRI examinations where mothers were asked about their alcohol consum ption habits using tools like PRAMS and TACE. The brains were denoised, moti on-corrected, and segmented for 3D reconstruction. Gyrification indic es were calculated, and asymmetry of the cerebral cortex was quantified. Ne urotypical alcohol-affected cases were compared with healthy control cases.

Results

or Findings: The study finally included 22 alcohol-exposed fetus es (mean gestational-age [GW] 27.61 ± 3.94 weeks) and 22 non-alcohol-exposed control fetuses (mean GW 27.57 ± 3.94 weeks), match ed in a 1:1 ratio. Of the mothers in the alcohol-exposed group, 17 consumed o nly small amounts of alcohol (<14 grams per week). The typical asymmetry observed in the temporal lobes of the fetal brain was significantly diminish ed (p=0.48, 95% CI -2.24 to - 0.01) in fetuses exposed to alcohol.

Conclusion

Alcohol consumption during pregnancy impacts the de velopment of the fetal brain, with alterations observable in temporal brain asymmetry through fetal MRI. These findings align with studie s linking prenatal brain asymmetry to language development in childhood, whi ch is frequently impaired in children diagnosed with FASD. These structural c hanges are found even in cases of small alcohol consumption.

Limitations

A potential limitation of the study is the relative ly small sample size, which may limit the ability to detect more su btle effects of low-level alcohol exposure on fetal brain development. Funding for this study: The sponsor is the Medical University of Vienna. Ethics committee - additional information: This study was approved by the local IRB. Saturday Abstract-based Programme 199 Author Disclosures: Georg Langs: Nothing to disclose Athena Taymourtash: Nothing to disclose Marlene Stuempflen: Nothing to disclose Johannes Tischer: Nothing to disclose Daniela Prayer: Nothing to disclose Patric Kienast: Nothing to disclose Gregor Kasprian: Nothing to disclose Intracranial Volumes of Preterm Born Infants and Ag e-matched Fetuses: A Comparison of In Utero versus Ex Utero Conditions *G. Abaci*, G. Biechele, S. Schläger, F. Obereisenb uchner, K. Förster, A. Flemmer, J. Ricke, A. Hilgendorff, S. Stöcklein; Munich/DE ([email protected]) Purpose or Learning Objective: The aim of our study was to compare intracranial volume (ICV), brain volume (BV), and t he volumes of the inner cerebral fluid (iCSF) and outer cerebral fluid spac es (oCSF) of preterm brorn infants with age-matched fetuses and to assess the impact of postnatal therapies including ventilation.

Methods

or Background: Preterm born infants and fetuses were matched for gestational age (± 4 days), ranging from 32 to 39 weeks' gestation (GW). 26 preterm infants and 26 age-matched fetuses without reported brain pathology underwent MRI. Anatomical T2-weighted brain images were analyzed by manual segmentation.

Results

or Findings: Preterm infants were characterized by reduced ICV ( p= 0.005), BV (p= 0.007) and oCSF volumes (p= 0.074) w hen compared to age- matched fetuses undergoing prenatal imaging, wherea s ICV-corrected BV (BV/ICV, p= 0.385), oCSF (oCSF/ICV, p= 0.568) and t otal CSF (total CSF/ICV, p= 0.274) did not differ significantly between the two groups. When considering postnatal therapies, more days of noninvasive posit ive pressure ventilation were associated with higher corrected values for oC SF (oCSF/ICV, p= 0.0392), and total CSF (total CSF/ICV, p= 0.0282) as well as lower corrected values for BV (BV/ICV, p= 0.0282).

Conclusion

Preterm infants showed reduced ICV, BV and oCSF vol umes compared to age-matched fetuses in utero. CSF volum es in preterm infants were impacted by postnatal therapy, potentially ind iciating that intrathoracic pressure might influence venous return and CSF reab sorption into the venous system. In conclusion, brain and CSF volumes are im pacted by prematurity and associated therapeutic strategies. Potential me chanisms underlying these efffects and their implications for ex utero versus in utero brain development need to be further investigated.

Limitations

Retrospective design Funding for this study: None Ethics committee - additional information: Lmu 207-33 Author Disclosures: Anne Hilgendorff: Nothing to disclose Andreas Flemmer: Nothing to disclose Gizem Abaci: Nothing to disclose Sarah Schläger: Nothing to disclose Kai Förster: Nothing to disclose Sophia Stöcklein: Nothing to disclose Florian Obereisenbuchner: Nothing to disclose Gloria Biechele: Nothing to disclose Jens Ricke: Nothing to disclose Machine Learning Analysis in Diffusion Kurtosis Ima ging for Discriminating Pediatric Posterior Fossa Tumors *I. P. Voicu*, C. D'Orazio, E. Piccirilli, F. Dotta , P. Toma, G. S. Colafati; Rome/IT ([email protected]) Purpose or Learning Objective: Differentiating pediatric posterior fossa (PF) tumors medulloblastoma (MB), ependymoma (EP) and pi locytic astrocytoma (PA) remains relevant, because of prognostic implic ations. Diffusion kurtosis imaging (DKI) has not been investigated for pediatr ic PF tumors. Whole-tumor based (VOI) segmentations may improve repeatability compared to conventional region-of-interest (ROI) approaches. O ur purpose was to compare repeatability between ROI and VOI measureme nts and assess DKI accuracy in discriminating among pediatric PF tumor s with machine learning (ML) techniques.

Methods

or Background: We retrospectively analyzed 34 children (M, F, mean age 7.48 years) with PF tumors who underwent p reoperative MRI on a 3 Tesla magnet, including DKI. For each patient, two neuroradiologists segmented the whole solid tumor, the ROI of area of maximum tumor diameter and a small 5 mm ROI. The automated analysis pipeli ne included inter- observer variability (coefficient of variation- COV ), and machine learning (ML) analyses. We estimated DKI accuracy with MANOVA ana lysis. We applied SMOTE to balance the dataset and performed a Random Forest (RF) machine learning classification analysis based on all DKI m etrics from the SMOTE dataset ( 70/30 for the training and testing cohort ).

Results

or Findings: Tumor histology included medulloblastoma (15), pilo cytic astrocytoma (14) and ependymoma (5). VOI-based meas urements presented lower variability than ROI-based measurements. DKI- derived metrics discriminated accurately between PF tumors. SMOTE g enerated a balanced dataset with 45 instances (34 original and 11 synth etic, 10 EP and 1 PA). ML analysis yielded accuracy of 0.928, correctly predi cting all but one lesion in the testing set.

Conclusion

VOI-based measurements presented improved repeatabi lity compared to ROI-based measurements. ML techniques b ased on DKI-derived metrics are useful for discrimination of pediatric PF tumors.

Limitations

The study was approved by Institutional Review Boar d (IRB) of Bambino Gesù Children’s Hos-pital (RAP-2024-0001). T Funding for this study: None Ethics committee - additional information: The study was approved by Institutional Review Board (IRB) of Bambino Gesù Ch ildren’s Hos-pital (RAP- 2024-0001). Author Disclosures: Claudia D'Orazio: Nothing to disclose Paolo Toma: Nothing to disclose Francesco Dotta: Nothing to disclose Ioan Paul Voicu: Nothing to disclose Giovanna Stefania Colafati: Nothing to disclose Eleonora Piccirilli: Nothing to disclose Ultra-low dose computed tomography as an alternativ e to radiographic shunt series in the diagnosis of mechanical ventric uloperitoneal shunt complications – an ex vivo phantom study *B. Yildirim*, R. J. Serger, S. Zensen, H. Styczen, M. Schüßler, M. Forsting, C. Deuschl, M. Opitz, D. Bos; Essen/DE ([email protected]) Purpose or Learning Objective: The standard modality for the diagnosis of ventriculoperitoneal (VP) shunt failure is the radi ographic shunt series (RSS). However, ultra-low dose computed tomography (ULD-CT ) offers lower radiation exposure compared to RSS. The aim of this study was to compare the radiation doses of RSS and ULD-CT on photon-cou nting CT (PCCT) in the diagnosis of mechanical shunt failure in human phan tom models and to demonstrate the diagnostic performance of ULD-CT.

Methods

or Background: VP shunts with different mechanical complications were placed on human phantom models corresponding t o ages of 1, 5, 10 and 30 years. RSS and ULD-CT on PCCT scans based on top ograms with tube currents ranging from 10 and 55 mAs (Sn100 kV) were performed on each phantom. Effective doses of RSS in pediatric phanto ms were estimated by using the conversion factors of Seidenbusch (2006, 2008 and 2009). Effective doses of ULD-CT were estimated according to ICRP 10 3.

Results

or Findings: ULD-CT demonstrated lower effective doses compared to RSS for phantoms representing ages of 5, 10 and 30 years, while successfully detecting mechanical VP shunt complica tions in all cases. However, higher effective doses were assessed for U LD-CT scans of the 1- year phantom in comparison to RSS. The effective do ses for radiographic RSS and ULD-CT (using the lowest dose by utilizing 10 m As topograms), respectively, were as follows: 1 year: 0.056 vs. 0. 104 mSv; 5 year: 0.186 vs. 0.092 mSv; 10 year: 0.240 vs. 0.082 mSv; 30 year: 0 .641 vs. 0.050 mSv.

Conclusion

ULD-CT is a potentially superior alternative to rad iographic shunt series for the detection of mechanical VP shunt com plications in patients aged 5 years and above, which is particularly relevant i n children due to reduction of the radiation risk.

Limitations

Phantom study. Funding for this study: No funding was received for this study. Ethics committee - additional information: Ethics committee approval was not required. Author Disclosures: Michael Forsting: Nothing to disclose Maximilian Schüßler: Nothing to disclose Hanna Styczen: Nothing to disclose Denise Bos: Nothing to disclose Cornelius Deuschl: Nothing to disclose Sebastian Zensen: Nothing to disclose Raya Juliane Serger: Nothing to disclose Berk Yildirim: Nothing to disclose Marcel Opitz: Nothing to disclose Saturday Abstract-based Programme 200 Living in a gold mining community: Assessing 3rd tr imester estimated foetal weight, prevalence of multimorbidity across body systems, and a 6-year trend in birth outcomes *A. D. Piersson*¹, E. K. Effah², B. Brusah³, S. T. Quartei⁴, H. Mumuni⁵, A. Akanchimayoro³, L. Jones⁶, K. Dzefi-Tettey⁴, M. R. Asamani⁷; ¹York/UK, ²Obuasi/GH, ³Cape Coast/GH, ⁴Accra/GH, ⁵Tamale/GH, ⁶Virginia, VA/US, ⁷Michigan, MI/US ([email protected]) Purpose or Learning Objective: In this study, we assessed the following – 3rd trimester estimated foetal weight (EFW), preval ence of multimorbidity across body systems, and a 6-year trend in birth ou tcomes in a mining community, exposed to Galamsey, a term used to defi ne illegal small-scale and artisanal gold mining.

Methods

or Background: We compared 3rd trimester estimated foetal weight on ultrasound systems between pregnant women ( ≥ 18 years old) with singleton pregnancies living in a mining (n = 181) and non-mining communities (n = 260). Then, we compared health problems across body systems in another two groups (mining, n = 507; non-mining, n = 127). Further, we analyzed a 6-year trend in birth outcomes in the mi ning community located in Obuasi, Ashanti Region in Ghana.

Results

or Findings: Overall, women living in the mining area showed significantly higher 3rd-trimester EFW than those l iving in a non-mining community (mining: 2.95+/-0.66 kg; non-mining: 2.72 +/-0.04 kg; p = 0.0002). Women living in the mining community showed a highe r prevalence of morbidities in 8 body regions/conditions than those living in the non-mining community. A sub-analysis of the prevalence of the 8 body regions/conditions among women in the two communities showed significa nt difference (p = 0.016). A 6-year trend in birth outcomes showed inc reasing rates of stillbirth varying between 1.69% and 2.90%, a relatively low r ecord of congenital anomalies, and a relatively high record of newborn complications.

Conclusion

Our preliminary findings suggest a potential link b etween maternal exposure to mining-related environmental f actors and adverse perinatal outcomes, including increased foetal grow th, a higher prevalence of maternal morbidities, rising stillbirth rates, and increased newborn complications. Further study is warranted to invest igate the underlying mechanisms and long-term health implications for bo th mother and child.

Limitations

Our study did not provide quantitative maternal blo od or environmental measurements of heavy metal exposures . Funding for this study: None Ethics committee - additional information: Ethical Clearance - UCCIRB/EXT/2022/30 Author Disclosures: Lashell Jones: Nothing to disclose Klenam Dzefi-Tettey: Nothing to disclose Benjamin Brusah: Nothing to disclose Albert Dayor Piersson: Nothing to disclose Augustine Akanchimayoro: Nothing to disclose Emmanuel Kofi Effah: Nothing to disclose Mercedes Rowe Asamani: Nothing to disclose Hanifatu Mumuni: Nothing to disclose Sarah Teiko Quartei: Nothing to disclose 08:00-09:00 Research Stage 4 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 1705 Human and machine factors in artificial intelligence Moderator E. Kotter; Freiburg/DE ([email protected]) Author Disclosures: Elmar Kotter: Advisory Board: contextflow; Author: thieme, springer; Share Holder: Siemens, Bayer; Speaker: siemens healthinee rs, bayer Visual acuity among participants of the European Co ngress of Radiology 2024: should visual assessment be recommended for r adiologists? *T. Van Nijnatten*¹, M. Smidt¹, J. E. Wildberger¹, M. Fuchsjäger², F. J. Gilbert³, F. Pediconi⁴, R. G. H. Beets-Tan⁵, F. Van Den Biggelaar¹, C. Catalano⁴; ¹Maastricht/NL, ²Graz/AT, ³Cambridge/UK, ⁴Rome/IT, ⁵Amsterdam/NL ([email protected]) Purpose or Learning Objective: Currently there is no recommendation regarding visual assessment for radiologists. The a im was to evaluate visual acuity among participants of the European Congress of Radiology (ECR) 2024.

Methods

or Background: Participants of ECR 2024 organized by the European Society of Radiology (February 28th-March 3rd 2024; Vienna, Austria) were asked to participate in an on-site vi sual assessment. Medical ethical approval was obtained. Each participant sig ned written informed consent. The assessment consisted of vision chart r eading (Sloan ETDRS Vision Chart) at 66 cm. Afterwards, auto-refraction was performed to determine refractive error. Finally, participants re-read a d ifferent vision chart, using on- site glasses to correct for refractive error. A Log MAR score of 0.0 (i.e. Snellen equivalent 1.0/100%) was considered an adequate vis ual acuity for radiology reporting.

Results

or Findings: 321 participants completed on-site visual assessmen t (41% (132/321) male and 59% (189/321) female). Repo rted professions were 114 consultant or board certified radiologists (36% ), 121 radiology residents (38%), 24 PhD students in radiology (7%), 37 medica l students (12%), 11 radiographers (3%), 3 medical physicists (1%) and 1 1 others (3%). Mean age was 30 years (range: 18-69). Of the 57% (182/321) p articipants who wore glasses/contact lenses, 171 (94%) wore glasses/cont act lenses during image interpretation tasks. Among all participants, 24 pa rticipants (7.5%) did not achieve logMAR score of 0.0 when reading the first vision chart. After auto- refraction measurements, 11 out of these 24 partici pants improved to a logMAR score of at least 0.0 using on-site glasses to correct for the refractive error.

Conclusion

A considerable percentage of radiologists has accur ate visual acuity at a radiology reporting distance of 66 cm. Yet, 7.5% of the participants of the on-site visual assessment did not achieve an adequate vision score. Visual assessment could be considered among radiolo gists.

Limitations

N/a Funding for this study: N/a Ethics committee - additional information: Metc 2023-0249. Author Disclosures: Marjolein Smidt: Nothing to disclose Thiemo Van Nijnatten: Nothing to disclose Michael Fuchsjäger: Nothing to disclose Fiona J. Gilbert: Nothing to disclose Joachim E. Wildberger: Nothing to disclose Regina G. H. Beets-Tan: Nothing to disclose Frank Van Den Biggelaar: Nothing to disclose Federica Pediconi: Nothing to disclose Carlo Catalano: Nothing to disclose Saturday Abstract-based Programme 201 Exploring how AI influences human gaze behaviour du ring mammography reading A. Taib¹, G. Partridge¹, P. Phillips², J. James¹, * Y. Chen*¹; ¹Nottingham/UK, ²Lancaster/UK ([email protected]) Purpose or Learning Objective: Most studies assess artificial intelligence’s (AI) diagnostic performance in mammography, but few examine its impact on human reader behaviour and decision-making. The aim was to investigate the influence of AI prompts on human performance, visua l search patterns, reader confidence and look for any interaction with reader s individual personality traits when reading standard 2D screening mammograms.

Methods

or Background: In this paired reader study, eight readers working in the UK breast screening programme evaluated a set o f 60 anonymised mammograms with and without AI (Lunit Insight MMG). Cases with false negative and false positive AI prompts were incorpo rated into the test set containing a mix of normal, benign and malignant ca ses. Readers initially assessed the mammograms without AI while their visu al search behaviour was monitored using eye-tracking equipment (SmartEyePro ). After a six-week washout period, readers reviewed the same cases wit h the addition of AI with eye tracking. For each read, clinical opinion was r ecorded using a scale (1- normal, 2-benign, 3-indeterminate, 4-suspicious, 5- malignant) and entered onto the Personal Performance in Mammographic Scree ning (PERFORMS) website. Each reader completed a specially designed psychological questionnaire.

Results

or Findings: A paired analysis at breast level, using pathologic al data as the ‘ground-truth’, determined how correct and i ncorrect AI prompts influenced diagnostic accuracy, gaze behaviour and reader confidence. The effect of different reader personality traits was a lso correlated with these outcomes.

Conclusion

There is little evidence exploring how AI influence s a reader’s visual search patterns during mammography interpret ation. This pilot study provides an insight into changes in reader behaviou r when using AI and will help guide further studies and recommendations on h ow radiologists should interact with AI when interpreting screening mammog raphy.

Limitations

The limited sample of human readers may lead to typ e two error. Funding for this study: By Lunit. Ethics committee - additional information: The study was approved by the institutional review board. Author Disclosures: George Partridge: Nothing to disclose Peter Phillips: Nothing to disclose Adnan Taib: Nothing to disclose Jonathan James: Nothing to disclose Yan Chen: Nothing to disclose Narrow AI as Double Edged Sword: effects of using A I for fracture detection on distributing attention among focal and peripheral tasks *F. Mol*¹, D. Pourhassan Gilkalaye¹, M. H. Rezazade Mehrizi¹, W. Grootjans²; ¹Amsterdam/NL, ²Leiden/NL ([email protected]) Purpose or Learning Objective: Exploring the impact of using AI on distributing attention among focal tasks (diagnosis ) versus peripheral tasks (detection of incidental findings) in shoulder radi ographs by radiographers.

Methods

or Background: 17 radiographers evaluated 255 shoulder radiographs from 15 outpatient trauma patients, wit h fracture detection as primary task. To assess impact of AI assistance, 17 0 cases were analysed using commercially available AI software for fractu re detection (Gleamer). Both fracture (204) and non-fracture (51) cases were inc luded. Additionally, 102 cases had incidental findings (e.g., pulmonary nodu les, bone cysts, rotator cuff calcifications). Eye-tracking (Tobii 5) and mouse-t racking (in-house) software were used to measure attention distribution. Data a re presented as mean ± standard deviation, and statistical differences wer e assessed using Wilcoxon signed-rank test, with significance defined as p<0. 05.

Results

or Findings: Participants spent more time on cases with AI assistance (n=170), averaging 47±28.24 seconds, com pared to 33 ± 28.24 seconds without AI (p < .001). Mouse clicks were 3. 85 ± 10.24 without AI, and 12.09 ± 10.24 with AI (p < .001). Eye-tracking data indicated greater attention to peripheral tasks with AI assistance (p = 0.024), while fracture detection was higher without AI (p = 0.038). Radiographers mentio ned 31% of incidental findings.

Conclusion

Use of narrow AI tools can increase sensitivity tow ards peripheral tasks, possibly resulting from enhanced cognitive a vailability by delegating the focal task to AI. Similarly, measured decrease in f racture detection with AI indicates decrease in attention towards the focal t ask. This positions narrow AI as a “double-edged sword”: while automation can fre e up cognitive resources, it can lead to over-reliance and reduced attention to focal tasks.

Limitations

Real-world clinical environment may not be simulate d completely in experimental setting. Eye and mouse tracking may not capture all attention distribution aspects. Funding for this study: N.a. Ethics committee - additional information: N.a. Author Disclosures: Dorsa Pourhassan Gilkalaye: Nothing to disclose Ferdinand Mol: Nothing to disclose Willem Grootjans: Nothing to disclose Mohammad Hosein Rezazade Mehrizi: Nothing to disclo se Colour map recommendations for MR relaxometry *B. D. Wichtmann*¹, M. Fuderer², N. Desouza³, F. Cr ameri⁴, V. Gulani⁵, N. Sollmann⁶, S. Weingärtner⁷, S. Mandija², X. Golay³; ¹Bonn/DE, ²Utrecht/NL, ³London/UK, ⁴Bern/CH, ⁵Ann Arbor, MI/US, ⁶Ulm/DE, ⁷Delft/NL ([email protected]) Purpose or Learning Objective: Quantitative imaging data may be colour coded and represented as a colour-map. However, com monly used schemes (e.g. rainbow, jet) lack perceptual uniformity, hav e the brightest colour mid- range and are not usable by colour-blind individual s. Furthermore, lack of standardization of colour-maps, makes comparisons a cross studies and institutions difficult and misleading. This work de scribes recently published recommendations for standardisation of MR relaxomet ry colour-maps (Fuderer, MRM 2024) in order to promote their adoption and dr ive the process for other biomarkers.

Methods

or Background: Recommendations were generated in 4 Delphi rounds. A multidisciplinary committee devised quest ions on key colour-map features, including the colour choice for T1/T2 map s, even colour gradient contrast, high overall colour and lightness contras t, intuitive and constant gradient magnitude, and recognizability. Questions were circulated to the ISMRM quantitative imaging group and European subsp ecialist society representatives. Respondents received feedback afte r each round to aid consensus. Responses on a 9-point Likert scale were summarised to Agree, Neutral, Disagree categories. 75% consensus was the threshold for items reaching recommendation. The proposed colour maps w ere based on previous proposals (Griswold, ISMRM 2018) but modified for p erceptual linearity and readability by colour-blind people.

Results

or Findings: 58 experts responded to Round 1; 48 (45% medical, 47% physicists) completed all 4 rounds. There was c onsensus that the logarithm-processed Lipari colour-map for T1 and th e logarithm-processed Navia colour-map for T2 were suitable. Colour bars were deemed mandatory as was a specific value indicating “invalidity”. Th ere was no consensus on whether to fix ranges by anatomy.

Conclusion

The logarithm-processed Lipari colour map for displ aying T1 and R1 values and the logarithm-processed Navia colour- map for displaying T2, R2, T2* and R2* are recommended for use in scientif ic reports.

Limitations

Future work will focus on range recommendations. Funding for this study: No funding was provided for this study. Ethics committee - additional information: No patient-sensitive data were processed in this study. Author Disclosures: Nandita Desouza: Nothing to disclose Vikas Gulani: Grant Recipient: I get research suppo rt from Siemens Healthineers. I have intellectual property licensed by Siemens Healthineers. Fabio Crameri: Nothing to disclose Miha Fuderer: Grant Recipient: NWO grant number 179 86, which has partly been funded by the company Philips Sebastian Weingärtner: Nothing to disclose Barbara Daria Wichtmann: Speaker: I have given scie ntific presentations for Philips GmbH, Lilly Deutschland, and Bender group/b .e.imaging GmbH on unrelated topics for which I received monetary comp ensation. Stefano Mandija: Nothing to disclose Xavier Golay: Consultant: Bioxydyn Founder: Gold St andard Phantoms: Founder, shareholder and employee Nico Sollmann: Nothing to disclose Evaluating the Impact of Explainable AI on Anchorin g and Automation Biases in Mammography Interpretation *F. Pesapane*, L. Nicosia, S. Carriero, L. Mariano, A. C. Bozzini, A. Latronico, L. Meneghetti, F. Abbate, E. Cassano; Milan/IT ([email protected]) Purpose or Learning Objective: This study investigates how AI support influences diagnostic biases (anchoring and automat ion) among radiologists with varying experience levels in breast imaging. I t evaluates whether explainable AI (XAI), using a heatmap, reduces thes e biases and improves diagnostic accuracy.

Methods

or Background: Six radiologists (2 low experience: 0-5 years, 2 medium: 5-10 years, 2 high: >10 years) participated . Each assessed 200 mammograms across two phases: (1) AI BIRADS score p resented before diagnosis (anchoring phase), and (2) AI score prese nted after an independent diagnosis (automation phase). A crossover design wa s used, with a 30% AI Saturday Abstract-based Programme 202 error rate. Two AI conditions were tested: standard (score only) and explainable (score with heatmap). Diagnostic change s, accuracy, and bias frequency were recorded, with subgroup analyses bas ed on experience.

Results

or Findings: In the anchoring phase, radiologists altered their diagnoses in 180/400 cases (45%) when AI was incorr ect; XAI reduced this to 100/400 (25%). In the automation phase, 220/400 cor rect diagnoses (55%) changed after AI input; XAI reduced this to 120/400 (30%). Low-experience radiologists showed higher susceptibility, particul arly in automation (260/400, 65% change rate). XAI improved accuracy in this gro up by 80/400 cases (20%). Experienced radiologists demonstrated minima l bias reduction with XAI, indicating experience as a moderating factor.

Conclusion

XAI reduces anchoring and automation biases, especi ally for less experienced radiologists when AI errors are present . Tailored AI solutions with explainability are crucial for unbiased decision-ma king in breast imaging.

Limitations

This study involved a small sample size, which may limit generalizability. Future studies should expand the sample and explore the long-term impact of XAI on diagnostic confidence. Funding for this study: N/A Ethics committee - additional information: Code of approval: UID 4810 Author Disclosures: Luciano Mariano: Nothing to disclose Luca Nicosia: Nothing to disclose Antuono Latronico: Nothing to disclose Anna Carla Bozzini: Nothing to disclose Serena Carriero: Nothing to disclose Francesca Abbate: Nothing to disclose Filippo Pesapane: Nothing to disclose Enrico Cassano: Nothing to disclose Lorenza Meneghetti: Nothing to disclose The diagnostic performance of an AI model in prosta te cancer detection decreased significantly in reduced scan-quality of biparametric MRIs, while radiologists’ performances did not decrease *E. H. P. Pooch*¹, G. Agrotis¹, A. Dehghanpour², R. G. H. Beets-Tan¹, T. Janssen¹, I. G. Schoots¹; ¹Amsterdam/NL, ²Rome/I T Purpose or Learning Objective: To assess the diagnostic performance of artificial intelligence (AI) model and radiologists in detecting Grade Group (GG) ≧2 disease in prostate cancer suspected men, on diag nostic biparametric MRI (bpMRI) scans, considering variations in scan quali ty as assessed by PI-QUAL scores.

Methods

or Background: A nnU-Net GG≧2 cancer segmentation model used 1500 bpMRI scans for training (cohort PI-CAI) and 8 9 scans for external validation (cohort PROMIS). The external cohort ana lysis included PI- RADSv2.1 assessment by two readers (R), while one a ssigned PI-QUALv1 (MRI-quality) scores. The outcome measurement was G G≧2 cancer, based on biopsies. MRI-positive scans were defined as PI-RAD S 3-5 scores. The model’s and radiologists' diagnostic performance (A UCs) were compared.

Results

or Findings: Overall, the trained model (AUC=0.888) achieved an AUC of 0.652(0.525-0.760) during external validatio n. At reduced scan quality (PI-QUAL 1-3), the model’s AUC dropped to 0.552(0.3 50-0.747), while at high- quality scans (PI-QUAL 4-5), the model’s AUC improv ed to 0.720(0.556- 0.855). In contrast, the AUCs of R1 and R2 were 0.7 33(0.631-0.829) and 0.711(0.614-0.803), respectively, showing a signifi cant difference to the AI model in the reduced-quality group (p0.99). The readers’ AUCs did not d rop at reduced scan quality (AUCs 0.723(0.576-0.875) and 0.727(0.576-0.862) and did not improve at high-quality scans (0.743(0.616-0.848) and 0.695(0. 562-0.812)), respectively.

Conclusion

The diagnostic performance of the AI model differed significantly between reduced- and high-quality scans. In contras t, radiologists maintained consistent diagnostic accuracy. To ensure optimal p erformance, consistently high-quality MRI scans are required for a successfu l implementation of AI in clinical practice.

Limitations

Limited sample size and only one radiologist provid ed PI-QUAL scores, which may limit the findings’ generalizabil ity. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was made using public data. Author Disclosures: Georgios Agrotis: Nothing to disclose Ailin Dehghanpour: Nothing to disclose Eduardo H. P. Pooch: Nothing to disclose Tomas Janssen: Nothing to disclose Regina G. H. Beets-Tan: Nothing to disclose Ivo Gerardus Schoots: Nothing to disclose Variability of classification labels is an importan t barrier to effective comparison of artificial intelligence software betw een vendors *A. Maiter*, E. Hesketh, P. Metherall, J. Taylor, S . Alabed, K. Dwivedi, W. Tindale, A. Swift, C. S. Johns; Sheffield/UK Purpose or Learning Objective: Comparing the performance of AI software between different vendors is important for guiding procurement and deployment decisions. This requires consistency in how software outputs are presented. We assessed the number, nature and termi nology of classification labels provided by commercially available software from seven vendors for the evaluation of chest radiographs.

Methods

or Background: The classification labels provided by the software from each vendor were appraised qualitatively and u sing descriptive statistics. Synonymous labels were reconciled by merging. Label s were categorised according to their intended purpose. Where relevant , label terminology was compared with the 2024 and 2008 Fleischner Society Glossary of Terms for Thoracic Imaging.

Results

or Findings: The median number of labels per vendor was 17 (IQR 7 to 100). Most labels were for the detection of path ology (median 88%, IQR 74% to 94%); these varied from non-specific signs ( e.g. ‘bronchovascular markings’) to specific diagnoses (e.g. ‘sarcoidosis ’). In some cases, individual vendors provided multiple labels with overlapping m eanings (e.g. ‘consolidation’, ‘air bronchogram’, ‘air space opac ification’ and ‘alveolar pattern opacity’). Fewer labels were for the detection of d evices (median 12%, IQR 0% to 23%); these also ranged from non-specific (e.g. ‘catheter’) to more specific with a decision on adequacy (e.g. ‘suboptimal nasog astric tube’). The median concordance of terminology with the Fleischner Soci ety Glossary was 58% (IQR 50% to 72%).

Conclusion

We identified considerable variability in the label s provided by software, including inconsistent adherence to estab lished terminology. This represents a barrier to effective comparison of per formance between vendors and potentially limits the clinical utility of soft ware outputs. Our study highlights the need for better harmonisation of output labels across the AI field.

Limitations

The interpretation of classification labels can be subjective and may differ between assessors. Funding for this study: This study was funded by the NHS South Yorkshire Integrated Care System. Ethics committee - additional information: This study did not involve patient data, and no ethics committee approval was required . Author Disclosures: Jonathan Taylor: Nothing to disclose Andrew Swift: Nothing to disclose Ahmed Maiter: Other: Bayer Chris S Johns: Nothing to disclose Samer Alabed: Nothing to disclose Peter Metherall: Nothing to disclose Eleanor Hesketh: Nothing to disclose Krit Dwivedi: Nothing to disclose Wendy Tindale: Nothing to disclose Rethinking Radiology Reports: The Perspective of Re ferring Physicians *P. Reschke*, L. D. Gruenewald, V. Koch, E. Höhne, T. J. Vogl, J. Gotta; Frankfurt/DE Purpose or Learning Objective: High report quality and completeness are essential for efficient patient management. However , the clarity and comprehensiveness of radiology reports are often a point of contention among referring physicians. This study aims to assess ref erring physicians’ perspectives on the quality and utility of radiolog y reports in clinical practice.

Methods

or Background: A prospective, anonymous online survey was conducted from June 2023 to June 2024, targeting pr acticing physicians in Germany, including internists, general practitioner s, surgeons.

Results

or Findings: A total of 149 participants were included: 40% inte rnists, 35,8% general practitioners, 24,2% surgeons. The av erage satisfaction score for radiology report completeness was 34.4 (±42.3) on a scale from -100 to +100. The primary reasons for incomplete reports we re a lack of clinical context (33.3%), missing prior imaging (18.6%), ina ppropriate imaging techniques (13.8%) and unclear clinical questions ( 11,6%). Nearly half of the respondents (48.9%) preferred concise reports, whil e 35.7% opted for medium- length, and only 15.4% favored detailed reports. A majority of participants preferred semi- or fully structured reporting forma ts (92.5%), with free-text being rarely chosen (7.5%), showing no significant differences across specialties (p=0.08). Most participants (84.1%) fou nd imaging in interdisciplinary case conferences valuable for und erstanding reports, with 35.9% rating them as “very helpful.”

Conclusion

Referring physicians strongly prefer structured rep orting and concise radiology reports. Integrating imaging into interdisciplinary meetings can further improve report comprehension. Saturday Abstract-based Programme 203

Limitations

As the survey was conducted exclusively in Germany, the results may not be directly applicable to other healthcare systems or international settings, where clinical practices and communicatio n standards may differ. The study focused on a limited number of specialties (i nternists, general practitioners, surgeons), potentially overlooking t he perspectives of other key stakeholders such as neurologists, oncologists, or emergency medicine physicians. Funding for this study: None. Ethics committee - additional information: Not applicable Author Disclosures: Leon D. Gruenewald: Nothing to disclose Vitali Koch: Nothing to disclose Philipp Reschke: Nothing to disclose Thomas J. Vogl: Nothing to disclose Jennifer Gotta: Nothing to disclose Elena Höhne: Nothing to disclose 08:00-09:00 Room G1 Research Presentation Session: Radiographers RPS 1714 Advancements in MR: safety, patient care, and technological innovation Moderators S. B. Grover; Greater Noida/IN ([email protected]) C. Tsiotsios; Limassol/CY ([email protected]) MRI safety education across Europe: perspectives on teaching and assessment J. Mcnulty¹, A. Cradock¹, A. Mcgee¹, *A. De Bock*², J. L. Portelli³, A. England⁴; ¹Dublin/IE, ²Brussels/BE, ³Msida/MT, ⁴Utrecht/NL Purpose or Learning Objective: Little has been published on the details of curricular content for radiographers in MRI safety despite the EFRS MRSO role descriptor benchmark document being published in 20 21. This survey aimed to capture the current status quo in terms of MRI safe ty education for radiographers across Europe in terms of content, pe dagogical and assessment approaches, challenges, and opportunities.

Methods

or Background: An online survey of educational institutions (EIs) involved in the delivery of MRI safety education to radiographers was designed. The survey was distributed via the EFRS t o all member EIs. Additionally, it was shared with all member nationa l societies and to committee and working group members for onward distribution t o EIs.

Results

or Findings: Responses were received from 69 EIs. 58 EIs confirm ed they deliver MRI safety education to radiographers (23 to undergraduate (UG) medical imaging (MI) students, 23 to UG combined MI and radiotherapy (RT) students, 5 to UG RT students, and 22 to Masters st udents. The most common MRI safety teaching methods for postgraduates (PGs) were in-person lectures (70.8%), small group discussions (61.9%), review of MRI safety scenarios (61.9%), and clinical placements (61.9%). The MRI s afety contact hours for PGs averaged just 6 (range: 0-30 hours). For PGs, a t the ‘advanced level’, the most infrequently taught topics were ‘Digital data safety’ (27.8%) and ‘Special MRI systems’ (33.3%). 65.2% of EIs were either ‘ver y satisfied’ or ‘satisfied’ with their PG MRI safety education provision.

Conclusion

Significant heterogeneity in content and levels of delivery of MRI safety education for radiographers was evident and highlights the potential for a common MRSO curriculum.

Limitations

This was a convenience sample of EIs across Europe accessed through the EFRS network and thus may not be repres entative of the population as a whole. Funding for this study: The ECSO-MRI Project is co-funded by the Erasmus+ Programme of the European Union (KA220: 20 22-1-BE02-KA220- HED-000085873). Ethics committee - additional information: Due to the nature of this study and the target population, the University College D ublin Human Research Ethics Committee – Sciences granted a formal Ethics Exemption (Reference: LS-LR-24-203-McNulty). Author Disclosures: Jonathan Loui Portelli: Nothing to disclose Andrea Cradock: Nothing to disclose Anke De Bock: Nothing to disclose Jonathan Mcnulty: Nothing to disclose Allison Mcgee: Nothing to disclose Andrew England: Nothing to disclose Vetting MR referrals, are radiographers as good as radiologists? *E. Kjelle*, I. Ø. Brandsæter, J. Porthun, B. M. Ho fmann; Gjøvik/NO ([email protected]) Purpose or Learning Objective: This study compares radiographers' and radiologists' vetting of justification for MRI refe rrals for low back pain and headaches.

Methods

or Background: 360 lower back MR referrals and 353 brain MR referrals in adult patients were collected from pri vate imaging centers in Norway. Four experienced radiologists and four expe rienced radiographers vetted the referrals using the Choosing Wisely reco mmendations for patients with low back pain and uncomplicated headaches. The assessors had three alternatives: 1)Justified, 2)Unjustified, and 3)Nee d more information. Data was analyzed using descriptive statistics, chi-square t est to compare groups, and Gwen's AC2 for inter-rater agreement analysis. Sign ificant level was p<0.05

Results

or Findings: On average, in brain MR, the radiographers rated 52 % of the referrals as justified, compared to 53% amon g the radiologists. The unjustified rate was 32% among radiographers and 41 % among radiologists. In the category need more information, the rate was 15 % for radiographers and 6% for radiologists. The difference was statistical ly significant, p<0.001. The interrater agreement was moderate 0.45 (95% CI:0.38 -0.52) among the radiographers and good 0.71 (95% CI:0.68-0.75) amon g the radiologists. In lower back MRI, the justified rate was 65% for radi ographers and 59% among radiologists. The unjustified rate was lower among the radiographers (18%) than among radiologists (26%). While the need more information rate was higher among radiographers. The difference was stat istically significant, p<0.001. The interrater variability was moderate in both groups, with 0.60 (95% CI:0.53-0.66) among the radiographers and 0.56 (95% CI:0.49-0.62) among the radiologists.

Conclusion

Radiographers need more referral vetting training t o avoid performing unjustified imaging. In this study, radi ographers often required more information or rated a referral as justified compar ed to radiologists.

Limitations

The general clinical practices and experience could influence the assessors' vetting; thus, these results are not nec essarily generalizable. Funding for this study: This study was funded by the Norwegian Research Council (Project number 302503). Ethics committee - additional information: Regional Committees for Medical and Health Research Ethics ref.no. 378396 Author Disclosures: Ingrid Øfsti Brandsæter: Nothing to disclose Bjørn Morten Hofmann: Nothing to disclose Elin Kjelle: Nothing to disclose Jan Porthun: Nothing to disclose Assessing MRI referrals' appropriateness for low ba ck pain post a radiology-initiated intervention *C. C. Chilanga*¹, M. Heggelund¹, E. Kjelle²; ¹Dram men/NO, ²Gjøvik/NO ([email protected]) Purpose or Learning Objective: To evaluate a pilot intervention to reduce low-value Magnetic Resonance Imaging (MRI) referral s for Low Back Pain (LBP).

Methods

or Background: The study evaluated MRI referrals for LBP before and after an intervention involving information cam paigns and return letters to clinicians whose referrals were declined. Four radi ologists and two radiographers assessed the referrals based on quali ty and justification. Justification was classified as justified, unjustif ied, or requiring more information. A point system rated quality on an 8-p oint scale, with scores above 5.5 marked as "good" and below 2.5 as "poor." Pre- and post- intervention variations were analysed using mixed m odel in Stata (Release 18). A p-value <.05 was considered significant.

Results

or Findings: A total n= 300 referrals (150 pre- and post-interve ntion) were assessed. Post-intervention, rated justified r eferrals increased from 63% to 68%, while unjustified referrals decreased from 19% to 17%. Those needing more information decreased from 19% to 16%. Poor-qu ality referrals decreased to 4% post intervention. Mixed model anal ysis estimated justified referrals at 61% (95% CI: 55.8–65.5) pre-interventi on and increased to 66% (95% CI: 61.5–70.9) post-intervention. Unjustified referrals fell from 20% (95% CI: 15.9–24.3) to 17% (95% CI: 13.5–21.4), Rated go od quality referrals increased from 7% (95% CI: 4.6–9.0) to 8% (95% CI: 5.5–10.7), poor-quality referrals decreased from 30% (95% CI: 25.4–35.3) to 27% (95% CI: 22.3– 31.7). Variations were not statistically significan t. Saturday Abstract-based Programme 204

Conclusion

The intervention is ongoing and needs further evalu ation. However, providing reasons for declined referrals c an serve as an educational tool for clinicians and contribute to the reduction of low value MRI for LBP in radiology departments.

Limitations

Adherence to referrers’ confidentiality during data sampling prevented confirmation of whether all post-interven tion referrals originated from clinicians who received return letters; however, re gion-wide campaigns likely mitigated this issue. Funding for this study: The Research council of Norway (Project number 302503) Ethics committee - additional information: Regional Committees for Medical and Health Research Ethics (REK) reference number 3 78396 and Norwegian Agency for Shared Services in Education and Researc h (SIKT) reference 261461. Author Disclosures: Mina Heggelund: Nothing to disclose Elin Kjelle: Nothing to disclose Catherine Chilute Chilanga: Nothing to disclose White matter hyperintensities and silent brain infa rcts in aortic valve repair: the PEARL Study *M. Bono*¹, M. Zanardo², V. Bari², B. Cairo², A. Po rta², F. Sardanelli², P. Vitali²; ¹Varese/IT, ²Milan/IT ([email protected]) Purpose or Learning Objective: We assessed the burden of white matter hyperintensities (WMH), a biomarker of chronic cere brovascular disease, and quantified ischemic lesions using diffusion-weighte d magnetic resonance imaging (DW-MRI) in patients undergoing surgical ao rtic valve replacement (SAVR) or transcatheter aortic valve implantation ( TAVI). Our goal was to investigate whether a relationship exists between c hronic cerebrovascular disease and acute ischemia in aortic valve repair p atients.

Methods

or Background: This prospective study involved brain MRI scans at 1.5-T performed within seven days following SAVR or TAVI. The semi- quantitative Fazekas scale was used to classify WMH severity into low, intermediate, and high categories of chronic cerebr ovascular disease. WMHs and DWI-positive ischemic lesions were quantified u sing automatic segmentation by Quantib® ND and semi-automatic segm entation adjusted by a neuroradiologist and a trained radiographer.

Results

or Findings: A total of 55 patients were included in the study, of whom 47 underwent SAVR (62±15 years) and 8 TAVI (83 ±3 years). The mean lesion count for SAVR patients was 2.11 ± 7.07, while the mean lesion count for TAVI patients was 7.25 ± 6.88(p = 0.081).An analysis of WMH using the Fazekas scale indicated that SAVR group had mean Fa zekas score of 1.30 ± 0.69, while TAVI group 2.25 ± 0.71 (p = 0.053).

Conclusion

While TAVI patients tended to have a higher burden of chronic cerebrovascular disease as indicated by the Fazekas score, the lack of significant differences in ischemic lesion counts b etween SAVR and TAVI suggests that both procedures pose similar acute is chemic risks. The WMH burden may enable risk-stratification of patients w ho undergo TAVI/SAVR and identify those that would benefit most from the ado ption of neuroprotective devices.

Limitations

Low number of TAVI patients enrolled. Funding for this study: Ricerca Finalizzata Code: RF-2016-02361069. Ethics committee - additional information: Ethics committee Lombardia Author Disclosures: Paolo Vitali: Nothing to disclose Vlasta Bari: Nothing to disclose Francesco Sardanelli: Nothing to disclose Martina Bono: Nothing to disclose Alberto Porta: Nothing to disclose Moreno Zanardo: Nothing to disclose Beatrice Cairo: Nothing to disclose Preliminary results on white matter hyperintensitie s and lesion counts in patients undergoing cardiac surgery with cardiopulm onary bypass: the PASCAL study A. Nocita¹, M. Zanardo², V. Bari², B. Cairo², P. Si ngh², A. Porta², F. Sardanelli², P. Vitali², *M. Bono*¹; ¹Varese/IT, ²Milan/IT ([email protected]) Purpose or Learning Objective: This preliminary study investigates changes in white matter hyperintensities (WMH) and lesion c ounts in patients undergoing cardiac surgery with cardiopulmonary byp ass (CPB). The primary

Objective

was to assess the differences in WMH volu mes and lesion counts pre- and post-surgery, measured through brain magne tic resonance imaging (MRI) Fluid-Attenuated Inversion Recovery (FLAIR) 3 D sequence.

Methods

or Background: The first fifteen patients of the PASCAL study were included. Two brain MRI scans were acquired before the surgical intervention with cardiopulmonary bypass (pre), and after the su rgical intervention (post) but within 10 days. Pre-operative and post-operativ e WMH volumes and lesion counts were evaluated using automatic segmentation by Quantib® ND and semi-automatic segmentation adjusted by a neuroradi ologist and a trained radiographer. A t-test was conducted for statistica l analysis, while reproducibility was evaluated using a Bland-Altman plot.

Results

or Findings: The mean pre-surgical WMH volume was 0.49 cm³, which increased slightly to 0.57 cm³ post-surgery. A paired t-test showed a statistically significant difference between pre- a nd post-surgical volumes (p = 0.044). The lesion count analysis showed a mean pre -operative count of 13, increasing to 14 post-surgery. However, this differ ence was not statistically significant (p = 0.09). The Bland-Altman analysis f or WMH volume demonstrated a mean difference of 0.04 cm³ with lim its of agreement ranging from -0.15 to 0.23 cm³. For lesion count, the mean difference was 1 with limits of agreement from -5 to 7, indicating moderate repr oducibility between the two methods.

Conclusion

The results suggest that while WMH volumes may incr ease after CPB surgery, the number of lesions does not signifi cantly change, although the sample size limits the generalisability of these fi ndings. Definitive results will be obtained once the estimated sample size is reached.

Limitations

Preliminary results. Funding for this study: Ricerca Finalizzata Code: GR-2021-12372037. Ethics committee - additional information: Ethics Committee Lombardia 1, code: 06/INT/2023 Author Disclosures: Paolo Vitali: Nothing to disclose Alba Nocita: Nothing to disclose Vlasta Bari: Nothing to disclose Pavandeep Singh: Nothing to disclose Francesco Sardanelli: Nothing to disclose Martina Bono: Nothing to disclose Alberto Porta: Nothing to disclose Moreno Zanardo: Nothing to disclose Beatrice Cairo: Nothing to disclose Hardware as a Predictor of Anxiety in Patients Unde rgoing Magnetic Resonance Imaging Examinations *D. A. A. Costa*, A. Grilo, E. Carolino, M. C. P. R ibeiro; Lisbon/PT ([email protected]) Purpose or Learning Objective: The objective of this study was to assess which type of Magnetic Resonance (MRI) examination, the coil used and the patient position (to scan the brain or the knee), m ost influence patient’s anxiety levels.

Methods

or Background: Hundred patients underwent MRI scan using a 1,5T Magneton Symphony by Siemens Healtineers. Fifty of them performed brain and the others knee scan. Before and after the MRI scan, the STAI Inventories Form Y-1 (state anxiety) and the Form Y-2 (trait an xiety) was applied. The physiological measurements of Blood Pressure and He art Rate were collected at the beginning and at the end of scan.

Results

or Findings: In brain studies, the initial mean anxiety levels ( ̅ =37.28 ± 12.446) are higher than those collected after examination (̅ =33.72 ± 13.389). In the knee examinations the initial mean anxiety values (̅ =31.96 ± 10.681) are higher compared to the final values ( ̅ =29.42 ± 8.094). When comparing brain and knee studies, the first shown h igher initial anxiety levels (̅ =37.28 ± 12.446) than patients who undergo knee sca n (̅ =31.96 ± 10.681). The mean values of Maximum BP ( ̅ =130.06 ± 3.808) and Final HR (̅ =125.10 ± 1.851), collected after brain studies, are higher than the values collected in knee (̅ =88.24 ± 2.720) and (̅ =81.12 ± 1.943).

Conclusion

Through Multiple Linear Regression (Stepwise method ), the examination type and the patient's assessment befor e MRI are predictors of anxiety, as their probability of occurrence is low. Higher and significant differences were found at the beginning of MRI scan . The appliance of different transceiver coils, reduces the surrounding space, c ontributing to increasing the levels of state anxiety.

Limitations

Lack information about other pathologies. Higher pe riods of questions and evaluation Funding for this study: Not Applied Ethics committee - additional information: Approval by Ethical Commission of School of Health Technology with the reference C E-ESTeSL-Nº.107-2022 Author Disclosures: Ana Grilo: Nothing to disclose Elisabete Carolino: Nothing to disclose Diogo André Arrais Costa: Nothing to disclose Margarida Carmo Pinto Ribeiro: Nothing to disclose Saturday Abstract-based Programme 205 Educational background in MRI Safety of healthcare professionals working in MRI departments: Insights from the ECSO- MRI Project J. Scheurleer¹, A. V. Diepen¹, *A. De Bock*², H. Bi jwaard¹; ¹Haarlem/NL, ²Brussels/BE Purpose or Learning Objective: This study aimed to assess the educational needs and preferences regarding MRI safety training among healthcare professionals working in MRI departments. Findings aimed to support the development of a standardised European MRI Safety O fficer (MRSO) curriculum as part of the funded ECSO-MRI project.

Methods

or Background: A survey among radiographers, radiologists, and other healthcare professionals across European MRI departments assessed educational backgrounds, MRI safety responsibilitie s, and training preferences, with 313 responses analysed. The study received loc al ethics committee approval and confirmed European data security requi rements.

Results

or Findings: Of the respondents, 71.6% had clinical experience i n MRI, but 49.5% did not regularly update their MRI s afety knowledge. Informal peer-to-peer training was the most commonly used le arning method (62.9%). Hands-on training was the most preferred approach ( 89.5%). Additionally, 70% of respondents considered the development of Standa rd Operating Procedures (SoP) an advanced MRI safety topic. Furthermore, 40 % highlighted the need for formal basic training for all radiographers wor king with MRI, while 63% indicated a need for formal advanced training for k ey users of MRI. Only 31% reported following local MRI safety guidelines, and 18.5% adhered to international standards, underscoring the need for more consistent safety protocols.

Conclusion

Significant gaps in MRI safety education, particula rly in advanced training, were identified. The ECSO-MRI project aim s to address these by developing a standardized curriculum in collaborati on with European institutions.

Limitations

The reliance on self-reported data may introduce bi as in assessing actual MRI safety practices and training needs. Funding for this study: This study was funded by the European Union’s Erasmus+ program. Ethics committee - additional information: Ethics Committee of University College Dublin Research Ethics Research Ethics Refe rence Number is: LS-LR- 24-220-McNulty Author Disclosures: Anja Van Diepen: Nothing to disclose Anke De Bock: Nothing to disclose Jelle Scheurleer: Nothing to disclose Harmen Bijwaard: Nothing to disclose Radiographers’ Knowledge and Attitudes differences towards Cardiac Implant Patients between radiographers in Magnetic Resonance Imaging C. Maloney, A. England, N. Moore, R. Young, G. A. C urran, *M. F. Mcentee*; Cork/IE ([email protected]) Purpose or Learning Objective: Radiographers in magnetic resonance imaging (MRI) are at the forefront of patient safet y, and patients with cardiac implantable electronic devices (CIED) have been ref erred for MRI imaging more frequently in recent years. A rise in MR-condi tional devices has resulted in CIED patients accessing MRI, and previously pati ents with CIEDs were classed as absolute contraindications. However, the magnetic fields from the MRI scanner can interfere with CIEDs causing damage to the device itself and/or the patient. This study evaluates CIED speci fic knowledge, confidence levels and attitudes amongst radiographers in MRI t owards cardiac implant patients.

Methods

or Background: A quantitative, online survey was conducted amongst qualified radiographers in MRI internationa lly to evaluate participants MRI qualification status, what CIED specific knowle dge is held by radiographers, confidence levels amongst radiograph ers scanning CIED patients, and determine common attitudes amongst ra diographers towards implant patients who present for an MRI scan.

Results

or Findings: 90 responses were recorded and overall, and 58% of participants held a postgraduate degree in MRI whil e 42% did not. 73% of radiographers highlighted they had opportunities to update their MRI safety in the work setting, while only 36% had opportunities to update their cardiac implant safety knowledge. Overall, radiographers co ncluded that their MRI and CIED safety knowledge was sufficient and possessed adequate knowledge of the CIED and magnetic field relationship. However, evident disagreement on confidence statements on explaining MR field impact s on CIED functioning, having insufficient knowledge on device workings to confidently scan and conflicting safety results indicate that this area requires addressing.

Conclusion

Inclusion of specific CIED safety training for MRI radiographers should be recommended as educational topics to equi p radiographers with the knowledge, skills and confidence to competently pro vide safe patient care.

Limitations

Survey distribution was in English. Funding for this study: None Ethics committee - additional information: Medical School Social Research Ethics Committee - University College Cork Author Disclosures: Mark F. Mcentee: Nothing to disclose Chloe Maloney: Nothing to disclose Niamh Moore: Nothing to disclose Rena Young: Nothing to disclose Gráinne Alison Curran: Nothing to disclose Andrew England: Nothing to disclose 09:30-11:00 Research Stage 1 Research Presentation Session: Interventional Radiology RPS 1809 What's new in genitourinary and female health interventions? Moderator V. Kostadinova; Ljubljana/SI P-RENAL: a new scoring system for the prediction of complications and recurrence in kidney malignancies percutaneous abla tion *G. Ferrillo*, D. Poretti, N. Buffi, P. Casale, M. Francone, V. Pedicini; Milan/IT ([email protected]) Purpose or Learning Objective: There is a lack of internationally validated scores for the assessment of outcomes in percutaneo us ablation of kidney tumours. We designed the P-RENAL score and compared it to the surgical standard score (RENAL).

Methods

or Background: We performed a retrospective analysis of 146 patients treated with RFA (N=90) or MWA (N=56) betw een 01/01/2016 and 31/03/2021 (minimum follow up of 6 months). RENALan d p-RENAL were calculated in both populations. Primary outcomes we re recurrence (presence of vital tumour >1 cm in the treated area) and comp lications (according to the CIRSE complications score). P-RENAL is derived from RENAL, it consists of 6 parameters, each assigned a value from 1 to 3: radi us (3cm), exophytic area and nearness to the collecting syste m (identic to RENAL score), location (1 interpolar mass; 2 lower polar; 3 apical tumours), side (1 posterior lateral; 2 medial posterior; 3 medial ant erior), proximity of external structure(>3 cm from an high risk structureas ileal or colic wall,renal artery, spleen or liver); 2 from 3 to 1 cm; 3 points if <1 cm.

Results

or Findings: P-RENAL showed higher sensitivity than RENAL score in predicting recurrence after ablation of kidney t umours expecially with MWA (AUC 0.86). There was no association of the RENAL s core with complications, while we observed an independent positive correlati on between p-RENAL and the presence of grade > 2 complications.

Conclusion

P-RENAL has higher sensitivity than RENAL in detect ion of recurrence and complications in percutaneous ablati on.

Limitations

Retrospective monocentric study. Funding for this study: None Ethics committee - additional information: Retrospective study board review. Author Disclosures: Vittorio Pedicini: Nothing to disclose Nicolò Buffi: Nothing to disclose Marco Francone: Nothing to disclose Dario Poretti: Nothing to disclose Paolo Casale: Nothing to disclose Giuseppe Ferrillo: Nothing to disclose Thulium laser enucleation of the prostate versus pr ostatic artery embolization for big prostates above 80gms: a multi centre study *N. M. A. Attia*¹, M. A. El Hamid², M. Abd El Wahab ², A. Salah²; ¹Assiut/EG, ²Cairo/EG ([email protected]) Purpose or Learning Objective: To compare the postoperative and functional outcomes of thulium laser enucleation of the prosta te (ThuLEP) and prostatic artery embolization (PAE) for the treatment of larg e-volume benign prostate hyperplasia (BPH) > 80 ml. Saturday Abstract-based Programme 206

Methods

or Background: We performed a retrospective, multicentric study of 120 patients consecutively treated with THuLEP (60 patients) or PAE (60 patients) for symptomatic large BPH >80gms between March 2022 and March 2023. These two groups were compared by treatment r esponse, postoperative recovery period and complication rate at baseline, 1, 3 and 6 months.

Results

or Findings: Intraoperative complications were significantly hig her in the THULEP group (13.5% vs 0%, p=0.006). There was no statistically significant difference in the postoperative complic ations between the two groups (10% vs 8.4%, p=0.287). There was significan t delay of catheter removal in the PAE group with subsequent delay in h ospital stay (25% vs 5%, p=0.002). International Prostate Symptom Score (4.4 ±0.9SD vs 10.7±1.5SD, p<0.001) and PSA (1.4±0.6SD vs 4.3±5.1SD, p<0.001) were significantly lower in the THuLEP versus PAE group after 6 months. Post void residual (PVR) was higher in the PAE group from baseline till 6months (12.5±3.9SD vs 20.35±4.8SD, p<0.001). Maximum urinary flow rate (Q max) was significantly higher in the THULEP group from baseline (37.93 ±4SD vs 27.3±8.1SD, p<0.001). Erectile dysfunction was significantly lo wer in the PAE group after 3 months.

Conclusion

Both ThuLEP and PAE relieve lower urinary tract sym ptoms with high efficacy and safety, however ThuLEP was superi or to PAE in improving the functional outcomes in the first 6months while PAE had lower intraoperative complications and erectile dysfuncti on.

Limitations

The quality of life was not assessed being a retros pective study. The sample size was relatively small. Longer follow -up time is needed to compare the durability of the symptomatic improveme nt from each procedure. Funding for this study: None Ethics committee - additional information: This study was approved by the Committee of Medical Ethics of the Faculty of Medic ine Assiut University with IRB no: 04-2024-300429 Author Disclosures: Mahmoud A. El Hamid: Nothing to disclose Noha Mohamed Ali Attia: Nothing to disclose Ahmed Salah: Nothing to disclose Mohamed Abd El Wahab: Nothing to disclose A decade of varicocele embolizations: Success and r ecurrence *M. Ni Mhiochain De Grae*¹, M. Alkhattab¹, A. Alkad himi², M. Springael², G. O' Sullivan¹; ¹Galway/IE, ²Dublin/IE ([email protected]) Purpose or Learning Objective: Varicoceles are vascular lesions of the pampiniform plexus, affecting 10-20% of the populat ion and found in 40% of men with primary infertility and 80% with secondary infertility. (1-4) Pain occurs in 2-10% of cases. (1,5) Treatment options include conservative management, percutaneous embolization, or surgery. (6) Percutan eous embolization, introduced in the late 1970s, has a technical failu re rate of around 13% and recurrence rates ranging from 0.6% to 45%. (7,8) Th is study evaluates success and recurrence rates of percutaneous varicocele emb olizations at Galway University Hospital and Galway Clinic between 2009- 2022, with a minimum follow-up of 18 months.

Methods

or Background: Data collected included patient age, procedure date, access site, side of occurrence, previous int erventions, treatment method, need for re-intervention, and recurrence ra tes. Technical success was defined as successful access to the gonadal vein an d insertion of coil/sclerosant. Clinical success was assessed thro ugh follow-up consultations and ultrasound when available.

Results

or Findings: The technical success rate was 94.7%, with 98.1% of successful embolizations achieving clinical success . Of 225 patients, 3.12% had prior failed surgeries, all treated successfull y with IR, and 0.89% required surgical intervention. Telephone follow-ups (42.7% response rate) revealed a recurrence rate of 10.7% with an average follow-up of 8.03 years. The complication rate was 1.78%. Fertility outcomes sho wed 51.35% of patients had successful pregnancies, while 6.9% experienced ongoing fertility issues.

Conclusion

Our study of 225 patients is the largest to date, s howing higher success rates (technical 94.7%, clinical 98.1%) and a 10.7% recurrence rate with long-term follow-up.

Limitations

This is a single centre retrospective review. 42.7% response to telephone consultation follow up. Funding for this study: No funding. Ethics committee - additional information: Local ethics approval in Galway University Hospital. Author Disclosures: Maia Springael: Nothing to disclose Amor Alkadhimi: Nothing to disclose Meadhbh Ni Mhiochain De Grae: Nothing to disclose Gerry O' Sullivan: Nothing to disclose Maha Alkhattab: Nothing to disclose The Clinical Study of Magnetic Resonance-Guided Hig h-Intensity Focused Ultrasound (MRgFUS) Treatment for Adenomyos is *Q. Zhang*; Shanghai/CN ([email protected]) Purpose or Learning Objective: Objective: To investigate the safety and efficacy of Magnetic Resonance-Guided High-Intensit y Focused Ultrasound (MRgFUS) treatment on adenomyosis.

Methods

or Background: After receiving approval from local Ethics Committee, 62 patients with adenomyosis underwent M RgFUS treatment from October 2018 to March 2021. The 62 patients with ad enomyosis were evaluated for adverse reactions during and after MR gFUS treatment. The lesion volume was measured before and after treatme nt using T2-weighted MRI images. Non-perfused volume (NPV) was calculate d using T1-weighted contrast-enhanced images, then the ablation rate wa s calculated by dividing NPV by the lesion volume. Visual analogue scale (VA S) was used to score dysmenorrhea in adenomyosis patients before and aft er treatment.

Results

or Findings: All 62 patients successfully underwent MRgFUS treatment without severe adverse reactions, achievi ng an average ablation rate of 69.12%±17.64%. The average lesion volume before ablation was 83.72±84.02 cm³. At 3 months post-ablation, the ave rage lesion volume was 62.66±66.31 cm³, at 6 months it was 56.09±66.67 cm³, and at 12 months it was 68.53±87.04 cm³. There were statistically significa nt differences in lesion volume (p<0.05) before and after ablation treatment . The preoperative VAS score for the 62 patients was 6.68±2.13. At 3 months post-treatment, the average dysmenorrhea score was 3.69±1.95, at 6 mont hs it was 2.72±1.40, and at 12 months it was 3.38±1.45. Post-treatment dysmenorrhea scores showed statistically significant decrease compared to pre-treatment (p<0.05).

Conclusion

MRgFUS is a safe, feasible and effective technique in treatment for adenomyosis.

Limitations

Serological examinations were not performed in pati ents with adenomyosis to assess changes in serologic markers after treatment in this study. Funding for this study: Not applicable Ethics committee - additional information: Ethics Committee of Huashan Hospital affiliated to Fudan University Author Disclosures: Qi Zhang: Nothing to disclose Treatment of cesarean scar pregnancy with systemic Methotrexate followed by intra-arterial Methotrexate injection a nd uterine arteries embolization: experience from a tertiary center *C. Intrieri*, C. Lanza, S. A. Angileri, J. Tintori , C. Ercolino, A. M. Ierardi, G. Carrafiello, V. Chiarpenello; Milan/IT ([email protected]) Purpose or Learning Objective: The aim of this retrospective study is to evaluate the efficacy and the safety of the treatme nt of cesarean scar pregnancy (CSP) based on a single or cyclic systemi c methotrexate (MTX) injection followed by intra-arterial MTX injection and uterine arteries embolization (UAE).

Methods

or Background: A pregnancy’s ectopic implantation on a uterine scar tissue following a previous cesarean section i s defined as CSP. Because of its possible complication in uterine rupture, ma ssive hemorrhage, placenta accreta spectrum, fetal and maternal death, CSP is considered a life- threatening condition. Therefore, treatment is fund amental. The cooperating group of Radiologists and Gynecolog ists at our institution evaluated all patients affected by CSP from 2013 to 2023 (n=21) treated with systemic MTX injection followed by intra-arterial M TX injection and UAE with gelatin sponge. The inclusion criteria were: labora tory tests (levels of beta- human chorionic gonadotropin and beta-HCG over the normal threshold), ultrasound diagnosis of CSP, gestational age ≤8 weeks, stable hemodynamic, at least one previous pregnancy and at least one de livery through cesarean section. Exclusion criteria were: contraindications to MTX administration, the presence of abdominal hemorrhage and significant sh ock.

Results

or Findings: Technical success of UAC and clinical success (reduction of the the sac size, lack of active vagi nal re-bleeding and declining beta-HCG values) was achieved in all procedures (10 0%). No peri-procedural complications were recorded.

Conclusion

In our experience, the use of systemic MTX and intr a-arterial injection of MTX plus UAC proved to be effective an d safe to achieve CSP termination. The use of gelatin sponge after intra- arterial MTX administration proved to be effective also in preventing massive b leeding in case of successive uterine curettage.

Limitations

Controlled randomized trial with a larger sample is needed to assess the efficacy and the safety of treatment. Funding for this study: No funding was received for this study. Saturday Abstract-based Programme 207 Ethics committee - additional information: All subjects gave their consent for inclusion in the present study. The study was c onducted in accordance with the World Medical Association (WMA) Declaration of Helsinki [WMA]. Author Disclosures: Vittoria Chiarpenello: Nothing to disclose Salvatore Alessio Angileri: Nothing to disclose Anna Maria Ierardi: Nothing to disclose Carolina Ercolino: Nothing to disclose Jacopo Tintori: Nothing to disclose Cristina Intrieri: Nothing to disclose Carolina Lanza: Nothing to disclose Gianpaolo Carrafiello: Nothing to disclose What is the safe observation period following image -guided renal biopsies? *F. Taylor*, K. Sehgal, M. Van Wees, K. Li, D. W. D e Boo, L-A. Slater; Melbourne/AU ([email protected]) Purpose or Learning Objective: To investigate the timing, type, and severity of complications following percutaneous, image-guid ed renal biopsy and to determine if the current observation period of 4 ho urs can be safely reduced.

Methods

or Background: Consecutive image-guided percutaneous targeted and non-targeted renal biopsies performed between 2 017-2022 in adult patients by radiology medical staff were included. The PACS imaging system and electronic medical records (EMR) were accessed to obtain relevant patient information and procedural reports. Retrospective a nalysis of the type and timing of complications was performed against patie nt demographics and biopsy-related variables. Probabilities were calcul ated for a range of observation periods to assess the proportion of com plications identified by shortening the observation period from 4 hours.

Results

or Findings: 332 percutaneous renal biopsies were included and 4 4 (13%) complications identified within the 4-hour ob servation. 29 complications were post-operative bleeding, of which 2 developed macroscopic haematuria, 25 peri-nephric haematoma and 2 had hemodynamic ins tability with either haematuria or peri-nephric hematoma. 64% of all com plications occurred within the first hour, 86% occurring within 2 hours. Of th e 6 complications occurring after 2 hours, 2 were post-operative bleeding in no n-targeted renal biopsies requiring admission, the other 4 did not require ad ditional observation/procedures.

Conclusion

The vast majority of complications after targeted a nd non- targeted renal biopsies tend to occur within the fi rst 2 hours of observation. Complications that occurred after 2-hour observatio n were often pain related, something that can be overcome with a standardised post-operative analgesic regime. It may be possible to safely reduce observa tion times following image- guided targeted renal biopsies.

Limitations

Single institute dataset, including targeted and no n-targeted biopsies. Biopsies were performed by Radiologists w ith varied experience and supervision, not ascertained during retrospective d ata collection. The low number of complications limits our analysis of asso ciated factors. Funding for this study: This research did not receive any specific grant fr om funding agencies in the public, commercial, or not- for-profit sectors. This study was not supported by any funding. Ethics committee - additional information: All procedures performed in studies involving human participants were in accord ance with the ethical standards of the institutional and/or national rese arch committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. For this type of study formal consent is not required. Ethical approval was obtained from the local Human Research Ethics Committee (HREC), reference number: RES-23-0000-015 Q and the need for informed consent was waived. Author Disclosures: Fergus Taylor: Nothing to disclose Matthew Van Wees: Nothing to disclose Kunal Sehgal: Nothing to disclose Diederick Willem De Boo: Nothing to disclose Kenny Li: Nothing to disclose Lee-Anne Slater: Nothing to disclose Prostate artery embolization (PAE): are there predi ctors indicating the technical success rate? *T. Lauenstein*, K. Scherschel, J. Boddenberg, F. V erfürth, N. Ziayee; Düsseldorf/DE ([email protected]) Purpose or Learning Objective: To assess if factors including patients’ age, prior history of cardiovascular disease or prostate size have an impact on the technical success rate of PAE.

Methods

or Background: Data of 163 consecutive patients undergoing PAE was analyzed. Technical success rate was defined as full when arteries of both prostate lobes could be probed, as partial (one lob e) or as absent . These findings were correlated with patients’ age (range 53-93 years; mean 72 years), prior history of cardiovascular disease (pr esent in 88 patients) and size of the prostate gland (range 36-265ml; mean 84ml).

Results

or Findings: Bilateral PAE could be achieved in 125 patients. In 29 patients only one lobe was embolized and PAE failed to be successful in 9 patients. Patient’s age failed to show a statistica lly significant difference between these groups (71.4±8.4 years for successful vs. 72.4±9.2 years for unsuccessful procedures, P=0.5576, unpaired ttest), as well as prior history of cardiovascular disease (52% for successful vs. 60% for unsuccessful procedures, P=0.4576. Fishers ttest). In contrast, prostate size was significantly higher in patients with successful bi lobar PAE: 88±40ml for successful vs. 68±39ml for unsuccessful procedures (Mann -Whitney-test; p=0.0021). A consecutive ROC analysis revealed a pr ostate size of 79ml as a cut-off value for the success rate of bilobar PAE.

Conclusion

Prostate size was found to be the only predictor fo r a successful completion of PAE. Patients with volumes of the pro state gland <79ml should be informed to have a higher risk of technical ther apy failure.

Limitations

Limitations of this study are related to the retros pective nature of data analysis. Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Frank Verfürth: Nothing to disclose Thomas Lauenstein: Nothing to disclose Naim Ziayee: Nothing to disclose Katharina Scherschel: Nothing to disclose Jörg Boddenberg: Nothing to disclose First intermediate term report of safety and outcom es of radiofrequency ablation of Renal tumours using an automated energy delivery-controlled system *G. Ferrillo*, E. Baldassarre, T. Sirugo, E. Lanza, S. Romano, N. Buffi, P. Casale, M. Francone, V. Pedicini; Milan/IT ([email protected]) Purpose or Learning Objective: To determine the outcomes (Primary efficacy and safety) of RF ablation of hystologically proven renal cell neoplasms using a new single tip internally cooled RF probe.

Methods

or Background: After ethical commettee approval we performed a retrospective analyisis of data reguarding the firs t patients (N= 85)treated with a new RF system whose electrode has an active tip ( 1-5 cm) that erogates high energies (400-2500mAs) with a pulsing algorith m optimized efficiency. This allows fast ablation times, comparable to MW, with a higher safety due to lower peak temperature reached (< 100 °C). Minimum follow up was 6 months, on average 12.3 months The outcomes were clinical o utcome (success, partial success or failure) and safety. Tecnichal success w as defined as complete ablation evaluated at any follow up CT-scan; partia l success if any area of suspicious recurrence > 1 cm was identified anytime during follow up. Safety was assessed according to the CIRSE complica tions classification

Results

or Findings: The primary success rate was 91%; secondary success rate of 100 % (7 patient retreated). Reguarding saf ety there was one case of a CIRSE >2 complication (grade 3, bleeding needing em bolization), and a total of 16% of CIRSE grade < 2 complications (N = 14). Ther e was no mortality. Interestingly, tumours were divided by location and there was a high percentage of central (N= 10) or endophytic lesions (N = 29), with superimposable outcomes of the peripheral lesions.

Conclusion

RFA performed with this new modality resulted safe and effective and it is associated with short procedural times. T he automatic energy-delivery controlled system might overcome limitations due to long procedural times and heat-sink effect in RFA. Future studies are needed to confirm these results on larger populations

Limitations

The study is etrospective and monocentric Funding for this study: None Ethics committee - additional information: Regional ethical committee Author Disclosures: Vittorio Pedicini: Nothing to disclose Silvio Romano: Nothing to disclose Tiziano Sirugo: Nothing to disclose Nicolò Buffi: Nothing to disclose Eleonora Baldassarre: Nothing to disclose Ezio Lanza: Nothing to disclose Marco Francone: Nothing to disclose Paolo Casale: Nothing to disclose Giuseppe Ferrillo: Nothing to disclose Saturday Abstract-based Programme 208 Predicting local tumour control in renal cell carci noma after cryoablation by 3D volumetric image features *C-Y. Huang*, J-A. Hong, N-W. Chang, C-C. Li, C-A. Liu, S-H. Shen; Taipei City/TW ([email protected]) Purpose or Learning Objective: Cryoablation has been shown to be a safe and effective treatment for renal cell carcinoma (R CC); however, there is currently no consensus on the predictors for local tumour control. The aim of this study was to implement automatic segmentation of renal structures on pretreatment CT images, extract relevant image feat ures, and identify potential predictors for local tumour control in RCC after cr yoablation.

Methods

or Background: A total of 124 patients with RCC managed with cryoablation were included. Pretreatment abdominal CT in nephrographic phase were obtained and segmented into masks of the kidney, tumour, cyst, and renal sinus using deep learning segmentation mo dels and post- processing. Relevant image features were extracted, and logistic regression was utilized to assess the correlation between thes e image features and response to cryoablation.

Results

or Findings: Among the 124 patients included, failure of local t umour control occurred in 15 (12.1%). Tumour volume (P = .005) and the contact area between the tumour and the renal sinus (P = .001) w ere significantly associated with local tumour control according to u nivariable logistic regression. A multivariable logistic regression dem onstrated that higher tumour volume (P=.051, OR=1.038, 95% CI=1.000-1.079) and a larger contact area between the tumour and the renal sinus (P=.005, OR= 1.316, 95% CI=1.088- 1.591) were associated with failure of local tumour control. The area under the ROC curve to predict local tumour control after cry oablation was 0.76.

Conclusion

This study demonstrated that volumetric image featu res, including tumour volume and the contact area betwee n the tumour and the renal sinus, significantly increase the risk of fai lure of local tumour control in RCC after cryoablation.

Limitations

This study was subject to the inherent shortcomings of its retrospective design. Additionally, it was conducte d at a single centre, so external validation is warranted. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study was approved by the Institutional Review Board of Taipei Veterans Gener al Hospital (2023-09- 014CC). Author Disclosures: Chien-An Liu: Nothing to disclose Chih-Ying Huang: Nothing to disclose Shu-Huei Shen: Nothing to disclose Chih-Chien Li: Nothing to disclose Nai-Wen Chang: Nothing to disclose Jia-An Hong: Nothing to disclose Efficacy of point of care cortical assay testing du ring adrenal venous sampling *A. Tinney*, K. Lau, J. Dorey; Clayton/AU Purpose or Learning Objective: The aim of this single centre retrospective study was to evaluate the QCA false negative preval ence during AVS.

Methods

or Background: Primary aldosteronism is the most common cause of secondary hypertension. Adrenal venous sampling (AVS) is the gold standard for subtyping primary aldosteronism and gu iding treatment. Quick cortisol assay (QCA) has proven to be a useful tool to confirm correct catheter position within the adrenal vein, increasing proced ure success rates. All consecutive AVS procedures on adult patients at our institution from July 2022 until March 2024 utilising intraoperative QCA were included. AVS without QCA were excluded. QCA analysis was performed in accord ance with manufacturer recommendations. Correct catheter position was conf irmed with dedicated angiographic techniques, and with formal biochemica l analysis.

Results

or Findings: 158 successful and 7 unsuccessful procedures were included. No QCA false positive result was found in unsuccessful AVS. 37 (23.4%) procedures produced a false negative result on QCA analysis despite correct catheter. 23/37 (62.2%) of these procedures had a bilateral false negative QCA, and 14/37 (37.8%) had an asymmetrical unilateral false negative QCA result. There was no significant diffe rence in patient demographics across cohorts. Of the false negative QCA tests, 10 individual test kits (3.2%) provided a negative result despite laboratory cortisol results returning a value above the manufacturer reference value of 828nmol/L.

Conclusion

QCA has been shown to be a useful tool intra-operat ively in increasing AVS sampling success rates. Our results demonstrate that false negative QCA results not infrequently occur and pro ceduralists must remain vigilant. Detailed patient history, dedicated angio graphic techniques and correlation with preoperative CT remain vital when interpreting QCA results. Ongoing research is required to assess patient fact ors which may contribute to false negative results.

Limitations

Retrospective data availability. Funding for this study: N/A Ethics committee - additional information: Local institute low risk ethics approval. Author Disclosures: James Dorey: Nothing to disclose Adrian Tinney: Nothing to disclose Ken Lau: Nothing to disclose 09:30-11:00 Research Stage 2 Research Presentation Session: Abdominal and Gastrointestinal RPS 1801 Innovative imaging in colorectal cancer and pelvic floor disorders Moderator S. Rafaelsen; Vejle/DK MRI evaluation of nodal status after neoadjuvant th erapy in rectal cancer with node-by-node pathological comparison *Q-Y. Li*, X-Y. Yan, D. Yang, Z. Guan, R-J. Sun, Q. Lu, L. X. Ting, X. Zhang, Y-S. Sun; Beijing/CN ([email protected]) Purpose or Learning Objective: To validate the performance of the ESGAR criteria of nodal status after neoadjuvant therapy (NAT) in rectal cancer and to investigate how morphological features and changes before and after NAT can help with node-by-node pathological comparison.

Methods

or Background: Rectal cancer patients who received radical surgery after NAT and had complete pre- and post-NAT MRI we re consecutively and prospectively enrolled. For the nodes that achieved node-by-node matched between MRI and pathology, their SADs were measured , and morphological features (i.e. shape, internal structure, and borde r) were determined on pre- and post-NAT axial T2WI.

Results

or Findings: 207 patients were included and 612 nodes achieved matched, including 471 (77.0%) benign and 141 (23.0 %) metastatic nodes. On the post-NAT MRI, the ESGAR criteria, i.e. SAD ≥5mm, yield an AUC, sensitivity, and specificity in determining nodal s tatus of 0.67, 52.5%, and 82.4%, respectively. All morphological features dif fered between benign and metastatic nodes, with AUCs ranging from 0.55 to 0. 65. The prediction model combined of the only morphological independent pred ictor, internal structure, and SAD didn’t result in an improved diagnostic per formance compared to SAD alone (P=.64). As the changes before and after NAT, there were differences in the size reduction rate and the chan ge in internal structure, but not in shape and border. The AUC of the only indepe ndent predictor, size reduction rate, was only 0.58.

Conclusion

SAD of 5mm is a feasible criterion for determining nodal status on post-NAT MRI and its diagnostic performance coul d not be improved by morphological features or changes before and after NAT, which suggests the limited efficacy of conventional features and the u rgent need for novel features in the future.

Limitations

The limitation is the lack of histopathological gol d standards for every visible node on MRI. Funding for this study: Funding was received from the National Natural Science Foundation of China (82271955) and Capital' s Funds for Health Improvement and Research (2024-1-1022). Ethics committee - additional information: This study was approved by the institutional review board (No. 2019KT76). Author Disclosures: Li Xiao Ting: Nothing to disclose Xin-Yue Yan: Nothing to disclose Rui-Jia Sun: Nothing to disclose Ding Yang: Nothing to disclose Ying-Shi Sun: Nothing to disclose Zhen Guan: Nothing to disclose Qing-Yang Li: Nothing to disclose Xiaoyan Zhang: Nothing to disclose Qiaoyuan Lu: Nothing to disclose Saturday Abstract-based Programme 209 A method of matching nodes between MRI and patholog y in rectal cancer patients *Q-Y. Li*, X-Y. Yan, D. Yang, Z. Guan, R-J. Sun, Q. Lu, L. X. Ting, X. Zhang, Y-S. Sun; Beijing/CN ([email protected]) Purpose or Learning Objective: To develop a method that enables node-by- node matching between preoperative MRI and postoper ative pathology in rectal cancer patients, thereby providing reliable node-based ground-truth labels for further radiological studies.

Methods

or Background: This methodological study prospectively enrolled 535 patients (59 ± 11 years; 326 males) with rectal cancer between 2021 and 2023. Target nodes were defined as nodal structures with a short-axis diameter (SAD) of ≥ 3 mm in the mesorectum or around the superior rect al artery on MRI. With relative location to the tumor, rectal wall and mesorectal fascia, each target node was localized in three dir ections. Combining the coordinates of each node, a 3D node map centered on the tumor and including all target nodes was constructed for each patient a nd used as a bridge enabling node-by-node matching between MRI and path ology.

Results

or Findings: 3,038 target nodes were detected on preoperative MR I, of which 2,220 (73.1%) achieved matched between MRI and pathology. Of the 1,707 matched benign nodes, 1,321 (77.4%), 3 78 (22.1%), and 8 (0.5%) had SADs of 9 mm, respectively. Whereas of the 513 matched metastatic nodes, 224 (43.7%), 254 (49.5%), and 35 (6.8%) had SADs of 9 mm, respectively. Pati ents with lower matching rates tended to have higher T-stages and m ore target nodes on MRI, whereas other factors, e.g. the BMI, therapeutic re gimen, tumor location, and time interval between MRI and pathological examinat ion did not show significant effect on the matching accuracy.

Conclusion

A matching method between MRI and pathology was dev eloped to label numerous nodes with precise statuses in re ctal cancer patients, which contributes to future radiological studies.

Limitations

The limitation is the small proportion of metastati c nodes compared to benign ones. Funding for this study: Funding was received from the National Natural Science Foundation of China (82271955) and Capital' s Funds for Health Improvement and Research (2024-1-1022). Ethics committee - additional information: This study was approved by the institutional review board (No. 2019KT76). Author Disclosures: Li Xiao Ting: Nothing to disclose Xin-Yue Yan: Nothing to disclose Rui-Jia Sun: Nothing to disclose Ding Yang: Nothing to disclose Ying-Shi Sun: Nothing to disclose Zhen Guan: Nothing to disclose Qing-Yang Li: Nothing to disclose Xiaoyan Zhang: Nothing to disclose Qiaoyuan Lu: Nothing to disclose Prognostic impact of MRI-detected risk factors in t otal neoadjuvant therapy for locally advanced rectal cancer *G. F. Cicala*, S. Parisi, F. De Cobelli, V. Burgio , M. Ronzoni, R. Rosati, U. Elmore, P. Passoni; Milan/IT ([email protected]) Purpose or Learning Objective: In Locally Advanced Rectal Cancer (LARC), Total Neoadjuvant Therapy (TNT) has shown to be a v alid therapeutic option leading to a significant reduction in recurrence, d istant metastases, and an improvement in disease-free survival. Magnetic reso nance imaging (MRI) plays a crucial role in the detection of extramural venou s invasion (EMVI), tumor deposits (TDs), mesorectal fascia invasion (MFI) an d helps identify patients who may benefit from a TNT.

Methods

or Background: A retrospective analysis was performed on prospectively collected data involving 109 patients who received TNT between 2009 and 2022. MRI scans were conducted both before and after TNT, with a focus on EMVI, TDs, MFI. Following TNT, viable EMVI and TDs were evaluated using a standardized five-point Likert sc ale.

Results

or Findings: Among the 109 patients, 95 patients met the inclusi on criteria, 64.2% were male, with a median age of 60. 3 years. Positive EMVI scores were observed in 47.4% of cases, while TDs a nd MFI in 15.7% and 24.7%, respectively. Positive EMVI, TDs, MFI, and L ikert scores significantly correlated with reduced time to progression (TTP) a nd poorer overall survival (OS). In multivariate analysis, the presence of TDs (HR 10.28, p=0.002) and a Likert score of 4 (HR 16.44, p=0.003) were strong i ndependent predictors of shorter TTP.

Conclusion

This study confirms the impact of MRI-detected risk factors as significant predictors of prognosis, supporting the role of a standardized five- point Likert scale in the stratification of patients for personalized treatment.

Limitations

Treatment changes after TNT can lead to false negat ives, necessitating a multidisciplinary evaluation for pa tients. Larger cohorts are needed to improve the Likert scale's reliability. Funding for this study: No funding was received for this study. Ethics committee - additional information: Pending confirmation from the ethics committee Author Disclosures: Giuseppe Franco Cicala: Nothing to disclose Valentina Burgio: Nothing to disclose Sergio Parisi: Nothing to disclose Monica Ronzoni: Nothing to disclose Riccardo Rosati: Nothing to disclose Francesco De Cobelli: Nothing to disclose Paolo Passoni: Nothing to disclose Ugo Elmore: Nothing to disclose Evaluation of diffusion-weighted imaging in predict ing response in locally advanced rectal cancer *P. N. Franco*, C. Maino, C. R. G. L. O. M. Talei F ranzesi, R. Corso, D. Ippolito; Monza/IT ([email protected]) Purpose or Learning Objective: To assess the performance of Diffusion- Weighted Imaging (DWI) and apparent diffusion coeff icient (ADC) values in predicting response to neoadjuvant chemoradiation t herapy (CRT) in patients with locally advanced rectal cancer (LARC).

Methods

or Background: Ninety-four patients with MRI pre- and post- neoadjuvant treatment were retrospectively enrolled . Three regions of interest (ROIs) were manually drawn on three different tumor slices for every DWI sequence. ROIs were automatically copied to the cor responding ADC maps and the system derived three different ADC values ( mean, maximum, and minimum), and the standard deviation (SD). Only mea n ADC values were considered. After surgical intervention, pTNM and M andard tumor-regression- grade (TRG) were obtained. Patients with TRG 1-2 we re classified as responders while patients with TRG 3-5 were classif ied as non-responders.

Results

or Findings: No correlation was found between pre-ADC values and TRG classes, while post-ADC and ΔADC values showed a significant correlation with TRG classes (r= -0.285, p=0.002 an d r= -0.290, p=0.019, respectively). Post-ADC values were statistically d ifferent between responders and non-responders (p=0.019). When considering the relation between overall survival (OS) and ADC values, pre-ADC showed a nega tive correlation with OS (r= -0.381, p=0.001) while a positive correlation w as found between ΔADC values and OS (r= 0.323, p=0.013). According to ΔADC values, the mean OS time between responders and non-responders showed a significant difference (p=0.030). A statistical difference was found betwe en TRG classes and OS (p=0.038) and by dividing patients into responders and non-responders (p=0.019).

Conclusion

The pre-ADC and ΔADC values could be used as useful predictors for patients' prognosis. Post-ADC values , due to their relationship with TRG classes, could be a useful tool to predict response.

Limitations

No correlation between imaging and surgical specime ns; the mean interval among CRT, restaging MRI, and surgery was variable among patients. Funding for this study: None Ethics committee - additional information: Upon reviewing the protocol, the local ethical committee deemed formal approval unne cessary, owing to the retrospective, observational, and anonymous nature of this study. Author Disclosures: Cesare Maino: Nothing to disclose Cammillo Roberto Giovanni Leopoldo Oreste Massimili ano Talei Franzesi: Nothing to disclose Rocco Corso: Nothing to disclose Paolo Niccolò Franco: Nothing to disclose Davide Ippolito: Nothing to disclose Efficacy of interventional transarterial treatment in locally recurrent or unresectable colorectal carcinoma: Therapy response and survival *T. J. Vogl*, A-I. Nica, C. Booz, L. S. Alizadeh, I . Yel, T. Biciusca, A. Gökduman, T. Gruber-Rouh, H. Adwan; Frankfurt/DE Purpose or Learning Objective: To evaluate the efficacy of transarterial chemoperfusion (TACP) and transarterial chemoemboli zation (TACE) as palliative and symptomatic treatment options for un resectable colorectal carcinoma (CRC) regarding local tumor response and survival.

Methods

or Background: Between January 2000 and October 2023, 318 TACP and 80 TACE procedures were performed in 67 pa tients with locally recurrent or unresectable CRC. Forty-eight patients were treated with TACP (mean 6.2 procedures/patient, range 2-22), 14 with TACE (mean 4.6 procedures/patient, range 2-11) and 5 patients rece ived a combination of both therapies (mean 4.5 procedures/patient, range 2-13) . Local tumor response Saturday Abstract-based Programme 210 was retrospectively evaluated using the RECIST crit eria and overall survival (OS) and progression-free survival (PFS) were calcu lated using the Kaplan- Meier estimator.

Results

or Findings: 49 (73.13%) of the 67 patients had stable disease ( SD), 15 (22.39%) progressive disease (PD) and 3 patients (4.48%) partial response (PR). Median OS was 16.17 months, median PFS was 11 .25 months. There was no statistically significant difference in OS ( P=0.598) and PFS (P=0.847) between patients either receiving TACP or TACE or b oth treatments. One year after the first procedure, 27 (40.3%) patients were still alive. Nine patients (13.4%) were still alive after 2 years and six pati ents (8.9%) were still alive after 3 years. No major complications were reported.

Conclusion

CONCLUSION: TACP and TACE are minimally invasive procedures that offer a treatment option for patien ts with locally recurrent or unresectable CRC, potentially preventing tumor prog ression and improving quality of life. However, their benefits in the tre atment of CRC warrant further investigation.

Limitations

Retrospective, single-center study Varying follow-u p intervals Short follow-up period Funding for this study: No funding Ethics committee - additional information: Approval of the ethics committee of the Johann Wolfgang Goethe University, Frankfurt Author Disclosures: Christian Booz: Nothing to disclose Ibrahim Yel: Nothing to disclose Teodora Biciusca: Nothing to disclose Andreea-Ioana Nica: Nothing to disclose Thomas J. Vogl: Nothing to disclose Aynur Gökduman: Nothing to disclose Tatjana Gruber-Rouh: Nothing to disclose Hamzah Adwan: Nothing to disclose Leona Soraja Alizadeh: Nothing to disclose Opportunistic Colorectal Cancer Screening in Comput ed Tomography: Exploration of the Colon-Liver Axis with Machine Le arning *S. Grosu*¹, A. Hinterberger², A. E. Sint¹, J. Rick e¹, M. Ingrisch¹, P. Wesp¹; ¹Munich/DE, ²Heidelberg/DE Purpose or Learning Objective: Colorectal cancer (CRC) detection in non- dedicated computed tomography (CT) examinations wit hout bowel preparation is challenging. Current research suggests that live r diseases are associated with an increased risk of colorectal cancer. The ai m of this study was to identify patients with CRC using machine learning (ML)-based opportunistic analysis of the liver in non-dedicated routine clinical CT scan s.

Methods

or Background: Patients 18 years or older with histologically prov en CRC or negative colonoscopy that underwent contrast -enhanced CT of the abdomen for various indications within 5 years or l ess to colonoscopy were included retrospectively. Patients were randomly di vided into a training set (75%) and a test set (25%), stratifying for age and sex. Deep learning-based automated liver segmentation on CT images was perfo rmed. Standardized Radiomic image features were extracted from the liv er segmentations. A random forest ML algorithm was trained on the train ing set to differentiate between patients with histologically confirmed CRC or negative colonoscopy (non-CRC). Algorithm performance was evaluated on t he test set using ROC- AUC, sensitivity and specificity.

Results

or Findings: The training set comprised 809 CT scans from 809 patients (mean age = 62.2 years; 42% female), 66 (8 %) with CRC. The test set comprised 270 CT scans from 270 patients (mean age = 62.2 years; 44% female), 23 (9 %) with CRC. The ROC-AUC for random forest-based differentiation between CRC and non-CRC patients wa s 0.62, with a sensitivity of 74% at a specificity of 50%.

Conclusion

Our results indicate that opportunistic analysis of routine clinical CT images of the liver might have the potential to detect patients with an increased risk for CRC without additional radiation exposure or examinations.

Limitations

Further refinement of the presented model is needed to further increase its diagnostic performance. Funding for this study: No funding was received for this study. Ethics committee - additional information: Ethics committe of the LMU University Hospital, LMU Munich, Munich, Germany Author Disclosures: Alena Elisabeth Sint: Nothing to disclose Sergio Grosu: Nothing to disclose Michael Ingrisch: Nothing to disclose Philipp Wesp: Nothing to disclose Anna Hinterberger: Nothing to disclose Jens Ricke: Nothing to disclose Prognostic value of lateral lymph node metastasis i n pretreatment MRI for rectal cancer in patients with neoadjuvant CRT and surgical resection without LLND: A systemic review and meta-analysis *T. Lee*¹, N. Horvat², M. J. Gollub², J. Garcia-Agu ilar², T. H. Kim²; ¹Seoul/KR, ²New York, NY/US ([email protected]) Purpose or Learning Objective: To systematically review and meta-analyze the prognostic significance of lateral lymph node m etastasis (LLNM) on pretreatment MRI in patients with rectal cancer who undergo neoadjuvant chemoradiation followed by curative surgical resect ion without lateral lymph node dissection (LLND).

Methods

or Background: We searched the MEDLINE and EMBASE databases until September 27, 2023, utilizing the f ollowing search terms: (rectal OR rectum OR colorectal) AND (lateral OR si dewall) AND (lymph OR node). The QUIPS tool was employed to evaluate meth odological quality. We pooled the association between LLNM on pretreatment MRI and outcomes such as local recurrence, distant metastasis, disea se-free survival, and overall survival using hazard ratio (HR) and odds ratio (OR ) based on random effects model.

Results

or Findings: We included 9 studies, encompassing 3180 patients. LLNM on pretreatment MRI revealed a significant ass ociation with increased local recurrence rates (HR: 4.11; 95 % CI: [1.87, 9 .02]) and elevated risks for both disease-free (HR: 1.70; 95 % CI: [1.42, 2.03]) and overall survival (HR: 1.76; 95 % CI: [1.44, 2.15]). As for distant metast asis, our analysis indicated a potential trend towards increased rates, though thi s did not reach statistical significance (HR: 1.67; 95 % CI: [0.85, 3.27]).

Conclusion

Our findings underscore the relationship between LL NM and increased local recurrence and compromised disease- free and overall survival. This emphasizes the potential limitations of relyin g solely on neoadjuvant chemoradiation and highlights the potential need to intensify treatment in select patients.

Limitations

Firstly, we included a relatively small number of s tudies. Second, a notable heterogeneity was observed in the criteri a used to define LLNM on MRI across different studies, which probably led to heterogenous proportion of LLNM-positive patients among included studies. Funding for this study: The National Cancer Institute Cancer Center Core Grant P30 CA008748 Ethics committee - additional information: This study was a systematic review and therefore exempt from requiring approval from our institutional review board. Author Disclosures: Taehee Lee: Nothing to disclose Julio Garcia-Aguilar: Nothing to disclose Marc Jeffrey Gollub: Nothing to disclose Natally Horvat: Nothing to disclose Tae Hyung Kim: Nothing to disclose Open magnetic field and MRI defecography *G. Sterlicchio*, I. Carbone, M. Rengo, C. L. Salet ti, D. Bellini; Latina/IT ([email protected]) Purpose or Learning Objective: To investigate the diagnostic value and image quality of Open MRI scanner (0.5T) on the eva luation of pelvic floor disfunctions.

Methods

or Background: Twenty one patients (20 women and 1 man) underwent MRI defecography using both Open MRI scan ner (0.5 T) and high filed MRI scanner (1.5 T). For both scanners, the s ame MRI protocol has been adopted, including morphological sequences and dyna mic sequences during rest and defecation. Contrast-to-noise ratio (CNR), and signal to noise ration (SNR) were calculated and compared among the differ ent data sets. Qualitative assessment of image quality was perform ed by 3 readers using 5 points Likert scale.

Results

or Findings: SNR was significantly lower using 0.5T compared to 1.5T (mean value 33.1 vs 95.2 for T2 sequences and 44.8 vs. 55.6 for dynamic sequences; P<0.05). CNR was significantly lower usi ng 0.5T compared to 1.5T (mean value 37.1 vs 57.6 for T2 sequences and 27.1 vs. 51.3 for dynamic sequences; P<0.05). However, the readers' image qua lity scores showed that open MRI scanner is not inferior to High magnetic f iled. Diagnostic accuracy was the same for both scanners.

Conclusion

Despite the low values of all quantitative quality metrics on 0.5T compared to i.5T, readers perception of image quali ty is the same between Open MRI scanner and High filed MRI scanner. Low ma gnetic field does not affect diagnostic accuracy.

Limitations

Small sample size. Funding for this study: None Saturday Abstract-based Programme 211 Ethics committee - additional information: Yes Author Disclosures: Iacopo Carbone: Nothing to disclose Marco Rengo: Nothing to disclose Giuseppe Sterlicchio: Nothing to disclose Carlos Leon Saletti: Nothing to disclose Davide Bellini: Nothing to disclose Assessment of puborectalis (PRM) and pubococcygeus (PCM) muscles thickness by RMI defecography: A promising adjuvant radiological parameter for the identification of spastic pelvic floor syndrome(SPFS) *M. Mandolini*, V. Carrozzo, A. Calculli, R. Falett i, A. Ferraris; Torino/IT ([email protected]) Purpose or Learning Objective: To investigate the association between radiological suspicion of SPFS and the thickness ra tio R(PRM/PCM) in patients with pelvic floor disorders.

Methods

or Background: The retrospective study involved 109 women (age 17-86 ys, BMI 16-35.5 kg/m2, past pregnancies 76%) who underwent RMI Defecography between November 2021 and September 20 24 at our institution. The ESGAR-ESUR recommended protocol was complemente d by measurements of PRM and PCM branches thickness made bilaterally at half the length of each bundle and averaged to obtain th e ratio R (PRM/PCM). Continuous independent variables were compared with Mann-Whitney’s Test when independent; Wilcoxon’s Test and Bland-Altman plot when correlated; dichotomic variables were studied with Fisher’sTest . PropensityScoreMatching (PSM) was used to reduce the effect of confounding covariates on the outcome.The ReceivingOperatingCurve (ROC) was used to estimate the performance of R with the Area Under The Curve (AUC ).

Results

or Findings: The 23 (21%) patients satisfying the radiological c riteria for SPFS recognized by scientific literature formed the Study Group.PSM extracted from the remaining 84 patients, a Control Group of 23 with baseline comparable (p>0.77, Standardized Mean Difference <0 .20) to that of the Study patients. The thickness ratio R was 3.2(3.0-3.5) fo r the former vs 2.5(2.2-2.75) for the latter (p<0.0001). The ROC curve assessed a good diagnostic ability of R for SPSF, (AUC=0.94) with threshold R≥3.0 (sensitivity 0.87, specificity 0.96, PPV=0.95, NPV=0.88).The measurements made by two re aders showed good intra- and inter-operator agreement.

Conclusion

The performance of the PRM/PCM thickness ratio is p resently being explored over a larger population; if the pos itive results will be confirmed, this indicator may become an additional tool for ra diological identification of SPFS.

Limitations

The main limitations of the study are the absence o f gold standard method for SPFS diagnosis and the small nu mber of our Study Group. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Being a retrospective preliminary study based on the use of anonymous data, it has no t yet been subjected to scrutiny by the ethics committee. Author Disclosures: Matilde Mandolini: Nothing to disclose Annarita Calculli: Nothing to disclose Valentina Carrozzo: Nothing to disclose Riccardo Faletti: Nothing to disclose Andrea Ferraris: Nothing to disclose 09:30-11:00 Research Stage 3 Research Presentation Session: Neuro RPS 1811 Brain in function Moderator N. Bargalló; Barcelona/ES ([email protected]) Constructing normative human brain atlases of R1, R 2, proton density, and myelin volume fraction using synthetic quantita tive MRI *H. M. H. Sbaihat*, A. K. Roenneke, D. Müller, T. L adopoulos, R. Schneider, B. Krieger, B. Bellenberg, C. Lukas; Bochum/DE ([email protected]) Purpose or Learning Objective: Quantitative MRI (qMRI) provides valuable insights into tissue-specific MR properties, extend ing the diagnostic capabilities of conventional MRI. We aimed to construct normativ e multimodal human brain atlases to serve as references for tissue alteratio ns in neurological disorders.

Methods

or Background: Fifty-eight healthy controls (HC) underwent qMRI of the brain at 1.5T using the QRAPMASTER sequence, re sulting in parameter maps for Myelin Volume Fraction (MVF, %), Proton De nsity (PD, %), and Relaxation Rates R1 and R2 (s⁻¹). We constructed four high-resolution (1mm³) atlases in standard space after bias-field correcti on, interpolation, normalization, and smoothing. These atlases allow v isual and quantitative comparisons with individual datasets and will be ma de available to the research community. Additionally, quantitative data were extracted from 26 white matter regions of interest (ROIs). The atlase s and extracted data were validated using z-score maps of three healthy contr ols and three multiple sclerosis (MS) patients. Group differences across t he ROIs between the qMRI atlases and the testing subjects were assessed usin g t-tests.

Results

or Findings: The resulting atlases demonstrated high anatomical accuracy, resolution, and comprehensive brain cover age. The ROI values for R1, R2, PD, and MVF were consistent with the publis hed literature. The z- score maps, particularly for R1 and MVF, accurately reflected individual lesion patterns and diffuse tissue changes in each MS pati ent. T-test results for R1, R2, PD, and MVF confirmed the alignment between the HC atlases and the HC testing group, while significant differences were o bserved with the MS testing subjects at a p-value of 0.01.

Conclusion

We successfully generated high-resolution, multi-mo dal qMRI atlases of the human brain, providing normative bas elines for R1, R2, PD, and MVF. The validation process underscores their poten tial for assessing microstructural brain alterations in individual pat ients with neurodegenerative diseases.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee of the Medical Faculty of the Ruhr-University Bochum, Germ any (Approval Np. 20- 7054-BR). Area of Interest: CNS, Neuroradiology brain. Author Disclosures: Ruth Schneider: Nothing to disclose Theodoros Ladopoulos: Nothing to disclose Barbara Bellenberg: Nothing to disclose Britta Krieger: Nothing to disclose Hasan M H Sbaihat: Nothing to disclose Anna Katharina Roenneke: Nothing to disclose Carsten Lukas: Nothing to disclose Dajana Müller: Nothing to disclose How does hippocampal volume in mesial temporal scle rosis affect brain networks during a functional MRI memory task? *S. B. Rosa*¹, B. Direito², F. Sales², D. J. Pereir a²; ¹Lisbon/PT, ²Coimbra/PT ([email protected]) Purpose or Learning Objective: Mesial Temporal Sclerosis (MTS) is the most frequent histopathological abnormality in drug-resi stant temporal lobe epilepsy. Anterior temporal lobectomy is a possible treatment , but postoperative memory deficits may follow. MTS patients activate extra-te mporal regions during memory-related tasks, possibly as part of compensat ory networks. Further research is required to understand these networks a nd how they may impact surgical outcomes. We aimed to understand how hippo campal atrophy affects connectivity during a memory fMRI task in MTS patie nts. Saturday Abstract-based Programme 212

Methods

or Background: We retrospectively included drug-resistant MTS patients admitted to CLHU (2019-2023) for surgical evaluation. Patients underwent MRI, including 3D T1WI and event-related memory fMRI. Hippocampal volume ratio was automatically calculat ed using volBrain. CONN toolbox was used to assess the effect of hippocampa l volume on seed-to-voxel functional connectivity of the right and left hippo campus in patients with right and left MTS, respectively, while performing a verb al memory task.

Results

or Findings: Thirteen patients were included (eight with lMTS, f ive with rMTS). Hippocampal volume ratio was 0.1047±0.0 406 for lMTS and 0.1144±0.0331 for rMTS patients (mean ± standard deviation). Seed-to-voxel analysis showed a positive association between hipp ocampal volume and connectivity with contralateral hippocampus and par ahippocampal gyri in both groups. In lMTS, hippocampal volume was negatively associated with connectivity to right superior parietal lobule and supramarginal gyrus (p＜0.05 cluster-size p-FDR corrected). No significant negat ive associations were found in rMTS.

Conclusion

As expected, greater hippocampal atrophy correspond ed to reduced connectivity with the parahippocampal gyri and contralateral mesial temporal regions. Surprisingly, lMTS patients with greater atrophy showed higher connectivity to right parietal areas, a comp ensatory network not previously described in this condition, to our know ledge.

Limitations

Small sample size, applying a resting-state fMRI an alysis method to task-based fMRI, assessing right hippocampal con nectivity with a left hippocampal task. Funding for this study: No funding. Ethics committee - additional information: All patients signed written consent for data usage and all data was anonymized. Author Disclosures: Francisco Sales: Nothing to disclose Sara Botelho Rosa: Grant Recipient: ESR-EIBIR seed grant Daniela Jardim Pereira: Nothing to disclose Bruno Direito: Nothing to disclose Unveiling the Brain’s Response to Valenced Sounds: Neural Correlates of Auditory Emotion *F. Aldhafeeri*; Hafar al-Batin/SA ([email protected]) Purpose or Learning Objective: Complex neural pathways that integrate auditory processing with emotional evaluation play a crucial role in the perception of emotional sounds. This study aimed to identify the neural circuits that differentially encode positive and negative va lence during the implicit processing of emotional stimuli.

Methods

or Background: A block-design fMRI experiment was conducted with thirty healthy participants. The study measure d blood oxygen level dependent (BOLD) signal changes in response to plea sant and unpleasant sounds from the International Affective Digitized S ounds (IADS), with each condition compared against a neutral baseline

Results

or Findings: Significant activation (pFDRcorrected <0.05) was fo und in the medial prefrontal cortex (mPFC), ventral ant erior cingulate cortex (vACC), and temporal lobe when contrasting pleasant sounds with neutral conditions. In response to unpleasant sounds, signi ficant activation (pFDRcorrected <0.05) was observed in the amygdala, nucleus accumbens, parahippocampal gyri, temporal lobe, visual cortex, PFC, insula, anterior cingulate gyrus, and cerebellum, compared to the ne utral condition

Conclusion

The neural correlates of pleasant and unpleasant st imuli involve a complex interplay between brain regions that regu late emotional responses. The auditory cortex, amygdala, and nucleus accumben s are key components of this process, with distinct activation patterns depending on the emotional valence of the auditory stimuli. Understanding thes e neural mechanisms enhances our insight into how sound influences emot ional experiences and may guide the development of therapeutic interventi ons for auditory-related emotional disorders

Limitations

The study was conducted with a relatively small sam ple size of thirty participants. While this is typical for many fMRI studies, the limited number of participants may affect the generalizabil ity of the findings. A larger sample size could provide more robust and reliable results. Funding for this study: None Ethics committee - additional information: Local Research Ethics Committee Author Disclosures: Faten Aldhafeeri: Nothing to disclose Amygdala multimodal reorganization as an indicator of affective dysfunction in tinnitus patients *Q. Chen*; Beijing/CN ([email protected]) Purpose or Learning Objective: This study aimed to systematically investigate structural and functional alterations i n amygdala subregions using multimodal MRI in patients with tinnitus with or wi thout affective dysfunction.

Methods

or Background: Sixty patients with persistent tinnitus and 40 heal thy controls (HCs) were recruited. Based on a questionn aire assessment, 26 and 34 patients were categorized into the tinnitus pati ents with affective dysfunction (TPAD) and tinnitus patients without affective dysf unction (TPWAD) groups, respectively. MRI-based measurements of gray matter volume, fractional anisotropy (FA), fractional amplitude of low-freque ncy fluctuations (fALFF), regional homogeneity (ReHo), degree centrality (DC) , and functional connectivity (FC) were conducted within 14 amygdala subregions for intergroup comparisons. Associations between the MR I properties and clinical characteristics were estimated via partial correlat ion analyses.

Results

or Findings: Compared with HCs, the patients exhibited significa nt structural and functional changes, with more pronou nced WMI changes in the TPAD group, predominantly within the left auxiliary basal or basomedial nucleus (AB/BM), right central nucleus, right later al nuclei (dorsal portion), and left lateral nuclei (ventral portion containing bas olateral portions). Moreover, the TPAD group exhibited decreased FC between the l eft AB/BM and left middle occipital gyrus and right superior frontal g yrus (SFG), left basal nucleus and right SFG, and right lateral nuclei (intermedia te portion) and right SFG. In combination, these amygdalar alterations exhibited a sensitivity of 65.4% and specificity of 96.9% in predicting affective dysfun ction in patients with tinnitus.

Conclusion

Although similar structural and functional amygdala remodeling were observed in the TPAD and TPWAD groups, the cha nges were more pronounced in the TPAD group. These changes mainly involved alterations in functionality and white matter microstructure in va rious amygdala subregions; in combination, these changes could serve as an ima ging-based predictor of emotional disorders in patients with tinnitus.

Limitations

This is a cross-section study Funding for this study: None Ethics committee - additional information: This study was approved by the Institutional Review Board of Beijing Friendship Ho spital, Capital Medical University (No. 2017-P2-134-01). Author Disclosures: Qian Chen: Nothing to disclose The impact of diet on brain structural and function al networks *R. Rajiah*, Q. Aziz, P. Nachev, J. K. K. Ruffle; L ondon/UK Purpose or Learning Objective: To delineate the association between diet, brain structure and function in patients undergoing multi-modal MRI.

Methods

or Background: Extensive evidence from lab-based and clinical studies support diet as an essential regulator of b rain function. However accompanying neuroimaging evidence is scarce and sm all in scale. We studied 518 healthy participants from the Cambridge Centre for Aging and Neuroscience (CAM-Can) repository (261 male and 257 female, mean age 53 years). Participants were clustered via their self- reported dietary intakes using K-means clustering. From MRI, resting-state functio nal, white matter tractography, and grey matter volumetry networks co nditional to dietary cluster were investigated using network-based statistics, w ith participant age and sex as nuisance covariates. Relevant clinical data comp rising BMI and Hospital Anxiety and Depression Score (HADS) were compared u sing Welch’s t-test.

Results

or Findings: Four dietary intake patterns were identified: balan ced, meat-predominant, vegetarian, and raw vegetable-pre dominant. Specific dietary types – balanced, meat-predominant and vege tarian – were associated with increased functional, grey matter or white mat ter connectivity between regions comprising the bilateral insula, frontomedi al cortex (FMC) and bilateral anterior cingulate cortex (ACC) (p=<0.04). A balanc ed diet was associated with significantly increased functional, structural, and white matter connectivity between brain regions, including the orbitofrontal cortex (OFC), bilateral insula and FMC compared to all others (p=<0.04). A balance d diet was associated with a significantly reduced depression score (HADS -D), compared to a meat- predominant one (p=0.03). The meat-predominant diet cluster correlated with increased structural connectivity at the bilateral ACC and increased BMI (p=0.01), when compared to vegetarian diets (p=0.03 ).

Conclusion

In the largest study of its kind, we reveal the str uctural and functional brain associations of diet. Our findings imply physiological correlates for diet-induced brain changes for future research.

Limitations

We use self-reported dietary intake, limited by dat a availability. Saturday Abstract-based Programme 213 Funding for this study: The Issac Shapiro grant supported RR. JKR was supported by the Medical Research Council (MR/X0004 6X/1).  PN is supported by the Wellcome Trust (213038/Z/18/Z) and the UCLH NIHR Biomedical Research Centre. Ethics committee - additional information: The Cam-CAN project was approved by Cambridgeshire Research Ethics Committe e (reference: 10/H0308/50). Author Disclosures: Parashkev Nachev: Nothing to disclose Qasim Aziz: Nothing to disclose Rebekah Rajiah: Nothing to disclose James Kelsey K Ruffle: Nothing to disclose Current European trends in the use of CTA as an anc illary test in the determination of brain death *H. Briody*, I. Alam, R. Bruen, P. Mchugh, P. Rohan , S. Looby; Dublin/IE ([email protected]) Purpose or Learning Objective: To establish the current trends in European literature on the use of computed tomography angiog raphy (CTA) as an ancillary test in the determination of brain death and to clarify the technical parameters and interpretative criteria currently in use.

Methods

or Background: Brain death is primarily a clinical diagnosis made by confirming the absence of brainstem reflexes and th e presence of apnoea in the setting of irreversible coma where confounding factors have been excluded. However, in specific circumstances where it is not possible to perform the required tests to satisfy the clinical criteria, ancillary testing may be required to support the diagnosis. The radiological study endorsed by most national guidelines remains four-vessel digital sub traction angiography although there is a move towards the less invasive, more accessible CTA.

Results

or Findings: CTA is currently endorsed as an ancillary test for the determination of brain death in multiple European c ountries including France, Germany, Spain, Switzerland, Poland and the United Kingdom. The Polish and UK guidelines advise a three-phase study consisting of a non-contrast phase and post-contrast phases at 20 and 40 seconds post intravenous contrast injection. Opacification of the superficial tempora l or facial artery is required to confirm correct contrast administration. Cessation of cerebral circulation is diagnosed when there is bilateral absence of contra st in the middle cerebral arteries and internal cerebral veins on the second post-contrast phase.

Conclusion

CTA is emerging as a feasible ancillary test for th e diagnosis of brain death and is currently endorsed by a number o f European guidelines. The most widely used technique involves a three-pha se study with evaluation of intracranial vessels in four anatomical location s.

Limitations

This study is limited by a paucity of literature an d lack of consensus guidelines on imaging in brain death. Funding for this study: None. Ethics committee - additional information: N/A Author Disclosures: Paul Mchugh: Nothing to disclose Richard Bruen: Nothing to disclose Imran Alam: Nothing to disclose Seamus Looby: Nothing to disclose Hayley Briody: Nothing to disclose Pat Rohan: Nothing to disclose Speed-reading-induced changes in functional brain n etworks: A connectivity-based analysis *T. A. Walpola*, C. Yang, N. Dilhani, R. Iseki, T. Makino, T. N. Hoang, C. D. Kulathilake, I. Ichiro, A. Senoo; Tokyo/JP ([email protected]) Purpose or Learning Objective: Speed-reading is a salient technique used among children to improve fast learning skills. The main aim of the present study is to assess the alterations of the brain’s f unctional connectivity networks identified in a speed-reading-trained native Sinhal ese cohort.

Methods

or Background: A cohort of 18 healthy native Sinhala-speaking adults (>18yrs) who volunteered to undergo fMRI (sc anner: 3.0 T SIGNA Premier) were selected ((male, 11: female, 7), Age (mean; Stdv: 31;4.0) and dominant hand: Right). Participants read a simple S inhala novel silently inside the scanner. Three scans at one-month intervals eac h were done; the first two were the control scans, and the third was the train ing scan after the intervention of speed-reading training. Image acqui sition; fMRI sequence using GRE EPI (TR: 1000(ms) TE: 30(ms), characters per ta sk block: 320-350). Data analysis was performed using CONN toolbox v.22.a.

Results

or Findings: The reading speed (mean (Stdv)) increased from 213 (87) to 712 (200) wpm in the training group. The se ed-based connectivity (SBC) results showed that the language network has increased significant connectivity, with 57 voxels (32%) covering 20% of Right Heschl’s gyrus and 68 (38%) voxels covering 15% of planum temporale (S ignificant increase threshold: p<0.05 cluster-size p-FDR corrected, vox el threshold: p<0.001 (p- uncorrected)). According to the generalized psychop hysiological interactions (gPPI) results, the individual ROI analysis showed increased connectivity in the left posterior temporal gyrus with the bilateral vi sual lateral and occipital networks meanwhile a decreased connectivity with th e visual medial network (p-FDR corrected <0.05).

Conclusion

The study concluded that the activation of Heschl’s gyrus; crucial for sound perception in reading, and planum tempora le; responsible for phonological processing highlights the importance o f some auditory processing regions related to quick recognition and comprehens ion of text during speed reading.

Limitations

Eye-tracking is not performed Funding for this study: None Ethics committee - additional information: The Ethics have been approved by the Ethics Review Committee of Tokyo Metropolita n University, Tokyo, Japan Approval No-22022 Author Disclosures: Niluka Dilhani: Nothing to disclose Chutian Yang: Nothing to disclose Tatsuya Makino: Nothing to disclose Thishuli Anujaya Walpola: Nothing to disclose Thanh Ngoc Hoang: Nothing to disclose Rinako Iseki: Nothing to disclose Atsushi Senoo: Nothing to disclose Chathura Darshana Kulathilake: Nothing to disclose Iso Ichiro: Nothing to disclose Illuminating Minds: The Transformative Role of Radi ology in Mental Health Diagnosis R. Praveenkumar, *F. Abubacker Sulaiman*, J. Lydia; Chennai/IN ([email protected]) Purpose or Learning Objective: This abstract examines the contributions of radiology to diagnosing and managing mental health disorders, emphasizing advanced imaging modalities' role in understanding neurobiological mechanisms.

Methods

or Background: A systematic review of literature was conducted, focusing on the use of magnetic resonance imaging ( MRI), computed tomography (CT), and positron emission tomography ( PET) in evaluating conditions such as schizophrenia, major depressive disorder, anxiety disorders, and post-traumatic stress disorder (PTSD ).

Results

or Findings: Key findings illustrate the impact of radiology on mental health care: Structural Imaging: MRI and CT scans i dentify neuroanatomical changes in mental health disorders. For example, st udies show reduced gray matter in the prefrontal cortex of patients with sc hizophrenia and altered hippocampal volumes in those with major depressive disorder. Functional Imaging: PET and functional MRI (fMRI) enhance unde rstanding of neural circuitry. Research indicates hyperactivity in the amygdala of individuals with anxiety disorders and altered connectivity in the d efault mode network in patients with depression. Biomarker Discovery: Radi ological imaging aids in identifying neurobiological biomarkers. Specific me tabolic patterns in PET scans may predict treatment response in major depre ssive disorder, allowing for personalized therapy. Therapeutic Monitoring: I maging techniques monitor treatment efficacy. In patients receiving transcran ial magnetic stimulation (TMS) or electroconvulsive therapy (ECT), imaging p rovides real-time assessments of changes in brain activity.

Conclusion

Radiology significantly enhances the understanding and management of mental health disorders through advan ced imaging techniques, improving diagnostic accuracy and facil itating personalized treatment strategies.

Limitations

The review highlights the need for standardized ima ging protocols and further research to correlate neuroim aging findings with clinical outcomes Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Rathinamoorthy Praveenkumar: Nothing to disclose Johnbosco Lydia: Nothing to disclose Farook Abubacker Sulaiman: Nothing to disclose Altered Structural-Functional Coupling in Parkinson ’s Disease Patients with Depression: Insights from Multimodal Brain Net work Analysis *M. Wang*¹, C. Tan², J. Hakumäki¹; ¹Kuopio/FI, ²Cha ngsha/CN Purpose or Learning Objective: This study integrates multimodal brain network data to explore alterations in structural-f unctional coupling across multi-scale brain networks in Parkinson’s disease p atients with depression (PDD). By identifying potential imaging biomarkers, this research aims to enhance sustainable diagnostic processes, improving healthcare efficiency through accurate and early detection of depression in Parkinson’s disease. Saturday Abstract-based Programme 214

Methods

or Background: A total of 106 consecutive, drug-naïve Parkinson’s disease patients were prospectively enrolled and di vided into PDD (n = 50, HAMD > 17, BDI > 10) and PDND (n = 56, HAMD < 7, BD I < 10) groups. All participants underwent diffusion tensor imaging (DT I) and resting-state MRI on a 3.0T scanner. Whole-brain functional and structur al networks were constructed. Structural-functional coupling, intra- and inter-module connectivity, as well as topological parameters (e.g., clustering coefficient, local efficiency), were compared between the two groups. ROC curve ana lysis was performed to evaluate the diagnostic performance of these par ameters in differentiating PDD from PDND.

Results

or Findings: PDD patients exhibited significantly lower whole-br ain structural-functional coupling compared to PDND pat ients (P = 0.017, TFCE- corrected). At the module level, PDD patients showe d increased structural- functional coupling within the default mode network (P = 0.032, TFCE- corrected). Additionally, PDD patients had signific antly lower clustering coefficients (P = 0.007) and reduced local efficien cy (P = 0.021) across the brain's structural network. ROC analysis combining these metrics demonstrated a sensitivity of 65% and a specificity of 77.7% for distinguishing PDD from PDND.

Conclusion

Altered multi-scale brain network structural-functi onal coupling in PDD suggests potential imaging biomarkers for more sustainable and precise diagnosis. These findings may help reduce unnecessa ry interventions and optimise healthcare resources by providing a non-in vasive tool for early detection of depression in Parkinson’s disease.

Limitations

The limitation is the relatively small sample size. Funding for this study: CSC funding No. 202306370155 Ethics committee - additional information: The study received institutional review board approval, and written informed consent was obtained from all participants. Author Disclosures: Changlian Tan: Nothing to disclose Juhana Hakumäki: Nothing to disclose Min Wang: Nothing to disclose Systematically altered connectome gradient in patie nt with Type2- diabetes mellitus: Potential effect on cognitive fu nction *H. Ran*, K. Huang, T. Zhang; ZunYi/CN ([email protected]) Purpose or Learning Objective: Type2-diabetes mellitus (T2DM) is known to affect brain networks and cognitive function. Conne ctome gradient studies have suggested a primary-to-transmodal gradient in functional brain network. However, whether this gradient structure is disrupt ed in patients with T2DM are still ambiguous. The aim of this study is to invest igate connectome gradient alterations and its potential contribution to cogni tive function in T2DM.

Methods

or Background: We recruited resting-state functional magnetic resonance imaging (rs-fMRI) data of 45 participants (24 T2DM patients and 21 healthy controls) and Montreal Cognitive Assessment Scale (MoCA) and Mini- Mental State Examination (MMSE) were performed in p atients with T2DM. We studied the related alterations in the principal an d secondary connectome gradient between T2DM and healthy controls at the v oxel and network levels. We further examined the associations between T2DM-r elated changes of connectome gradients and clinical variables. The RV R algorithm was employed to assess the predictive capacity of princ ipal gradients in relation to cognitive function scores among patients diagnosed with T2DM

Results

or Findings: Relative to the controls, T2DM patients extended gradient at different network-level and voxel-level . In the principal gradient, the left rolandic operculum, precuneus gradient score w as negatively correlated with MMSE score and duration, respectively (r =-0.4 79, -0.481, P=0.028, 0.027)，and the connectome gradient alterations in ventral attention network was negatively correlated with MMSE and MoCA (r=-0. 484, -0.435, P=0.026, 0.049). Patients’ principal gradient maps significa ntly predicted their MMSE (r = 0.465, P = 0.022).

Conclusion

We reported a systematically disrupted functional g radient in patients with T2DM and its negative correlation wit h cognitive function. These findings improve our comprehension the neurobiologi cal mechanisms that underlie cognitive function and offer potential ima ging biomarkers for the assessment of cognitive function in T2DM.

Limitations

The sample size of this study was relative small. Funding for this study: This study was supported by Intelligent Medical Imaging Engineering Research Center of Guizhou High er Education Institutions project (Grant No. Qianjiaoji [2023] 0 38). Ethics committee - additional information: Ethics approval of this research was granted by the Ethic Committee of the Affiliate d Hospital of Zunyi Medical University[KLL-2024-512] Author Disclosures: Tijiang Zhang: Nothing to disclose Kexin Huang: Nothing to disclose Haifeng Ran: Nothing to disclose Brain Structural Connectivity alteration and its ro le in Language processing in Post-stroke Aphasia *N-T. Hoang*, T. Hada, C. D. Kulathilake, T. A. Wal pola, N. Dilhani, A. Senoo; Tokyo/JP ([email protected]) Purpose or Learning Objective: Our main objectives are to evaluate the brain connectivity in post-stroke aphasia (PSA) and its c orrelation with language ability.

Methods

or Background: Tractography of 20 PSA was reconstructed by using deterministic algorithms. Brodmann atlas was used to brain parcellation. Connectivity analysis with following network measur es: density (den), clustering coefficient (Cc), transitivity (trans), characteristic path length (CPL), small worldness (Sw), global efficiency (G_eff), ra dius (r), diameter (d), assortativity coefficient (A_coeff), rich-club coef ficient (club_coeff) were extracted. All statistical analysis were performed using SPSS version 27.

Results

or Findings: There are significantly lower value of CPL (p = 0.0 46), A_coeff (p = 0.026), diameter (p = 0.031), and high er value of G_eff (p = 0.046) in the high-level repetition group (n = 6) compared to the low-level repetition group (n = 14). This findings indicate the brain ne tworks of the high-level group are more efficient in transferring information. Add itionally, the network connections in the high-level group are more random or diverse regarding node characteristics, whereas networks in the low-l evel group may display more organized connections between similar nodes. T here are strong positive correlations between the speaking sentence repetiti on and d (r = 0.593, p = 0.006), Sw (r = 0.580, p = 0.007), G_eff (r = 0.566 , p = 0.009), and strong negative correlation between the speaking sentence repetition and CPL (r = 0.571, p = 0.009).

Conclusion

The high-level repetition group transfers informati on across regions more efficiently than the low-level group. It indicates the potential for rehabilitation strategies and emphasizing the need for personalized approaches in aphasia treatment.

Limitations

This research have some limitations. Our small samp le cannot represent for PSA population. Additionally, the lac k of longitudinal research is insufficient to determine the impact of network con nectivity on language recovery. Funding for this study: Research grant was provided by the Tokyo Global Partner Scholarship Program Ethics committee - additional information: This study is being approved by the ethics committee Author Disclosures: Niluka Dilhani: Nothing to disclose Thishuli Anujaya Walpola: Nothing to disclose Ngoc-Thanh Hoang: Nothing to disclose Atsushi Senoo: Nothing to disclose Chathura Darshana Kulathilake: Nothing to disclose Takuya Hada: Nothing to disclose Exploring brain region changes following acute caff eine intake based on ASL and OEF *Z. Shu*; Zhejiang, Hangzhou/CN ([email protected]) Purpose or Learning Objective: This study aims to employ Arterial Spin Labeling (ASL) and Oxygen Extraction Fraction (OEF) to assess the impact of caffeine consumption on local cerebral blood flow p erfusion and brain oxygen metabolism.

Methods

or Background: A prospective study was conducted involving 18 healthy young volunteers with no habitual caffeine consumption. ASL and OEF images were acquired both prior to and 90 minutes f ollowing the administration of 200 mg of caffeine. Voxel-based analysis was emp loyed to quantify alterations in brain oxygen metabolism across vario us cerebral regions before and after caffeine consumption. Regional cerebral b lood flow perfusion was assessed in areas exhibiting changes in brain oxyge n metabolism, with a focus on evaluating the extent of perfusion variation sub sequent to caffeine intake. Additionally, the correlation between OEF indices i n regions with modified cerebral blood flow and shifts in Karolinska Sleepi ness Scale (KSS) scores was analyzed.

Results

or Findings: In regions exhibiting alterations in brain oxygen metabolism following caffeine consumption, signific ant reductions in cerebral blood flow perfusion were identified within the med ial occipital-temporal areas of both hemispheres, the right anterior cingulate g yrus, the left postcentral gyrus, and the left cerebellum (P<0.05). Notably, o nly the left cerebellum demonstrated an increase in brain oxygen metabolism . A negative relationship between brain oxygen metabolism and cerebral blood flow perfusion in the left cerebellum after caffeine intake (r=-0.5, p<0.05). Furthermore, a negative correlation was observed between changes in brain o xygen metabolism and shifts in KSS scores before and after caffeine cons umption (r=-0.738, p<0.05).

Conclusion

The increase in brain oxygen metabolism in the left cerebellum may compensate for the reduced cerebral blood flow perfusion following acute caffeine intake, which could be a key factor in red ucing drowsiness. Saturday Abstract-based Programme 215

Limitations

The research sample size only focuses on young peop le Funding for this study: The work was supported by the National Natural Science Foundation of China (Grant No.82101983) Ethics committee - additional information: Ethics Committee of Zhejiang Provincial People's Hospital Author Disclosures: Zhenyu Shu: Nothing to disclose 09:30-11:00 Research Stage 4 Research Presentation Session: Chest RPS 1804 Technological advancements in chest imaging: MRI, photon counting CT and more Moderator P. Ciet; Rotterdam/NL ([email protected]) Author Disclosures: Pierluigi Ciet: Advisory Board: European Medicine A gency (EMA); Board Member: European Society of Pediatric Radiology (ES PR) Research Committee, AI and Cardiothoracic Taskforces; Consul tant: Siemens Healthineers, Vertex Pharmaceutical; Employee: Eras mus MC - Sophia Children's Hospital; Grant Recipient: Research Dutc h Council, Horizon Pathfinder; Research Grant/Support: General Electri cs; Speaker: ECR, ESPR, ERS, InSpIrEd. Characterization of Interstitial Lung Abnormalities and Prediction of Disease Progression with MRI *D. Kütting*, J. A. Luetkens, D. Thomas, A. M. C. B oehner, T. Dell, A. Faron; Bonn/DE ([email protected]) Purpose or Learning Objective: Interstitial lung abnormalities (ILA) impact survival and quality of life, yet predictive imagin g markers for progression are lacking. MRI holds promise in enhancing ILA phenoty ping, potentially enabling tailored follow-up strategies while reducing repeti tive CT imaging and associated radiation exposure.

Methods

or Background: Assessment of detectability and estimation of disease progression of ILA in a single center, lung cancer screening cohort (224 participants) receiving same day CT/MRI. Radio logists independently evaluated chest images for presence of ILA using st andardized criteria. Follow- up exams were reviewed for disease progression. MRI sequences included T2- TSE MVXD, T2-STIR, and diffusion-weighted imaging ( DWI). Statistical analyses evaluated the agreement between CT and MRI findings and MRI's diagnostic performance for ILA detection and progre ssion prediction.

Results

or Findings: Among the 224 participants (mean age 58.5 ± 5.7 yea rs; 45% female), 26 exhibited ILA on baseline CT, with 65% categorized as subpleural fibrotic. Baseline CT findings served as the reference standard. MRI detected ILA in 30 cases, 20 of which were congruen t with CT findings, yielding a sensitivity of 76.9% and a specificity o f 94.9% (McNemar's test, p=0.3173). MRI detected ILA in 19/26 cases using T2 -TSE MVXD, 20/26 using T2-STIR (7/20 with hyperintense signal), and 6/26 u sing DWI (3/6 with hyperintense signal). Seven subjects showed progres sive disease on follow- up, with 6 of the subjects initially having a subpl eural fibrotic pattern. Hyperintense signals in STIR and DWI sequences pred icted progression, with hazard ratios of 6.79 and 5.43, respectively. The c ombination of hyperintense signals in STIR and DWI had a positive predictive v alue of 100%.

Conclusion

MRI reliably detected ILA and predicted disease pro gression, particularly in the fibrotic subtype. MRI offered v aluable insights for ILA monitoring and phenotyping, potentially improving p atient management and reducing radiation exposure.

Limitations

Limited amount of patients Funding for this study: No funding Ethics committee - additional information: University Hospital Bonn Author Disclosures: Julian Alexander Luetkens: Nothing to disclose Tatjana Dell: Nothing to disclose Anton Faron: Nothing to disclose Daniel Thomas: Nothing to disclose Alexander Marc Christian Boehner: Nothing to disclo se Daniel Kütting: Nothing to disclose Visualized quantitative evaluation of regional vent ilation and perfusion in patients with COPD using MRI phase-resolved functio nal lung imaging (PREFUL) Z. M. Xie, X. Gao, J. Gu, Z. Zhang, H. Yu, *L. Zhu* ; Shanghai/CN Purpose or Learning Objective: To evaluate the clinical value of phase- resolved functional lung imaging(PREFUL)in diagnosi ng and regional specificity assessment of ventilation and perfusion status of chronic obstructive pulmonary disease (COPD) patients of different seve rity.

Methods

or Background: 100 healthy volunteers, 40 patients with COPD(18 as GOLD1, 10 as GOLD2, 7 as GOLD3, and 5 as GOLD4 ) underwent MRI using 3D PREFUL under free breathing at 3.0 T(Free- breathing 1H MRI acquisition, no contrast agent administration). The PREFUL postprocessing

Method

was used for the extraction of dynamic perfu sion and ventilation parameters. Mean ventilation and perfusion maps, ve ntilation flow-volume loops(FVL) correlation, ventilation defect percenta ge(VDP), perfusion defect percentage (QDP), map of ventilation/perfusion defe cts (V/Q defects), and matched defect percentage on both perfusion and ven tilation maps (VQM) were calculated.

Results

or Findings: Compare to the homogenous ventilation and perfusion maps of healthy volunteers, COPD patients showed si gnificant heterogeneity. The mean ventilation and perfusion percentage in CO PD patients were significantly lower than the healthy volunteers (P< 0.01), while the FVL is statistically higher in COPD patients (P<0.01). The ventilation map showed regional differences in visual agreement with emphy sema on CT and all 3D PREFUL-derived ventilation parameters correlated wi th FEV1 and FEV1/FVC in the patients with COPD(all P<0.05). Besides, our data showed a trend of correlation between different GOLD grades and VDP ( P<0.05) in the COPD patients, but no significant difference in QDP amon g groups. Compared to patients with low-grade COPD (GOLD 1-2), severe COP D (GOLD 3-4) had higher VQM which indicated a better consistency of regional defect in ventilation and perfusion maps(P<0.01).

Conclusion

MRI PREFUL plays a promising role in evaluating the severity of COPD and visually predicting regional ventilation a nd perfusion defect in 3D lung imaging.

Limitations

The study included a relatively small sample size o f COPD patients Funding for this study: This work was supported in part by the National Natural Science Foundation of China (8207070786), Y oung Scientists Fund of the National Natural Science Foundation of China (8 2302188), Shanghai Pujiang Program (22PJD069). Ethics committee - additional information: Shanghai Chest Hospital ethics committee Author Disclosures: Zi Ming Xie: Nothing to disclose Junfeng Gu: Nothing to disclose Zhengqi Zhang: Nothing to disclose Xiaokun Gao: Nothing to disclose Hong Yu: Nothing to disclose Lin Zhu: Nothing to disclose AI-Enhanced 3D Gradient Echo MRI: An Alternative fo r Lung Nodule Detection and Assessment A. W. Marka¹, M. Steinhardt¹, M. Graf¹, L. Rahn¹, K . Weiss², M. R. Makowski¹, D. C. Karampinos¹, J. Gawlitza¹, *S. Ziegelmayer*¹; ¹Munich/DE, ²Hamburg/DE Purpose or Learning Objective: Recent years have seen significant progress in pulmonary MR imaging for lung nodule detection t hrough optimization and new sequences. This study evaluates the capabilitie s of a 3D gradient echo MRI sequence for detection and classification of pu lmonary nodules, specifically in relation to the Lung CT Screening R eporting and Data System (Lung-RADS).

Methods

or Background: In this prospective trial, patients with benign and malignant lung nodules admitted between December 20 21 and July 2024 underwent low-dose chest CT and pulmonary MRI using a 3D gradient echo sequence, accelerated by parallel imaging, compress ed sensing, and deep learning (CSAI). Three radiologists (4, 9, and 10 y ears of experience), blinded to clinical information, independently evaluated th e MR images. Nodule detection, characterization (size, morphology), and Lung-RADS assessment were performed for all patients. To quantify interr eader agreement, intraclass correlation coefficient (ICC) for nodule measuremen ts and Cohen’s kappa for Lung-RADS classifications were calculated.

Results

or Findings: A total of 75 patients (mean [SD] age, 65±12 years; 33 women [44%]) with 135 pulmonary nodules were includ ed and analyzed. Nominal scan time was 3:53 min. The CSAI sequence a chieved a detection rate of 96,3%, with 5 missed nodules all being ≤4mm. The mean nodule diameter for MRI deviated from CT by 0.1 mm (1.96 S D5.87mm; -1.96 SD- 5.67mm). Nodule size for CT and MRI showed excellen t inter-rater agreement (ICC-CT: 0.995, CI95: 0.993, 0.996; ICC-CSAI: 0.993 , CI95: 0.991, 0.995). Lung-RADS category agreement between CT and MRI was almost perfect for Saturday Abstract-based Programme 216 Reader 2 (k=0.86) and Reader 3 (k=0.90), while Read er 1 showed substantial agreement (k=0.69).

Conclusion

Pulmonary MRI with an accelerated 3D gradient echo sequence showed high detection rates for pulmonary nodules w ith comparable Lung- RADs scores and morphological assessments to CT.

Limitations

-Heterogeneous cohort -No follow-up scans Funding for this study: None Ethics committee - additional information: It was approved by the local ethical review board (protocol number 692/21S). Author Disclosures: Markus Graf: Nothing to disclose Marcus R. Makowski: Nothing to disclose Alexander Wolfgang Marka: Nothing to disclose Sebastian Ziegelmayer: Nothing to disclose Kilian Weiss: Nothing to disclose Leonie Rahn: Nothing to disclose Dimitrios C. Karampinos: Nothing to disclose Joshua Gawlitza: Nothing to disclose Manuel Steinhardt: Nothing to disclose Conjugate Gradient and Deep Learning Reconstruction s: Utility for Lung MRI with Ultra-Short TE to Reduce Acquisition Time with Keeping Image Quality and Nodule Detection Capability *Y. Ohno*, H. Nagata, T. Ueda, M. Nomura, T. Yoshik awa, D. Takenaka, Y. Ozawa; Toyoake/JP ([email protected]) Purpose or Learning Objective: To determine capability of Conjugate gradient reconstruction (CG-recon) and deep learnin g reconstruction (DLR) for reducing acquisition time with keeping image qualit y and nodule detection performance on UTE-MRI.

Methods

or Background: 35 patients with lung nodule underwent UTE-MRI obtained with CG-recon and grid-reconstruction (Gri d-recon) by original (UTEoriginal), 1/2 (UTE1/2) and 1/4 (UTE1/4) sampli ng spoke numbers at 1.5T and 3T systems. Then, each UTE-MRI was reconstructe d with and without DLR. Standard protocol in this study was UTEorigina l obtained by Grid-recon and reconstructed without DLR. In each patient, sta ndard reference for nodule was determined by thin-section CT. To determine the influence of sampling spoke number reduction and reconstruction method di fferences, signal-to- noise ratios (SNRs) of lung and nodule, overall ima ge quality and nodule presence probability were assessed by ROI measureme nts or 5-point scales. SNRs and overall image quality were compared betwee n each UTE-MRI and standard protocol by Student’s t-test or Wilcoxon’s signed rank test. Then, ROC analysis was performed to compare nodule detect ion capability between each UTE-MRI and standard protocol.

Results

or Findings: DLR was significantly improved SNRs of all UTEoriginals and UTE1/2 obtained by CG-recon as com pared with standard protocol (p<0.05). Overall image qualities of each UTE1/4 and all UTE1/2s except that obtained by CG-recon and reconstructed with DLR were significantly lower than that of standard protocol (p<0.05). Area under the curve (Az) of standard protocol (Az=0.97) was significant ly larger than that of all UTE1/4s (0.82<Az<0.92, p<0.0001) and UTE1/2 obtaine d by Grid-recon and reconstructed without DLR (Az=0.94, p=0.03), althou gh it was significantly smaller than that of UTEoriginals obtained by CG-re con (Az=0.98, p<0.05).

Conclusion

CG-recon and DLR can reduce acquisition time withou t degradation of image quality and nodule detection o n UTE-MRI.

Limitations

Limited study population and nodule numbers Funding for this study: Canon Medical Systems Corporation Ethics committee - additional information: Fujita Health University Hospital Author Disclosures: Yoshiyuki Ozawa: Research/Grant Support: Smoking Re search Foundation Research/Grant Support: Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, Sports, Sc ience and Technology Masahiko Nomura: Nothing to disclose Takahiro Ueda: Research/Grant Support: Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Daisuke Takenaka: Nothing to disclose Hiroyuki Nagata: Research/Grant Support: Canon Medi cal Systems Corporation Research/Grant Support: Grants-in-Aid f or Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Takeshi Yoshikawa: Nothing to disclose Yoshiharu Ohno: Research/Grant Support: Canon Medic al Systems Corporation Research/Grant Support: Smoking Researc h Foundation Visual analysis of dynamic oxygen-enhanced MRI (OE- MRI): Comparison with V/Q SPECT and MR perfusion in chronic thromboe mbolic pulmonary hypertension *G. Agarwal*¹, D. Gopalan¹, M. Naik¹, N. Soneji¹, B . Statton¹, B. Ariff¹, M. Tibiletti², G. Parker², S. Copley¹; ¹London/UK, ²Manchester/UK ([email protected]) Purpose or Learning Objective: Chronic thromboembolic pulmonary hypertension (CTEPH) is an under-recognized conditi on associated with significant morbidity yet is surgically treatable. Oxygen-enhanced MRI (OE- MRI) is an emerging tool for quantifying and mappin g regional gas delivery and uptake without the need for hyperpolarized gas. Thi s study evaluates the accuracy of visual analysis of OE-MRI for diagnosin g CTEPH compared to the current standard of care: V/Q SPECT and MR perfusio n (MR-P).

Methods

or Background: Prospective study conducted from 2018 to 2023. Participants with a clinical suspicion of CTEPH und erwent T1-weighted dynamic OE-MRI on a 1.5T scanner, V/Q SPECT and MR- P. Four consultant radiologists (2 MRI, 2 NM specialists) scored relev ant scans independently (positive, negative, or indeterminate) for CTEPH bl inded to clinical data with differences resolved by consensus. The reference st andard was the multidisciplinary team diagnosis of CTEPH.

Results

or Findings: A total of 58 patients were included, with 49 under going both OE-MRI and V/Q SPECT, and 48 undergoing all th ree modalities. Studies considered indeterminate for CTEPH were excluded fr om sensitivity and specificity analyses (2/49 V/Q, 1/48 MR-P, and 1/49 OE-MRI). The sensitivity of OE-MRI, V/Q SPECT and MR-P were 0.932 (95% CI 0. 78-0.981), 0.964 (95% CI 0.823-0.994) and 0.931 (95% CI 0.78-0.931) respectively. Specificities were 0.789 (95% CI 0.567-0.915), 0.947 (95% CI 0.75 4-0.991) and 0.833 (95% CI 0.608-0.942) respectively. There was no statisti cally significant difference between OE-MRI and V/Q or OE-MRI and MR-P using McN emar test (P > .05).

Conclusion

Visual analysis of OE-MRI maps is a valuable adjunc t in diagnosing CTEPH with similar sensitivity but lower specificity than VQ SPECT and MR-P. Further analysis of quantitative data is required to fully assess the role of this technique.

Limitations

Limitations are a relatively small sample size (n=5 8) and only visual (not quantitative) analysis. Funding for this study: Study funded by NIHR Imperial Biomedical Research Centre (BRC) grant Ethics committee - additional information: Informed consent obtained from all patients for the extra oxygen-enhanced MRI sequ ence, the other investigations were standard of care. Author Disclosures: Deepa Gopalan: Nothing to disclose Susan Copley: Nothing to disclose Mitesh Naik: Nothing to disclose Geoff Parker: Employee: Bioxydyn Ltd Board Member: Quantitative Imaging Ltd and Bioxydyn Ltd Shareholder: Quantitative Imag ing Ltd and Bioxydyn Ltd Ben Statton: Nothing to disclose Girija Agarwal: Nothing to disclose Marta Tibiletti: Board Member: Bioxydyn Limited Sha reholder: Quantitative Imaging Ltd Employee: Bioxydyn Limited Neil Soneji: Nothing to disclose Ben Ariff: Nothing to disclose Early clinical experiences for chest imaging with a new photon counting CT system combining cadmium zinc telluride detector s and super-high resolution deep-learning image reconstruction S. S. Schalekamp, L. J. Oostveen, M. Simmelink, W-J . Van Der Woude, P. P. P. Van Der Tol, M. Prokop, *E. J. Smit*; Nijm egen/NL ([email protected]) Purpose or Learning Objective: To assess the image quality of chest scans acquired using a photon-counting CT (PCCT) scanner with cadmium-zinc- telluride (CZT) detectors and super-high resolution deep-learning image reconstruction (SHR-DLR).

Methods

or Background: We analyzed the chest images from two consecutive cohorts of 18 and 25 patients who under went imaging on a prototype PCCT scanner for various indications. Ima ges were reconstructed using both normal resolution (NR: 0.62mm sections, 512-matrix, hybrid- iterative-reconstruction) and super-high resolution (SHR: 0.21mm sections, 1024-matrix, deep-learning-reconstruction) protocol s. An experienced chest radiologist assessed image quality using a 5-point scale (poor to excellent) for overall quality, sharpness, detail visibility, nois e, and artifacts. A homogeneous region in the left ventricle was used to measure im age noise. The number of visible bronchial branching generations was quantif ied in three lung regions: the upper-right-lobe(1R), the upper-left-lobe(5L), and the right-lower-lobe(10R). Bronchus volumes were automatically calculated (fir st cohort only). Statistical significance was determined using a signed rank tes t (p<0.05). Saturday Abstract-based Programme 217

Results

or Findings: The SHR-DLR images were rated to have higher overal l image quality (4.7 vs 3.6), image sharpness (4.8 vs 3.3), detail visibility (4.7 vs 3.6) compared to the NR images, while having lower perceived image noise (4.3 vs 3.3; all p<0.01). No image artifacts were o bserved with either protocol (both 4.0). Although measured noise levels were sim ilar between SHR-DLR (33.2 HU) and NR (34.2 HU), SHR-DLR images demonstr ated a finer noise texture. SHR-DLR enabled visualization of finer bro nchial details, with 1.2 more peripheral branches visible on average (p<0.01). Au tomatic segmentation showed larger bronchus volumes in the SHR-DLR image s (54.1 mL) compared to the NR images (47.2 mL;p<0.01).

Conclusion

PCCT with CZT detectors and SHR-DLR reconstruction provides excellent spatial resolution and superior visualiza tion of the bronchial system while maintaining low image noise.

Limitations

No comparison to conventional CT. Funding for this study: None Ethics committee - additional information: Waived Author Disclosures: Pieternel P P Van Der Tol: Nothing to disclose Steven S Schalekamp: Nothing to disclose Luuk J. Oostveen: Nothing to disclose Mathias Prokop: Research/Grant Support: Canon Medic al Systems Speaker: Canon Medical Systems Willem-Jan Van Der Woude: Nothing to disclose Mirte Simmelink: Nothing to disclose Ewoud J. Smit: Speaker: Canon Medical Systems Feasibility study on Photon-Counting CT-derived vir tual non-contrast images substitute for true non-contrast images L. Lei, *Y. Zhou*; Zhengzhou City/CN ([email protected]) Purpose or Learning Objective: To explore the feasibility of the virtual non- contrast images derived from Photon-Counting CT (PH CT) substitute for true non-contrast images.

Methods

or Background: 40 patients underwent pre-and arterial-venous dual- phase post-contrast chest imagining on a PHCT and h ad previously undergone a chest CT with a standard energy-integrating detec tor system (EID-CT) scanner were retrospectively included in this study . The images were retrospectively analyzed. The arterial VNC images ( VNC-A) and venous VNC images (VNC-V) were derived from raw datasets using dedicated software respectively. Two radiologists assessed image quali ty using a five-point Likert scale and performed measurements of vessels and lun g parenchyma for signal-to-noise ratio (SNR), contrast-to-noise rati o (CNR), and in the case of solid lung masses-to-lung parenchyma contrast ratio .

Results

or Findings: The image noise of all tissues among the four kinds of images had significant differences, images noise of VNC-A and VNC-V images were lower than TNC images and EID-CT images (P ＜0.05), and VNC-V images had the highest SNR and CNR. Good equivalenc e between VNC and TNC images was observed in all relevant tissues wit h Bland-Altman analysis. Image quality subjective scoring of EID-CT, TNC, VN C-A, and VNC-V were 5.00(1.00), 5.00(1.00), 5.00(0.75), 5.00(1.00), res pectively which had no significant differences (P=0.20).

Conclusion

The VNC image derived from PHCT enhanced image migh t be used as a substitute for the TNC image, and the ima ge quality is higher than conventional EID-CT images.

Limitations

Not applicable Funding for this study: Not applicable Ethics committee - additional information: This study was approved by the Human Research Ethics Committee. Author Disclosures: Limin Lei: Nothing to disclose Yuhan Zhou: Nothing to disclose Quantitative Lung Imaging using Ultra High-Resoluti on Spectral Capabilities of CZT-based Photon-Counting Detector CT: A Feasibility Study *S. Sharma*¹, S. Ross¹, T. Labno¹, R. Zhang¹, X. Zh an¹, R. Thompson², Z. Yu¹, A. Pourmorteza³; ¹Vernon Hills, IL/US, ²Cleveland, OH/US, ³Atlanta, GA/US ([email protected]) Purpose or Learning Objective: To evaluate ultra-high-resolution spectral capabilities of CZT-based photon-counting detector CT (PCD-CT) for quantitative lung imaging.

Methods

or Background: A COPDGene2 phantom, with three reference foams (20-lb, 12-lb, and 4-lb with HU-120kVp of -70 3, -824, and -937) and airways (inner-diameter (ID): 2.5-6 mm, wall-thickn ess (WT): 0.4-1.5 mm), was scanned on a CZT-based PCD-CT (120 kVp, 0.4x0.5 mm focal-spot, and CTDIvol =12.8 mGy). Scans were reconstructed in nor mal-resolution (NR) and ultra-high-resolution (UHR) spectral modes (pixel s ize (PS): 0.125 mm, slice thickness (ST): 0.2 and 0.6 mm, respectively), with a lung kernel (FC52) and iterative denoising, followed by generation of 40-1 50 keV VMIs. For evaluation, the following were quantified: (1) HU bias between UHR and NR VMIs (ST=3.0 mm), (2) contrast (C) and contrast-to-noise ratios (CNRs) for ground-glass nodules (GGNs) and emphysema (ES) (ST=3.0 mm) (20-l b, 12-lb, and 4-lb foams were surrogates for GGN, normal lung, and ES) , and (3) IDs and WTs in VMI with maximum CNR(GGN) and CNR(ES) (ST=0.6 mm).

Results

or Findings: Bias between UHR-VMI and NR-VMI was found to be <5 HU for all materials. Noise was greater in UHR-VMI (8.2-22.6%) than NR-VMI. C(GGN) and C(ES) improved in UHR-VMI compared to no n-spectral UHR images at 60 keV (max: 3 .5 HU), respectively. The 70 keV VMI was optimal for both CNR(GGN) and CN R(ES) due to minimal noise amplification. For airways, errors (mean±σ) (mm) in 70 keV UHR-VMI were lower: (UHR/NR) 0.06±0.09/0.15±0.19 (WT) and -0.18±0.17/-0.29±0.31 (ID), with highest improvements for smaller airways .

Conclusion

Spectral-UHR imaging with CZT-based PCD-CT offers diagnostically-relevant contrast improvements betwe en diseased and normal lung over non-spectral images, with reduced measure ment errors for airways.

Limitations

Study used a phantom approximating compositional si milarity to normal and diseased lung instead of real patients. Funding for this study: N/A Ethics committee - additional information: N/A Author Disclosures: Zhou Yu: Employee: Canon Medical Research USA, Inc. Steven Ross: Employee: Canon Medical Research USA, Inc. Richard Thompson: Employee: Canon Healthcare USA, I nc. Amir Pourmorteza: Nothing to disclose Tom Labno: Employee: Canon Medical Research USA, In c. Ruoqiao Zhang: Employee: Canon Medical Research USA , Inc. Xiaohui Zhan: Employee: Canon Medical Research USA, Inc. Shobhit Sharma: Employee: Canon Medical Research US A, Inc. Ultra low dose Photon Counting CT versus low dose p hoton counting CT in patients with cystic fibrosis *L. G. Murkes*, M. Lidegran, M. Sund, P. Hillergren , S. Diaz; Stockholm/SE ([email protected]) Purpose or Learning Objective: To introduce ultra low dose photon counting CT (ULDPCCT) as the main diagnostic follow up metho d in patients with cystic fibrosis (CF) and thereby reduce radiation dose wit h preserved or improved diagnostic value.

Methods

or Background: Patients with CF undergo lifelong yearly follow up alternating CT and chest radiography from an early age. CT is the method of choice and provides important information about the course of the disease, making low and ultralow dose methods imperative. Th is prospective study included 71 CF patients between 7 and 66 years of a ge. A specific study protocol was set up on a photon counting detector C T, Siemens Naeotom Alpha. All patients included underwent an inhaled a nd exhaled ULDPCCT and low dose PCCT (LDPCCT) examinations at their yearly follow up. Radiation doses were collected for each scan and patient. The median was also calculated. Images from all scans were assessed sep arately by two paediatric radiologists with different years of experience usi ng a modified Bhalla scoring system. Interobserver agreement was calculated with Cohens’ kappa coefficient. P-values of <0.05 were considered stat istically significant.

Results

or Findings: The effective dose median (IQR) was 0,11 mSv (0.1- 0.13) for the ULDPCCT and 0,77 mSv (0.66-0.87) for the LDPCCT respectively. There was no statistically significan t difference between the Bhalla scoring when comparing ULDPCCT versus LDPCCT scans with p-value 0,71 (0,37;1,04). Interobserver agreement was subst antial (Kappa value 0.65 for ULDPCCT and 0.71 for LDPCCT)

Conclusion

A tailored ULDPCCT scan protocol might be used as y early follow up diagnostic tool in CF patients with detai led diagnostic value and reduction of the radiation dose at approximately 1/ 7th of a regular LDCT, thereby reducing the accumulative dose contribution to the patient.

Limitations

Relatively limited amount of patients. Funding for this study: Funding was provided by " Riksförbundet Cystisk Fibros " for the statistical analysis Ethics committee - additional information: The study was approved by the swedish ethics commitee. According to the declarati on of Helsinkii. Dnr 2023- 01227-01 Author Disclosures: Marika Lidegran: Nothing to disclose Pierre Hillergren: Nothing to disclose Lena Gordon Murkes: Nothing to disclose Sandra Diaz: Nothing to disclose Marie Sund: Nothing to disclose Saturday Abstract-based Programme 218 AI-based body composition analysis of COPD patients ’ CT scans – a multicentric study *B. K. Budai*¹, S. Hettinger¹, V. M. Wagner¹, V. Pa lm¹, R. Hosch², F. Nensa², O. Von Stackelberg¹, H-U. Kauczor¹, J. Biederer¹; ¹ Heidelberg/DE, ²Essen/DE Purpose or Learning Objective: This study focused on AI-based CT body composition analysis (BCA) as an alternative to bio electrical impedance analysis (BIA) for identifying COPD patients at hig h risk of sarcopenia. We aimed to construct CT-based linear regression model s for predicting the patients’ fat mass (FM), fat-free mass (FFM), skele tal muscle mass (SMM), and total muscle mass (MM).

Methods

or Background: A total of 571 COPD patients (349 males (61.1%), aged 65.5 ± 8.6y) from a prospective multicentric study (COSYCONET) underwent baseline chest CT scans and BIA. The AI-b ased BCA of inspiratory chest CTs was performed by the “Body and Organ Anal ysis” (BOA) algorithm. Volumes of muscles, bones, and fatty tissues were c onverted to mass in kg using standard human tissue densities. Linear regre ssion with estimated thorax weight fitted to patient weight was used to extract residuals which combined with the respective CT-based measures, age, sex, we ight, and height were used for predicting BIA-based results. The reliabil ity of CT-predicted body composition measures was evaluated with intraclass correlation coefficients (ICC). The performance of the CT-based FFMI in iden tifying high-risk sarcopenia patients was assessed with ROC curve ana lysis.

Results

or Findings: The CT-based estimated body composition measures correlated well with BIA with ICCs of 0.90 (FM), 0. 94 (FFM), 0.92 (SMM), and 0.92 (MM). The CT-based FFMI achieved an ICC=0.88 a nd predicted high-risk sarcopenia patients with an AUC, accuracy, sensitiv ity, and specificity of 0.903, 88.3%, 88.7%, and 88.2%, respectively. Gwet’s AC1 o f 0.82 suggested excellent agreement between the two approaches.

Conclusion

AI-based body composition analysis of chest CT scan s could be used to assess BIA-based body composition measures of COPD patients and to identify patients at high risk of sarcopenia.

Limitations

The limitation of the study is that no external val idation was performed. Funding for this study: Funding was provided by the German Federal Ministry of Education and Research (BMBF) (Projektt räger: DLR e.V. Bonn, Funding ref. 01GI0884) Ethics committee - additional information: This study was approved by the local ethics committee and the central ethics commi ttee of the multicenter study. Author Disclosures: Jürgen Biederer: Nothing to disclose Sophia Hettinger: Nothing to disclose Hans-Ulrich Kauczor: Nothing to disclose Oyunbileg Von Stackelberg: Nothing to disclose Verena Maria Wagner: Nothing to disclose Viktoria Palm: Nothing to disclose René Hosch: Nothing to disclose Bettina Katalin Budai: Nothing to disclose Felix Nensa: Nothing to disclose Artificial intelligent based automated detection of chest x-ray abnormalities as a support for young radiologists *L. Giuliani*, G. M. Masci, N. Landini, P. Giuliani , V. Panebianco, C. Catalano; Rome/IT ([email protected]) Purpose or Learning Objective: To investigate whether AI represents an added value for chest X-ray (CXR) interpretation.

Methods

or Background: A dataset of CXR performed between March 2023 and January 2024 were retrospectively selected from the institutional PACS by a senior thoracic radiologist. All CXR were evaluat ed by two young radiologists with 1 year of experience in chest imaging, who ass essed the presence of several findings (consolidation, nodule, atelectasi s, fibrosis, calcification, pneumothorax, cardiomegaly, pleural effusion, media stinal enlargement, pneumoperitoneum). All examinations were then analy zed with an AI tool (Lunit INSIGHT CXR, Version 3.110) which assessed t he same features. Finally, an additional AI-assisted evaluation was p erformed by the two radiologists. The ground truth was established by t he radiologist in charge of image selection who classified the abnormalities as either visible or not visible on the radiograph.

Results

or Findings: A total of 548 CXR examinations were selected. From the 500 CXR analyzed, a total of 876 findings were reported either from the radiologists or from the AI tool. The two radiologi sts showed a sensitivity/specificity of 80.6%/93.2% and 87.2%/95 .3%, respectively. AI showed a sensitivity/specificity of 96.9%/64.3%. Wi th AI assistance, the sensitivity of the two radiologists increased to 82 .9% (+2.3%) and 89.3% (+2.1%), while specificity decreased to 87.2% (-6%) and 89.5% (-5.8%). The abnormalities for which the radiologists showed hig her disagreement with AI were fibrosis and calcification (p<0.0001), whilst the abnormalities for which the radiologists more often changed interpretation afte r AI evaluation were nodules, calcification, and pneumothorax (p<0.0001) .

Conclusion

AI does not show significant increase of the diagno stic performance compared to standard radiological evalu ation of CXR. Particularly, AI slightly increases sensitivity but at cost of a significant decrease in specificity. Therefore, standard radiological in terpretation still remains the gold standard for CXR.

Limitations

It was conducted retrospectively Funding for this study: No funding was received for this study. Ethics committee - additional information: Sapienza University, Rome, Rif.7226, Prot.0473/2024 Author Disclosures: Valeria Panebianco: Nothing to disclose Nicholas Landini: Nothing to disclose Paolo Giuliani: Nothing to disclose Giorgio Maria Masci: Nothing to disclose Carlo Catalano: Nothing to disclose Luca Giuliani: Nothing to disclose Dark-field Chest Radiography for Pneumothorax Asses sment *F. T. Gassert*, H. Bast, T. Urban, M. Lochschmidt, L. Kaster, T. Koehler, M. R. Makowski, F. Pfeiffer, D. Pfeiffer; Munich/DE Purpose or Learning Objective: Conventional imaging techniques have

Limitations

in early detection of pneumothorax, par ticularly for small pneumothoraces. Therefore, the purpose of this stud y was to evaluate the potential of dark-field chest radiography in improv ing the detection and assessment of pneumothorax.

Methods

or Background: This study included 100 participants, comprising 36 patients with clinically diagnosed pneumothorax and 64 healthy controls. All participants underwent dark-field X-ray chest radio graphy using a prototype system that simultaneously acquires attenuation-bas ed and dark-field images. Sensitivity, specificity, accuracy, reading time, a nd diagnostic confidence were compared between attenuation-based radiographs and the combination of attenuation-based radiographs with dark-field image s (dark-field overlays).

Results

or Findings: Dark-field radiography increased sensitivity for pneumothorax detection from 84.2% (attenuation-base d radiographs) to 87.4% (dark-field overlays; p = .26), while specificity r emained constant (97.3% vs. 97.4%). The median reading time was significantly r educed from 30.8 seconds to 10.3 seconds (p < .001), and diagnostic confiden ce improved significantly across all readers (p < .001).

Conclusion

Dark-field chest radiography enhances the detection of pneumothorax, significantly reducing reading time a nd increasing diagnostic confidence without compromising specificity.

Limitations

Inclusion criteria and the imaging modality they we re assessed on were different for pneumothorax patients and contro ls. While for pneumothorax patients, a conventional radiograph showing a pneum othorax was sufficient, controls had to show a normal CT scan to be include d. Funding for this study: We acknowledge financial support through the European Research Council (AdG 695045), the Center for Advanced Laser Applications (CALA), the Federal Ministry of Educat ion and Research (BMBF) and the Free State of Bavaria under the Excellence Strategy of the Federal Government and the Länder, the German Research Foun dation (GRK2274), as well as by the Technical University of Munich–In stitute for Advanced Study. This work was carried out with the support of the K arlsruhe Nano Micro Facility (KNMF, KNMF, www.kit.edu/knmf), a Helmholtz Researc h Infrastructure at Karlsruhe Institute of Technology (KIT). Ethics committee - additional information: Ethics committee of the Technical University of Munich Author Disclosures: Henriette Bast: Nothing to disclose Florian Tilman Gassert: Nothing to disclose Thomas Koehler: Employee: Philips Marcus R. Makowski: Nothing to disclose Theresa Urban: Nothing to disclose Lennard Kaster: Nothing to disclose Franz Pfeiffer: Nothing to disclose Maximilian Lochschmidt: Nothing to disclose Daniela Pfeiffer: Nothing to disclose Saturday Abstract-based Programme 219 12:30-13:30 Research Stage 1 Research Presentation Session: Breast RPS 1902

Introduction

of artificial intelligence in breast screening Moderator S. Schiaffino; San Donato Milanese/IT ([email protected]) Author Disclosures: Simone Schiaffino: Board Member: European Journal o f Radiology, Eurorad, European Radiology Experimental; Speaker: GE Health care Impact on quality performance indicators after impl ementing AI in a breast cancer screening program in Germany K. Hamm¹, *A. Rodriguez Ruiz*², T. Jordan¹, B. Vett er¹, C. Engel³, C. Entrup⁴, M. Engelke⁵; ¹Chemnitz/DE, ²Nijmegen/NL, ³Leipzig/DE, ⁴Koblenz/DE, ⁵Hamburg/DE Purpose or Learning Objective: To evaluate breast cancer screening quality indicators after implementation of an AI system for support reading mammograms.

Methods

or Background: Two prospective and consecutive collected cohorts of women attending breast cancer screening with mam mography in a region of Germany where identified, just before and after imp lementation of an AI decision support system to aid radiologists reading mammograms (Transpara version 1.7, ScreenPoint Medical). Before AI implem entation, all mammograms were double read without AI. Afterwards, mammograms were double read using AI as concurrent decision support. All mammog rams were acquired with same devices (Siemens Mammomat Inspiration). A tota l of X radiologists assessed the exams in this screening program. Scree ning quality indicators (cancer detection rate, recall rate, false positive rate, PPV2) were compared in the cohorts of women before and after implementatio n of AI using multivariate logistic models adjusted for age, breast density, a nd interval from previous examination.

Results

or Findings: 59.676 women attending screening before AI implementation (2020-2021) and 58.546 women after A I implementation (2022- 2023) were included in the analysis. Average age wa s 60 years old in both cohorts. Average number of months between rounds wa s 838 days in the no- AI cohort and 815 days in the AI cohort. After impl ementing AI, cancer detection rate increased (349 screen-detected cance rs, 6.0/1000 vs 286 screen-detected cancers, 4.8/1000, p=0.01), recall rate remained stable (2.5% vs 2.6%, p=0.29), false positive rate was reduced ( 1.9% vs 2.1%, p=0.002), and PPV2 increased (69%, 349/509 vs 60%, 286/477, p =0.009).

Conclusion

Implementing AI to support radiologists reading mam mograms in a breast cancer screening program in Germany is saf e and effective, improving cancer detection rates and reducing false positives .

Limitations

This prospective study has a non-paired non-randomi zed design. Funding for this study: None. Ethics committee - additional information: Approved by local ethics committee. Author Disclosures: Alejandro Rodriguez Ruiz: Employee: ScreenPoint Med ical Martin Engelke: Nothing to disclose Klaus Hamm: Nothing to disclose Torsten Jordan: Nothing to disclose Bert Vetter: Nothing to disclose Christoph Engel: Nothing to disclose Christian Entrup: Nothing to disclose Breast Cancer Characteristics after the introductio n of Artificial Intelligence-supported double-reading in a Mammogra phy Screening Program: comparison of baseline and subsequent roun ds *C. M. Weiss*, E. Di Gaetano, E. Cattarin, R. Cerni ato, G. Soppelsa, I. Vinci; Treviso/IT ([email protected]) Purpose or Learning Objective: To analyse the prognostic factors of breast cancers (BCs) detected with artificial intelligence -supported double-reading (AI-DR) and attempt to determine the long-term impa ct of these changes, particularly on possible overdiagnosis.

Methods

or Background: AI-DR was applied to all digital screening mammograms (DSM) from November 2021 to June 2024: 9 9320 in the AI- baseline-screen (AIBS) and 21237 in the AI-subseque nt-screen (AISS). The collected data were compared by retrospective analy sis to determine whether AIBS screen-detected BCs differed from AISS. We use d the Z-test to compare the proportions of the data between AIBS and AISS.

Results

or Findings: With a total of 1093 screen-detected BCs (AIBS: 944/99320; AISS: 149/21367), the study revealed a d ecrease (-26.6%) in the cancer detection rate (CDR) per 1000 in AISS compar ed to AIBS (6.97vs9.5). The recall rate (RR) was lower (-42.3%) in AISS tha n in AIBS (1.8%vs3.1%). No significant differences were found in the percen tage of invasive BCs (AIBS82%vsAISS81.9%) and in situ BCs (AIBS18%vsAISS 18.1%). Higher percentages of luminal BCs were observed in the AIB S than in AISS (91.1%vs82.6%), while in AISS, there were higher pe rcentages of high-grade (AIBS25.3%vs AISS33.6%), HER2positive (AIBS 8.3%vsA ISS12.4%) and triple-negative BCs (AIBS2.6%vsAISS 5%).

Conclusion

The reduction of RR and CDR in AISS aligns with the expectation of later screening focusing on disease onset or pro gression cases. In AISS, compared to AIBS, more aggressive BCs were detected , while less aggressive BCs were reduced. This might suggest an improved pe rformance in the second round of screening, with a positive impact o n the reduction of overdiagnosis.

Limitations

Data on interval and advanced cancers are lacking, which would allow for an analysis of long-term clinical outcome s. Funding for this study: No funding Ethics committee - additional information: Not requested Author Disclosures: Giorgia Soppelsa: Nothing to disclose Eleonora Di Gaetano: Nothing to disclose Elisa Cattarin: Nothing to disclose Ivana Vinci: Nothing to disclose Claudia Maria Weiss: Nothing to disclose Roberta Cerniato: Nothing to disclose Multicenter Analysis of AI Assessed Mammography Tec hnologist Positioning Variability Between Breast Screening Pr ograms *G. Spear*¹, L. R. Margolies², J. Payne³, S. E. E. Iles³, J. Seely⁴, N. Sharma⁵, S. H. Heywang-Köbrunner⁶, T. W. W. Vomweg⁷, M. Abdolell³; ¹Chicago, IL/US, ²New York, NY/US, ³Halifax, NS/CA, ⁴Ottawa, ON/CA, ⁵Leeds/UK, ⁶Munich/DE, ⁷Koblenz/DE ([email protected]) Purpose or Learning Objective: Variability in mammographic positioning quality, both between breast screening programs (BS Ps) and across different positioning errors, presents a challenge to establi shing standardized mammography quality service delivery. Although high -quality mammography helps ensure diagnostic accuracy, and training can enhance image quality, there is a lack of supporting population-level empi rical data. This study aims to quantitatively assess mammography technologists’ po sitioning error rates across BSPs.

Methods

or Background: The MAMMO.IQ study encompassed a total of 249,817 screening mammograms acquired between Decem ber 1, 2019, and February 28, 2021, from seven BSPs across North Ame rica and Europe. The positioning errors assessed included: exaggeration, portion cut off, posterior tissues missing, nipple not in profile, too high on IR, pectoralis shape/position, sagging, IMF missing/obscured, PNL difference, and compression. The Coefficient of Variation (CV) assessed variability in error rates, (1) between BSPs, and (2) between positioning errors. The withi n-BSS CV for each unmet positioning criterion was computed using rates for all technologists within a BSS.

Results

or Findings: Images acquired by 310 technologists were analyzed. Over/under compression had the lowest variability ( CV=16.44%) indicating consistent practices. Too High on IR exhibited the highest variability (CV=71.52%) reflecting a high level of inconsistenc ies. The MLO Inadequate Pectoralis Length had a CV of 50.45%, representing the median level of variability.

Conclusion

This study highlights variability in mammography te chnologists’ positioning errors between and within BSPs. While s ome positioning criteria show consistent practices, others may benefit from improved standardization. Understanding inconsistencies in mammography servic e delivery helps identify opportunities to standardize positioning practices and reduce variability, leading to more equitable, high-quality care and fe wer positioning errors.

Limitations

Missing data on technologist experience, staffing, and COVID-19 response measures limits the understanding of facto rs driving disparities. Funding for this study: No funding was received for this study. Ethics committee - additional information: Ethics approvals were obtained from participating BSPs (NSHA-REB#1026590). Saturday Abstract-based Programme 220 Author Disclosures: Nisha Sharma: Advisory Board: Densitas Laurie R. Margolies: Nothing to disclose Sylvia H. Heywang-Köbrunner: Nothing to disclose Sian E. Elizabeth Iles: Nothing to disclose Jean Seely: Nothing to disclose Jennifer Payne: Other: Densitas Toni Werner W Vomweg: Nothing to disclose Georgia Spear: Nothing to disclose Mohamed Abdolell: CEO: Densitas Characteristics of Breast Cancers before and after the introduction of Artificial Intelligence-supported double reading in a Mammography Screening Program *C. M. Weiss*, E. Di Gaetano, E. Cattarin, R. Cerni ato, G. Soppelsa, I. Vinci; Treviso/IT ([email protected]) Purpose or Learning Objective: Screening supported by artificial intelligence (AI) increased cancer detection compared to screeni ng without AI. However, it is still unclear whether the additional cancer dete ction improves outcomes or leads to overdiagnosis of breast cancers (BCs).

Methods

or Background: From January 2019 to October 2021, 134259 women underwent digital screening mammography (DSM) with human-double- reading (HDR) and from November 2021 to June 2024, 131406 DMS with AI- supported HDR (AI-HDR) DM. The collected data (canc er detection rate [CDR], recall rate [RR] and tumour characteristics) were c ompared by retrospective analysis to determine whether the BCs detected by s creening differed between HDR and AI-HDR. We used the Z-test to compare the p roportions of the data between HDR and AI-HDR.

Results

or Findings: With a total of 2044 screen-detected BCs (HDR:938/134259; AI-HDR:1106/131406), the study rev ealed a significant increase (+20.5%) in CDR per 1000 with AI-HDR compa red to HDR (8.42vs6.99 per 1000, respectively). The RR was low er (-14.9%) with AI-HDR than with HDR (2.6%vs3.1%). The AI-HDR showed the f ollowing differences in BCs rates compared to HDR: lower for invasive BCs ( 82.1%vs84.1%), and higher for in situ BCs (17.9%vs15.9%); higher moder ate-grade BCs (65.5%vs61.5%), and lower high-grade BCs (26.4%vs 2 7.9%); higher luminal BCs (88.2%vs83.2%), lower HER2positive (8.7%vs13.2% ), and lower triple negatives (2.9%vs3.6%).

Conclusion

It can be concluded that the use of AI-HDR produced statistically significant differences in detecting various tumour subtypes compared to HDR. In particular, it seems to have increased the detec tion of less aggressive BCs and reduced unnecessary recalls.

Limitations

Data on interval and advanced cancers are lacking, which would allow for an analysis of long-term clinical outcome s. Funding for this study: No fundings Ethics committee - additional information: Not requested Author Disclosures: Giorgia Soppelsa: Nothing to disclose Eleonora Di Gaetano: Nothing to disclose Elisa Cattarin: Nothing to disclose Ivana Vinci: Nothing to disclose Claudia Maria Weiss: Nothing to disclose Roberta Cerniato: Nothing to disclose Artificial intelligence as an initial reader for do uble reading in breast cancer screening: A prospective initial study of 32 ,822 mammograms of the Egyptian population S. A. Mansour, R. M. Kamal, M. M. Gomaa, E. Sweed, S. Hussien, E. Abdalla, *Y. M. Nada*, G. Mohamed, A. F. I. Moustafa; Cairo/ EG ([email protected]) Purpose or Learning Objective: Although artificial intelligence (AI) has potential in the field of screening of breast cance r, there are still issues. It is vital to make sure AI doesn't overlook cancer or ca use needless recalls. The aim of this work was to investigate the effectivene ss of indulging AI in combination with one radiologist in the routine dou ble reading of mammography for breast cancer screening.

Methods

or Background: The study prospectively analyzed 32822 screening mammograms. Reading was performed in a blind-paired style by i) two radiologists and ii) one radiologist paired with AI . A heatmap and abnormality scoring percentage were provided by AI for abnormal ities detected on mammograms. Negative mammograms and benign-looking lesions that were not biopsied were confirmed by a 2-year follow-up.

Results

or Findings: Double reading by the radiologist and AI detected 1 324 cancers (6.4%); on the other side, reading by two r adiologists revealed 1293 cancers (6.2%) and presented a relative proportion of 1·02 (p<0·0001). At the recall stage, suspicion and biopsy recommendation w ere more presented by the AI plus one radiologist combination than by the two radiologists. The interpretation of the mammogram by AI plus only one radiologist showed a sensitivity of 94.03%, a specificity of 99.75%, a p ositive predictive value of 96.571%, a negative predictive value of 99.567%, an d an accuracy of 99.369% (from 99.252% to 99.472%). The positive likelihood ratio was 387.260, negative likelihood ratio was 0.060, and AUC “area under the curve” was 0.969 (0.967 to 0.971).

Conclusion

AI could be used as an initial reader for the evalu ation of screening mammography in routine workflow. Implemen tation of AI enhanced the opportunity to reduce false negative cases and supported the decision to recall or biopsy.

Limitations

The study is a single institute work. Funding for this study: No source of funding. Ethics committee - additional information: The study has been approved by the Baheya Charity Hospital research center. Author Disclosures: Sahar Abdelkhalek Mansour: Nothing to disclose Samar Hussien: Nothing to disclose Mohammed Mohamed Gomaa: Nothing to disclose Engy Abdalla: Nothing to disclose Enas Sweed: Nothing to disclose Yasmin Mohamed Nada: Nothing to disclose Ghada Mohamed: Nothing to disclose Rasha Mohamed Kamal: Nothing to disclose Amr Farouk Ibrahim Moustafa: Nothing to disclose Simulating single reading for high-risk examination s in the randomized controlled Mammography Screening with Artificial In telligence trial (MASAI) *V. Josefsson*, D. Schmidt, H. Sartor, O. Hagberg, K. Lang; Malmö/SE ([email protected]) Purpose or Learning Objective: To assess the value of double reading of high-risk examinations in the MASAI trial.

Methods

or Background: In the randomised controlled MASAI trial AI supported screening was compared to standard double reading. AI was used to triage exams to single or double reading dependi ng on malignancy risk and as detection support. Of the 53 048 participants in the intervention arm, 3800 exams were high risk and underwent double reading w hile the remaining exams underwent single reading. In this retrospecti ve study, we assessed the relative performance in the intervention arm, compa ring simulated single and factual double reading of high-risk exams and its e ffect on cancer detection, recalls, and false positives. Cancers solely detect ed by the second reader were described.

Results

or Findings: The simulated single reading scenario resulted in 8 .9% (308 vs. 338) fewer detected cancers and 5.9% fewer recalls (1045 vs. 1110) compared to the factual outcome in the intervention arm. Corresponding simulated vs. factual rates were 5.8/1000 vs. 6.4/1 000 for cancer detection, 2.0% vs. 2.1% for recalls and 1.4% vs. 1.5% for fal se positive. Of the 30 cancers solely detected by the second reader, 24 (8 0.0%) were invasive and 21 (70.0%) were classified as T1. Of the invasive c ancer, 23 (95.8%) were lymph-node negative and 8 (33.3%) non-luminal A, of which four were triple negative.

Conclusion

Double reading of high-risk exams improved cancer d etection without unduly increasing false positives. The addi tional cancers detected were mostly small, lymph-node negative invasive cancers, including those of significant prognostic subtypes. These findings sup port the continued use of double reading for high-risk exams in AI-supported screening.

Limitations

Single-institution trial. Retrospective simulation. Funding for this study: The Swedish Cancer Society Regional Cancer Centers in Collaboration Lund University ALF-funds Ethics committee - additional information: The Swedish Ethical Review Authority 2020-04936 2023-026848-02 Author Disclosures: Kristina Lang: Nothing to disclose Viktoria Josefsson: Nothing to disclose Hanna Sartor: Nothing to disclose Oskar Hagberg: Nothing to disclose David Schmidt: Nothing to disclose Saturday Abstract-based Programme 221 Using prior mammograms to improve specificity of an AI system for breast cancer detection: a large-scale retrospectiv e multi-site validation *A. Rodriguez Ruiz*¹, S. Pires¹, R. Peeters¹, G. Ro driguez-Esteban¹, D. Sperber¹, C. De Wolf², J. L. Raya Povedano³, S. Romero Martin³, R. Mann¹; ¹Nijmegen/NL, ²Geneva/CH, ³Cordoba/ES Purpose or Learning Objective: To investigate how the use of prior mammograms impacts breast cancer detection performa nce of an AI system.

Methods

or Background: Mammograms from women attending three European screening programs were collected based on availability of prior images and at least 2 years follow-up, including or iginal radiologists assessments. Each case was analyzed by a breast can cer detection AI product (Transpara, ScreenPoint Medical, v2.1), res ulting in two cancer risk scores: using as input the current mammogram alone and using prior mammograms. AI specificity was compared between usi ng priors or not, matching the single radiologist sensitivity . Subse quently, the combination of a single radiologist and AI was modelled and compared to double human reading. P-values using McNemar and binomial confid ence intervals were computed.

Results

or Findings: 37,148 cases were included (20,300 from Switzerland , 916 from Spain 15,932 from The Netherlands), with 1 ,034 recalled cases (2.8%), 247 screen-detected cancers (6.6/1000), and 59 interval cancers (1.6/1000). 56% of cases had 1 prior mammogram, 44% had 2 or 3. Images were acquired with Hologic, Siemens, GE, Planmed an d Philips machines. At the average sensitivity of a single radiologist (71 .2%), AI specificity increased when using prior images, from 98.1% (98.0-98.2%) to 98.8% (98.7-98.9%), representing a 37% reduction in false positives (fr om 1.9% to 1.2%, P<0.001). Combining AI using priors with a radiologist achiev ed comparable sensitivity (83.0% vs 82.0%, P=0.66) and higher specificity (96 .3%, 96.0-96.5%) than double human reading before consensus (95.4%, 95.2- 95.6%), representing 20% fewer false positives (P<0.001). The improved s pecificity was higher for cases with breast density C/D (+0.8%, P<0.001) than for A/B (+0.3%, P=0.005).

Conclusion

A higher specificity was achieved by a breast cance r detection AI system using prior mammograms, potentially offering better aid to radiologists in breast cancer screening.

Limitations

Retrospective design. Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Santiago Pires: Employee: ScreenPoint Medical Daan Sperber: Employee: ScreenPoint Medical Alejandro Rodriguez Ruiz: Employee: ScreenPoint Med ical Ruud Peeters: Employee: ScreenPoint Medical Sara Romero Martin: Nothing to disclose Jose Luis Raya Povedano: Nothing to disclose Gonzalo Rodriguez-Esteban: Employee: ScreenPoint Me dical Chris De Wolf: Nothing to disclose Ritse Mann: Nothing to disclose 12:30-13:30 Research Stage 2 Research Presentation Session: Vascular RPS 1915 Carotid and intracranial artery imaging Moderator V. Silvestri; Seclin/FR ([email protected]) Sub-1-minute Relaxation-Enhanced Angiography withou t Contrast and Triggering of the Extracranial Arteries *J. P. Janssen*¹, K. Kaya¹, R. A. Terzis¹, J. Trist ram¹, R. J. Gertz¹, L. Goertz¹, L. Pennig¹, C. H. Gietzen¹, K. Weiss²; ¹Cologne/DE, ²Hamburg/DE ([email protected]) Purpose or Learning Objective: To evaluate the acceleration of a 3D isotropic flow-independent non-contrast MRA (REACT) of the neck using Compressed SENSE (CS) combined with deep learning-b ased reconstruction (CS-AI).

Methods

or Background: Thirty-four volunteers received cervical REACT at 3T ((acquired threefold: (1) CS factor 7 (scan time : 1:20 min), (2) CS factor 10 (0:55 min), and (3) CS-AI factor 10 (0:55 min)). Tw o radiologists rated the image quality of seven arterial segments and overal l image noise. Additionally, a pairwise forced-choice comparison was conducted. Apparent signal- (aSNR) and contrast-to-noise ratios (aCNR) were measured, and image sharpness was assessed by calculating the edge rise distance (ERD). Multiple t-tests and non-parametric tests with Bonferroni correction wer e performed for comparison to CS7, which was considered as the current clinica l standard.

Results

or Findings: Compared to CS7, CS10 showed lower image quality scores (p<0.001) while CS10-AI obtained higher resu lts (p=0.010). Image noise was similar between CS7 and CS10 (p=0.138) wh ile CS10-AI yielded a lower noise (p=0.008). Forced choice revealed prefe rences for CS7 over CS10 (p0.999). Compared to CS7, aSNR and aCNR were lower in CS10 (p<0.001) and the ERD was longer (p=0.004), while CS10-AI provided better aSN R and aCNR (p=0.001) and showed no difference in ERD (p=0.776).

Conclusion

CS-AI enables the acquisition of cervical REACT in less than one minute without compromising image quality. Further studies are required to confirm these results in patients and to evaluate t he diagnostic performance regarding vascular findings such as stenosis or dis section.

Limitations

No pathologies were assessed. No comparison was mad e with established reference standards. Funding for this study: Not applicable. Ethics committee - additional information: Approved by our institutional review board (reference number: 20-1296_1) and regi stered in the national registry for clinical trials (DRKS00030210). Author Disclosures: Lenhard Pennig: Speaker: Guerbet GmbH Speaker: Phil ips Healthcare Carsten H. Gietzen: Nothing to disclose Juliana Tristram: Nothing to disclose Roman Johannes Gertz: Speaker: Guerbet GmbH Speaker : Philips Healthcare Kenan Kaya: Nothing to disclose Kilian Weiss: Employee: Philips GmbH Robert Angelo Terzis: Nothing to disclose Jan Paul Janssen: Nothing to disclose Lukas Goertz: Nothing to disclose Carotid artery assessment in dual-source photon-cou nting CT: impact of low-energy virtual monoenergetic imaging on image q uality, vascular contrast and diagnostic assessability *A-I. Nica*¹, C. Booz¹, G. M. Bucolo¹, L. S. Alizad eh¹, T. Vogl¹, T. D'Angelo², H-L. Kaatsch³, D. Overhoff³, S. Waldeck³; ¹Frankfur t/DE, ²Messina/IT, ³Koblenz/DE Purpose or Learning Objective: The purpose of this study is to evaluate the impact of low-energy VMI reconstructions on quantit ative and qualitative image quality, vascular contrast, and diagnostic assessab ility of the carotid arteries in photon-counting CTA.

Methods

or Background: A total of 122 patients (67 male) who had undergone dual-source photon-counting CTA scans of the carotid artery were retrospectively analyzed in this study. Standard 12 0 kV CT images and low- keV VMI series from 40 to 100 keV with an interval of 15 keV were reconstructed. Quantitative analyses included the e valuation of vascular CT numbers, signal-to-noise ratio (SNR), and contrast- to-noise ratio (CNR). CT number measurements were performed in the common, e xternal, and internal carotid arteries. Qualitative analyses were perform ed by three board-certified radiologists independently using five-point scales to evaluate image quality, vascular contrast, and diagnostic assessability of the carotid arteries.

Results

or Findings: Mean attenuation, CNR and SNR values were highest i n 40 keV VMI reconstructions (HU, 1362.32 ± 457.81; CNR, 33.19 ± 12.86; SNR, 34.37 ± 12.89) followed by 55-keV VMI reconstructions; all three mean values at these keV levels were significantly higher compa red with standard 120 kV CT series (HU, 154.43 ± 23.69; CNR, 16.34 ± 5.47; SNR, 24.44 ± 7.14) (p < 0.0001). The qualitative analysis showed highest ra ting scores for 55 keV VMI reconstructions followed by 40 keV and 70 keV V MI series with a significant difference compared to standard 120 kV CT images regarding image quality, vascular contrast, and diagnostic as sessability of the carotid arteries (all comparisons, p < 0.01).

Conclusion

Low-keV VMI reconstructions at a level of 40–55 keV significantly improve image quality, vascular contrast, and the d iagnostic assessability of the carotid arteries compared with standard CT seri es in photon-counting CTA.

Limitations

Single-center retrospective study Funding for this study: No funding was received Ethics committee - additional information: The local IRB approved this study. Author Disclosures: Christian Booz: Speaker: Siemens Healthineers Thomas Vogl: Nothing to disclose Daniel Overhoff: Nothing to disclose Andreea-Ioana Nica: Nothing to disclose Hanns-Leonhard Kaatsch: Nothing to disclose Stephan Waldeck: Nothing to disclose Tommaso D'Angelo: Speaker: Philips Speaker: Bracco Leona Soraja Alizadeh: Nothing to disclose Giuseppe Mauro Bucolo: Nothing to disclose Saturday Abstract-based Programme 222 A Novel Approach in Vascular Imaging: AI-Driven 3D Reconstruction of Carotid Arteries for Enhanced Stroke Risk Assessmen t K. Gasbarrino¹, *A. Benjamin*¹, T. Beiko¹, J. Ramir ez-Garcia Luna¹, R. Khan¹, L. H. Gonzalez Torres¹, S. Levasseur¹, S. Taj², K. Khan¹; ¹Montreal/CA, ²Columbia, MD/US Purpose or Learning Objective: The standard approach to assessing stroke risk via 2D carotid ultrasound is limited by operat or variability, a lack of 3D vessel visualization, and subjective interpretation , resulting in a nearly 30% misclassification rate. To address these challenges , we developed AI-powered software that transforms 2D ultrasound images into precise 3D models of carotid arteries and automates vessel measurements.

Methods

or Background: We applied a multi-class U-Net AI model, trained on ~4000 2D ultrasound images from 113 North American patients with cardiovascular risk factors. Two independent sonogr aphers annotated these images, identifying key vascular structures, includ ing medial-adventitial boundary, intimal-luminal boundary, and plaque. 3D reconstructions were achieved by integrating 2D image segmentations with positional data captured from an electromagnetic sensor (Northern Digital In c, Canada) during a single B-mode sweep of the carotid artery. Algorithms were developed for automated measurement of vessel diameter, artery stenosis, an d classification of disease severity. Validation was conducted using a carotid artery phantom with a predefined 70% stenosis (R.G. Shelley Ltd, Canada), along with clinical evaluation in 8 patients to compare performance aga inst the current standard of care.

Results

or Findings: The AI model demonstrated strong performance, achieving a DICE coefficient of 0.86 in detecting v essel structures. The software successfully generated 3D models, with vas cular metrics showing a 99% agreement with the known stenosis in the phanto m model. Intra-operator variability was minimal, with stenosis measurements showing only minor deviations (71.42±3.42%). In the clinical study, a 90% reduction in ultrasound scan was achieved, while maintaining diagnostic acc uracy equivalent to that of a vascular radiologist with >10 years of experience .

Conclusion

Our software represents a significant advancement i n carotid artery imaging, delivering a ten-fold improvement i n scan efficiency while achieving expert-level diagnostic accuracy with min imal variability.

Limitations

N/A Funding for this study: Ontario Brain Institute; Québec's Ministère de l'Économie, de l'Innovation et de l'Énergie Ethics committee - additional information: The study was approved by Advarra IRB (Pro00068778) Author Disclosures: Luis H. Gonzalez Torres: Nothing to disclose Rafia Khan: Consultant: Sonaro Inc Jose Ramirez-Garcia Luna: Shareholder: Sonaro Inc Sophie Levasseur: Nothing to disclose Karina Gasbarrino: Founder: Sonaro Inc Thierry Beiko: Employee: Sonaro Inc. Kashif Khan: Founder: Sonaro Inc Sabir Taj: Advisory Board: Sonaro Inc Alex Benjamin: Research/Grant Support: Sonaro Association Between Pericarotid Fat Density and Hem orrhagic Transformation After Endovascular Therapy for Acute Extracranial Internal Carotid Artery Occlusion *Z. Cui*, J. Zhang; Shanghai/CN ([email protected]) Purpose or Learning Objective: This study aimed to investigate the association between pericarotid fat density (PFD) a round the occlusion and hemorrhagic transformation (HT) risk and functional outcome in acute ischemic stroke (AIS) patients with extracranial internal ca rotid artery (e-ICA) occlusion who underwent endovascular thrombectomy (EVT).

Methods

or Background: This multicenter retrospective study included a cohort of patients with e-ICA occlusion after EVT b etween June 2019 and March 2024. PFD was assessed using semi-automated q uantitative software at pre-operation neck CT angiography (CTA). The ass ociations between PFD and HT, and functional outcome (mRS score 0-3 vs 4- 6) were analyzed using multivariable logistic regression. A mediation anal ysis was conducted to explore whether HT mediates the relationship betwee n PFD and functional outcome. Additionally, We sought the association be tween PFD and admission C-reactive protein (CRP) levels.

Results

or Findings: 101 patients were included and divided into an HT g roup (n=36) and a non-HT group (n=65). PFD was independe ntly associated with HT (adjusted odds ratio [aOR]: 1.84, 95% CI: 1.28 t o 2.66, P<0.001) and unfavorable functional outcome (1.41, 1.04 to 1.91, p=0.030). The AUC values were 0.79 (95% CI, 0.70 to 0.89) and 0.68 (0.57 to 0.78), indicating a favorable predictive performance for the HT risk and unfavora ble prognosis prediction. Mediation analysis revealed that HT explained more than 60% of the relationship between PFD and worse functional outco me. In addition, higher PFD was positively correlated with elevated CRP lev els.

Conclusion

PFD is an independent predictor of HT and a worse f unctional outcome at 90 days in patients with AIS and e-ICA o cclusion who underwent EVT. Evaluation of PFD provides opportunities for H T risk stratification and outcome prediction.

Limitations

Given the sample size, we were unable to apply the more stringent inclusion criteria of restricted to the e xtracranial ICA. Funding for this study: No funding was received for this study. Ethics committee - additional information: Informed consent was waived for all participants due to the retrospective nature of the study. Author Disclosures: Jun Zhang: Nothing to disclose Zhimeng Cui: Nothing to disclose The influence of visual signals on blood flow in th e central retinal and internal carotid arteries M. Beraia, D. Gachechiladze, *A. Siradze*, N. Eliav a, M. Lazarashvili, N. Nikabadze, S. Siradze, J. Giorgelashvili, L. Udu mashvili; Tbilisi/GE ([email protected]) Purpose or Learning Objective: The retina offers a unique window into brain structure and functional disorders due to its anato mical, physiological, and embryological similarities with the brain. This res earch explores the influence of visual-verbal/nonverbal stimuli on blood flow in bo th the retina and brain.

Methods

or Background: A duplex Ultrasound study was conducted with 25 volunteers (11 males, 14 females, aged 21–35), exam ining blood flow in the central retinal artery (CRA) and internal carotid a rtery (ICA) under two types of visual stimuli: verbal irritation (Shakespeare’s so nnets) and nonverbal (pictures – find the hidden figures). Blood flow parameters ( Vsys, Vdia, PI, and RI) were measured in intervals 1-15 and 25-40 seconds after stimuli initiation, to assess the nature (neuro/humoral) of blood flow regulation (RBC circulation time: 20 seconds). Initial 5sec for the baseline images. CE- MRI angiography (TOF) was used to exclude vascular anomalies.

Results

or Findings: In the CRA, Vsys increased from 7.5–11.5 cm/sec to 10.7–14.3 cm/sec, with Vdia at 4.1–4.9 cm/sec. RI r ose from 0.53 to 0.64, and PI from 0.71 to 0.94. In the ICA, Vsys rose from 80 –130 cm/sec to 95–170 cm/sec, with Vdia at 24–45 cm/sec. RI increased fro m 0.67 to 0.74, and PI from 1.01 to 1.17. Changes in the CRA and ICA were unidirectional (r = 0.7). In nonverbal cases, Vsys and RI were higher in the CRA (P < 0.05) and ICA (P < 0.01). The time of the stimuli initiation did not c hange the results. These details indicate the mostly sympathetic regulation of cereb ral blood flow.

Conclusion

Quantitative blood circulation studies in these art eries may be used as the functional ultrasound diagnostics of th e retina and brain.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee notification can be found under the number UID 2438. Author Disclosures: Nino Eliava: Nothing to disclose Luarsabi Udumashvili: Nothing to disclose Aneta Siradze: Nothing to disclose Dudana Gachechiladze: Nothing to disclose Mariam Lazarashvili: Nothing to disclose Salome Siradze: Nothing to disclose Merab Beraia: Nothing to disclose Jano Giorgelashvili: Nothing to disclose Nini Nikabadze: Nothing to disclose Intracranial arterial calcification detection; a co mparison between ultra- high-resolution photon-counting CT, conventional en ergy-integrating CT and micro-CT *J. Van Der Bie*¹, B. P. Berghout¹, R. P. J. Budde¹ , J. Gutierrez², M. Van Straten¹, D. Bos¹; ¹Rotterdam/NL, ²New york, NY/US ([email protected]) Purpose or Learning Objective: To assess the performance of photon- counting detector CT (PCD-CT) in detecting and quan tifying intracranial arterial calcifications and comparing the performance to con ventional energy- integrating CT (EID-CT), using micro-CT (µCT) as th e reference standard.

Methods

or Background: Thirty histopathological cross-sections of intracranial arteries were scanned with PCD-CT, EID -CT, and µCT. µCT was optimized for image quality (reference standard), w hile clinical protocols were used for PCD-CT and EID-CT. Various reconstruction kernels (EID-CT: Hv40/Hv49/Hv59; PCD-CT: Hv40/Hv48/Hv56/Hv64/Hv72/Hv 89) were used to enhance spatial resolution. Two experienced observe rs independently evaluated the presence of calcifications in all acq uisitions and were compared by Cohen’s Kappa and concordance percentages. For o bjective analysis, mass scores were used to assess both the detection and m ass. The objective measure was analyzed using Bland-Altman plots. Saturday Abstract-based Programme 223

Results

or Findings: Observer 1 detected calcifications in 24 samples an d Observer 2 in 23 samples using µCT (90% concordance , κ=0.706). EID-CT with Hv59 showed the highest interobserver agreemen t (97% concordance, κ=0.911), but low detection rates (observer 1: 27%, observer 2: 25%) compared to µCT. PCD-CT yielded better detection wi th Hv48 (observer 1: 70% detection rate, 90% concordance, κ=0.706) and Hv56 (observer 2: 80% detection rate, 77% concordance, κ=0.314). Mass scores indicated the highest detection with PCD-CT Hv64, though with increased n oise. The Hv48/Hv56 kernels were deemed optimal, yielding sensitivity, and specificity (for observers 1/2) of 83%/92%, 50%/83%, respectively, and accurac y of 77%/90%.

Conclusion

PCD-CT outperformed EID-CT in detecting intracrania l calcifications. Nevertheless, small calcifications sometimes go undetected compared to µCT. Hv48 and Hv56 kernels are recommen ded for optimal results, balancing detection rates and noise.

Limitations

In the observer study, detection was assessed on a sample basis, leading to potential oversight of smaller calcifica tions in segments where larger ones were present, which is reflected in the mass s cores. Funding for this study: This study has received funding by Smart*Light is partially funded by the Interreg V Flanders-Netherl ands program with financial support from the European Regional Development Fund (ERDF). Ethics committee - additional information: All brain sources were approved by the IRB at their respective institutions. The In stitutional Review Board waived written informed consent: Author Disclosures: Ricardo P. J. Budde: Speaker: Siemens Healthineers Advisory Board: Bayer Marcel Van Straten: Nothing to disclose Daniel Bos: Nothing to disclose Bernhard P. Berghout: Nothing to disclose Jose Gutierrez: Nothing to disclose Judith Van Der Bie: Nothing to disclose Cerebral blood flow alterations and host genetic as sociation in individuals with long COVID: A transcriptomic-neuro imaging study *Y. Wang*¹, F. Zhou²; ¹Nanchang/CN, ²NanChang City, Jiangxi Province/CN ([email protected]) Purpose or Learning Objective: Neuroimaging studies have indicated that altered cerebral blood flow (CBF) was associated wi th the long-term symptoms of long COVID. long COVID were found to be strongly associated with host gene expression. Nevertheless, the relationships be tween altered CBF, clinical symptoms, and gene expression in the central nervou s system (CNS) remain unclear in individuals with long COVID

Methods

or Background: First, CBF pattern was computed from arterial spin labeling sequence in long COVID. Next, using CNS ge ne expression data from the AHBA transcriptomic dataset, we conducted the s patial correlation between CBF and gene expression to defined the CBF-related genes. Functional enrichment analyses were applied to understand the biological functions of CBF-related genes. The cell type-specific expressio n analyses is utilized to identify the CNS cell types most closely associated with long COVID-19

Results

or Findings: Lower CBF in left frontal-temporal gyrus was associ ated with higher fatigue and worse cognition in long COV ID. This CBF pattern was spatially associated with the expression of 2,178 g enes, which significantly overlap with the genes reported to interact with SA RS CoV-2 proteins (odds ratio= 1.60, P= 0.0036). Functional enrichment anal yses indicated these 2,178 genes were enriched in the molecular functions and biological pathways of COVID-19. Additionally, these genes were strongly a ssociated with the oligodendrocyte progenitor cells, astrocytes, and m yelinating oligodendrocytes of the cortex. ALL above results were corrected for multiple comparisons

Conclusion

Lower CBF is associated with persistent clinical sy mptoms in long COVID individuals, possibly as a consequence o f the complex interactions among multiple COVID-19-related genes, which contri butes to our understanding of the impact of adverse CNS outcomes and the trajectory of development to long COVID.

Limitations

The heterogeneity of clinical symptoms at the time of scanning may affect gene expression in individuals with long COVID. Funding for this study: The author(s) disclosed receipt of the following financial sup port for the research, authorship, an d/or publication of this article: This work was supported by the COVID-19 Research Pr oject of the leading medical discipline in Jiangxi Province, Jiangxi Pro vince Double Thousand Talent Plan (jxsq2023201039), Clinical Research Cen ter for Medical Imaging In Jiangxi Province (20223BCG74001) and Jiangxi Pro vince Key Laboratory for Precision Pathology and Intelligent Diagnosis ( 2024SSY06281). Ethics committee - additional information: The present study was approved by the Ethics Committee of the First Affiliated Hos pital of Nanchang University (IIT2023018) Author Disclosures: Yao Wang: Nothing to disclose Fuqing Zhou: Nothing to disclose 12:30-13:30 Research Stage 3 Research Presentation Session: Paediatric RPS 1912 Imaging the growing brain Moderator C. Carducci; Rome/IT ([email protected]) Children's brain development is linked to the traje ctory of epigenetic- based inflammatory scores J. Chuah, A. M. A. Manahan, S. Y. Chan, H. Pei, M. Fortier, M. Meaney, *A. P. Tan*; Singapore/SG ([email protected]) Purpose or Learning Objective: Dysregulation of immune activation has been consistently shown in patients with mental hea lth disorders. In this study, we mapped the trajectories of DNA-methylation based inflammation scores and examined how these trajectories relate to expos ure to maternal depression, subsequent cognitive outcomes, and brai n development at multiple levels.

Methods

or Background: Inflammation scores were calculated for 293 children from DNA methylation data based on epigeno me-wide association studies of serum C-reactive proteins at ages 9 and 48 months. We stratify these children into quartiles based on their inflam mation scores at baseline and map the trajectories of inflammation scores for eac h quartile. Next, we examined if children with different inflammation sc ore trajectories have different exposure to maternal depression, executive function performance, and brain changes evaluated using multimodal MRI at ages 4.5, 6.0, and 7.0 years.

Results

or Findings: We observed a decreasing trend of our DNA-methylati on based inflammation scores, primarily driven by chil dren with higher levels of inflammation scores at baseline. Children with lowe r levels of inflammation scores at 9M and a slower decrease in inflammation scores between 9M and 48M were exposed to higher levels of maternal depre ssive symptoms and showed poorer executive function performance at age s 4.5 and 7. Children with different inflammation score trajectories exhi bit significantly different brain structure and function, involving predominantly bra in regions involved in executive function performance, emotion, and reward processing.

Conclusion

Children with lower baseline inflammation scores an d slower rate of decrease across childhood are exposed to higher levels of maternal depression, possibly related to a blunted immune re sponse from chronic stress exposure. This has a significant downstream impact on cognitive and brain development at multiple levels.

Limitations

Evaluation of executive function performance with q uestionnaires Funding for this study: This research was supported by grants NMRC/TCR/004-NUS/2008 and NMRC/TCR/012-NUHS/2014 fr om the Singapore National Research Foundation (NRF) under the Translational and Clinical Research Flagship and grant OFLCG/MOH-0005 04 from the Open Fund Large Collaborative Grant Programmes and admin istered by the Singapore Ministry of Health’s National Medical Res earch Council (NMRC), Singapore. In RIE2025, GUSTO is supported by fundin g from the NRF’s Human Health and Potential (HHP) Domain, under the Human Potential Programme. Additional funding was provided by the S ingapore Institute for Clinical Sciences, Agency for Science Technology an d Research (A*STAR), Singapore. MJM is supported by funding from the Hop e for Depression Research Foundation, USA, the Toxic Stress Network of the JPB Foundation, USA, the Jacobs Foundation, Switzerland, and the NR F and A*STAR’s Human Potential Programme (H22P0M0001), Singapore. SYC is supported by funding from the NMRC Open Fund – Young Individual Research Grant (MOH-001149- 00). EHT is supported by the NMRC Clinician-Scienti st Award (CSA) (MOH- 001415). APT is supported by funding from the NMRC Transition Award (MOH- 001273-00) and A*STAR (Brain-Body Initiative, iGran ts call ID #21718). Ethics committee - additional information: The study was approved by the National Healthcare Group Domain Specific Review Bo ard (D/2009/021 and B/2014/00411) and the SingHealth Centralized Instit utional Review Board (D/2018/2767 and A/2019/2406). All investigations w ere conducted according to the principles expressed in the Declaration of H elsinki. Written consent was obtained from all guardians on behalf of the enroll ed children. Author Disclosures: Ai Peng Tan: Nothing to disclose Marielle Fortier: Nothing to disclose Michael Meaney: Nothing to disclose Shi Yu Chan: Nothing to disclose Jasmine Chuah: Nothing to disclose Huang Pei: Nothing to disclose Aisleen M. A. Manahan: Nothing to disclose Saturday Abstract-based Programme 224 Geometric microstructural changes of white matter i n infants with periventricular white matter injury and spastic cer ebral palsy *M. Wang*¹, H. Zhu², T. Huang³, J. Cheng², H. Jiang ¹; ¹Wuxi/CN, ²Beijing/CN, ³Zhengzhou/CN ([email protected]) Purpose or Learning Objective: To investigate the geometric microstructural changes in white matter in infants with periventric ular white matter injury and spastic cerebral palsy (PWMI-SCP) and facilitate ea rly prediction.

Methods

or Background: PWMI is a high-risk factor for SCP. However, early identification of PWMI-SCP infants remains challeng ing. A novel mathematical framework, “Director Field Analysis” (DFA), reflect s changes in the microstructural geometry of white matter and offers new insights into the pathological mechanisms of PWMI-SCP. DFA provides s pecific quantitative metrics, including splay, twist, bend, and total di stortion index. Analysis of variance, correlation analysis, and receiver operat ing characteristics analysis were performed. Corrected p-values < 0.05 were cons idered significant.

Results

or Findings: The PWMI-SCP group exhibited significantly elevated DFA metrics, primarily in the corpus callosum, post erior thalamic radiata, and corona radiata, comparing to the PWMI without SCP g roup, which were associated with enlarged lateral ventricles, reduce d deep nuclear volumes and motor dysfunction. Mediation analysis indicated tha t increased geometric microstructure in the corpus callosum partially med iates the relationship between the lateral ventricles and motor function. A multi-parameter model based on DFA metrics can effectively predict PWMI-S CP with an AUC of 0.95.

Conclusion

Abnormal increases in white matter geometric micros tructure in the sensorimotor circuit may be one of the neural s ubstrates underlying the manifestation of SCP in PWMI infants. Monitoring fi ber-orientational alterations may provide new insights into early prediction of P WMI-SCP.

Limitations

First, this study focused on infants aged 6−36 mo n ths ; the re fo re , future research should investigate changes in white matter geometric microstructure in infants aged 0-6 months to improv e the early diagnosis rate of PWMI-SCP. Additionally, biomolecular research is required to understand the interactions between brain morphology, tissue m echanics, and white matter geometric microstructure to elucidate the pathophys iological mechanisms of PWMI-SCP. Funding for this study: This study was financially supported by grants from the STI 2030 - Major Project (Grant No. 2022ZD02090 00); the Hong Kong global STEM scholar scheme; the internal fund of th e Hong Kong Polytechnic University; the Wuxi Municipal Health Commission's 'Double Hundred' Medical Health Young Elite Talent Project (Grant No. BJ2023 088) and the Wuxi Municipal Science and Technology Bureau's Medical a nd Health Tackling Project (Grant No. Y20232012); the Jiangsu Province Graduate Research Innovation Project (KYCX24_2647); National Natural Science Foundation of China (Grant No. 61971017, No. 82204933); Open Rese arch Fund of the State Key Laboratory of Cognitive Neuroscience and Learni ng (Grant No. CNLZD2101). Ethics committee - additional information: The study received institutional review board approval, and written informed consent was obtained from all participants. Author Disclosures: Jian Cheng: Nothing to disclose Hua Zhu: Nothing to disclose Tingting Huang: Nothing to disclose Miaoyan Wang: Nothing to disclose Haoxiang Jiang: Nothing to disclose Altered coupling of cerebral blood perfusion and ne uronal activity in children with MRI-negative drug refractory Epilepsy *H. Ran*, T. Zhang; Zunyi/CN ([email protected]) Purpose or Learning Objective: Drug refractory left temporal lobe epilepsy (DRLTLE) often give rise to neuronal activity and c erebral vascular hemodynamics changes, which may result in neurovasc ular decoupling. However, neuroimaging evidence on neurovascular dec oupling remains scarce, this study aimed to assess the manifestatio n of neurovascular coupling (NVC) in childhood DRLTLE using resting-state fMRI (rs-fMRI) and arterial spin labeling imaging (ASL).

Methods

or Background: Based on the collected rs-fMRI and ASL imaging data, degree centrality (DC) and cerebral blood flo w (CBF) were calculated respectively. Across voxel CBF-DC correlations used to evaluate the NVC within whole brain, and NVC of brain region was ass essed by the CBF/DC ratio. We performed correlation analysis to evaluat e the relationship between the variables. Finally, we explored classification problems between DRLTLE and healthy control (HC).

Results

or Findings: Compared HC group, the DRLTLE children with higher across voxel CBF-DC correlations. The brain regions of abnormal CBF, DC, and CBF/DC ratio in predominantly in the default mo de and the executive control network, the abnormally CBF, DC values in s ome brain regions were significantly correlated cognitive function. The cl assification model using CBF/DC ratio as features achieved the 72.8% accurac y, 0.764 area under the curve, 68.5% sensitivity, 87.5% specificity, the cl assification accuracy were higher than the model using CBF or DC feature.

Conclusion

The study reveals the cerebral blood perfusion, neu ronal activity, global and regional NVC alteration in children with MRI-negative DRLTLE non- invasively, associated with lower cognitive perform ance. These findings indicating that NVC-based study can better integrat e information of neuronal activity and cerebral hemodynamics, offering a new insight into the neuropathological mechanisms of DRLTLE, and NVC may help clinical classification for childhood DRLTLE.

Limitations

The sample size was relatively small and the potent ial impact of antiepileptic drugs could not totally eliminated. Funding for this study: This study was supported by National Natural Science Foundation of China (Grant Nos .82171919) a nd Intelligent Medical Imaging Engineering Research Center of Guizhou High er Education Institutions project (Grant No. Qianjiaoji [2023] 0 38) Ethics committee - additional information: The ethic committee of Zunyi Medical University reviewed and granted ethical app roval of this research (Ethical Batch Number: lunshen [2021] 1-080) Author Disclosures: Tijiang Zhang: Nothing to disclose Haifeng Ran: Nothing to disclose Magnetic resonance imaging of children with the use of an incubator P. D. Mika, J. Cydejko, *B. Rowinski*, D. Świętoń, E. Szurowska; Gdańsk/PL ([email protected]) Purpose or Learning Objective: Important limitation of pediatric MRI imaging is requirement of general anesthesia (GA) in the yo ungest group of patients. Imaging in GA is not only stressful for children bu t also extends the time of the procedure. The solution is to use an incubator dedi cated to work in a magnetic field. The MRI dedicated incubator allows a full sp ectrum of MRI imaging. Analysis of examinations with the use of an MRI inc ubator performed at the Department of Radiology of the University Clinical Center in Gdansk.

Methods

or Background: A retrospective evaluation of MRI examinations performed using an incubator carried out in the you ngest group of patients. The study includes an analysis of the tests perform ed in terms of the child's age and the examined area. The paper presents the b enefits of using an MRI incubator. Sequences prepared specifically for test protocols in which the incubator is used are discussed Patients were teste d in the "feed and sleep" protocol.

Results

or Findings: Between 2020 and 2023, a total of 146 MRI examinations were performed. Most of the studies we re brain scans without the administration of a contrast agent. Only in 1 c ase supplementary general sedation was necessary.

Conclusion

1. The study showed a sharp increase in the number of studies using an incubator. 2. The incubator dedicated for MRI examinations allows to limit the amount of anesthesia in the youngest pati ents to 4 kg of body weight. 3. Increase in the cost of effectiveness of MRI exa minations in the youngest group of patients due to the lack of anesthesia cos ts.

Limitations

Movement artifacts Weight over 4kg Implants not allowed for MRI Funding for this study: No Ethics committee - additional information: No Author Disclosures: Paulina Danuta Mika: Nothing to disclose Edyta Szurowska: Nothing to disclose Bartosz Rowinski: Nothing to disclose Dominik Świętoń: Nothing to disclose Joanna Cydejko: Nothing to disclose Brain Perfusion Imaging by Arterial Spin Labelling Predicts Postsurgical Seizure Freedom in Pediatric Focal Lesional Epileps y *A. G. Gennari*, L. Gaito, D. Cserpan, R. Kottke, R . Tuura O’Gorman, G. Ramantani; Zurich/CH ([email protected]) Purpose or Learning Objective: In children with pharmacoresistant focal lesional epilepsy, lesion-associated brain perfusio n changes captured by arterial spin labelling (ASL) are an emerging imagi ng tool improving lesion detection. However, their correlation with postsurg ical seizure outcomes is still unexplored. This study aims to determine whether in cluding ASL-derived perfusion changes in surgical planning is associate d with favorable postsurgical seizure outcomes in children with foca l cortical dysplasia (FCD) or low-grade epilepsy-associated tumors (LEAT).

Methods

or Background: We retrospectively analyzed MRI scans from 18 children (median age at MRI: 4.8 years, IQR: 1.9–11 .5) who underwent surgical resection for pharmacoresistant epilepsy a nd had at least 1 year of post-surgical follow-up. All patients received pres urgical ASL imaging along Saturday Abstract-based Programme 225 with pre- and postsurgical structural MRI. Image po stprocessing, including segmentation and coregistration, was used to qualit atively and quantitatively evaluate the completeness of resection of both the anatomical lesion and the ASL-detected perfusion changes. The DICE similarity index was adopted in quantitative analysis to grade the segmentations’ a lignment. These findings were then correlated with seizure outcomes.

Results

or Findings: Fourteen (78%) patients achieved complete seizure freedom. Qualitative analysis showed that complete resection of the ASL- detected perfusion changes significantly correlated with seizure freedom (p=0.009). Quantitative analysis indicated that hig her degrees of alignment between perfusion and resection cavity segmentation s, as measured by DICE score, were associated with seizure freedom (p=0.04 3), while lesion volume inclusion was not (p=0.44).

Conclusion

Including ASL perfusion imaging in the presurgical evaluation can help better define the epileptogenic zone, improvin g postsurgical seizure outcomes, and supporting it as a complementary tool in surgical planning for pharmacoresistant pediatric focal lesional epilepsy .

Limitations

Limitations: - small sample size; - focus on MRI-vi sible lesions only, limiting the generalizability to MRI-negative patients; - manual segmentation, which may limit the reproducibility o f our results. Funding for this study: We thank the National Science Foundation (SNSF: 208184) (to G.R.), the Anna Mueller Grocholski Foun dation, and the Theodor und Ida Herzog-Egli-Stiftung (to A.G.G.) for fundin g. The funders had no role in the design or analysis of the study. Ethics committee - additional information: The collection and analysis of patient data were approved by and performed accordi ng to the guidelines and regulations of the local ethics committee (KEK-ZH P B-2024-00298). All parents gave written, informed general consent to reuse cli nical data for research. Author Disclosures: Ruth Tuura O’Gorman: Nothing to disclose Antonio Giulio Gennari: Research/Grant Support: The Anna Mueller Grocholski Foundation and the Theodor und Ida Herzog-Egli-Stif tung funded Dr. Gennari post-doc Dorottya Cserpan: Nothing to disclose Georgia Ramantani: Research/Grant Support: Prof Ram antani received a grant from the National Science Foundation (SNSF: 208184) Raimund Kottke: Nothing to disclose Luca Gaito: Nothing to disclose Refining Diagnostic Accuracy in Pediatric Metabolic Brain Disorders: Integrating MRI, Proton Spectroscopy, and Diffusion -Weighted Imaging *R. Agarwal*¹, U. Gupta², N. Jha²; ¹Bengaluru/IN, ² Ghaziabad/IN ([email protected]) Purpose or Learning Objective: This study aims to assess MRI signal abnormalities in pediatric metabolic brain disorder s, focusing on diffusion- weighted imaging (DWI) and proton magnetic resonanc e spectroscopy (MRS). The goals are to identify specific imaging patterns , correlate them with clinical, biochemical, and genetic data, and enhance diagnost ic accuracy for early intervention.

Methods

or Background: Metabolic brain disorders in children are inherited conditions leading to progressive neurodegeneration , where early diagnosis is critical. This study evaluated 30 pediatric patient s (aged 0-12 years) with suspected metabolic brain disorders using MRI, DWI, and proton MRS, along with biochemical and genetic testing. MRI findings were categorized based on the involvement of white matter, grey matter, or bo th.

Results

or Findings: Of the 30 patients, 24 (80%) were diagnosed with metabolic brain disorders, predominantly in males ( 79.17%) and in the 0-3 years age group (62.5%). Common symptoms included r egression of developmental milestones (70.83%) and seizures (58. 33%). MRI showed white matter involvement in 10 cases, grey matter in 5, a nd both in 9. Diagnoses included X-linked adrenoleukodystrophy, metachromat ic leukodystrophy, Leigh disease, Wilson disease, glutaric aciduria type I, and neuronal ceroid lipofuscinosis. MRI findings included symmetrical T 2 hyperintensities in 66.67%, diffusion restriction in 41.67%, and distin ctive MRS peaks such as elevated N-acetylaspartate in Canavan disease and l actate in Leigh disease.

Conclusion

DWI and proton MRS are crucial for early diagnosis of pediatric metabolic brain disorders. Identifying characterist ic imaging patterns enhances diagnostic precision and facilitates timely interve ntion, improving patient outcomes.

Limitations

The study's small sample size may limit the general izability of the findings, and the cross-sectional design does not a ddress disease progression over time. Future research should involve larger co horts and longitudinal studies to validate these results and assess the lo ng-term efficacy of imaging- based diagnostic methods. Funding for this study: None Ethics committee - additional information: Not required Author Disclosures: Ritika Agarwal: Nothing to disclose Narendran Jha: Nothing to disclose Ujjwal Gupta: Nothing to disclose Dynamic network dysfunction in children with idiopa thic generalized epilepsy and its association with cognitive impairm ent and gene expression profiles *H. Ran*, K. Huang, T. Zhang; Zunyi/CN ([email protected]) Purpose or Learning Objective: Idiopathic generalized epilepsy(IGE) has been considered as a network disease, recurrent sei zures may result in network reconfiguration and cognitive impairments. The dynamic changes in functional network in IGE children and the relation ship with cognitive impairment and gene expression profiles needs to be explored.

Methods

or Background: 26 IGE children and 35 healthy controls(HC) were recruited, the modular variability(MV) of construct ed time-varying multi-layer network was calculated and compared between groups based on rs-fMRI. The correlation analysis was performed between MV and c ognitive function scores and clinical variables. Allen Human Brain Atlas wer e used to identify gene sets associated with dynamic network remodeling in IGE. The associated biological processes, pathways were identified by gene enrichm ent tools.

Results

or Findings: Compared to HC, IGE children demonstrated changed MV mainly located in the sensorimotor areas, salien ce/ventral attention, and default mode network, and at the sub-network level, children with IGE exhibited increased MV in the default mode network(p<0.05, FD R). MV changes in the left prefrontal, precuneus cortex were negatively c orrelated with the verbal IQ, full scale IQ and performance IQ scores, respective ly(r=-0.400, -0.419, -0.408; p=0.042, 0.032, 0.038), while MV in the right orbit ofrontal cortex was positively correlated with the verbal IQ and full scale IQ sco res, respectively(r=0.488, 0.442; p=0.011, 0.023). Gene expression profiles wa s associated with dynamic network dysfunction in IGE(r=0.499, pperm  < 0.05). Enrichment analysis indicated that the genes related to dynamic network reorganization were principally enriched in dendrite, axon, and nervous system development.

Conclusion

In IGE children, altered dynamic functional network s has been identified and correlated with cognitive function a nd gene expression, revealing the complex relationship between the dynamic change s of macroscopic modules and genetic pathological mechanisms in IGE patients.

Limitations

The sample size of this study was relative small. Funding for this study: This study was supported by National Natural Science Foundation of China (Grant Nos .82171919) a nd Intelligent Medical Imaging Engineering Research Center of Guizhou High er Education Institutions project (Grant No. Qianjiaoji [2023] 0 38) Ethics committee - additional information: Ethics approval of this research was granted by the Ethic Committee of the Affiliate d Hospital of Zunyi Medical University[KLL-2021-347] Author Disclosures: Tijiang Zhang: Nothing to disclose Kexin Huang: Nothing to disclose Haifeng Ran: Nothing to disclose 12:30-13:30 Research Stage 4 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 1905 Artificial intelligence in abdominal and oncological imaging Moderator G. P. Raval; Rajkot/IN ([email protected]) Total Segmentator: Integration and validation into PACS workstation for abdominal CT scans. A feasibility study *G. Lappas*¹, N. Patlakas¹, P. Giannikopoulos¹, M. Triantafyllou², G. I. Kalaitzakis³, M. Klontzas², K. Petropoulos¹; ¹Athens/GR, ²Crete/GR, ³Heraklion/GR Purpose or Learning Objective: This study aims to validate the performance of the open-source Total Segmentator for the segmen tation of abdominal organs in CT scans and integration into the existin g PACS workstation.

Methods

or Background: The model's segmentation capability was quantified using Dice Similarity Coefficient (DSC) and Normali zed Surface Distance (NSD) across five datasets (N=1243). Data variabili ty including statistics radiomics analysis was performed. Grad-CAM and Mont e-Carlo were deployed focusing on the understanding of model decision-mak ing and robustness, respectively. Οne assistant professor of radiology and one senior radiology resident rated the predicted segmentations’ quality from Greek hospital CT scans (N=100) and integration into the clinical rou tine. Saturday Abstract-based Programme 226

Results

or Findings: The model demonstrated high accuracy in segmenting most organs, e.g., DSC - CI: 0.85-0.97, while showe d lower performance on gallbladder, pancreas and prostate, e.g., DSC – CI: 0.71-0.85. Those results depict the robustness of the model across the five datasets which present high variability depicted by normalized volume ( μ±σ: 0.24±0.18) and normalized intensity (μ±σ: 0.39±0.21) coming in agreement with radiomics fin dings. Grad- CAM and Monte-Carlo noise results provided insights into the model's decision-making process and highlighting areas for potential improvement. The clinical feasibility study indicates a promising to ol generating high quality segmentations with 97% of those requiring minimal m anual adaptations.

Conclusion

Total Segmentator showed robust performance in segm entation of the major abdominal organs with poorer performan ce for gallbladder, pancreas and prostate. The integration into PACS wo rkstation proves the model's potential for routine use in Greek hospital s marking a significant step towards more efficient radiological flow.

Limitations

The model had lower performance for gallbladder, pa ncreas and prostate while 7% of the validated cases had disagr eements larger than 50% volume-wise. The available number of CT scans used for rating by radiologists was limited. Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Michail Klontzas: Nothing to disclose Konstantinos Petropoulos: Board Member: IKnowHealth S.A. Nektarios Patlakas: Nothing to disclose Georgios Ioannis Kalaitzakis: Nothing to disclose Petros Giannikopoulos: Nothing to disclose Georgios Lappas: Nothing to disclose Matthaios Triantafyllou: Nothing to disclose

Results

of the ULS23 Challenge on automatic 3D univ ersal lesion segmentation in computed tomography M. J. J. De Grauw¹, E. Scholten¹, E. J. Smit¹, M. J . Rutten², B. Van Ginneken¹, M. Prokop¹, *A. Hering*¹; ¹Nijmegen/NL, ²'S-Hertoge nbosch/NL ([email protected]) Purpose or Learning Objective: Generalizable automatic segmentation

Methods

are well-suited for application in the dive rse clinical contexts encountered during tumour follow-up in CT. The ULS2 3 challenge establishes the state-of-the-art in automatic 3D lesion segment ation quality, measurement accuracy, and prediction robustness.

Methods

or Background: Current benchmarks often focus on organ-specific lesion segmentation, yet diverse clinical cases req uire fast, generalist models. The ULS23 challenge focuses on universal 3D lesion segmentation across chest-abdomen-pelvis CT, with 38,693 diverse lesion s in the training dataset. The evaluation dataset contains 775 clinically rele vant lesions from 284 patients across two Dutch tertiary care centers. We developed a strong baseline method based on the nnUnet and invited the research community to submit their solutions to the challenge. Post-chall enge, we conducted experiments to explore lesion type influence, uncer tainty, and robustness.

Results

or Findings: The ULS23 challenge encouraged over 50 internationa l researchers to develop solutions for automatic 3D l esion segmentation. During the official challenge period, 153 submissions were recorded during the development phase with seven teams submitting to th e final leaderboard. The U-mamba framework achieved the highest challenge sc ore, excelling in segmentation quality (70.8% ± 23.5% Dice), axial measurement accuracy (Long-axis 10.3%, Short-axis 11.8% Symmetric Mean A bsolute Percentage Error), and robustness when evaluated using repeate d segmentation (79.7% ± 24.2% Dice). Bone, pancreas and colon lesions remai n challenging, and while segmentation consistency improves with performance, models still showed significant variability, affecting measurement accu racy.

Conclusion

The results of the ULS23 challenge demonstrate the potential of 3D universal lesion segmentation using large, aggre gated datasets as a viable alternative to organ-specific models. However, sign ificant variability in segmentation performance, particularly for bone, pa ncreas and colon lesions, indicates the need for further improvements. Addres sing these performance inconsistencies could help reduce variance and enha nce overall clinical applicability.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: N/A Author Disclosures: Mathias Prokop: Patent Holder: Mevis Medical Soluti on Consultant: Canon Medical Systems, Siemens Healthineer Shareholder: T hirona Ernst Scholten: Nothing to disclose Max Jacobus Johannes De Grauw: Nothing to disclose Alessa Hering: Nothing to disclose Bram Van Ginneken: Founder: Plain, Thirona Ewoud J. Smit: Nothing to disclose Matthieu J.C.M. Rutten: Nothing to disclose Automated Matching of Lesions in Cancer Follow-Up U sing 3D Siamese Neural Networks and CT *A. Vergara*, A. Jimenez-Pastor, A. Alberich-Bayarr i; Valencia/ES ([email protected]) Purpose or Learning Objective: There is a lack of automated and reliable tools to automatically track lesion changes over ti me, as well as to streamline treatment response reporting, such as RECIST-1.1 an d others. The primary goal of this work was to address the automatic matc hing of lesions in the thoracoabdominal region across timepoints through 3 D-Siamese Neural Networks (SNN).

Methods

or Background: A retrospective dataset of 253 longitudinal CT exams from metastatic NSCLC patients was used, with a high variability in lesion location, including lymph nodes/lungs/liver/ adrenal glands (43.87%/34.02%/11.82%/6.17%, respectively) and volu me (0.1-1000cm3). A 3D-SNN architecture was used to compare the similar ity between tumor pairs and matching the same lesion across consecutive sca ns. The final model resulted from a two-step training process, initiall y evaluating 144 hyperparameter settings and subsequently retraining using the top-performing ones. Different configurations were created combini ng input information (CT image, CT image and lesion segmentation mask, and C T image with several HU windows), learning rate (LR), LR schedulers, con volutional block complexity (1/2 convolutional layers), and loss fun ctions (contrastive and BCE). An 80/20 training/test split ensured comparable les ion size and location heterogeneity in both sets.

Results

or Findings: The best model employed the CT image as input, a LR of 10e-4, a step LR scheduler, BCE loss, and 1-conv olutional-layer block, delivering 92.80%/91.10%/92.4% in accuracy/precisio n/recall in the test set. The accuracy decomposed by location was 93.79%/88.3 2%/92.50%/94.35% for lymph nodes, lung, liver and adrenal gland, res pectively. Considering lesion volumes, the accuracy was 90.57%/93.35%/94.48% for small (100cm3) ones.

Conclusion

3D SNNs are a promising and reliable technique to a ccurately match tumor lesions over consecutive timepoints, re gardless of lesion size and region.

Limitations

Current work focuses on matching lesions individual ly, not the entire patient, which is future work. Funding for this study: None Ethics committee - additional information: N/A Author Disclosures: Alejandro Vergara: Nothing to disclose Ana Jimenez-Pastor: Nothing to disclose Angel Alberich-Bayarri: Nothing to disclose Comparative Analysis of Language Models for Automat ed Interpretation of Longitudinal TACE Reports in Hepatocellular Carc inoma *E. Can*¹, E. Kotter¹, K. Vogt¹, M. Brönnimann², A. Elkilany³, W. Uller¹, K. Bressem⁴, L. C. Adams⁴; ¹Freiburg/DE, ²Bern/CH, ³Leipzig/DE, ⁴Munich/DE ([email protected]) Purpose or Learning Objective: The increasing complexity of radiology data in hepatocellular carcinoma (HCC) requires innovati ve solutions to ensure consistent and efficient interpretation. This study aims to evaluate the performance of four leading language models (GPT, G emini, Llama, and Llama405b) in extracting and interpreting key clini cal data from longitudinal TACE reports. By automating this process, we seek t o reduce the burden on radiologists, enhance decision-making accuracy, and improve workflow efficiency in interventional radiology.

Methods

or Background: We analyzed the performance of each model on 50 anonymized TACE reports. The models were assessed f or accuracy in extracting clinical data (diagnosis, BCLC staging, mRECIST assessment), as well as vascular involvement and metastases identif ication. A detailed error analysis was conducted to evaluate consistency and precision across tasks. Performance metrics included diagnosis accuracy, li ver segment identification, and error severity.

Results

or Findings: All models demonstrated high accuracy (90-100%) in basic tasks such as diagnosis identification and pr ocedure date extraction. However, in complex tasks like mRECIST assessment a nd lymph node status evaluation, performance varied significantly. Gemin i outperformed in segment identification (4.6/5) and vascular involvement (2. 9/5), while all models struggled with mRECIST accuracy (0-10%). Llama and Llama405b exhibited slightly higher error rates (7.7, 8.0) compared to GPT (6.5) and Gemini (6.6).

Conclusion

While language models show promise in automating TA CE report interpretation, challenges remain in specialized me dical tasks. Gemini showed the best overall performance, but significant impro vements are needed in mRECIST assessment and anatomical reporting. Furthe r research is required to refine these models for clinical application, po tentially reducing radiologists’ workload and enhancing decision-making processes.

Limitations

The models' generalizability across other forms of interventional radiology remains to be validated. Funding for this study: No funding provided. Saturday Abstract-based Programme 227 Ethics committee - additional information: Approved by the ethics committee of the University Medical Center Freiburg . Author Disclosures: Elif Can: Nothing to disclose Lisa C. Adams: Nothing to disclose Aboelyazid Elkilany: Nothing to disclose Katharina Vogt: Nothing to disclose Michael Brönnimann: Nothing to disclose Keno Bressem: Nothing to disclose Wibke Uller: Nothing to disclose Elmar Kotter: Nothing to disclose CT-based Foundation Model Enhanced Preoperative Pre diction of Microvascular Invasion in Hepatocellular Carcinoma: a multicenter study *L. Deng*, W. Xia, J. Xia, W. Dai, F. Yan, R. Li; S hanghai/CN ([email protected]) Purpose or Learning Objective: To develop CT-based foundation model and multi-instance learning (MIL) framework to preopera tively predict microvascular invasion (MVI) in hepatocellular carcinoma (HCC).

Methods

or Background: CT-based foundation models were developed through self-supervised learning using public CT im age datasets. Patients with pathologically proven HCC were included from two ce nters, and the CT image sequences of non-contrast, arterial and portal veno us phase were acquired. The features from slices of HCC tumor region were e xtracted by foundation models, and the features were aggregated by MIL to predict MVI status. The predictions of all sequences were combined to obtai n a final prediction by logistics regression. The performance of proposed m ethod was evaluated by area under the receiver operating characteristic cu rve (AUC).

Results

or Findings: The CT image patches of 36,811 lesions from DeepLesion, LiTS and 3D-IRCADb were used for founda tion model development. A total of 617 HCC patients (median ag e, 69 years; IQR, 52-67 years; men 510) were included and divided into trai ning set (center 1, n=493) and independent test set (center 2, n=124). By usin g a few slices of HCC tumor region (median number, 3; IQR, 2-4), the prop osed method achieved AUCs of 0.80, 0.78, 0.74, and 0.83 for non-contrast , arterial phase, portal venous phase, and all sequences combined, respectiv ely. The proposed

Method

significantly outperformed the method withou t foundation model (AUC=0.72, 0.67, 0.71 and 0.64 for each sequence, P <.05) and the radiomics model (AUC=0.72, P<.05).

Conclusion

By leveraging a few representative CT slices and av oiding the need for full tumor delineation, the foundation mod el-based method achieves improved performance compared to previous methods, demonstrating the crucial role of the foundation model for accurate M VI prediction and facilitating more efficient clinical decision-making.

Limitations

Not applicable. Funding for this study: Not applicable. Ethics committee - additional information: Not applicable. Author Disclosures: Wei Xia: Nothing to disclose Ruokun Li: Nothing to disclose Fuhua Yan: Nothing to disclose Ji Xia: Nothing to disclose Wenwen Dai: Nothing to disclose Lin Deng: Nothing to disclose Artificial intelligence for evaluation of magnetic resonance imaging- detected extramural vascular invasion in rectal can cer *H. Huang*, K. Zhao, Z. Liu, C. Liang; Guangzhou/CN ([email protected]) Purpose or Learning Objective: Extramural vascular invasion (EMVI) is a detectable magnetic resonance imaging (MRI) marker that reflects both tumor invasive and metastatic potential. Since EMVI+ is s een as an independent indicator of worse prognosis, clinicians may pursue more intensive treatment strategies for affected patients. However, EMVI ass essment was influenced by observer experience and subjective factors, limitin g its practical effectiveness. This study aims to develop and validate an interpre table deep learning-based approach for automated mrEMVI identification throug h voxel-level segmentation, providing objective and consistent de tection method.

Methods

or Background: This is a multicenter study that included a total o f 2,501 rectal cancer patients, with 1,830 in the tra ining cohort and 671 in the validation cohorts. Dice similarity score was used to measure segmentation performance, while the inter-reader agreement of mr EMVI was calculated using Cohen’s Kappa (κ). The prognostic value of mrEMVI statuses identifi ed by the artificial intelligence (AI) model was evalu ated by Kaplan–Meier curves and the Cox model.

Results

or Findings: Our model demonstrated excellent performance in identifying mrEMVI, achieving accuracy of 81.54% an d 84.72% in the two validation cohorts. The model demonstrated a high l evel of inter-reader consistency with senior radiologists in identifying mrEMVI status (κ: 0.713– 0.736). AI-mrEMVI+ patients have significantly shor ter 3-year disease-free survival (DFS) and 5-year overall survival (OS) com pared to AI-mrEMVI− patients (DFS: 62.23% vs 84.91%, HR=2.67 95% CI: 1. 95–3.66, P<0.001; OS: 68.71% vs 87.14%, HR=2.64 95% CI: 1.75–3.97, P<0.00 1).

Conclusion

We provide a more objective and consistent approach for the detection of mrEMVI, demonstrating potential in pro gnostic prediction, and offering promising contributions to optimizing the treatment of rectal cancer patients.

Limitations

the model exhibited some false positives, primarily because the model mistakenly identified larger blood vessels in the mesorectal area as mrEMVI. Funding for this study: National Science Foundation for Young Scientists of China (82202267). Ethics committee - additional information: This study has obtained approval from the ethics review committees of all participat ing hospitals. Considering the retrospective design of the study, the requirement for written informed consent from patients was waived. Author Disclosures: Changhong Liang: Author: Administrative support, ma nuscript review Zaiyi Liu: Author: Administrative support, manuscri pt review Haitao Huang: Author: Study design, analyzed the da ta, manuscript Writing Ke Zhao: Author: Study design, analyzed the data PROVIZ Proof-of-Technology: Performance of a machin e learning software for detection of clinically significant pr ostate cancer on biparametric MRI in a prospective clinical study *R. Segre*, M. Sunoqrot, G. Nketiah, P. Davik, S. L angorgen, M. Elschot, T. Frost Bathen; Trondheim/NO ([email protected]) Purpose or Learning Objective: PROVIZ is a machine learning software designed to detect clinically significant prostate cancer (csPCa, defined as GGG > 1) on MRI as a reference for targeted biopsy. The aim of this study is to evaluate feasibility (technical issues in < 10% of processed cases), safety (absence of Serious Adverse Device Effects, SADEs), and performance of PROVIZ.

Methods

or Background: Prospective, proof-of-technology study on 80 consenting men. Inclusion criteria: biopsy-naive me n undergoing MRI for suspicion of prostate cancer. MR images were first delineated according to PI- RADS v. 2.1 by an experienced radiologist. Subseque ntly, automated detection (up to three suspicious areas) was performed using PROVIZ. Delineations of all lesions were used as a reference for targeted b iopsies, providing the ground truth.

Results

or Findings: To date, 73/80 patients have completed the study. Regarding feasibility, one technical issue was expe rienced among the participants. In terms of safety, no SADEs were obs erved. As to performance, PROVIZ scored a high AUROC (92.3%) and can be retro spectively tuned to reach the same patient-level sensitivity of the rad iologist at PI-RADS 3 (94.6%), while gaining an improvement in specificit y (72.2% vs 52.8%). At this operating point, PROVIZ would have referred 7 less patients to biopsy than the radiologist (45/73 vs 52/73). On a lesion-level, PR OVIZ showed a slightly lower sensitivity than the radiologist (42.9% vs 45.5%) b ut fewer false positives per case (30.1% vs 41.1%).

Conclusion

Preliminary results of this prospective study indic ate that PROVIZ is feasible and safe to use, with a performance for detection of csPCa comparable to an experienced radiologist. As a supp ort tool for the radiologist, PROVIZ shows potential for reducing false positive predictions, therefore minimizing unnecessary biopsies.

Limitations

Incomplete data collection. Single-center study. Funding for this study: Norwegian University of Science and Technology (NTNU), Research Council of Norway (Grant Number 29 5013), The Liaison Committee between the Central Norway Regional Healt h Authority and the Norwegian University of Science and Technology (Gra nt Numbers 983005100, 982992100 and 90368401), St. Olavs Hospital - Trond heim University Hospital, Central Norway Regional Health Authority. Ethics committee - additional information: REK (Regionale komiteer for medisinsk og helsefaglig forskningsetikk) approval no. 479272. Author Disclosures: Sverre Langorgen: Nothing to disclose Matthijs Elschot: Nothing to disclose Rebecca Segre: Nothing to disclose Tone Frost Bathen: Nothing to disclose Mohammed Sunoqrot: Nothing to disclose Petter Davik: Nothing to disclose Gabriel Nketiah: Nothing to disclose Saturday Abstract-based Programme 228 Diagnostic performance of a fully automated AI algo rithm for lesion detection and PI-RADS classification in patients wi th suspected prostate cancer *H. Engel*¹, A. Nedelcu¹, R. Grimm², H. Von Busch², A. Sigle³, T. Krauß¹, J. Weiß¹, M. Benndorf⁴, B. Oerther¹; ¹Freiburg im Breisgau/DE, ²Forchheim /DE, ³Freiburg/DE, ⁴Detmold/DE Purpose or Learning Objective: To evaluate the diagnostic performance of a fully automated AI algorithm with lesion detection and PI-RADS classification in a cohort of consecutive patients verified by target ed and extensive systematic biopsies.

Methods

or Background: This retrospective, single-centre study included consecutive patients who underwent 3T multiparametr ic prostate magnetic resonance imaging (MRI) performed between 05/2017 a nd 05/2020, followed by targeted transperineal ultrasound-fusion guided and systematic biopsy. The AI algorithm (syngo.via Prostate MR, VB60S HF01, Si emens Healthineers) was described in previous publications and is based on axial T2- and diffusion- weighted imaging sequences. The results of the AI a lgorithm were compared with those of human readers and the diagnostic perf ormance was determined.

Results

or Findings: The evaluation of 272 patients resulted in 436 targ et lesions. 135 patients (49.5%) had clinically signif icant prostate cancer (csPCa), 35 (12.8%) had clinically insignificant prostate ca ncer (ISUP=1) and 102 (37.5%) were benign. Patient-level cancer detection rates (CDRs) of csPCa for AI versus human reading were 11%/18% for PI-RADS ≤2, 24%/11% for PI- RADS 3, 54%/41% for PI-RADS 4, and 74%/92% for PI-R ADS 5. The accuracy of the AI was significantly better (0.74 versus 0.6 3 at a threshold of PI-RADS ≥4, p <0.01). 62 patients with human reading PI-RADS ≥3 were correctly classified as true negative by AI.

Conclusion

The AI algorithm proved to be a reliable and robust tool for lesion detection and classification. Furthermore, the CDRs and distribution of PI- RADS assessment categories of the AI are consistent with the results of recent meta-analyses, indicating precise risk stratificati on.

Limitations

The limitations of our study are mainly its retrosp ective and monocentric design. Additionally, the study design based on histopathological verification implies an under-representation of neg ative MRI scans and a cohort that is not fully representative of the wide r patient population. Funding for this study: The licence of the AI algorithm was part of an unrestricted collaboration agreement between Siemen s Healthineers and the Department of Radiology, Medical Center - Universit y of Freiburg, Faculty of Medicine, University of Freiburg. While Siemens pro vided technical support, the study conception and design, as well as the ana lysis and interpretation of the data, were conducted independently. August Sigl e received research support within the Berta-Ottenstein-Programme. Othe r than that, the authors declare that no funds, grants, or other support wer e received during the preparation of this manuscript. Ethics committee - additional information: Approval was granted by the Ethics Committee of the University of Freiburg (No. 20-1256). Author Disclosures: Tobias Krauß: Nothing to disclose Benedict Oerther: Nothing to disclose August Sigle: Nothing to disclose Jakob Weiß: Nothing to disclose Robert Grimm: Nothing to disclose Andrea Nedelcu: Nothing to disclose Matthias Benndorf: Nothing to disclose Hannes Engel: Nothing to disclose Heinrich Von Busch: Nothing to disclose 14:00-15:30 Research Stage 1 Research Presentation Session: Cardiac RPS 2003 Cardiac applications of photon-counting CT Moderator F. Catapano; Milan/IT ([email protected]) Ultrahigh-resolution photon-counting detector CT de tects a significantly lower coronary plaque volume than energy-integratin g detector CT *M. Vecsey-Nagy*¹, G. Tremamunno¹, C. Gnasso¹, E. Z sarnóczay², D. Kravchenko², B. Szilveszter², P. Maurovich-Horva t², A. Varga-Szemes¹, T. S. Emrich¹; ¹Charleston, SC/US, ²Budapest/HU ([email protected]) Purpose or Learning Objective: To evaluate the effect of photon-counting detector (PCD)-CT on coronary plaque quantification and characterization on coronary CT angiography (CCTA) series compared to e nergy-integrating detector (EID)-CT.

Methods

or Background: Consecutive patients undergoing clinically indicate d CCTA on EID-CT (192×0.6 mm collimation) were enroll ed for an ultrahigh- spatial-resolution (UHR) PCD-CT scan (120×0.2 mm co llimation) within 30 days. PCD-CT was acquired using equivalent or lower CT dose index and equivalent contrast media volume as the clinical sc an. Total, calcified, fibrotic, and low-attenuation coronary plaque volumes were qu antified and compared between scanners. Intra- and inter-reader reproduci bility was assessed on both systems.

Results

or Findings: A total of 164 plaques from 48 patients were segmen ted on both scans. Total plaque volume was lower on PCD -CT compared to EID- CT (723.5 [interquartile range: 500.6–1184.7] vs. 1 084.7 [IQR: 710.7–1609.8] mm3, p<0.001). UHR-based segmentations produced low er fibrotic plaque volumes than EID-CT-based measurements (325.4 [IQR: 151.7–519.2] vs. 627.7 [IQR: 385.8–795.1] mm3, p<0.001), while low-a ttenuation (72.1 [IQR: 38.6–161.9] vs. 58.1 [IQR: 23.4–102.3] mm3, p=0.052 ) and calcified plaque volumes (IQR: 344.5 [174.3–605.7] vs. 342.1 [IQR: 1 80.4–607.5] mm3, p=0.50) did not differ significantly between PCD-CT and EID-CT. Total, calcified, and fibrotic plaque volumes demonstrated excellent agreement between repeated measurements and between readers f or both PCD-CT and EID-CT (all intraclass correlation coefficients >0. 90). While low-attenuation plaque volume had strong intra- (ICC: 0.84 [95%CI, 0.57–0.94]) and inter- reader (ICC: 0.92 [95%CI, 0.81–0.97]) agreements fo r PCD-CT, EID-CT showed only moderate (ICC: 0.62 [95%CI, 0.11–0.86]) and poor (ICC: 0.47 [95%CI, 0.01–0.79]) intra- and inter-reader reprodu cibility.

Conclusion

Compared to EID-CT, PCD-CT UHR imaging reduces segm ented coronary plaque volume by nearly one-third and impr oves the reproducibility of low-attenuation plaque measurements.

Limitations

Lack of invasive reference. Funding for this study: The study was funded by a research grant from Siemens Healthineers. Ethics committee - additional information: Local Ethics Committee approved the present study. Author Disclosures: Emese Zsarnóczay: Nothing to disclose Pál Maurovich-Horvat: Research/Grant Support: Sieme ns Healthineers Milán Vecsey-Nagy: Nothing to disclose Giuseppe Tremamunno: Nothing to disclose Dmitrij Kravchenko: Nothing to disclose Bálint Szilveszter: Nothing to disclose Tilman Stephan Emrich: Research/Grant Support: Siem ens Healthineers Chiara Gnasso: Nothing to disclose Akos Varga-Szemes: Research/Grant Support: Siemens Healthineers Reducing Variability in Coronary Plaque Characteriz ation with Photon- Counting CT A. Choux¹, S. Sharma², S. Ross², R. Thompson², Z. Y u², *A. Pourmorteza*¹; ¹Atlanta, GA/US, ²Vernon Hills, IL/US ([email protected]) Purpose or Learning Objective: Photon-counting detector CT (PCD-CT) has demonstrated significant radiation dose reduction c apabilities compared to energy-integrating detector CT (EID-CT), particular ly for coronary artery calcium scoring (CACS). Alternatively, the dose sav ings from PCD-CT can be Saturday Abstract-based Programme 229 leveraged to improve the reproducibility of CACS by reconstructing images with thinner slice thicknesses to reduce variabilit y from partial volume effects (PVE). This ex vivo study aims to evaluate and quan tify the effect of PVE on reproducibility of calcified plaque volume measurem ents.

Methods

or Background: Six excised human hearts with varying degrees of calcification were placed inside a chest phantom an d scanned on a CdZnTe- based PCD-CT (120 kVp and 250 mAs). To simulate var iability in the scan and reconstruction ranges, 10 image volumes were recons tructed for each heart using identical parameters, with the only variation being the starting slice location, which was incremented by 1/10 of the slic e thickness for each reconstruction. Images were reconstructed using FBP (soft kernel, 0.3 mm in- plane pixel size) at both 3-mm (recommended for CAC S) and 1.5-mm slice thicknesses. Calcium volume was measured using 130 HU threshold (as per SCCT guidelines). Variability was quantified as the ratio of std deviation and mean of calcium volume measurements for each heart and a paired t-test (alpha=0.05) was used for establishing statistical significance.

Results

or Findings: Variability was found to be 36.2% for conventional 3-mm images, which was significantly higher than the 16. 6% variability for 1.5-mm images (p<0.001).

Conclusion

Slight variations in scan or reconstruction range l eads to significant variability in coronary plaque characte rization. Taking advantage of thinner slices provided by PCD-CT (matched in radia tion dose to thick EID slices) can mitigate this variability drastically.

Limitations

This was a small sample size ex-vivo study. Funding for this study: Sponsored research agreement with Canon Medical Research USA, Inc. Ethics committee - additional information: Ex vivo study, did not require ethics committee approval Author Disclosures: Zhou Yu: Employee: CMRU Steven Ross: Employee: CMRU Richard Thompson: Employee: CMRU Arnaud Choux: Research/Grant Support: Canon Medical Research USA, GE HealthCare Amir Pourmorteza: Grant Recipient: Canon Medical Re search USA, GE HealthCare Shobhit Sharma: Employee: CMRU Quantification of Coronary Plaque Components with P hoton-counting CT: Analyzing Software Consistency and Variability *M. Gruber*, D. Beitzke, C. Loewe, D. Beitzke; Vien na/AT ([email protected]) Purpose or Learning Objective: Coronary computed tomography angiography (CCTA) enables non-invasive quantificat ion of plaque burden and composition. The aim of this study was to evaluate the intra-reader and inter- platform reproducibility of coronary plaque volume and composition measurements from two plaque analysis software solu tions using a first- generation Photon-counting CT (PCCT) system.

Methods

or Background: Twenty plaques from thirteen patients who underwent CCTA with a slice thickness of 0.4 mm wer e analyzed. Plaque quantification was performed using two dedicated so ftware solutions (Software 1: Syngo.via Frontier CT Coronary Plaque Analysis, Siemens Healthineers; Software 2: QAngio CT, Medis Medical Imaging System s). Volumes of coronary lumen, total plaque and plaque components were assessed at two time points with each software using the following attenuation thresholds for plaque classification in Hounsfield units (HU) : ne crotic core: -30 to 30 HU, fibrotic plaque: 31 to 350 HU and calcified plaque: >351 HU.

Results

or Findings: Intraobserver variability, determined by the Pearso n correlation coefficient showed a strong positive co rrelation (r = 0.94, p< 0.001), indicating a high consistency in repeated measures. Furthermore, the paired- samples t-test showed no statistically significant difference between the two time points for Software 1 t(19) = 0.77, p= 0.45 or Software 2 t(19) = -0.81, p= 0.43). A repeated-measures ANOVA revealed no signif icant main effect for the software type, but differences were observed for lu men volume and calcified plaque volume with F(11.78, 33.85) =12.81, p< 0.001 and F(1.29, 24.56) = 13.89, p < 0.001, respectively.

Conclusion

Quantification measurements of plaque with PCCT are feasible and highly reproducible, indicating that software c omparisons should take into account potential differences in measurements of sp ecific components.

Limitations

Limitations of this study are the relatively small sample size, as well as the reliance on a single reader. Funding for this study: Beitzke Daniela as a research radiographer is supported by a research grant from Siemens Healthin eers. Ethics committee - additional information: The study is approved by the local Ethics board. All patients gave written and i nformed consent. Author Disclosures: Manuel Gruber: Nothing to disclose Dietrich Beitzke: Nothing to disclose Christian Loewe: Nothing to disclose Daniela Beitzke: Research/Grant Support: Siemens He althineers Semiquantitative Metrics of Coronary Artery Disease Burden: Intra- Individual Comparison between Ultrahigh-Resolution Photon-Counting Detector CT and Energy-Integrating Detector CT *G. Tremamunno*¹, A. Varga-Szemes², U. J. Schoepf², D. Kravchenko², M. T. Hagar², A. Laghi¹, T. S. Emrich², M. Vecsey-N agy²; ¹Rome/IT, ²Charleston, SC/US ([email protected]) Purpose or Learning Objective: To assess the impact of ultrahigh-resolution (UHR) photon-counting detector (PCD)-CT on the semi quantitative evaluation of coronary artery disease (CAD) compared to energy -integrating detector (EID)-CT.

Methods

or Background: Patients undergoing coronary CT angiography (CCTA) on an EID-CT system were prospectively enrol led for UHR PCD-CT scan within 30 days. Both datasets were visually ev aluated using five established semiquantitative scores: Segment Involv ement Score (SIS), Segment Stenosis Score (SSS), Multivessel Aggregate Stenosis Score (MVAS), CCTA-adapted Leaman score (CT-LeSc), and Co ronary Artery Disease Reporting and Data System (CAD-RADS). Addit ionally, the total number of detected plaques and high-risk features w ere reported (positive remodeling, spotty calcification, low-attenuation, and napkin-ring sign).

Results

or Findings: The cohort comprised 46 patients (37 men, 68.4±6.9 years). When assessing stenosis severity, PCD-CT sh owed lower SSS (3.5 [1.3-5.0] vs 6.5 [3.0-9.8], p<0.001), MVAS (5.5 [4. 0-7.0] vs 7.0 [5.0-9.0], p<0.001), and CT-LeSc (10.4 [8.5-13.9] vs 11.2 [8.8 -15.4], p=0.032). Furthermore, 52% (24/46) of patients were reclassif ied to a lower CAD-RADS category compared to EID-CT. In terms of CAD extent , PCD-CT demonstrated higher SIS (8.0 [6.0-9.0] vs 7.0 [6.0-8.8], p=0.018 ) and plaque count (9.0 [7.0- 13.8] vs 7.0 [7.0-9.8] p<0.001). Positive remodelin g was less frequent in PCD- CT datasets (2.0 [1.0-4.3] vs 1.0 [0.0-3.0], p=0.01 2), with no significant differences in other high-risk features.

Conclusion

The use of UHR PCD-CT results in the detection of l ess severe, but more extensive CAD in the same patient compared to EID-CT. The effect of such CCTA-based differences on individual risk s tratification needs further investigation.

Limitations

Only a limited number of patients were enrolled and the characterization and extent of plaques were not val idated against invasive reference. Funding for this study: This study received funding by Siemens Healthineers Ethics committee - additional information: Name of Institution: Medical University of South Carolina Code/Number: Pro00108359 Date of Approval: 4/13/2021 Author Disclosures: Muhammad Taha Hagar: Nothing to disclose Milán Vecsey-Nagy: Nothing to disclose Uwe Joseph Schoepf: Research/Grant Support: Siemens Healthineers Giuseppe Tremamunno: Nothing to disclose Dmitrij Kravchenko: Nothing to disclose Andrea Laghi: Nothing to disclose Tilman Stephan Emrich: Research/Grant Support: Siem ens Healthineers Akos Varga-Szemes: Research/Grant Support: Siemens Healthineers Cost-effectiveness of ultrahigh-resolution photon-c ounting detector coronary CT angiography for the evaluation of stabl e chest pain *M. Vecsey-Nagy*¹, T. S. Emrich¹, G. Tremamunno¹, D . Kravchenko¹, M. T. Hagar¹, B. Szilveszter², P. Maurovich-Horvat² , A. Varga-Szemes¹, J. A. Decker³; ¹Charleston, SC/US, ²Budapest/HU, ³A ugsburg/DE ([email protected]) Purpose or Learning Objective: To simulate the cost-effectiveness of ultrahigh-resolution (UHR) photon-counting detector (PCD)-CT in stable chest pain patients undergoing coronary CT angiography (C CTA).

Methods

or Background: A decision and simulation model was developed using Monte Carlo simulations with 1,000 bootstrap resamples to estimate the costs associated with PCD-CT in lieu of EID-CT for CCTA and the referral for subsequent testing. The model was constructed using the diagnostic accuracy metrics of 55 coronary lesions of patients who unde rwent CCTA on both CT systems and subsequent invasive coronary angiograph y (ICA). Sensitivity and specificity were defined for each Coronary Artery D isease Reporting and Data System category. The aggregate healthcare expenditu res were derived from the hospital billing system.

Results

or Findings: Assuming a projected cohort of 15,000 patients over the lifetime of the PCD-CT, its implementation resulted in a 18.9% reduction in the number of functional follow-up tests (6330.3 ± 59.5 vs. 5135.7 ± 60.6, p<0.001), a 6.0% reduction in performed ICAs (1,447 .7 ± 36.2 vs. 1,360.2 ± 34.7, p<0.001), and a 9.4% decrease in major proced ure-related complications. Over a 10-year expected life expecta ncy, PCD-CT led to an average cost saving of $794.50 ± 18.50 per patient and an overall cost difference of $11,917,500 ± 4,350,169. Saturday Abstract-based Programme 230

Conclusion

PCD-CT has the potential to reduce the financial bu rden on healthcare systems and procedure-related complicati ons for stable chest pain patients with coronary calcification when compared to EID-CT.

Limitations

Limited number of lesions included. Funding for this study: The study was funded by a research grant from Siemens Healthineers. Ethics committee - additional information: Local Ethics Committee approved the present study. Author Disclosures: Muhammad Taha Hagar: Nothing to disclose Pál Maurovich-Horvat: Research/Grant Support: Sieme ns Healthineers Milán Vecsey-Nagy: Nothing to disclose Giuseppe Tremamunno: Nothing to disclose Dmitrij Kravchenko: Nothing to disclose Bálint Szilveszter: Nothing to disclose Tilman Stephan Emrich: Research/Grant Support: Siem ens Healthineers Josua A. Decker: Nothing to disclose Akos Varga-Szemes: Research/Grant Support: Siemens Healthineers Intra-individual differences in pericoronary fat at tenuation index measurements between photon-counting and energy-int egrating detector computed tomography G. Tremamunno¹, M. Vecsey-Nagy¹, M. T. Hagar¹, U. J . Schoepf¹, J. O'Doherty¹, J. A. Luetkens², A. Varga-Szemes¹, T . S. Emrich¹, *D. Kravchenko*¹; ¹Charleston, SC/US, ²Bonn/DE Purpose or Learning Objective: Pericoronary adipose tissue (PCAT) fat attenuation index (FAI) predicts major adverse card iac events, but is known to be influenced by multiple factors, such as kernel s harpness, slice thickness, and tube potential. The objective of this study was to explore intra-individual differences in PCAT FAI between PCD- and energy-int egrating detector (EID)- CT.

Methods

or Background: Patients were prospectively enrolled for a PCD-CT research scan after EID coronary CT angiography. Re constructions were performed using a Qr36 kernel at 0.6 mm slice thick ness (EID and PCD-down- sampled [DS]) and at 0.2 mm ultra-high resolution ( UHR) for the PCD-CT. Data was processed either with no use of iterative recon struction, using a weighted filter back projection, or set to a strength level of Advanced Modeled Iterative Reconstruction 3 for the EID-CT and Quantum Iterati ve Reconstruction 4 for the PCD-CT. PCAT FAI of the right coronary artery ( RCA), left anterior descending artery, and left circumflex artery (LCX) was measured automatically using established thresholds of -190 to -30 HU at a set distance and radius. Statistical testing was performed using repeated-measures ANOVA and Bonferroni’s multiple comparison tests (p<.003) .

Results

or Findings: 40 patients (mean age 68±8 years, 32 males [80%]) were analyzed. Absolute FAI measurements differed s ignificantly for all vessels between all reconstructions in the ANOVA comparison (all p<.001). The mean FAI when using iterative reconstruction did not dem onstrate significant differences on multiple comparisons (e.g. LCX: EID: -65.7±8.5; PCD-DS: - 66.0±7.4; PCD-UHR: -67.8±7.0 HU, respectively; all p >.05).

Conclusion

Intra-individual absolute PCAT FAI measurements dif fer significantly between EID- and PCD-CT when controll ing for reconstruction kernel and slice thickness. However, the use of ite rative reconstruction minimizes most differences in FAI, enabling inter-s canner comparability.

Limitations

Small population size and not all patients underwen t both examinations at the same tube potential. Funding for this study: In part supported by a research grant from Siemens Healthineers. Ethics committee - additional information: This HIPAA compliant single- center study received approval from the Institution al Review Board of the Medical University of South Carolina. Author Disclosures: Julian Alexander Luetkens: Nothing to disclose Muhammad Taha Hagar: Nothing to disclose Milán Vecsey-Nagy: Nothing to disclose Jim O'Doherty: Employee: Siemens Uwe Joseph Schoepf: Consultant: Keya Medical Consul tant: HeartFlow Consultant: Bracco Consultant: Bayer Grant Recipien t: Siemens Consultant: Elucid Consultant: Guerbert Giuseppe Tremamunno: Nothing to disclose Dmitrij Kravchenko: Speaker: Philips Tilman Stephan Emrich: Speaker: Siemens Akos Varga-Szemes: Consultant: Elucid Grant Recipie nt: Siemens Correlation between hemodynamically significant ste noses and spectral photon-counting CT first-pass myocardial perfusion imaging compared with dual-energy CT in very-high risk patients *G. Fahrni*¹, S. A. Si-Mohamed², R. Wiemker³, D. C. Rotzinger¹, A. Houmeau², C. Prieur², P. C. Douek², S. Boccalini²; ¹Lausanne/ CH, ²Lyon/FR, ³Hamburg/DE Purpose or Learning Objective: To assess the capabilities of first-pass myocardial perfusion imaging (MPI) with Spectral Ph oton Counting CT (SPCCT) Coronary Angiography to detect fractional-f low reserve (FFR) positive coronary artery stenoses, compared with dual-energy CT (DECT) in a very- high cardiovascular risk population.

Methods

or Background: 18 very-high cardiovascular risk patients referred for CCTA after diagnostic invasive coronary angiogr aphy (ICA) were prospectively included (15 men, 3 women). ECG-gated CCTA was performed with DECT (IQon CT and CT7500, Philips Healthcare) and SPCCT (Philips clinical SPCCT prototype, Philips Healthcare), with in 3 days. First-pass perfusion images were reconstructed with high-resol ution (DECT) and ultra- high-resolution (SPCCT) parameters. Myocardial terr itories were visually labelled as normal or showing hypoperfusion. These labels were then compared to hemodynamic positivity on diagnostic IC A, indicated by either by significant stenosis (>70%) or FFR-positive stenosi s. Myocardial attenuation was measured in normal and hypoperfusion territorie s.

Results

or Findings: A total of 54 coronary artery territories were incl uded. At ICA, significant stenosis was found in 15 (26%) art eries (8 LAD, 3 CX, 4 RCA). 11 lesions were significant stenoses, 4 were FFR-po sitive. Seventeen (31%) coronary artery territories were labelled as hypope rfusion for DECT and 18 (33%) for SPCCT. There was a significant difference between normal and hypoperfusion segments both for DECT (mean: 97±43 H U versus 80±35 HU, p<0.001) and SPCCT (mean: 92±86 HU versus 76±74 HU, p<0.001). A sensitivity for hemodynamically significant stenose s detection of 73% and specificity of 69% were found for DECT, versus 60% ad 72% for SPCCT.

Conclusion

In very-high cardiovascular risk population, SPCCT first-pass myocardial perfusion imaging was comparable to DECT in detecting myocardial hypoperfusion in significant or FFR-posi tive coronary artery stenosis territories.

Limitations

Small cohort of patients and predominance of LAD le sions. Funding for this study: This work was supported by the European Union Horizon 2020 grant No. 643694. G.F. is supported by a research grand from the Swiss Society of Radiology (SSR, Luzern, Switze rland) and Lausanne University hospital (CHUV, Lausanne, Switzerland). Ethics committee - additional information: Hospices Civils de Lyon, approval number: 2019-A02945–52, SPEQUA study Author Disclosures: Salim Aymeric Si-Mohamed: Nothing to disclose Cyril Prieur: Nothing to disclose Angele Houmeau: Nothing to disclose Guillaume Fahrni: Nothing to disclose David Christian Rotzinger: Nothing to disclose Rafael Wiemker: Employee: Philips Innovative Techno logies, 22335 Hamburg, Germany Philippe Charles Douek: Nothing to disclose Sara Boccalini: Nothing to disclose Myocardial Extracellular Volume Using Photon-Counti ng Detector CT with Synthetic Hematocrit Derived from Virtual Non- Contrast Images *V. Mergen*, N. Ehrbar, L. J. Moser, R. Manka, H. A lkadhi, M. Eberhard; Zurich/CH ([email protected]) Purpose or Learning Objective: To evaluate the accuracy of myocardial extracellular volume (ECV) calculation using a synt hetic hematocrit derived from virtual non-iodine images (VNI) and virtual no n-contrast images (VNC) with photon-counting detector (PCD) CT.

Methods

or Background: In this retrospective study, 125 consecutive patien ts examined by a PCD coronary CT angiography (CCTA) an d a cardiac late enhancement (LE) scan, and having a recent blood he matocrit were included. In the derivation cohort (first 75 patients), CCTA and LE scans were reconstructed as VNI at 60, 70, and 80keV and as VN C with quantum iterative reconstruction (QIR) strengths 2, 3, and 4. Mean bl ood pool attenuation (BPmean) was correlated to the blood hematocrit. In the validation cohort (next 50 patients), BPmean served to calculate a syntheti c hematocrit. Myocardial ECV was computed using the synthetic hematocrit and compared with the ECV using the blood hematocrit as reference. Saturday Abstract-based Programme 231

Results

or Findings: In the derivation cohort (49 men, mean age 79±8 yea rs, mean BMI 26±5 kg/m2), correlation between BPmean an d blood hematocrit ranged from poor for VNI of CCTA at 80keV, QIR2 (R2 =0.12) to moderate for VNI of LE at 60keV, QIR4; 70keV, QIR3 and 4; and VN C of LE, QIR3 and 4 (all, R2=0.58). In the validation cohort (29 men, a ge 75±14 years, mean BMI 26±5 kg/m2), BPmean from VNC of the LE scan with QIR3 served to calculate the synthetic hematocrit. Median ECV was 26.9% (int erquartile range (IQR), 25.5%,28.8%) using the blood hematocrit and 26.8% ( IQR, 25.4%,29.7%) using the synthetic hematocrit (VNC, QIR3; mean dif ference,-0.2%; limits of agreement,-2.4%,2.0%; p=0.33).

Conclusion

Myocardial ECV calculation with PCD-CT, using a syn thetic hematocrit derived from VNC images, enables accurat e ECV measurements.

Limitations

Limited number of patients and lack of comparison w ith the clinical reference standard cardiac MRI Funding for this study: None Ethics committee - additional information: Kantonale Ethikkommission Zürich Author Disclosures: Victor Mergen: Research/Grant Support: institutiona l grants from Bayer, Canon, Guerbet, and Siemens Matthias Eberhard: Research/Grant Support: institut ional grants from Bayer, Canon, Guerbet, and Siemens Nicolas Ehrbar: Research/Grant Support: institution al grants from Bayer, Canon, Guerbet, and Siemens Lukas Jakob Moser: Research/Grant Support: institut ional grants from Bayer, Canon, Guerbet, and Siemens Robert Manka: Research/Grant Support: institutional grants from Bayer, Canon, Guerbet, and Siemens Hatem Alkadhi: Research/Grant Support: institutiona l grants from Bayer, Canon, Guerbet, and Siemens Quantification of extracellular volume (ECV) with P hoton Counting CT to identify transthyretin-related cardiac amyloidosis A. Clemente, *A. Marcucci*, C. De Gori, M. Occhipin ti, M. Muca, F. Pignatelli, D. Cioni, E. Neri; Pisa/IT Purpose or Learning Objective: Cardiac amyloid deposition causes interstitial expansion, thereby increasing myocardi al extracellular volume (ECV). The need for quantification of amyloid burde n in transthyretin-related cardiac amyloidosis (TTR-CA) is currently met in pa rt through semi-quantitative bone scintigraphy or with measurement of ECV throug h Cardiovascular Magnetic Resonance(CMR). Although Photon-counting C T (PCCT) provides comprehensive spectral data with every acquisition of the heart, to date few studies analysed ECV in cardiac amyloidosis by usin g it. We evaluated the accuracy of extracellular volume (ECV) quantificati on with PCCT in TTR-CA.

Methods

or Background: We prospectively enrolled 15 patients referred to our centre for suspected cardiac amyloidosis and at ypical symptoms who underwent a complete diagnostic work-out including bone scintigraphy and PCCT. Iodine maps were created by using multienergy late scan. Volumetric ROIs of at least 2 cm3 were manually positioned usi ng a 17 segments model of the left ventricle and global ECV was calculated as average of all segments. Then, the study population was divided into TTR-CA cases and non-amyloid heart disease cases (NCA).

Results

or Findings: One patient was excluded due to light chain (AL) amyloidosis. Among the 14 patients included (77 yea rs±5;10 men) 7 were diagnosed with TTR-CA. All TTR-CA cases showed posi tive scintigraphy (Perugini score ≥2), except for one, where the diagnosis was made th rough fat tissue biopsy and consistent cardiac CMR findings. Global ECV was significantly higher in patients with TTR-CA (40.50 ±8.97%) than in NCA patients (27.00±3.03%;P<0.01). The accuracy of myoc ardial global ECV to identify occult TTR-CA was high (AUC=0.87; 95%CI, 0 .65-1.00).

Conclusion

Preliminary results on myocardial tissue characteri zation based on ECV quantification with PCCT iodine maps show th is is a promising method to detect TTR-CA.

Limitations

Preliminary results with small sample. Funding for this study: No funding received for this study. Ethics committee - additional information: No Author Disclosures: Francesca Pignatelli: Nothing to disclose Mariaelena Occhipinti: Nothing to disclose Emanuele Neri: Nothing to disclose Matilda Muca: Nothing to disclose Alessandro Marcucci: Nothing to disclose Alberto Clemente: Investigator: study MYOAMY-CT Carmelo De Gori: Nothing to disclose Dania Cioni: Nothing to disclose Accuracy of Iodine Maps from Photon-Counting Detect or CT for Detecting Myocardial Late Enhancement – A Compariso n to LGE-MRI G. Tremamunno¹, A. Varga-Szemes², D. Kravchenko³, A . Laghi¹, F. Bamberg⁴, M. Vecsey-Nagy⁵, T. S. Emrich⁶, *M. T. Hagar*⁴; ¹Rome/IT, ²Charleston, SC/US, ³Bonn/DE, ⁴Freiburg im Breisgau/DE, ⁵Budapest/HU, ⁶Mainz/DE ([email protected]) Purpose or Learning Objective: To assess the feasibility and determine the diagnostic accuracy of iodine maps from photon-coun ting detector (PCD) CT in detecting and characterizing myocardial late enhanc ement (LE), compared to late gadolinium enhancement (LGE) MRI.

Methods

or Background: This IRB-approved retrospective analysis of a prospective study cohort included subjects who unde rwent cardiac MRI followed by late iodine enhancement (CT-LE) using a PCD-CT system (NAEOTOM Alpha, Siemens Healthineers). CT-LE scans were performed 5 minutes after administering 100 mL of contrast medi a (Ultravist, 370 mg I/mL; Bayer Healthcare) with an ECG-triggered sequential protocol, and full spectral capabilities. Iodine maps were reconstructed at a s ection thickness of 1.0 mm, using a quantitative kernel (Qr40), and iterative r econstruction (QIR level 3). Two blinded and independent radiologists interprete d all images. Diagnostic accuracy was evaluated per-patient and per-segment using LGE-MRI as the standard of reference. Inter-reader agreement was a ssessed with Cohen’s kappa.

Results

or Findings: The final cohort comprised 27 patients (52% female, mean age 52.9 ± 17.2 years). Twelve patients (44%) had positive LGE on MRI (3 ischemic, 9 non-ischemic). Per-patient sensitivi ty was 100% and 91.7%, with specificities of 73.3% and 80.0%, and accuracies of 85.2%, respectively for both readers. Per-segment sensitivity was 74.7% and 66.7%, with specificities of 94.9% and 96.4%, and accuracies of 91.1% and 90. 7%. Cohen’s kappa was 0.70 (patient level) and 0.63 (segment level), resp ectively.

Conclusion

Iodine maps from PCD-CT provide high diagnostic acc uracy for myocardial LE detection with substantial inter-read er agreement. PCD-CT may serve as an alternative to LGE-MRI in selective cas es, such as a contraindication to MRI or anxiety.

Limitations

Our study's limited sample size mandates confirmato ry research. Our study design introduces a selection bias, so fu rther studies are needed to evaluate the diagnostic role across diverse patient populations. Funding for this study: This study was partially supported by an unrestrict ed research grant from Siemens Healthineers. Ethics committee - additional information: The study protocol, compliant with the Health Insurance Portability and Accountab ility Act, received approval from the local institutional review board at the re spective academic medical center, and all subjects provided written informed consent. Author Disclosures: Muhammad Taha Hagar: Speaker: Siemens Healthineers Milán Vecsey-Nagy: Nothing to disclose Giuseppe Tremamunno: Nothing to disclose Fabian Bamberg: Nothing to disclose Dmitrij Kravchenko: Nothing to disclose Andrea Laghi: Nothing to disclose Tilman Stephan Emrich: Nothing to disclose Akos Varga-Szemes: Research/Grant Support: Siemens Healthineers Ultra-low dose (0.4 mSv) coronary computed tomograp hy angiography using photon-counting detector computed tomography *S. Araki*, S. Nakamura, M. Takafuji, Y. Ichikawa, H. Sakuma, K. Kitagawa; Tsu/JP ([email protected]) Purpose or Learning Objective: Photon-counting detector computed tomography (PCD-CT), which allows for the exclusion of electronic noise, shows promise for significant dose reduction in cor onary computed tomography angiography (CCTA). This study aimed to assess the radiation dose and image quality of CCTA using PCD-CT, combin ed with electrocardiogram (ECG)-triggered prospective high- pitch helical scanning and an ultra-low tube potential of 70 kVp and to invest igate the effect of a sharp kernel on image quality and stenosis assessment in such an ultra-low dose CCTA setting.

Methods

or Background: Forty patients (65% male) with stable heart rates and no prior coronary interventions were included. Data on CT dose index volume (CTDIvol) and dose-length product (DLP) were collected, with effective radiation dose estimated using a conversion factor of 0.014. Images were reconstructed using kernels of Bv64 and Bv40 for im age quality and stenosis assessment. Saturday Abstract-based Programme 232

Results

or Findings: The mean CTDIvol, DLP, and effective dose of CCTA were 1.72±0.38 mGy, 29.1±6.8 mGy·cm, and 0.41±0.09 mSv, respectively. Image quality was similar (p=0.75) between the two kernels, with over 95% of segments achieving a rating of good image quality f or both kernels. Additionally, 42% of non-calcified plaques showed a n increased stenosis severity from Bv40 to Bv64 (p<0.001), while 25% of calcified plaques exhibited a decreased severity (p<0.001).

Conclusion

PCD-CT technology with ECG-triggered prospective hi gh-pitch helical scanning and the tube potential of 70kVp ca n provide CCTA with ultra- low radiation exposure (0.4 mSv), offering a safer and more effective method for diagnosing coronary artery disease. The noise r eduction capability of PCD- CT allows the use of a sharp kernel even in this lo w-dose CCTA setting without compromising image quality, potentially improving t he evaluation of coronary artery stenosis.

Limitations

There is no reference standard such as coronary ang iography. Funding for this study: No funding was provided for this study Ethics committee - additional information: Clinical Research Ethics Review Committee of Mie University Hospital (approval No. H2019-207) Author Disclosures: Kakuya Kitagawa: Nothing to disclose Hajime Sakuma: Nothing to disclose Suguru Araki: Nothing to disclose Yasutaka Ichikawa: Nothing to disclose Masafumi Takafuji: Nothing to disclose Satoshi Nakamura: Nothing to disclose Benefits of photon counting CT for the assessment o f native heart valves C. Mayard, S. A. Si-Mohamed, A. Houmeau, L. Boussel , P. C. Douek, *S. Boccalini*; Lyon/FR ([email protected]) Purpose or Learning Objective: To assess the benefits of photon counting CT (PCCT) on image quality of cardiac valves as com pared to conventional CT (conv-CT).

Methods

or Background: Patients were prospectively included to undergo a clinically indicated coronary CT angiography with P CCT and conv-CT within three days. All the components of each valve were s ubjectively scored by two observers with a 4-point scale for sharpness and co nspicuity. The number of nodules and of mitral chordae was noted. The number and the localisation of the calcifications relative to the thickness of the aortic leaflets were assessed. Furthermore, the full width at mid weight (FWMH) of the attenuation profile of a line perpendicular to the commissure between the le ft and right coronary cusps of the aortic valve was calculated.

Results

or Findings: Thirty-three patients were included (88% men; 62±13 years). 33 pairs of aortic and mitral valves and 18 pairs of pulmonary valves could be analysed. Conspicuity of aortic, mitral, a nd pulmonary valvular structures was increased with PCCT except for one c ommissure of the aortic and pulmonary valves (p=0.06 and p=0.07). Sharpness was superior for all structures of the aortic and mitral valves, and for 2/3 edges of the pulmonary valve. More fine structures (nodules, chordae) and calcifications of the aortic and mitral valves were visible with PCCT. The preci se localisation of the calcifications could be assessed with PCCT in most cases while it remained doubtful in many cases with conv-CT (p=0.02). FWMH was lower with PCCT (1.7(IQ=1.1) vs 2.5mm (IQ=1.3); p<0.01). The radiat ion dose was lower with PCCT (567.8 ±67.7 vs 681.8 ±159.6 mGy*cm; p<0.01).

Conclusion

PCCT yielded better objective and subjective image quality of cardiac valves as compared to conv-CT and more comp onents of the valve structures were visible.

Limitations

Small cohort Funding for this study: European grant H2020 Ethics committee - additional information: Approved Author Disclosures: Salim Aymeric Si-Mohamed: Speaker: Philips Angele Houmeau: Nothing to disclose Philippe Charles Douek: Speaker: Philips Loïc Boussel: Speaker: Philips Sara Boccalini: Speaker: Philips Charles Mayard: Nothing to disclose 14:00-15:30 Research Stage 2 Research Presentation Session: Musculoskeletal RPS 2010 Imaging of injuries and instabilities of the peripheral joints Moderator A. J. Shah; Ahmedabad/IN ([email protected]) MRI findings of shoulder injury related to vaccine administration (SIRVA) following COVID-19 vaccination: A cross-sectional s tudy *N. M. I. Obeidat*¹, R. Khasawneh¹, S. Bani Essa¹, M. Alkhatatba¹, A. Abdel Kareem¹, M. Al-Na'Asan¹, Y. Alshgerat¹, M. Aljarrah¹, L. Sawalha²; ¹Irbid/JO, ²Amman/JO ([email protected]) Purpose or Learning Objective: To investigate the MRI findings of patients presenting with SIRVA after COVID-19 vaccination an d to assess the associations between these findings and patient dem ographics, clinical symptoms, and vaccine-related factors.

Methods

or Background: A retrospective cross-sectional study involved patients who reported shoulder disorders following COVID-19 vaccination between 1 May 2021 and 1 May 2022. Data collected i ncluded demographics, clinical symptoms, vaccination details, and MRI fin dings. Statistical analyses assessed associations between MRI findings and pati ent demographics, clinical symptoms, and vaccine-related factors. Chi -square tests and t-tests were utilized, with statistical significance set at p < 0.05 and trends noted at p < 0.1.

Results

or Findings: 35 patients were diagnosed with SIRVA (3 had bilate ral involvement; mean age 53.6 ± 9.0 years; 54.3% females). The majority developed symptoms within 24 hours post-vaccination (88%), most of them resolved within a week (84%). Pain was the predomin ant symptom (51.4%). MRI findings were subacromial bursitis (89.5%), ent hesial erosions (63.2%), tendinopathy (47.4%), rotator cuff tears and change s of adhesive capsulitis (each 36.8%), effusion (23.7%), as well as muscle a nd bone marrow edema (10.5% and 7.9%, respectively). Statistically signi ficant associations were found between tendinopathy and both, older age (p=0 .024) and AstraZeneca vaccine (p=0.033), subacromial bursitis with both f emale gender (p=0.013) and higher BMI (p=0.023), and between changes of adhesi ve capsulitis and Sinopharm vaccine (p=0.029). Non-diabetics, females , and patients with tendinopathy, were more likely to have persistent s ymptoms after 3 years following vaccination.

Conclusion

SIRVA following COVID-19 vaccination results in dif ferent shoulder pathologies apparent on MRI, many of which are strongly related to patient demographics and type of vaccine administer ed. Awareness of SIRVA changes among radiologists is paramount, especially in seasons of vaccination campaigns such as in early winter (flu-vaccines).

Limitations

Retrospective study and small patient sample. Funding for this study: No funding was received for this study. Ethics committee - additional information: Retrospective study. Author Disclosures: Leen Sawalha: Nothing to disclose Mohammad Al-Na'Asan: Nothing to disclose Mohammad Alkhatatba: Nothing to disclose Suhaib Bani Essa: Nothing to disclose Naser Mohammad Issa Obeidat: Nothing to disclose Majed Aljarrah: Nothing to disclose Ruba Khasawneh: Nothing to disclose Ali Abdel Kareem: Nothing to disclose Yahya Alshgerat: Nothing to disclose Low-dose four-dimensional ct in diagnosing wrist in stability *I. Blom*¹, N. Mathijssen², G. Kraan²; ¹Delft/NL, ² Zoetermeer/NL ([email protected]) Purpose or Learning Objective: To assess the ability to lower the radiation dose in four-dimensional computed tomography (4DCT) for assessment of scapholunate (SL) instability without compromising diagnostic quality. Saturday Abstract-based Programme 233

Methods

or Background: Scapholunate (SL) instability can be diagnosed using 4DCT. However, CT comes with increased radiat ion exposure compared to other diagnostic imaging tools. Therefore, six r andomly chosen cadaveric human hand specimens were scanned using an automati c device simulating radial-ulnar deviation. Parameters affecting radiat ion dose—scanning time, scan range, tube current (mAs), and tube voltage (k Vp)—were varied and compared to clinical settings. Effective dose (ED) and image noise were assessed for all performed scans. Image noise was m easured in soft tissue and cortical bone, since cortical bone was used for segmentation. Three medical specialists analyzed image quality and diag nostic value using a 5-point Likert scale.

Results

or Findings: The ED was 0.081 mSv under standard clinical settin gs. Reducing the scan range to 80 mm (including all car pal bones) decreased the ED to 0.038 mSv. Further reduction to 0.004 mSv was achieved by shortening the scanning time and lowering the tube current. Lo wering tube current increased image noise in soft tissue, but reduced n oise in cortical bone. Adjusting scanning time and scan range did not affe ct image noise. Image quality was deemed diagnostically acceptable for lo w dose 4DCT by all observers, even 22 percent of the scans were deemed as excellent image quality.

Conclusion

A low-dose 4DCT protocol for SL instability seems f easible without compromising diagnostic image quality compa red to other imaging tools. Further research is needed to explore low-do se 4DCT for other indications.

Limitations

The limitations of the study are a human cadaveric study and a questionaire based on 1 question. Funding for this study: No fundings were received for this study. Ethics committee - additional information: None. Author Disclosures: Nina Mathijssen: Nothing to disclose Gerald Kraan: Nothing to disclose Ian Blom: Nothing to disclose Enhancing hip replacement assessment: Integrating i terative metal artefact reduction (iMAR) algorithm with cinematic volume rendering technique (cVRT) in photon-counting CT *X. Liu*; Zhengzhou, Henan Province, China/CN ([email protected]) Purpose or Learning Objective: To explore the potential of combining the iterative metal artifact reduction (iMAR) algorithm with cinematic volume rendering technique (cVRT) in photon-counting CT fo r assessing hip replacements.

Methods

or Background: A retrospective study was conducted on 120 patients who underwent hip arthroplasty exams using the photon-counting CT scanner (NAEOTOM Alpha). Reconstruction of CT image s employing conventional methods, volume rendering (VR), and cV RT, both with and without iMAR. Measurements of CT numbers and standa rd deviations (SDs) in regions of interest (ROIs) were obtained. Objective image quality and subjective scores were assessed using established s cales. Statistical analyses included paired T tests, Mann-Whitney U tests, and Kappa tests.

Results

or Findings: Compared with the non-iMAR group, the iMAR group showed significantly decreased and increased CT num bers in hyperattenuating and hypoattenuating areas, respectively, as well as lowered artifact and image noise (p<.001). Qualitatively, the iMAR group showe d superiority to the non- iMAR group in both image quality and diagnostic con fidence, with scores increases of 2.70 and 2.88 points, respectively (p< 0.05). iMAR combined with cVRT received the highest subjective score (p<0.05) among the four series of post-processing images, followed by iMAR with VR im ages(P<0.05), cVRT and VR images in the non-iMAR group both received the l owest scores.

Conclusion

The iMAR algorithm in photon-counting CT effectivel y reduces artifacts and image noise, enhancing both image qua lity and diagnostic confidence in post-hip metal replacement assessment s. When combined with cVRT, it provides a more intuitive visualization of metal implant stability and the relationship between implants and adjacent tissues.

Limitations

Not applicable. Funding for this study: No funding was received for this study. Ethics committee - additional information: The ethics committee notification can be found under the number 2021-KS-HNSR115 Author Disclosures: Xing Liu: Nothing to disclose Complications of hip prostheses *S. L. Chung*¹, M. S. Sait²; ¹Oxford/UK, ²Kings Uni versity Hospital/UK ([email protected]) Purpose or Learning Objective: To familiarise the different types of hip prostheses. To be able to interpret normal imaging findings post surgery in mainstay modalities such as radiograph, ultrasound, CT and MRI. To understand the limitations of each imaging modaliti es and when to consider further imaging such as SPECT-CT scans or PET/CT. T o be aware of the possible false positive findings such as positive t racer uptake in nuclear scans within 1 year post operation is still within normal physiological limits. To increase awareness the numerous types of complicati ons from most common to least and the time intervals it occurs.

Methods

or Background: Review of current literature of optimisation of imaging modalities to reduce artefact caused by hip prostheses enabling better assessment of the joint.

Results

or Findings: Slice-encoding for metal artefact correction (SEMAC )-is a relatively new MRI sequence particularly STIR and T1-weighted SEMAC sequences help reduce artefacts caused by hip prost hesis best and can help exclude aseptic loosening. Understand the scopes of SPECT-CT, PET/CT and MRI including when to use adjunct imaging to diagno se.

Conclusion

MRI is the best imaging modality for reviewing hip prosthesis but accessibility is dependent on institution. MR SEMAC protocol should be routinely applied to reduce artefacts and increasin g diagnostic assessment.

Limitations

Limited articles on updated nuclear studies/protoco ls. Funding for this study: Nil Ethics committee - additional information: Not Applicable Author Disclosures: Mohammed Saif Sait: Nothing to disclose Siok Li Chung: Nothing to disclose How routine day-one radiographs affect patient mana gement after hip and knee arthroplasty or internal fixation *A. Jonkergouw*, P. Tukker, W. De Monye; Haarlem/NL Purpose or Learning Objective: This study aims to evaluate the clinical consequences of radiographs performed routinely on the first day after arthroplasty or internal fixation of the hip and kn ee.

Methods

or Background: We conducted a retrospective search of our imaging database for day-one post-operative radiogr aphs after total hip arthroplasty, hip hemi-arthroplasty, dynamic hip sc rew fixation, gamma nail fixation, total knee arthroplasty, and unicompartme ntal knee arthroplasty, starting from 1 January 2023 until 500 radiographs for hip surgeries and 500 for knee surgeries were included. For each case, th e radiological report was reviewed to determine if any immediate post-operati ve abnormalities were detected. Additionally, we recorded the type of pro sthesis, the patient’s age and sex, and information on excessive pain from pat ient records.

Results

or Findings: Of the 500 patients who underwent hip surgery, 388 (77.6%) received total hip arthroplasty, 27 (5.4%) hemi-arthroplasty, 34 (6.8%) dynamic hip screw fixation, and 51 (10.2%) gamma na il fixation. Of the 500 patients who underwent knee surgery, 420 (84%) rece ived total knee arthroplasty, and 80 (16%) unicompartmental knee ar throplasty. Across the entire cohort, post-operative abnormalities were no ted in 33 patients (3.3%). Five patients (0.5%) required additional medical ca re due to a fracture, all of whom had undergone hip surgery. Three (0.3%) underw ent revision surgery, and 2 (0.2%) received weight-bearing restrictions. Of the 5 patients with a fracture, 4 (80%) reported significant pain prior t o undergoing radiological examination. Gender did not influence the occurrenc e of post-operative abnormalities.

Conclusion

Given that only 0.5% of all patients required addit ional medical care after detection of post-operative abnormalitie s, the necessity for routine post-operative radiographs appears limited.

Limitations

Based on the available data, no definite predictive factor has been identified in relation to radiological abnorma lities after hip or knee surgery. Funding for this study: None Ethics committee - additional information: The ACLU gave permission for this study. This committee advises the Board of Dir ectors on new scientific studies at Spaarne Gasthuis. Author Disclosures: Anne Jonkergouw: Nothing to disclose Wouter De Monye: Nothing to disclose Paul Tukker: Nothing to disclose Increased Lateral Tibial Plateau Slope (LTPS) and e xtreme Intercondylar Notch Slope (INS): parameters to predict Anterior C ruciate Ligament (ACL) High-grade Injury (HgI) and Mucoid Degenerati on (MD)? *A. Cutaia*, R. Faletti, P. Fonio, E. La Paglia; Tu rin/IT ([email protected]) Purpose or Learning Objective: To assess the reliability of LTPS and INS to predict ACL HgI and MD.

Methods

or Background: 110 patients (Mean Age 46,1years) with MR evidence of pathological ACL were retrospectively s elected. Three groups: HgI (Mean age 31,58years), Low-grade injury (LgI) (Mean age 45,85years) and MD (Mean age 59,33years). LTPS was measured on sagitta l T1-TSE images: Firstly selecting a slice comprehending tibial atta chment of posterior cruciate ligament and intercondylar eminence, tracking the t ibial axis with craniocaudal Saturday Abstract-based Programme 234 circles method. Secondly, in a slice immediately me dial to the head of the fibula, drawing a line perpendicular to the tibial axis. Thirdly, measuring the angle between this line and the one parallel to the posterior tibial plateau. LTPS was reported as increased when greater than 10 °. INS was measured drawing the Blumensaat Line, then evaluating in whi ch section of the tibial plateau it landed: anterior third type 1, middle th ird type 2, posterior third type 3. INS was extreme when classified as group 2 or 3. Statistical significance was measured using Z and T test for quantitative va riables and ChiSquare for qualitative variables.

Results

or Findings: General Group (GG) mean LTPS was 6,87° +-3,09. Patients with an increased LTPS (Mean LTPS 10,69° + -0,73) showed significantly higher prevalence (66% vs.36%) of ACL HgI compared to GG (p<0.005). HgI group’s mean LTPS was 7,83° +-3,02, significantly higher than LgI group (p<0.025). MD group showed significantly higher prevalence (55% vs.27%) of extreme INS compared to HgI (p<0.025).

Conclusion

Increased LTPS should be added in MR reports as it could highlight a predisposition to ACL HgI. INS should b e assessed as it could play a pivotal role in the onset of ACL MD on the basis of chronic impingement.

Limitations

The study is retrospective and monocentric. Funding for this study: None. Ethics committee - additional information: None. Author Disclosures: Aldo Cutaia: Nothing to disclose Riccardo Faletti: Nothing to disclose Ernesto La Paglia: Nothing to disclose Paolo Fonio: Nothing to disclose Is It Necessary To Add Soft Tissue Injury to the Cl assification in Tibial Plateau Fracture Management? M. Tunçez¹, I. Akan¹, F. Seyfettinoğlu², *H. Çetin Tunçez*¹, B. Dirim Mete¹, C. Kazımoğlu¹; ¹Izmir/TR, ²Adana/TR ([email protected]) Purpose or Learning Objective: Current classification systems have generally been developed based on the type of fract ure patterns. The most commonly used is the Schatzker classification syste m, which includes six types of tibial plateau fractures. While this classificat ion evaluates the fracture in two dimensions, three-dimensional classifications were introduced after the spread of CT. To our knowledge, the classification system regarding soft tissue injuries based on MRI findings has not been established yet. Therefore, in this study, we aimed to evaluate the usability of a new modifie d classification by evaluating soft tissue injuries in tibial plateau f ractures.

Methods

or Background: A total of 36 patients with tibial plateau fracture s were included in the study. Patients’ age, gender, and affected sides were recorded. Injuries to the medial meniscus, lateral meniscus, anterior cruciate ligament, posterior cruciate ligament, medial colla teral ligament, and lateral collateral ligament were examined with preoperative magnetic resonance imaging. Soft tissue injuries were arranged accordi ng to the novel modified classification based on the Schatzker classificatio n.

Results

or Findings: The mean age of the study participants was 45 (19-7 6) years; 72% of the patients were men and 28% were wo men. Moreover, 44% and 56% of the patients had broken the right and le ft tibial plateaus, respectively. At least one soft tissue injury was d etected in 29 (81%) patients. In 14 (39%) patients, two or more soft tissue injur ies were observed. All patients were arranged according to the novel modif ied classification regarding ligament and meniscus injuries.

Conclusion

With this novel modified classification system, we think that having better information about the preoperative co ndition of the soft tissue injuries can change the surgical strategy in patien ts with tibial plateau fractures.

Limitations

Patients with minimal displacement and those requir ing conservative treatment were excluded. Funding for this study: No funding was received for this study. Ethics committee - additional information: This retrospective study was approved by the Institutional Review Board of our h ospital. Author Disclosures: Hülya Çetin Tunçez: Author: Data collection Author: STUDY DESIGN Cemal Kazımoğlu: Author: literature search Ihsan Akan: Author: data collection Mahmut Tunçez: Author: Study Design Berna Dirim Mete: Author: data collection Fırat Seyfettinoğlu: Author: literature search Validation of mRUST as an endpoint for fracture hea ling studies in lower- limb fractures D. Deppe¹, *M. Gabriele*¹, E. K. Simşek², A. Ordas-Bayon³, M. Leskovar¹, A. Trepczynski¹, S. Zachow¹, G. Duda¹, M. Heyland¹; ¹Berlin/DE, ²Ankara/TR, ³Madrid/ES ([email protected]) Purpose or Learning Objective: Determining the best treatment for lower limb fractures is challenging due to multiple facto rs affecting bone healing, requiring precise monitoring for optimal care. A ra diographic assessment of the end point of bone healing is difficult, prompting t he use of scores such as the modified Radiographic Union Score for Tibial fractu res (mRUST) that represents one of the most used scores to evaluate radiographic fracture healing. However, the use of mRUST across different locations and treatment options has not yet been thoroughly demonstrated. T he objective of this study is to validate the robustness of mRUST as a reliabl e measure during follow-up in lower limb fractures using various treatment mod alities.

Methods

or Background: Six international investigators (five orthopaedic surgeons and one radiologist) independently assesse d the mRUST in 166 patients with extra-articular lower-limb fractures for different follow-up timepoints. Inter-rater reliability was assessed fo r location (femur/ tibia), treatment option (plate fixation/ nail fixation) an d for different treatment options in different fracture locations using intraclass co rrelation coefficients (ICC).

Results

or Findings: 166 patients (63 femur fractures, 103 tibia fractur es; 32.52% female) with a total of 1136 follow up time points were included. Overall inter-rater reliability for mRUST was good (0.77), regardless of fixation

Method

(0.79, for both nail and plate fixation) or anatomical location (0.78 in tibia fractures, 0.81 in femur fractures). On corte x level, reliability varied for different location within in the bone with highest inter-rater agreement for the medial cortex (0.70-0.74) and lowest for the poster ior cortex (0.65-0.74)

Conclusion

The mRUST-Score proves to be a robust scoring tool for healing monitoring in lower-limb fractures treated with dif ferent fixation methods in different parts of the bone.

Limitations

Images were presented in chronological order, which limited the

Results

of this study. Funding for this study: None. Ethics committee - additional information: Local ethics committee approval was granted for this retrospective study (EA4/099/2 2). Author Disclosures: Georg Duda: Nothing to disclose Alejandro Ordas-Bayon: Nothing to disclose Stefan Zachow: Nothing to disclose Dominik Deppe: Nothing to disclose Mark Heyland: Nothing to disclose Matteo Gabriele: Nothing to disclose Marko Leskovar: Nothing to disclose Ekin Kaya Simşek: Nothing to disclose Adam Trepczynski: Nothing to disclose Correlation of Osteochondral Lesions of the Talar D ome with Tears of the Superior and Inferior Bands of the Anterior Talofib ular Ligament and the Calcaneofibular Ligament: A Retrospective Study *S. Rajan*, J. S. Chatha, H. Mahajan; New Delhi/IN ([email protected]) Purpose or Learning Objective: Osteochondral lesions of the talar dome (OLT) can result from recurrent ankle microinstabil ity, traumatic events, or anatomical abnormalities. This study aims to evalua te the correlation between OLT and single, double, or triple ligament tears wi thin the lateral collateral ligament complex.

Methods

or Background: A retrospective review was conducted on 212 consecutive ankle MRIs performed for pain or instab ility, sourced from the PACS server. Examination under anesthesia and arthr oscopic confirmation were obtained in 23 cases. The data were anonymized and analyzed by two radiologists with extensive experience. The status of the superior and inferior ATFL bands, the calcaneofibular ligament (CFL), and the superficial and deep deltoid ligaments were assessed. Osteochondral lesi ons were classified using the Anderson classification. Statistical significan ce was evaluated using the Chi-square test to assess the relationship between ligament tears and OLT.

Results

or Findings: OLT was identified in 74 patients (36.6%), while 12 6 patients (62.4%) had no OLT. Among those with OLT, 72 had lateral collateral ligament pathology, with only 2 cases showing no de finitive tears. Of the 74 patients with OLT, 26 had tears of the superior ATF L with inferior band scarring, 36 had tears of both ATFL bands, and 19 h ad tears of both ATFL bands and the CFL. Statistical analysis revealed a significant correlation (p < 0.01) between the presence of ligament tears and th e occurrence of OLT. Saturday Abstract-based Programme 235

Conclusion

This study highlights a significant correlation bet ween osteochondral lesions of the talar dome and lateral ligament tears, emphasizing the need for comprehensive assessment o f ligament integrity in patients with OLT.

Limitations

Arthroscopic proof of ligament tears was not presen t in all cases Funding for this study: None Ethics committee - additional information: Retrospective study Author Disclosures: Jagneet Singh Chatha: Nothing to disclose Sriram Rajan: Nothing to disclose Harsh Mahajan: Nothing to disclose Evaluating peroneus tendon motion using dynamic mag netic resonance imaging – a pilot study K. Bokwa, D. G. Mocanu, N. Solidakis, *P. Szaro*; G othenburg/SE ([email protected]) Purpose or Learning Objective: Previous studies indicate that peroneus brevis instability and split tears may be missed in up to half of patients clinically due to unclear clinical signs and on conventional m agnetic resonance (MRI) because of its static nature. We hypothesize that d ynamic imaging may improve the diagnosis of peroneus brevis instabilit y and split tears. However, no studies have evaluated whether dynamic MRI can v isualize peroneus tendon motion. The aim of this study is to assess w hether dynamic MRI can be used to evaluate the motion of the peroneal tendons .

Methods

or Background: Study design: observational pilot study. We performed dynamic MRI using two small flexible coil s in a 3T machine, assessing the axial plane at the lateral malleolus with Dynamic Balanced Fast Field Echo (BFFE). Ten participants received moveme nt training from a radiology nurse before imaging. Two radiologists (r aters) evaluated the image quality. We included only examinations with clear t endon outlines, visible motion without artifacts. Raters assessed the posit ion of peroneus brevis relative to peroneus longus in neutral, plantar, an d dorsal flexion, reaching a consensus. Radiologists measured the distance betwe en the tendons' central points in each position, reporting the mean values.

Results

or Findings: All examinations were included in the analysis. Preliminary analysis revealed that dynamic MRI allo ws visualisation of peroneus tendons motion in good quality. The mean d istance between the tendons was 2.1 mm (SD 0.3 mm) in the neutral posit ion, 4.8 mm (SD 0.3 mm) in dorsal flexion, and 2.0 mm (SD 0.2 mm) in planta r flexion.

Conclusion

Dynamic magnetic resonance allows the evaluation of peroneus tendon motion, offering a novel approach for evalua ting stability that may improve the accuracy of peroneus diagnostics.

Limitations

Small sample size, only dorsal and plantar flexion in the ankle. Funding for this study: The study was founded by Stiftelsen Tornspiran 934: 2023-12-01. Ethics committee - additional information: The Swedish Ethical Review Authority approved the study: 2023-07231-01. Author Disclosures: Dan Gheorghe Mocanu: Nothing to disclose Nektarios Solidakis: Nothing to disclose Katarzyna Bokwa: Nothing to disclose Pawel Szaro: Nothing to disclose Top five MRI findings of professional soccer player s in pre-season medical examinations E. Höhne, I. Yel, A. Gökduman, S. Bernatz, *M. Dimi trova*, C. Booz, T. Vogl, S. Mahmoudi; Frankfurt/DE Purpose or Learning Objective: In professional soccer extensive musculoskeletal assessments are conducted prior to player transfers to evaluate the current state and future risk of injur y. Magnetic resonance imaging (MRI) is essential in this process revealing muscul oskeletal findings even in the absence of symptoms. This analysis presents the fiv e most frequent MRI findings in a cohort of professional soccer players and aims to improve understanding of the physical condition of elite at hletes.

Methods

or Background: This retrospective study included comprehensive musculoskeletal 3T MRI scans obtained during medica l checks of professional soccer players from August 2019 to September 2024. Clinical data were extracted from medical records and supplemented wit h further functional information.

Results

or Findings: MRI scans of 44 professional soccer players were analysed. Among the players, five were left-footed, one was two-footed, and the remaining players were right-footed. The averag e age at the time of examination was 22.8 years (± 4,3). The most common finding was a secondary cleft on the left side, observed in 18 pl ayers (40.9%). A bulging or protruding disc at the L5/S1 level was found in 31. 8% of the cases. Degenerative changes in the labrum were identified in 27.3% of players on the left side and in another 27.3% on the right side. C hondropathy of the left knee was present in 27.3% of cases.

Conclusion

This retrospective analysis revealed several notabl e findings, particularly given the young average age of the ath letes. The main findings included a secondary cleft and labral degeneration, which may be associated with the common occurence of groin pain in soccer p layers. A better understanding of these associations could enhance t he development of more effective diagnostic and preventive strategies for musculoskeletal injuries.

Limitations

The limited sample size restricts the generalizabil ity of findings. Funding for this study: None. Ethics committee - additional information: The local ethics committee has approved this retrospective study. Author Disclosures: Christian Booz: Nothing to disclose Ibrahim Yel: Nothing to disclose Mirela Dimitrova: Nothing to disclose Thomas Vogl: Nothing to disclose Scherwin Mahmoudi: Nothing to disclose Aynur Gökduman: Nothing to disclose Simon Bernatz: Nothing to disclose Elena Höhne: Nothing to disclose 14:00-15:30 Research Stage 3 Research Presentation Session: Hybrid, Molecular and Translational Imaging RPS 2006 Hybrid and molecular imaging in oncology: clinical and translational studies Moderator P. M. Kazmierczak; Munich/DE Ga-68-FAPI PET/CT in Malignant Mesothelioma: Prospe ctive Single- Center Observational Trial *L. Kessler*, B. M. Schaarschmidt, J. Siveke, L. Um utlu, M. Schuler, M. Stuschke, K. Herrmann, W. Fendler, H. Hautzel; E ssen/DE ([email protected]) Purpose or Learning Objective: Mesothelioma are rare tumors mostly affecting the pleura and are associated with overal l poor prognosis. Mesothelioma subtypes have shown to express fibrobl ast-activation-protein (FAP) in tumor cells, suggesting FAP as a promising target for imaging and therapy. Thus, novel radiolabeled FAP-inhibitors (F API) are of interest for future theranostic approaches. The FAPI-PET observa tional trial (NCT04571086) evaluates Ga-68-FAPI PET imaging in c ancer patients and here we present data on Ga-68-FAPI in patients with mesothelioma.

Methods

or Background: Forty-one patients underwent Ga-68 FAPI-PET imaging and F-18-FDG PET. The primary endpoint was correlation of Ga-68- FAPI-PET uptake (SUVmax and SUVpeak) with histopath ological FAP expression. Secondary objectives included detection rate and diagnostic performance (sensitivity, specificity, positive/neg ative predictive values and accuracy) compared to F-18-FDG PET validated by his topathology or a compound reference standard (histopathology, altern ative imaging or follow-up imaging).

Results

or Findings: SUVmax and SUVpeak values showed a significant correlation with histopathological FAP expression ( SUVmax r = 0.49, p = 0.04; SUVpeak r = 0.51, p = 0.03). Overall Ga-68-FAPI sho wed high diagnostic performance (SE 98%, SP 81%, PPV 88% and NPV 97%). Ga-68-FAPI had similiar sensivity compared to F-18-FDG on both per -patient (100.0% vs. 97.3%) and per-region (98.0% vs. 95.9%) basis but s howed increased Specificity (81.1% vs. 36.8%) and positive predicti ve value (87.5% vs. 66.2%) in per-region analysis, indicating superior perform ance. This discrepancy was attributed to a higher number of false positive reg ions on F-18-FDG (FAPI, N = 7 vs. FDG, N = 31).

Conclusion

This is the first study to show correlation between Ga-68-FAPI uptake and histopathological FAP expression and sup erior diagnostic value compared to F-18-FDG in mesothelioma patients. Thes e findings highlight the potential of Ga-68-FAPI as a potential tool in clin ical practice.

Limitations

The limitations are single center, observational co hort with heterogenous patients. Funding for this study: No funding was received for this study. Ethics committee - additional information: University Duisburg-Essen permits 19-8991-BO and 20-9485-BO Saturday Abstract-based Programme 236 Author Disclosures: Ken Herrmann: Consultant: Bayer, SOFIE Biosciences, SIRTEX, Adacap, Curium, Endocyte, BTG, IPSEN, Siemens Healthineers, GE Healthcare, Amgen, Novartis, ymabs, Aktis Oncology, Theragnosti cs, Pharma15, Debiopharm, AstraZeneca, Janssen. Hubertus Hautzel: Consultant: Urenco Consultant: Ro che Wolfgang Fendler: Consultant: SOFIE Biosciences, Ja nssen (consultant, speaker), Calyx (consultant, image review), Bayer ( consultant, speaker, research funding) Martin Stuschke: Nothing to disclose Benedikt Michael Schaarschmidt: Research/Grant Supp ort: DFG Jens Siveke: Consultant: AstraZeneca, Bayer, Boehri nger Ingelheim, Bristol- Myers Squibb, Immunocore, MSD, Novartis, Roche/Gene ntech and Servier Board Member: Pharma15 Research/Grant Support: Abal os Therapeutics, Boehringer Ingelheim, Bristol-Myers Squibb, Celgene , Eisbach Bio, and Roche/Genentech Lukas Kessler: Nothing to disclose Lale Umutlu: Nothing to disclose Martin Schuler: Nothing to disclose Diagnostic accuracy and molecular characterization of endometrial cancer using fully hybrid [18F]FDG PET/MRI *T. Russo*, C. Bezzi, C. Sabini, G. Candotti, G. Ir oni, F. De Cobelli, P. Mapelli, A. Chiti, M. Picchio; Milan/IT ([email protected]) Purpose or Learning Objective: This study aims at evaluating the effectiveness of fully hybrid [18F]FDG PET/MRI in E C staging, assessing its diagnostic accuracy and prognostic role in predicti ng features of EC aggressiveness, including p53abn MMRd for the new m olecular classification.

Methods

or Background: This prospective study involved 80 patients with biopsy-confirmed EC who underwent preoperative [18F ]FDG PET/MRI for staging purposes. The PET/MRI scans were independen tly reviewed by a radiologist and a nuclear medicine physician, asses sing the diagnostic accuracy (ACC), sensitivity (SN), specificity (SP), and positive and negative predictive value (PPV, NPV). Imaging and clinical p arameters were then investigated for their correlation (Spearman's rank correlation) and analyzed through Fisher’s exact test, and ROC analysis. Kapl an-Meier survival curves, Log-rank tests and Cox proportional hazards models were used to evaluate the prognostic value of parameters for predicting tumor relapse.

Results

or Findings: PET/MRI provided ACC=98.75%, SN=98.75%, and PPV=100% for primary tumor detection, and ACC = 92. 31%, SN = 84.62%, SP = 93.85%, PPV = 73.33%, NPV = 96.83% for LN detecti on. PET/MR parameters were able to predict LVSI (AUC= 80.16%), deep MI, p53abn and MMRd (AUC>70%). Less accurate predictions were foun d for EC histotype (AUC=68.04%) and infiltration pattern (AUC=65.19%). Finally, quantitative parameters could also predict both disease relapse (AUC=81.63%), with MTV and Size_CC showing the highest prognostic value, a nd the need to administer post-operative adjuvant therapy (AUC=74.63%).

Conclusion

[18F]FDG PET/MRI show good accuracy in the staging of EC primary tumor and LN metastases. Moreover, PET and MRI-derived parameters have a potential role in the characteriz ation tumor aggressiveness and molecular alterations, as well as tumor recurre nce prediction, crucial information for an optimal patient treatment and ma nagement in clinical practice.

Limitations

Molecular characterization not available for all pa tients. Funding for this study: None Ethics committee - additional information: The study received approval from the Institution’s Ethics Committee (protocol n umber 85/INT/2019) and informed consent was obtained from all patients in accordance with EC guidelines . All procedures were carried out in acc ordance with the Declaration of Helsinki (1964) and its later amendments. Author Disclosures: Maria Picchio: Nothing to disclose Gabriele Ironi: Nothing to disclose Carlotta Sabini: Nothing to disclose Arturo Chiti: Nothing to disclose Giorgio Candotti: Nothing to disclose Carolina Bezzi: Nothing to disclose Tommaso Russo: Nothing to disclose Paola Mapelli: Nothing to disclose Francesco De Cobelli: Nothing to disclose Implmentation of Diffusion Weighted Imaging for who le body staging of lymphoma patients *A. Milosevic*¹, M. Chodyla¹, H. Steinberg¹, L. Kes sler¹, B. M. Schaarschmidt¹, L. Umutlu¹, J. Grueneisen²; ¹Essen/DE, ²Munich/DE ([email protected]) Purpose or Learning Objective: To asses the feasability of Diffusion Weighted Imaging (DWI) in staging of lymphoma patie nts to establish a radiation-free alternative to FDG-PET.

Methods

or Background: A total of 181 lymphoma patients (mean age: 30.9 ± 19.1 years. 75 female and 106 male) undergoing clin ically indicated 18F-FDG PET/MR examinations were retrospectivly assessed. 7 45 target lesions were assessed regarding Tracer-uptake (Standardized Upta ke Values, SUV), diffusion restriction (Apparent Diffusion Coefficie nt, ADC), size and localization. Each of the target lesions was assigned a Deauville score. SUVs and ADC values were then compared using Spearman's rank cor relation test. ROC- analysis was employed in order to find appropriate thresholds to distinguish between vital (score 4-5) and non-vital (score 1-3) manifestations in ADC- measurements,

Results

or Findings: Calculated mean values for the ADCmin and ADCmean of targets with a Deauville score of 4 and 5 were s ignificantly lower when compared to those lesions with a score of 1-3. Acco rdingly, ADCmean displayed a strong inverse correlation with the SUV s (r = -0.83). Furthermore, ROC analysis displayed an AUC of 0.91, 0.98 and 0,8 7 with a sensitivity of 87%, 93%, and 80% for ADCmin, ADCmean and ADCmax, r espectively.

Conclusion

We highly recommend considering DWI an adjunct para meter for staging and restaging of lymphoma patients. DWI can be particularly helpful for individuals suffering from subtypes with low avidit y to FDG and patient groups susceptible to radiation.

Limitations

Lack of proper gold standard for reference tissue i n ADC meassurements. Thus, threshold levels were calculat ed manually. Funding for this study: No funding. Ethics committee - additional information: committee of university of Duisburg-Essen Author Disclosures: Johannes Grueneisen: Nothing to disclose Benedikt Michael Schaarschmidt: Nothing to disclose Lukas Kessler: Nothing to disclose Lale Umutlu: Nothing to disclose Aleksandar Milosevic: Nothing to disclose Michal Chodyla: Nothing to disclose Hannah Steinberg: Nothing to disclose In pursuit of an appropriate use criteria for the u se of 18-F FES PET CT in the management of ER positive breast cancer : work in progress *P. S. Choudhury*, S. Chowdhury, M. Gupta, R. Kumar ; Delhi/IN ([email protected]) Purpose or Learning Objective: Oestrogen receptor (ER) is highly expressed in 70-80% of breast malignancies (BC). ER expressio n or absence plays a central role in its oncogenesis and is a prognostic and predictive biomarker. Molecular imaging with 18-F Fluroestradiol (FES) PE T-CT targets ER and may have higher incremental value in guiding management by increasing specificity.

Methods

or Background: We enrolled 57 female and 1 male breast cancer patient during initial staging and restaging as a p art of an ongoing prospective study and performed 18-F FDG and 18-F FES within 1 week. Whole body FDG and FES PET-CT scan done from base of skull to mid thigh. Image of the breasts performed in prone position by hanging tech nique. The study was approved by scientific committee (Res/SCM/53/2022/6 7) and IRB (RGCIRC/IRB-BHR/112/2022). Lesion detection sensiti vity was compared for a total number of lesions by McNemar test. FES was ta ken as reference in indeterminate lesions. Incremental value was report ed by identifying FES exclusive lesions. Spearman rank test was used to c o-relate ER expression and SUV max.

Results

or Findings: FDG was more sensitive in lesion detection (80.3% v s 61.2% p<0.001) However FES detected more lesions in lobular variety (81.5% vs 56.2% p0.09). Significant co-relation seen betwe en ER+ve and FES uptake. Significant incremental value of FES seen in 27% of patients with indeterminate lesions characterised by FES. Overall change in management noted in 21.1% (5.2% surgical and intent of managem ent 15.8%).

Conclusion

Potential clinical applications of FES PET CT could be to select appropriate patients for hormonal therapies, resolv ing ER status of lesions non-invasively, solving clinical dilemmas when resu lts of other investigations are inconclusive, systemic staging of breast cancer s with low metabolic activity and selecting optimal doses for current or novel ER targeted therapies.

Limitations

Work in progress Funding for this study: Radiopharmaceuticals were procured by the institution and the equipments used belongs to the institution. No other source of funding was used Saturday Abstract-based Programme 237 Ethics committee - additional information: Ethics Committee letter

Reference

IRB (RGCIRC/IRB-BHR/112/2022) Author Disclosures: Rajiv Kumar: Nothing to disclose Manoj Gupta: Nothing to disclose Suchita Chowdhury: Nothing to disclose Partha S Choudhury: Nothing to disclose [1-11C]acetate PET/CT distinguishes aggressive crib riform Gleason score 7 prostate cancer and is mechanistically informed b y spatial metabolomics *N. Sushentsev*¹, G. Hamm¹, R. Manavaki¹, D. Solovi ev², D. Lewis², L. Aloj¹, R. Goodwin¹, F. A. Gallagher¹, T. Barrett¹; ¹Cambri dge/UK, ²Glasgow/UK ([email protected]) Purpose or Learning Objective: We aimed at identifying a clinical metabolic imaging technique to differentiate Gleason score 7 (GS7) prostate tumours with dominant cribriform and non-cribriform Gleason pattern 4 (GP4) based on their comparative metabolic pathway enrichment anal ysis (MPEA).

Methods

or Background: 28 prostate cancer (PCa) patients with n=39 GS7 lesions on prostatectomy were recruited, of which n =27 and n=12 harboured non-cribriform and cribriform GP4, respectively. Th e patients were divided into three sub-cohorts (A, B, and C), each encompassing n=13 GS7 lesions (n=9 non-cribriform; n=4 cribriform). In cohort A, n=39 fresh-frozen tumour samples were used for spatial metabolomics imaging to enabl e comparative MPEA between cribriform and non-cribriform GP4 epitheliu m. In cohort B, formalin- fixed-paraffin-embedded samples were immunohistoche mically stained for fatty acid synthase (FASN) to corroborate the findings fr om cohort A. In cohort C, we determined standardised uptake value (SUVbw) for [1-11C]-acetate PET/CT in cribriform and non-cribriform GS7 lesions as a marker of fatty acid synthesis.

Results

or Findings: In cohort A, MPEA highlighted fatty acid biosynthes is as the most significantly enriched pathway in cribrifo rm GP4 epithelium compared to non-cribriform glands (fold change 4.2; Padj<0.0 001). In cohort B, this aligned with a significantly increased expression o f FASN in cribriform GS7 lesions compared to non-cribriform tumours (P=0.001 ). In cohort C, this corresponded to a significant increase in mean SUVb w of cribriform lesions compared to non-cribriform tumours (P<0.05 for all timepoints up to 60min post-injection). Conversely, the comparison of tumo ur-to-urine 1H-MRI ADCratio derived from the whole cohort showed no di fference between the two GS7 phenotypes (P=0.56).

Conclusion

Clinical imaging of lipid metabolism is a biologica lly informed way of characterising cribriform and non-cribriform GS7 PCa, which is a challenge for 1H-MRI.

Limitations

Modest sample size dictated by study complexity. Funding for this study: Prostate Cancer UK, Cancer Research UK, AstraZeneca Ethics committee - additional information: National Research Ethics Service Committee East of England, Cambridge South, Research Ethics Committee; study numbers: 16/EE/0205, 03/018. Cambr idge University Hospitals Local Ethics Committee (CUH/15/EE/0213), and the Administration of Radioactive Substances Advisory Committee (ARSAC , certificate reference RPC/83/400/33606). Author Disclosures: Luigi Aloj: Nothing to disclose Richard Goodwin: Employee: AstraZeneca Roido Manavaki: Nothing to disclose Nikita Sushentsev: Nothing to disclose David Lewis: Nothing to disclose Gregory Hamm: Employee: AstraZeneca Tristan Barrett: Nothing to disclose Ferdia Aidan Gallagher: Nothing to disclose Dmitry Soloviev: Nothing to disclose The Role of [ 68 Ga]Ga FAPi PET/CT in Staging and R estaging in Breast Cancer with Low FDG Uptake N. Alan Selcuk, *G. Beydagi*, K. Akcay, B. B. Oven, S. Celik, L. Kabasakal; Istanbul/TR ([email protected]) Purpose or Learning Objective: The aim of this study is to assess the potential efficacy of [68Ga]Ga FAPi PET/CT in stagi ng and restaging in breast cancer patients with FDG-negative or low FDG uptake lesions.

Methods

or Background: Between October 2020 and February 2024, 25 female patients with breast cancer were prospective ly enrolled. These patients underwent [68Ga]Ga-FAPi and [18F]-FDG PET/CT imagin g within one week for staging or restaging. The maximum standard upta ke values (SUVmax) of the primary tumor areas and metastases in the [68Ga ]Ga-FAPi and [18F]-FDG PET/CT images were recorded and statistically compa red using the paired t- test.

Results

or Findings: 25 female patients with suspicious primary malignan cy recurrence or metastases but low FDG affinity were imaged with [68Ga]Ga- FAPi PET/CT. The mean age was 57.1±11.7 years. Hist opathologic examination available for 20 patients revealed lobu lar carcinoma in 10 cases, ductal carcinoma in 8 cases, signet ring cell carci noma in one patient and squamous cell carcinoma in one patient. In six pati ents (24%), neither the [18F]-FDG nor the [68Ga]Ga-FAPi PET/CT revealed any findings indicating recurrence or metastasis. Disease stage increased i n 36% (n=9) of patients after [68Ga]Ga-FAPi PET/CT imaging, with 8 of them showing no pathologic findings on [18F]-FDG PET/CT. 60% (n=6) of the lobu lar carcinomas were upstaged after [68Ga]Ga-FAPi PET/CT. The detection of lymph nodes and distant metastases in lobular carcinoma was higher with [68Ga]Ga-FAPi PET/CT than with [18F]-FDG PET/CT. Furthermore, [68 Ga]Ga-FAPi PET/CT showed a higher SUVmax in primary tumor foci and me tastases (p<0.05).

Conclusion

[68Ga]Ga-FAPi PET/CT has been shown to be superior for staging in breast cancer, especially for lobular ca rcinoma with low FDG affinity. It is anticipated that [68Ga]Ga-FAPi PET/CT will be included in future guidelines for staging in breast cancer patients, e specially in patients with lobular carcinoma.

Limitations

None Funding for this study: None Ethics committee - additional information: Ethics committee approval no: 1576 Author Disclosures: Levent Kabasakal: Nothing to disclose Bala Basak Oven: Nothing to disclose Kaan Akcay: Nothing to disclose Gamze Beydagi: Nothing to disclose Serkan Celik: Nothing to disclose Nalan Alan Selcuk: Nothing to disclose Can 18F-FDG-PET CT Predict Mediastinal Lymph Node M etastases in NSCLC Patients Undergoing Neo-Adjuvant Immuno-Chemo therapy? *D. Kifjak*, M. J. Hochmair, A. Korajac, S. Pochepn ia, R-I. Milos, K. Sinn, A. Hoda, H. Prosch, L. Beer; Vienna/AT Purpose or Learning Objective: To evaluate the association between 18F- FDG-PET CT quantitative imaging markers of surgical ly resected mediastinal lymph nodes and histopathologic results in patients with operable non-small cell lung cancer (NSCLC) treated with neo-adjuvant combined immuno- chemotherapy.

Methods

or Background: This preliminary analysis of a prospective, single- center study included 20 patients (8 male, 12 femal e) with NSCLC who were treated with neo-adjuvant combined immune-chemother apy and underwent pre-operative 18F-FDG-PET-CT. We semi-automatically extracted the following parameters for each mediastinal lymph nod e station: metabolic tumor volume (MTV) and total lesion glycolysis (TLG) at p re-operative scans. The histological results of mediastinal lymph nodes wer e retrieved from patients’ records. A Mann-Whitney-U-Test was calculated to co mpare MTV/TLG and histological results for each mediastinal lymph nod e station.

Results

or Findings: A total of 191 mediastinal lymph nodes were surgica lly removed. Four mediastinal lymph node metastases wer e found in three patients. In contrast 17 patients showed no tumor c ells in their mediastinal lymph nodes. The median MTV for positive lymph node s was 4.2 (range: 0-7.4) compared to 0 (range 0-4.3) for negative lymph node s. The median TLG for positive lymph nodes was 6.3 (range 0-42) and for n egative lymph nodes it was 0 (range 0-24). A statistically significant ass ociation between MTV and mediastinal lymph node metastases was observed, p=0 .046. While there was a noticeable trend, no statistically significant asso ciation was identified between TLG and lymph node metastases, p=0.052.

Conclusion

High MTV values and to a lesser degree TLG are asso ciated with residual lymph node metastases. However, both false positive (inflammation) and negative findings (micrometastatis) occur, thus limiting the ability of 18F- FDG-PET CT to predict histopathological response.

Limitations

Single-center study. Selection bias, only patients with histologically confirmed lymph nodes were part of t he study. Funding for this study: The Austrian Federal Ministry for Digital and Economic Affairs, the National Foundation for Resea rch, Technology and Development and the Christian Doppler Research Asso ciation is gratefully acknowledged. Ethics committee - additional information: Local IRB # 1521/2015 Author Disclosures: Katharina Sinn: Nothing to disclose Maximilian Johannes Hochmair: Nothing to disclose Helmut Prosch: Nothing to disclose Lucian Beer: Nothing to disclose Aida Korajac: Nothing to disclose Daria Kifjak: Nothing to disclose Ruxandra-Iulia Milos: Nothing to disclose Alireza Hoda: Nothing to disclose Svitlana Pochepnia: Nothing to disclose Saturday Abstract-based Programme 238 Integrated diagnostics for survival prediction in p atients with GEP-NET undergoing PRRT *F. Herr*, C. A. Dascalescu, M. P. Fabritius, M. Br endel, C. Auernhammer, C. Spitzweg, J. Ricke, M. Heimer, C. C. Cyran; Muni ch/DE ([email protected]) Purpose or Learning Objective: Integrated biomarkers of survival for patients with gastroenteropancreatic neuroendocrine tumors ( GEP-NETs) receiving peptide receptor radionuclide therapy (PRRT) are st ill limited. This study aims to identify predictors of progression-free survival (PFS) in patients with GEP- NETs undergoing PRRT.

Methods

or Background: This single-center retrospective study included 178 patients with GEP-NETs (G1 and G2) who received at least two consecutive cycles of PRRT with [177Lu]Lu-DOTA-TATE and underwe nt somatostatin receptor (SSTR) PET/CT before and after therapy. At baseline, an assessment was conducted in accordance with the Krenning score , and clinical parameters, including chromogranin A (CgA), neuron-specific eno lase (NSE), hemoglobin, Ki-67, erythrocytes, C-reactive protein (CRP) and a lbumin were also collected. PFS was defined by a GEP-NET multidisciplinary team assessment.

Results

or Findings: In univariate analysis at baseline, Krenning score 3, elevated levels of CgA (> 200 ng/dl) and NSE (>25 n g/dl) were significantly (p 5 %, primary tumor in the pancreas, levels of erythrocytes > 4 Mio/ µl, CRP > 1 mg/dl and albumin < 4.1 g/dl at baseline were also significantly (p 2 00 ng/ml, NSE > 35 ng/ml, and Ki-67 index > 5 % at baseline were significantl y (p < 0.05) associated with shorter PFS. Including the Krenning score at baseli ne leads to a significant improvement of the cox regression model (p<0.05). O nly the Ki-67 index (z=2.55) showed a higher z-score than the Krenning score at baseline (z = 2.41).

Conclusion

This study demonstrates the additional prognostic v alue of the Krenning score in conjunction with clinical paramet ers for patients with GEP- NET undergoing PRRT.

Limitations

Limitations of this study are its retrospective sin gle-center design and the lack of multimodality imaging biomarkers. Funding for this study: Wilhelm Vaillant Stiftung Ethics committee - additional information: Ethics Committee of LMU- Munich – Project number: 20-1077 Author Disclosures: Matthias Philipp Fabritius: Nothing to disclose Clemens C. Cyran: Nothing to disclose Christine Spitzweg: Nothing to disclose Maurice Heimer: Nothing to disclose Felix Herr: Nothing to disclose Christoph Auernhammer: Nothing to disclose Matthias Brendel: Nothing to disclose Christian Alexander Dascalescu: Nothing to disclose Jens Ricke: Nothing to disclose Multi-radiotracer PET/CT for the evaluation of caro tid atherosclerotic plaque vulnerability: A systematic review *T. R. Readford*, P. Kench, M. Ugander, S. Patel, N . Giannotti; Sydney/AU ([email protected]) Purpose or Learning Objective: Carotid atherosclerosis is a major contributor to the burden of cerebrovascular diseases. Conventi onal imaging is limited in interrogating the biological and functional charact eristics that may increase plaque vulnerability. Multiple positron emission to mography/computed tomography (PET/CT) radiotracers provide novel diag nostic insights into plaque vulnerability and identify patients at highe r risk of cerebrovascular events. This systematic review investigated the cli nical role of PET radiotracers in identifying vulnerable carotid atherosclerotic p laque.

Methods

or Background: A systematic review of the existing literature was performed using the following search strategy: ‘car otid’, ‘PET’, ‘atherosclerosis’, ‘plaque’ and ‘vulnerability’. On ly original research articles were included. Covidence was used for data screenin g and data extraction.

Results

or Findings: Thirty-nine studies were included that used 18F- fluorodeoxyglucose (18F-FDG), 18F-sodium fluoride ( 18F-NaF), 18F- fluoromisonidazole (18F-MISO), 68Ga-DOTATATE, 68Ga- Pentixafor and 11C- Acetate to target plaque metabolism-related inflamm ation, microcalcification, hypoxia, activated macrophages, C-X-C motif chemoki ne receptor4 and fatty acid synthesis, respectively. Seven studies used du al PET radiotracers with time intervals between scans ranging from one day t o 4.8 months. Correlation between PET imaging and histology post-carotid enda rterectomy was available in 17 studies. Authors noted agreement between macr ophage-driven plaque inflammation by PET/CT and vulnerability-related mo rphological changes by MRI, suggesting complementary roles of combined MRI and PET/CT in detecting vulnerable plaque. Significant variabilit y was observed in reported PET/CT acquisition techniques, injected radiotracer dose [18F-FDG: 185- 925MBq, 18F-NaF: 125-370 MBq, 68Ga-DOTATATE: 148-15 7 MBq] and uptake times [18F-FDG: 50-180min ,18F-NaF: 60-180mi n, 68Ga-DOTATATE: 60-120min].

Conclusion

The use of multiple PET radiotracers may provide ad vanced diagnostic insights into carotid atherosclerotic pl aque vulnerability. Further research is necessary to establish consensus on wha t constitutes a standard approach for the evaluation of vulnerable carotid p laque by PET/CT.

Limitations

Further quantitative analysis was limited by the va riability of imaging parameters used across studies in this revi ew. Funding for this study: No funding was received for this study. Ethics committee - additional information: This was a systematic review of existing literature. Author Disclosures: Peter Kench: Nothing to disclose Martin Ugander: Nothing to disclose Nicola Giannotti: Nothing to disclose Thomas Ramsay Readford: Nothing to disclose Sanjay Patel: Nothing to disclose Lymphoma Therapy Response Assessment with Low-Dose [18F]FDG Total-Body PET/CT *C. Mingels*¹, K. J. Chung¹, H. Nalbant¹, A. Willey ¹, L. K. Shiyam Sundar², Y. G. Abdelhafez¹, R. Badawi¹, B. A. Spencer¹, L. N ardo¹; ¹Sacramento, CA/US, ²Vienna/AT Purpose or Learning Objective: Our aim was to identify the lower limit of injected dose for [18F]FDG Total-Body (TB) PET/CT i n lymphoma therapy response assessment.

Methods

or Background: In this prospective study 24 patients with biopsy- proven lymphoma were enrolled for interim or end-of -treatment TB PET/CT after 1h and 2h of the injection of ~3.0MBq/Kg [18F ]FDG. Lower injected activities (1.0 MBq/kg, 0.5MBq/kg, 0.25MBq/kg, 0.12 5MBq/kg) were simulated. Lesions were segmented by an artificial intelligenc e-aided software and confirmed by an expert. Standardized-uptake values (SUVmax/mean/peak), metabolic tumor volume (MTV) and total-lesion glyco lysis (TLG) were calculated. Additionally, total MTV (TMTV) was asse ssed for each patient. Liver and mediastinal blood-pool were used to calcu late tumor-to-background ratio (TBR) and contrast-to-noise level (CNR). Ther apy response assessment was performed by Deauville criteria.

Results

or Findings: In total, 182 lymphoma lesions were analyzed. SUVmax/mean/peak, MTV, TLG, TBR and TMTV were not s ignificantly different between reference standard and low-dose i mages. Image noise increased significantly with lower doses. CNR decre ased significantly. Clinical therapy response assessment by Deauville Score was significantly lower between 0.125MBq/kg and reference standard (p<0.01) for 1h p.i. imaging. All other low-dose reconstructions revealed no signific ant differences. For 2h p.i. there was a significant difference in Deauville Sco re for 0.5MBq/kg, 0.25MBq/kg and 0.125MBq/kg compared to the referenc e standard (p<0.01). Deauville Scores for 1MBq/kg at 2h were not signifi cantly different to the

Reference

standard (p=0.16).

Conclusion

Dose reduction in therapy response assessment with TB PET/CT is possible to a lower limit of 0.25MBq/kg for 1h p .i. imaging and 1.0MBq/kg for 2h p.i. TB PET/CT. However, lower injected activiti es are at risk to underestimate the metabolic activity of the lymphom a lesions due to higher noise levels. TMTV and TLG were not different in ul tra-low-dose [18F]FDG TB PET/CT.

Limitations

Small cohort, simulated low dose images Funding for this study: The work was also supported by the In Vivo Translational Imaging Shared Resources with funds f rom NCI P30CA093373 and by the Fred and Julia Rusch Foundation for Nucl ear Medicine Research and Education. Hande Nalbant’s funding is partially provided by United Imaging Health’s UIH Fellowship Gift. Ethics committee - additional information: This study was approved by the UC Davis institutional review board (IRB1470016). W ritten informed consent for inclusion was obtained. The study was performed in accordance with the Declaration of Helsinki. Author Disclosures: Hande Nalbant: Nothing to disclose Kevin J. Chung: Nothing to disclose Clemens Mingels: Nothing to disclose Ramsey Badawi: Nothing to disclose Lorenzo Nardo: Nothing to disclose Andrew Willey: Nothing to disclose Lalith Kumar Shiyam Sundar: Nothing to disclose Benjamin A. Spencer: Nothing to disclose Yasser Gaber Abdelhafez: Nothing to disclose Saturday Abstract-based Programme 239 14:00-15:30 Research Stage 4 Research Presentation Session: Abdominal and Gastrointestinal RPS 2001 Shining a spotlight on HCC and liver transplant Moderator W. K. Jeong; Seoul/KR ([email protected]) Author Disclosures: Woo Kyoung Jeong: Consultant: Guerbet; Research Gra nt/Support: GE Healthcare; Speaker: GE Healthcare The predictive value of iodine map histogram analys is of non-tumorous hepatic parenchyma for post-hepatectomy Liver Failu re after narrow- margin hepatectomy in hepatocellular carcinoma *Y. Xu*, J. Liu, J. Zhou; Lanzhou City/CN ([email protected]) Purpose or Learning Objective: Post-hepatectomy liver failure (PHLF) is a severe postoperative complication with a high incid ence and mortality rate, particularly in patients with narrow-margin (NM). T his study aims to predict PHLF in NM-hepatocellular carcinoma (HCC) using iod ine map histogram analysis of non-tumorous hepatic parenchyma.

Methods

or Background: A retrospective analysis was conducted on the clinical and imaging data of 107 patients with NM-H CC who underwent hepatectomy, divided into those with PHLF (n=45) an d without PHLF (n=62). Histogram parameters of non-tumorous hepatic parenc hyma were calculated from iodine map derived from the portal venous phas e of spectral CT, including Min, Max, Mean, SD, Skewness, Kurtosis, Entropy, an d percentiles (V10-V95), along with the future liver remnant volume (FLV) an d standardized future residual liver volume ratio (SFLV%). Logistic regre ssion analyse was used to identify independent predictors of PHLF, and a comp rehensive model was developed. The performance of the new comprehensive model was assessed using ROC curve analysis and was compared with ALBI and MELD scores.

Results

or Findings: Significant intergroup differences were observed in the iodine map histogram analysis of non-tumorous hepat ic parenchyma for Skewness, Kurtosis, V50, V75, V90, V95, FLV, and SF LV% (P < 0.01 to P = 0.04). Multivariate logistic regression analysis re vealed that V95, Kurtosis, and SFLV% were independent risk factors for predicting PHLF. The comprehensive model (ModelALL), developed by combining these inde pendent risk factors, exhibited the highest predictive efficacy for PHLF, with an AUC of 0.77 (0.67- 0.87), outperforming both the ALBI and MELD scores, which had AUCs of 0.70 (0.58-0.81) and 0.62 (0.49-0.74), respectively.

Conclusion

The model which combines the iodine map histogram p arameters of non-tumorous hepatic parenchyma (V95 and Kurtosi s) with SFLV%, aids in the preoperative prediction of PHLF in NM-HCC patie nts and outperforms conventional scoring systems.

Limitations

Not applicable Funding for this study: This study has received funding by grants of Natura l Science Foundation of China (82260361, 82371914), L anzhou University Second Hospital Second Hospital “Cuiying Technology Innovation Plan” (CY2022-QN-A10), Lanzhou University Second Hospital "Cuiying Postgraduate Instructor Cultivation Program Project (CYDSPY20200 3) and Outstanding Young Talents and Backbone Talents Project of Gansu Provincial Health Industry Research Program (GSWSQN2023-04). Ethics committee - additional information: The present study was approved by our hospital ethics committee (No. 2022A-112) an d performed according to the ethical guidelines of the 1975 Declaration of H elsinki. Author Disclosures: Junlin Zhou: Nothing to disclose Yuan Xu: Nothing to disclose Jianli Liu: Nothing to disclose Diagnostic performance of an abbreviated magnetic r esonance protocol for surveillance of hepatocellular carcinoma in can didates for liver transplant *M. Mattone*, F. Quintarelli, A. Napoli, C. Catalan o; Rome/IT ([email protected]) Purpose or Learning Objective: To evaluate the per-patient diagnostic performance of an abbreviated magnetic resonance pr otocol for hepatocellular carcinoma (HCC) surveillance in cirrhotic liver.

Methods

or Background: Retrospective study including cirrhotic patients wi th HCC who are candidates for liver transplant enrolle d in a surveillance program based on the use of MRI with hepatobiliary contrast agent. Two different sets of images for each patient are provided to two radi ologists, who interpret the images independently, to simulate an abbreviated pr otocol. Interobserver variability was assessed using Cohen's kappa coeffi cient. A reference standard based on histologic examination or radiologic crite ria (LIRADS at least equal to 4) was used to determine the diagnostic accuracy of each set of images.

Results

or Findings: A total of 200 patients who underwent MRI for HCC surveillance were included in this study. One set o f images consisted of the complete protocol (Set1) and one set of images cons isted of T2WI with fat saturation, DWI and hepatobiliary sequences (Set2). The sensitivity, specificity, and accuracy of Set of readers 1 and 2 were 91.5%/9 0.5%, 88.6.4%/87.6% and 87.5%/85.0%, respectively. The sensitivities of the sets were not significantly different. Inter-reader agreement was substantial. Ascites, previous surgical and interventional radiology trea tments, and small tumor size are associated with lower sensitivity.

Conclusion

An abbreviated MRI protocol including T2WI with fat saturation, DWI and hepatobiliary sequences is highly sensitive and can be a valid method for HCC surveillance in cirrhotic liver in patients candidates to liver transplant.

Limitations

Low number of readers Funding for this study: No founds were used Ethics committee - additional information: Not applicable Author Disclosures: Monica Mattone: Nothing to disclose Fabio Quintarelli: Nothing to disclose Carlo Catalano: Nothing to disclose Alessandro Napoli: Nothing to disclose Impact of CT-defined sarcopenia on survival in pati ents undergoing orthotopic liver transplant *D. Tore*¹, C. Guarnaccia¹, A. Depaoli², C. Gaetani ¹, M. Anna Pia¹, M. Visciano¹, M. Dini¹, F. Tandoi³, P. Fonio¹; ¹Tur in/IT, ²Ivrea/IT, ³Bari/IT Purpose or Learning Objective: To evaluate the impact of CT-defined sarcopenia on survival in patients undergoing ortho tropic liver transplant (OLT).

Methods

or Background: Monocentric retrospective study. 440 patients (70 females, 370 males) who underwent OLT at our Instit ution within 01.01.2014 and 31.12.2019 with an abdomen CT scan acquired wit hin six month before surgery were selected. Skeletal muscles segmentatio ns at the level of L3 and L4 were performed using open source software 3D Sli cer using thresholds (-29 to 150 HU); psoas muscle area (PMA), skeletal muscl e area (SMA), psoas muscle index (PMI) and skeletal muscle index (SMI) were calculated at both levels. Optimal cut-offs to dichotomize between sar copenic and non- sarcopenic patients divided in females and males we re calculated using Youden's J statistic and ROC curves. Survival analy sis was performed using Kaplan-Meyer's curves.

Results

or Findings: Male patients classified as sarcopenic according to SMI criterion at L3 level presented an increased risk o f mortality compared to non- sarcopenic with an hazard ratio of 1.63 (p=0.03). M ale patients classified as sarcopenic according to SMA criterion at L3 level p resented an increased risk of mortality compared to non-sarcopenic with an haz ard ratio of 1.87 (p=0.008). In the female group none of the sarcopenia definiti ons tested highlighted an increased risk of mortality with p values always >0 .05.

Conclusion

PMI and SMA CT-defined sarcopenia at L3 level repre sented a negative prognostic factor for male patients surviv al after OLT. The use of such definitions of sarcopenia may identify more fragile subjects in this setting and improve care and, possibly outcomes, for such patie nts. The absence of statistically significant differences between sarco penic and non-sarcopenic female subjects may be due to the relatively limite d sample size for such group of patients.

Limitations

Monocentric retrospective study. Funding for this study: No funds were received for this work. Ethics committee - additional information: Not applicable, retrospective study Saturday Abstract-based Programme 240 Author Disclosures: Micheletti Anna Pia: Nothing to disclose Alessandro Depaoli: Nothing to disclose Francesco Tandoi: Nothing to disclose Carla Guarnaccia: Nothing to disclose Massimiliano Dini: Nothing to disclose Davide Tore: Nothing to disclose Paolo Fonio: Nothing to disclose Clara Gaetani: Nothing to disclose Michele Visciano: Nothing to disclose Hepatic functional reserve score based on two-dimen sional shear wave elastography for evaluation of preoperative hepatic functional reserve and prediction of post-hepatectomy liver failure in HCC *X. Zhong*; Guangzhou/CN ([email protected]) Purpose or Learning Objective: To establish a Hepatic Functional Reserve (HFR) score based on liver stiffness (LS) and liver volume, and to explore its feasibility in evaluating preoperative hepatic rese rve and predicting postoperative remnant hepatic function in hepatocel lular carcinoma (HCC). Additionally, the study aimed to determine a safe f uture liver remnant (FLR) volume threshold to prevent post-hepatectomy liver failure (PHLF).

Methods

or Background: The study included 345 HCC patients, with 265 in the training group and 80 in the test group. Preope rative LS was measured using two-dimensional shear-wave elastography. Tota l liver volume (TLV), tumor volume (TuV), and FLR were simulated using th e IQQA system. The median LS of patients without significant fibrosis (7.6kPa) and standard liver volume (SLV) were used as normal references for liv er quality and quantity. The preoperative HFR score (HFR-PRE) was calculated as 7.6/LS×(TLV- TuV)/SLV, and its correlation with liver function i ndicators was assessed. The postoperative HFR score (HFR-POST) was calculated a s 7.6/LS×FLR/SLV, and its effectiveness in predicting PHLF was evalua ted. The optimal FLR threshold was determined using the maximum Youden i ndex.

Results

or Findings: HFR-PRE showed significant correlation with preoperative liver function indicators such as Chil d-Pugh, MELD, ALBI scores, and ICG-R15 (all p<0.05). The AUC for HFR-POST in p redicting PHLF was 0.864 in the training group and 0.857 in the test g roup. HFR-POST outperformed ALBI, MELD, and ICG-R15 scores (all p< 0.05) in predicting PHLF. The minimum FLR/SLV threshold was 5/76*LS to prevent PHLF, with a sensitivity of 81.5% and specificity of 77.3%.

Conclusion

The HFR score effectively evaluates preoperative li ver function and predicts PHLF in HCC patients. It also helps de termine a safe residual liver volume to reduce the risk of PHLF.

Limitations

LS measurement acquired from one liver segment migh t not represent the quality of the whole liver. Funding for this study: None Ethics committee - additional information: The study protocol received approval from the Institutional Review Board of The First Affiliated Hospital of Sun Yat-sen University (IRB approval number: [2019] 046). Author Disclosures: Xian Zhong: Nothing to disclose Comparative Analysis of Semi-Automated CT Volumetry and Intraoperative Graft Weight in Living Donor Liver T ransplantation in a tertiary care centre *A. Garg*, J. Chowdhary, P. K. Sukhani, S. P. Joshi ; Jaipur/IN ([email protected]) Purpose or Learning Objective: To evaluate the accuracy of semi-automated CT volumetry (CTV) in estimating liver volume in po tential living donor liver transplant (LDLT) donors, compared to the gold stan dard of intraoperative graft weight measurement. Aim: 1. Inform transplant surge ons and radiologists about the reliability of semi-automated CTV in LDLT donor evaluation. 2. Highlight the potential benefits of using CTV in pr eoperative planning, improved donor selection and reduced surgical compl ications. 3. Discuss the implications of accurate volume assessment on LDLT outcomes and patient safety.

Methods

or Background: Background: Living Donor Liver Transplantation (LDLT) is a life-saving procedure for patients with end-stage liver disease. Accurate liver volume assessment is crucial in LDLT to ensure sufficient graft volume for recipient survival, prevent small-for-si ze graft syndrome and reduce donor morbidity and mortality. Methods: This retros pective study included 70 potential LDLT donors who underwent: Semi-automated CTV using AW VolumeShare 7 on 128 slice - GE Healthcare, Optima. Inclusion Criteria: Adult LDLT donors (>18 years) Exclusion Criteria: Previou s liver surgery/disease Variant vascular/biliary anatomy, not allowing safe resection

Results

or Findings: Primary Outcome: Strong correlation (r = 0.92, p < 0.001) Sensitivity: 93.2% (95% CI: 85.1-97.5) Speci ficity: 90.5% (95% CI: 81.2- 95.8) Mean difference between CTV and intraoperativ e graft weight: 25.6 ± 57.8 grams Secondary Outcomes: Bland-Altman analysi s showed good agreement between CTV and intraoperative graft weig ht, with 95% limits of agreement (-89.2 to 140.4 grams) CTV accurately pre dicted graft weight within 10% of actual weight in 85% of cases (n = 70)

Conclusion

Semi-automated CTV demonstrates high accuracy and r eliability in estimating liver volume in LDLT donors. CTV accu rately predicts graft weight with good agreement with intraoperative measurement s thus helping in preoperative evaluation and surgical planning for L DLT.

Limitations

Single-center experience, Operator dependence, Intr aoperative graft weight measurement variability Funding for this study: No external funding Ethics committee - additional information: Approved by the Institutional Ethics Committee, Mahatma Gandhi Medical College an d Hospital, India Author Disclosures: Swati Purohit Joshi: Nothing to disclose Jai Chowdhary: Nothing to disclose Paresh Kumar Sukhani: Nothing to disclose Ashna Garg: Nothing to disclose LI-RADS Nonradiation Treatment Response Algorithm V ersion 2024: Diagnostic Performance and Added Value of Ancillary Features *S. Zhou*, Y-C. Wang; Nanjing/CN Purpose or Learning Objective: We aimed to evaluate the diagnostic performance of the MRI-based Nonradiation LR-TRA v2 024 and the value of incorporating ancillary features (AFs) on improving diagnostic precision.

Methods

or Background: This retrospective study included patients with HCC who underwent local-regional therapy (LRT) followed by curative treatments between January 2017 and December 2022. Each treate d lesion was evaluated according to the LR-TRA v2024, LR-TRA v20 17, and modified Response Evaluation Criteria in Solid Tumors (mRECI ST) criteria, with pathologic response serving as the reference. The s ensitivity, specificity, and accuracy of different treatment response criteria w ere compared using the McNemar test.

Results

or Findings: A total of 231 patients (198 males; median age, 56 years; IQR, 50-63 years) with 306 treated lesions ( 249 incomplete pathologic necrosis) were evaluated. LR-TRA v2024-Viable (with out AFs) exhibited comparable sensitivity and accuracy than LR-TRA v20 17-Viable and mRECIST (sensitivity: 81.1% [95%CI: 75.8, 85.8], 79.5% [95% CI: 74.0, 84.4], and 80.3% [95% CI: 74.8, 85.1]; accuracy: 80.1% [95% CI: 75.1 , 84.4], 79.4% [95% CI: 74.4, 83.8] and 75.2% [95% CI: 69.9, 79.9], respect ively) for predicting incomplete pathologic tumor necrosis. Both LR-TRA v 2024 (without AFs)- Viable and v2017-Viable exhibited significantly hig her specificity than mRECIST (both P ≤ .001). LR-TRA v2024 (with AFs)-Viable incorporatin g ancillary features exhibited the highest sensitivit y (85.9% [95% CI: 81.0, 90.0]) and accuracy (83.7% [95% CI: 79.0, 87.6]) among dif ferent treatment response criteria, and showed a statistic difference compare d to LR-TRA v2017-Viable (both P = .006), without sacrificing specificity (7 3.7% [95% CI: 60.3, 84.5]).

Conclusion

LR-TRA v2024 (without AFs) demonstrates good diagno stic performance and ease of use. AFs significantly enha nce diagnostic sensitivity and accuracy through category adjustments, without sacrificing specificity. LR- TRA v2024 combined with AFs is strongly recommended for use in clinical practice.

Limitations

Selection bias; only based on MRI Funding for this study: This study has received funding by four National Natural Science Foundation of China (NSFC, No. 8227 1978, 92359304, 82330060, 823B2040); Zhongda Hospital Affiliated to Southeast University, Jiangsu Province High-Level Hospital Pairing Assist ance Construction Funds (No. zdyyxy09). Ethics committee - additional information: This study was approved by our institutional review board, and the requirement for written informed consent was waived for the retrospective data [No. 2022ZDSY LL410-P01]. Author Disclosures: Yuan-Cheng Wang: Nothing to disclose Shuwei Zhou: Nothing to disclose Short MRI Surveillance (SMS) for hepatocellular car cinoma screening: first results on image quality of the SMS-HCC study *C. Van De Braak*, F. Willemssen, F. Smits, R. De M an, A. Van Der Lugt, D. Bos, R. S. Dwarkasing; Rotterdam/NL ([email protected]) Purpose or Learning Objective: Current guidelines recommend biannual US screening for patients with high risk of developing hepatocellular carcinoma (HCC). It was reported that the sensitivity of US f or detecting early-stage HCC in these patients is merely 47%. Our aim is to vali date a Short MRI Surveillance (SMS)-protocol in current surveillance patients and compare it to US in a prospective, multicentre study. Here, we pr esent our first results on image quality. Saturday Abstract-based Programme 241

Methods

or Background: From November 2023, patients from the current HCC surveillance programme were invited to undergo paired US-MRI screening. The MRI was performed on 1.5/3.0-T syste ms using a dedicated 8- 16 channel range body coil. The protocol consisted of T1W in-out phase, T2W with fat saturation and DWI. One radiologist evalua ted the US- and MRI images, while a second and third radiologist solely evaluated the MRI images. US was reported according to the LIRADS-US surveill ance classification. The current analyses were restricted to the first 50 st udy participants and revolve around descriptive analyses of the image quality an d presence of lesions.

Results

or Findings: The most common indication for the HCC surveillance programme was Hepatitis B (35/50, 70%), followed by cirrhosis (16/50, 32%). Image quality was rated as good in 64% of the US ex aminations and 94% of the MRI examinations. Based on US, a total of 16 le sions were found in eight patients, whilst on MRI this total was 109 lesions in 21 patients. Reported lesions were benign (e.g. cysts), except for one le sion that was noted on SMS, but undetected on US, which proved to be LIRADS-3 o n subsequent contrast- enhanced MRI.

Conclusion

Our preliminary results show the potential of MRI f or HCC surveillance, through better image quality and a hi gher detection rate of focal lesions.

Limitations

No limitations were identified. Funding for this study: Funding was provided by a grant of the Dutch Cancer Society (grant number: 2021-2/13803). Ethics committee - additional information: This study was approved by the institutional review board and written consent was obtained from all participants. Author Disclosures: Aad Van Der Lugt: Nothing to disclose Rob De Man: Nothing to disclose Roy S. Dwarkasing: Nothing to disclose Fokko Smits: Nothing to disclose François Willemssen: Nothing to disclose Daniel Bos: Nothing to disclose Céline Van De Braak: Nothing to disclose Enhanced prediction of microvascular invasion in he patocellular carcinoma: a comparative study between intraoperati ve ultrasound and CT radiomics *F. Rizzetto*, S. Tortora, E. Rondi, P. Carboni, M. M. B. Barabino, A. Vanzulli; Milan/IT ([email protected]) Purpose or Learning Objective: Intraoperative ultrasound (IOUS) can identify signs of microvascular invasion (MVI) in hepatocell ular carcinoma (HCC) during surgery. We evaluated whether preoperative C T-derived radiomics can offer a less invasive alternative.

Methods

or Background: All patients who underwent surgical resection with IOUS for histologically confirmed HCC lesions were retrospectively included. For those with available preoperative triphasic CT scans, HCC nodules were segmented across the arterial, venous, and delayed phases. Using PyRadiomics, radiomic features (RFs) were extracted from each segmentation. After dimensionality reduction, the selected RFs fr om each phase were used to train and validate various predictive models based on Support Vector Machine (SVM) algorithm, with histopathological confirmatio n of MVI as ground truth. The best performing model was tested on an independ ent dataset. Qualitative IOUS features were selected through multivariate re gression and used to build a corresponding SVM model. Performance for detectin g MVI was assessed using the area under the Receiver Operating Charact eristic curve (AUC-ROC), with model comparisons made using the DeLong test.

Results

or Findings: A total of 124 patients, each with a single HCC les ion, were selected, with preoperative CT scans acquired in nearly 30 different institutions. Of them, 86 patients were assigned to the training and validation dataset (80:20 split), while the remaining 38 serve d as independent test dataset. The most performing radiomic model include d 10 RFs extracted from arterial phase. ROC analysis in the independent tes t yielded an AUC-ROC of 70% (95% confidence interval[CI]: 52-88%; p=0.030) for radiomic model and 70% (95%CI: 51-87%; p=0.020) for IOUS model, with n o statistical difference (p=1.00).

Conclusion

Preoperative CT-derived radiomics demonstrated equi valent predictive performance to IOUS in assessing MVI in HCC, offering a non- invasive preoperative alternative for MVI risk stra tification.

Limitations

Retrospective design is the main study limitation. Funding for this study: No fundings were received for this study. Ethics committee - additional information: Institutional Review Board approved the retrospective data collection in anony mous form Author Disclosures: Francesco Rizzetto: Nothing to disclose Pierluigi Carboni: Nothing to disclose Angelo Vanzulli: Nothing to disclose Silvia Tortora: Nothing to disclose Elisa Rondi: Nothing to disclose Matteo Mario Bruno Barabino: Nothing to disclose A CT-based radiomics model to predict P53-mutated h epatocellular carcinoma *Y. Shi*, M. Li, Y. Pei, W. Li; Changsha/CN ([email protected]) Purpose or Learning Objective: To evaluate the diagnostic performance of a CT-based radiomics model for predicting P53-mutated hepatocellular carcinoma (HCC).

Methods

or Background: In this retrospective single-center study, patients with histopathologically confirmed HCC who underwen t preoperative contrasted-enhanced CT examination and surgery betw een November 2017 and July 2022 were recruited. HCC was classified in to P53-mutated HCC and non-P53-mutated HCC using the gene sequencing. Radi ological features were analyzed and clinical information were collected. R adscore was based on radiomics features extracted from the plain scan, a rterial phase and portal vein phase images using the random forest method. Univar iable and multivariable logistic regression analyses were used to identify variables that were significantly and independently associated with P53 -mutated HCC, which were furtherly to develop a model. The model performance was evaluated with the area under the receiver operating characteristic cu rve (AUC), and the log-rank test was used to analyze recurrence-free survival ( RFS).

Results

or Findings: A total of 109 patients were enrolled, and assigned randomly (8:2) into training (87 patients) and vali dation sets (22 patients). The univariable analysis showed that the presence of en hancing capsule and intratumoral artery, and the radscore were signific ant risk factors for P53- mutated HCC. Further multivariable analysis identif ied that only the radscore was the independent predictor (odds ratio (OR), 3.0 5 [95% CI, 2.00–4.67], p<0.001) for P53-mutated HCC, and was used to devel op the radiomics model. The model showed excellent performance in predictin g P53-mutated HCC, with AUC of 0.899 (95% CI: 0.836-0.962) in the training set and 0.744 (95% CI: 0.523-0.964) in the validation set, Patients with p redicted P53-mutated HCC has shorter RFS than those with predicted non-P53-m utated HCC (p=0.03).

Conclusion

A CT-based radiomics model could accurately predict P53- mutated HCC.

Limitations

A single-center retrospective study with a small sa mple. Funding for this study: Funding: This work was supported by the National Natural Science Foundation of China (82071895, Wenz heng Li; 82271984, Wenzheng Li), the Natural Science Foundation of Hun an Province (2023JJ30903, Wenzheng Li; 2022JJ30950, Yigang Pei) , the Natural Science Foundation for Youth of Hunan Province (2024JJ6665, Mengsi Li; 2023JJ40970, Wenguang Liu), the Youth Science Found ation of Xiangya Hospital (2023Q06, Mengsi Li) and Postdoctoral Fell owship Programof CPSF (GZC20242047, Mengsi Li). Ethics committee - additional information: Institutional Review Board approval was abtained. Author Disclosures: Wenzheng Li: Nothing to disclose Yuting Shi: Nothing to disclose Yigang Pei: Nothing to disclose Mengsi Li: Nothing to disclose The role of gadoxetic acid-enhanced MRI-derived res idual relative enhancement index (RREI) in quantifying liver funct ion in hepatocelulary carcinoma patients *U. Eryürük*, M. N. Tasdemir, E. Cakir, S. Aslan; G iresun/TR ([email protected]) Purpose or Learning Objective: To investigate the efficacy of the residual relative enhancement index (RREI), derived from gad oxetic acid-enhanced MRI, in estimating liver function in patients with hepatocellular carcinoma (HCC), by validating with the albumin-bilirubin (AL BI) grade.

Methods

or Background: We retrospectively analyzed 41 patients with HCC who underwent gadoxetic acid-enhanced MRI. Enhancem ent ratio (ER) was calculated using the formula ER=(SI_HBP20-SI_pre)/S I_pre, where SI_HBP20 represents the signal intensity in the hepatobiliar y phase and SI_pre represents the pre-contrast signal intensity. RREI was calculated as RREI=residual liver volume (RLV)×ER. Receiver opera ting characteristic curve analysis was performed to determine optimal cut-off values of RLV, ER, and RREI for predicting ALBI grades. Spearman's rank co rrelation was used to evaluate correlations between RLV, ER, RREI, and AL BI scores. Intraclass correlation coefficient (ICC) was used to assess in tra-reader reliability and inter-reader agreement for RLV, ER, and RREI measur ements. Saturday Abstract-based Programme 242

Results

or Findings: ROC analysis showed that the optimal RREI cut-off f or predicting ALBI grade 1 was 680-698, and for ALBI g rade 3 was 537-496, for reader 1 and reader 2, respectively. RREI demonstra ted good performance in predicting ALBI grade 1, with accuracy of 85.3%-95. 1%, sensitivity of 89.4%- 94.7%, and specificity of 81.8%-95.4%. In different iating ALBI grade 3 from other grades, RREI showed excellent performance, wi th accuracy of 92.7%- 97.6%, sensitivity of 90.6%-96.9%, and specificity of 100%. Strong correlations were observed between RREI and ALBI scores, with co rrelation coefficients of -0.852 and -0.839 for both readers. Intra-reader an d inter-reader reliability was almost perfect, with ICC values of 0.975 and 0.937, respectively.

Conclusion

RREI exhibited a strong correlation with ALBI score s for assessing liver function in HCC patients and showed good accuracy in predicting ALBI grades, indicating its potential as a reliable radiological tool for evaluating liver function.

Limitations

This was a retrospective study. Funding for this study: None. Ethics committee - additional information: The study was approved by the local ethics commitee. Author Disclosures: Uluhan Eryürük: Nothing to disclose Merve Nur Tasdemir: Nothing to disclose Ertugrul Cakir: Nothing to disclose Serdar Aslan: Nothing to disclose Sarcopenia improves after transjugular intrahepatic portosystemic shunt (TIPS) placement in patients with cirrhosis *J. Kittinger*, T. Müllner-Bucsics, L. Hartl, L. Re ider, F. Wolf, M. Trauner, M. Mandorfer, T. Reiberger, K. Lampichler; Vienna/A T Purpose or Learning Objective: Transjugular intrahepatic portosystemic shunt (TIPS) is used to treat complications of port al hypertension in patients with cirrhosis. Sarcopenia in cirrhosis has been li nked to worse patient outcomes but may improve after TIPS. We aimed to ev aluate the prevalence of sarcopenia in patients undergoing TIPS and its evol ution after TIPS.

Methods

or Background: Retrospective analysis of the Vienna TIPS cohort from 01/2004 to 03/2022. Transversal psoas muscle t hickness (TPMT) was evaluated on cross-sectional abdominal imaging both (i) prior to TIPS (time interval 18 months) – by two independent radiologists. Sar copenia was defined by previously published height-corrected TPMT cut-offs at the level of L3 (men < 12mm/m; women < 8mm/m).

Results

or Findings: 199 patients were included (mean age 55.7±11 years, 69.3% male; median MELD: 11 (9-18) ; history of var iceal bleeding 33.2% and ascites 79.9%). FU imaging was available in 70 pati ents (FU 1) and 57 patients (FU 2), respectively. Interrater reliability of pso as muscle measurements was excellent (κ=0.985). Sarcopenia was highly prevalent in patient s undergoing TIPS (42.7%, 85/199). 4/45 (8.9%) of non-sarcopenic patients at baseline developed sarcopenia after TIPS implantation at FU 1 and 5/35 (14.3%) at FU 2, respectively. Transplant-free survival (TFS) of patients without sarcopenia was favourable (median: 59 months, IQR: 15 months – ∞), while mortality was high among patients with sarcopenia at baseline (TF S: median 21, IQR 1.8 – 106 months; p < 0.001). Resolution of sarcopenia wa s linked to a significant survival benefit (HR 0.121; p<0.001; adjusted to ag e and MELD score).

Conclusion

Sarcopenia is prevalent in TIPS patients and linked to worse outcome.

Limitations

CT-based diagnosis of sarcopenia without functional testing. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Approval of local ethics committee (Medical University of Vienna) was obtain ed: EK 1760/2014, 14- 264-VK, EK 1943/2017. Author Disclosures: Mattias Mandorfer: Speaker: AbbVie, Gilead, Collect ive Acumen, and W. L. Gore & Associates, Takeda Advisory Board: AbbVie, G ilead, Collective Acumen, and W. L. Gore & Associates, Takeda Theresa Müllner-Bucsics: Nothing to disclose Michael Trauner: Speaker: Bristol-Myers Squibb (BMS ), Falk Foundation, Gilead, Intercept and Merck Sharp & Dohme (MSD) Adv isory Board: from Abbvie, Albireo, Boehringer Ingelheim, BiomX, Falk Pharma GmbH, GENFIT, Gilead, Hightide, Intercept, Janssen, MSD, Novartis , Phenex, Pliant, Regulus and Shire; travel grants from AbbVie, Falk, Gilead, Intercept and Janssen; and research grants from Albireo, Alnylam, CymaBay, Fal k, Gilead, Intercept, MSD, Takeda and Ultragenyx Katharina Lampichler: Nothing to disclose Jakob Kittinger: Nothing to disclose Thomas Reiberger: Speaker: AbbVie, Bayer, Boehringe r Ingelheim, Gilead, Intercept, MSD, Roche, Siemens, and W. L. Gore & As sociates Florian Wolf: Nothing to disclose Lukas Hartl: Nothing to disclose Lukas Reider: Nothing to disclose 14:00-15:30 Room G1 Research Presentation Session: Radiographers RPS 2014 Green innovations: radiographers pioneering sustainable healthcare Moderators G. D'Anna; Milan/IT ([email protected]) M. F. McEntee; Cork/IE ([email protected]) Sustainability in medical imaging and radiotherapy education and practice: a survey of the student perspectives in a Portuguese Allied Health School *J. M. Saude*, N. Adubeiro, L. Nogueira, I. Ribeiro , A. Xavier, C. Carvalhais; Porto/PT ([email protected]) Purpose or Learning Objective: Higher education institutions play an important role as they prepare the professionals, d ecision-makers, and democratic citizens of the future. In Allied Health education, it is crucial to develop a professional that, in addition to the tec hnical and clinical content, also have a critical sense regarding environmental and social aspects. This study aims to survey medical imaging and radiothera py (MIR) students’ perceptions and knowledge about environmental susta inability in MIR education and practice.

Methods

or Background: A cross-sectional study was carried out, using a self-designed questionnaire, partially adapted and translated to Portuguese from previous studies. The final version of the que stionnaire, after a pilot test, included twenty-six questions and was distributed a mongst MIR bachelor’s degree students attending 1st, 2nd, 3rd and 4th aca demic years in September and October of 2024.

Results

or Findings: A total of 175 students participated in the study. Results showed that almost everyone believes they possess g eneral knowledge about Sustainability, however 23,0% demonstrate a lack of awareness about the negative impact that clinical practice has on the e nvironment. Data also revealed that students considered there is a need f or more teaching about sustainability in the course (49,0%). In general, p articipants demonstrated little knowledge about sustainable practices in the profes sion (34,0%), although, in daily life reported high adoption of sustainable pr actices (85,0%). Strong connection between their course and SDGs 3 (91,4%), 4 (53,7%), and 8 (38,8%), was reported.

Conclusion

Our findings revealed that there is room for curric ula adjustments. Early exposure to sustainability concepts has the p otential to promote the development of environmentally conscious profession als. By integrating sustainability into their education will potentiall y increase the knowledge and skills to make informed decisions that can reduce t he negative environmental impacts of clinical practices.

Limitations

No limitations were identified. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is educational and received a favourable statement from the institutio nal Data Protection Officer. Author Disclosures: Luisa Nogueira: Nothing to disclose Nuno Adubeiro: Nothing to disclose Jose Miguel Saude: Nothing to disclose Inês Ribeiro: Nothing to disclose Carlos Carvalhais: Nothing to disclose Ana Xavier: Nothing to disclose European Radiology Departments: Radiographers' Pers pectives on Environmental and Energy Sustainability *A. Roletto*¹, D. Catania¹, L. Rainford², A. Savio³ , M. Zanardo¹, G. R. Bonfitto¹, S. Zanoni³; ¹Milan/IT, ²Dublin/IE, ³Brescia/IT ([email protected]) Purpose or Learning Objective: Energy consumption of radiology equipment, life cycle of consumables, waste generation, and CO 2 emissions caused by staff travel are factors that influence the environ mental impact of radiology departments. Through an international survey, the p erception and knowledge Saturday Abstract-based Programme 243 of European radiographers on environmental sustaina bility issues was investigated.

Methods

or Background: In March 2024, an online survey was developed and shared with radiographers and therapeutic radio graphers across Europe. The survey consisted of 43 questions, covering demo graphic information and participants' views and actions related to environm ental sustainability, energy consumption and waste generation in healthcare.

Results

or Findings: In total, 253 responses were collected from 27 Euro pean countries. Most respondents highlighted the importa nce of environmental sustainability in healthcare. According to 63.6% (n =161) of participant, the main source of environmental impact in radiology co mes from energy consumption of radiological equipment. Additionally , 44.7% (n=113) suggested that conducting diagnostic exams remotely could red uce the CO2 emissions caused by staff commuting. Regarding workplace prac tices, over 70% (n=192) reported that they turn-off devices after use. Amon g the possible obstacles to turning off radiological equipment, respondents ide ntified long shutdown and/or restart times, loss of clinical/technical data and non-applicability to the radiographers’ role. Although recycling is practice d, it is limited to paper, plastic, and glass. Lack of environmental sustainab ility strategies in the workplace was reported by 66% (n=167). Meanwhile, 9 6.1% (n=243) believe that radiographers could play an active role in pro moting environmental sustainability in their departments.

Conclusion

This study offers a comprehensive analysis of Europ ean radiographers' awareness and opinions on environmen tal sustainability. Although radiographers recognize the importance of creating eco-friendly radiology departments, there are still gaps in impl ementation of sustainable practices within radiology practices.

Limitations

Potential biases may have been introduced in relati on to the complexity of the topics covered. Funding for this study: Not applicable Ethics committee - additional information: Ethical exemption was granted by the host institution University College Dublin f or this anonymous survey (LS- LR-24-25-Catania-Rainford) Author Disclosures: Anna Savio: Nothing to disclose Diego Catania: Nothing to disclose Louise Rainford: Nothing to disclose Andrea Roletto: Nothing to disclose Moreno Zanardo: Nothing to disclose Giuseppe Roberto Bonfitto: Nothing to disclose Simone Zanoni: Nothing to disclose State-of-Play in Artificial Intelligence Sustainabi lity in Medical Imaging: a scoping review M. Champendal¹, B. Lokaj², J. Zaghir³, V. Durand De Gevigney², C. Lovis³, H. Müller⁴, J. Schmid², *R. S. T. Ribeiro*¹; ¹Lausanne/CH, ²G enève/CH, ³Geneva/CH, ⁴Sierre/CH ([email protected]) Purpose or Learning Objective: To synthesize the existing literature on how the environmental sustainability of artificial inte lligence (AI) in medical imaging is being addressed and to identify specific strateg ies that have been used.

Methods

or Background: A scoping review was conducted following the Joanna Briggs Institute methodology. Comprehensive literature search was performed in MEDLINE, Embase, CINAHL, and Web of Sc ience, targeting publications from 2014 to 2024 in English or French . The search used a combination of keywords and MeSH terms related to e nvironmental sustainability, AI, and medical imaging modalities. Three independent reviewers screened abstracts, titles and full texts for eligibility.

Results

or Findings: The search identified 2812 results, of which 11 met the inclusion criteria. The selected papers comprised 8 research articles, 3 reviews. Three key themes emerged: energy consumpti on (n=10), carbon footprint (n=4), and computational resources (n=4). The metrics CO2 equivalent, carbon intensity, training time, power use effectiveness, equivalent distance travelled by car were proposed to assess p otential AI impact on the environment. Most energy-efficient techniques invol ved data, AI modelling and training such as data augmentation, data quantisati on, lightweight model development, reduction of parameters. Identified st rategies to enhance efficiency and reduce environmental impact include (i)integrating energy and carbon metrics in AI evaluation in addition to accu racy assessments, (ii)developing an ecolabel for AI tools, (iii)transitioning to cloud computing and (iv)developing lightweight AI models.

Conclusion

This review identified critical metrics and actiona ble strategies used to assess and improve sustainable practices in AI for medical imaging which include the integration of specific sustainab ility-related metrics, cloud computing adoption and development of efficient AI models.

Limitations

The limitations of this review include not assessin g the quality of the articles, which is standard practice in scoping reviews, and restricting the search to only four databases. Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Ricardo Silva Teresa Ribeiro: Nothing to disclose Jamil Zaghir: Nothing to disclose Henning Müller: Nothing to disclose Christian Lovis: Nothing to disclose Jerome Schmid: Nothing to disclose Valentin Durand De Gevigney: Nothing to disclose Belinda Lokaj: Nothing to disclose Mélanie Champendal: Nothing to disclose A life-cycle assessment framework for quantifying t he carbon footprint of diagnostic imaging *A. Roletto*¹, A. Savio², B. Marchi², S. Zanoni²; ¹ Milan/IT, ²Brescia/IT ([email protected]) Purpose or Learning Objective: Environmental sustainability topic is increasingly relevant in the radiology sector, whic h accounts for about 10% of healthcare sector's carbon footprint. Life-cycle as sessment (LCA) is one of the main tools for analysing the environmental impact o f processes. This study aims to review existing LCA approaches in radiology available in literature, identify their characteristics and limitations to p ropose a more comprehensive LCA framework for diagnostic imaging.

Methods

or Background: Through a literature review on the topic of LCA in diagnostic imaging, performed according to the PRIS MA statement, a novel LCA framework specifically for diagnostic imaging w as developed. The framework includes all the features described by th e various selected articles and overcoming their limitations.

Results

or Findings: Regarding the literature review, an LCA approach wa s described by 5 articles. All studies electricity co nsumption in their framework. Usage of consumables and auxiliary equipment was in cluded in 80% (4/5) of the studies. Equipment production, staff commuting, and waste generation were included in only 40% (2/5). Only two articles have considered a cradle-to- grave system approach. Subsequently, a novel LCA fr amework was designed to overcome the limitations by covering all aspects that can be included in the three phases: 1) Preclinical, production of imaging equipment and consumables, staff commuting; 2) Clinical, use of d evices and consumables during diagnostic imaging; 3) Post-clinical, waste generation and data storage for the images produced.

Conclusion

This literature review provides an overview of stud ies focusing on LCA methodology for diagnostic imaging. The results culminated in the proposal of a comprehensive framework, which aims t o overcome the identified

Limitations

by providing a complete analysis of the environmental footprint of radiological procedures.

Limitations

Still sparse reporting of LCA in Diagnostic imaging limited the strength of the conclusions of this study. Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Anna Savio: Nothing to disclose Andrea Roletto: Nothing to disclose Beatrice Marchi: Nothing to disclose Simone Zanoni: Nothing to disclose Leading Radiographers into a Greener Future: A Syst ematic Review of Green Transformational Leadership and Sustainable P ractices *L. Federico*¹, A. Roletto², D. Catania², S. Zanoni ³, S. Durante¹; ¹Bologna/IT, ²Milan/IT, ³Brescia/IT ([email protected]) Purpose or Learning Objective: The healthcare system requires many resources to maintain the standards necessary to ca re for the population. It is the task of effective leadership to ensure the envi ronmental sustainability clinical activities. The aim of this study is to ex amine the role of Green Transformational Leadership (GTL) in radiography, f ocusing on the mitigation strategies adopted by radiographer managers and the ir impact on promoting environmental sustainability in radiology departmen ts.

Methods

or Background: A systematic literature was conducted in accordance with the PRISMA statements in several da tabases, targeting studies on leadership in radiography, green innovat ion, and environmental responsibility. The search strategy employed keywor ds such as "green transformational leadership," "radiographer manager ," and "environmental leadership in healthcare." Two reviewers independen tly screened titles and abstracts to select studies based on inclusion crit eria.

Results

or Findings: Among the retrieved articles, 6 met the inclusion c riteria. The analysis highlighted how GTL can influence the implementation of sustainable practices in radiography. Radiographer managers using GTL play a key role in promoting sustainable practices, such as reducing energy consumption, encouraging use of reusable imaging eq uipment, and minimising waste generation. Interdisciplinary collaboration c ould be decisive in increasing effectiveness of green initiatives. Studies reviewe d highlighted the importance Saturday Abstract-based Programme 244 of successful communication, measurable goals, and staff training for adopting green practices. Radiography service managers with emotional intelligence who fostered a climate of trust were more successfu l in promoting a culture of sustainability.

Conclusion

Radiography leaders who prioritise green strategies can significantly reduce the environmental impact of ra diology departments. Investing in leadership development, promoting emot ional intelligence, and encouraging interdisciplinary collaboration are cri tical steps in creating a green healthcare system that aims towards environmental s ustainability without reducing quality patient care.

Limitations

Sparce literature of GTL limit the strength of the conclusions of this study. Funding for this study: Not needed Ethics committee - additional information: Not needed Author Disclosures: Diego Catania: Nothing to disclose Stefano Durante: Nothing to disclose Andrea Roletto: Nothing to disclose Simone Zanoni: Nothing to disclose Laura Federico: Nothing to disclose Sustainable practices in nuclear medicine: A scopin g review *D. Fonseca Ribeiro*¹, K. Borg Grima², A. Geão³, C. Andersson⁴, S. Murphy⁵, P. S. Costa⁶, C. Baun⁷, A. Karangelis⁸, M. Champendal⁹; ¹London/UK, ²Naxxar/MT, ³Montijo/PT, ⁴Uppsala/SE, ⁵Dublin/IE, ⁶Porto/PT, ⁷Odense/DK, ⁸Patra/GR, ⁹Lausanne/CH Purpose or Learning Objective: Sustainable development seeks to balance economic growth, environmental impact, and social i nclusion, which is critical in high-tech fields like nuclear medicine. This stu dy reviewed published literature on sustainable practices in nuclear medi cine, focusing on the three main pillars of sustainability: environmental, soci al, and economic.

Methods

or Background: The scoping review was conducted in accordance with the Joanna Briggs Institute methodology. The s earch was performed on PubMed, Embase, Cinhal and Web of Science in Novemb er 2023 and included studies in English. The research equation combined keywords and Medical Subject Heading terms (MeSH) related to sustainabil ity in Nuclear Medicine. Three independent review authors screened all abstr acts and titles, and four reviewers conducted the data extraction and analysi s.

Results

or Findings: Thirty-two studies met the inclusion criteria for t his scoping review, with most articles having been publ ished in 2022 and 2023. Spain contributed the highest number of publication s. Studies were categorised according to procedure type (n = 32; 31 % therapy, 9% PET, 35% diagnostic & therapy, 19% SPECT) and study location (hospital based, non- hospital based, or both). The studies primarily foc used on strategies related to the three sustainability pillars, with an emphasis on the environmental impact.

Conclusion

This review highlighted the growing interest in sus tainability within the nuclear medicine field, especially in re lation to environmental factors. However, significant knowledge gaps emerged on the impact of economic and social factors within nuclear medicine practices. R ecommended strategies included proper radioactive waste management, resou rce optimisation, and fostering collaborative environments to ensure sust ainable nuclear medicine practices.

Limitations

The limitations of the study are the exclusion of a rticles not available in the English language and the quality o f the articles included was not assessed according to the methodology of a scop ing review. Funding for this study: No funding was provided for this study. Ethics committee - additional information: This study was a scoping review. Author Disclosures: Daniela Fonseca Ribeiro: Nothing to disclose Karen Borg Grima: Nothing to disclose Apostolos Karangelis: Nothing to disclose Ana Geão: Nothing to disclose Camilla Andersson: Nothing to disclose Christina Baun: Nothing to disclose Pedro Silva Costa: Nothing to disclose Shauna Murphy: Nothing to disclose Mélanie Champendal: Nothing to disclose Towards sustainable radiography: exploring percepti ons and overcoming barriers *M. A. Rawashdeh*¹, A. England², M. F. Mcentee², M. Ali¹, M. E. S. Zakaria¹; ¹Ajman/AE, ²Cork/IE ([email protected]) Purpose or Learning Objective: Previous research has examined the attitudes and behaviors of various professions rega rding environmental sustainability. However, there is a paucity of stud ies specifically addressing the perspectives of radiographers. This study aims to i nvestigate radiographers' perceptions, practices, and barriers to change rela ted to environmental sustainability within the field of radiology.

Methods

or Background: Ethical approval was obtained and data collection was conducted using Google Forms. The survey target ed 104 practicing radiographers across multiple countries. Questions were organized into five domains to gather insights into demographics, train ing related to global warming and climate change. Data analysis employed both descriptive and inferential statistics.

Results

or Findings: A total of 104 radiographers completed the survey. Female respondents exhibited a significantly higher participation rate in environmental protection campaigns (P=0.01). The ma jority of participants (68%) acknowledged their awareness of climate chang e and its impact on the natural environment. Furthermore, 74% of respondent s indicated a belief in the necessity to enhance sustainability practices. The most frequently employed strategies to reduce energy consumption and emissio ns included low-energy lighting (60%), real-time power monitoring tools (4 1%), and energy-efficient heating systems (32%). A notable concern regarding sustainability emerged among respondents, with time constraints (50%) and a lack of leadership (48%) identified as prevalent barriers.

Conclusion

Participants recognize the significance of environm ental sustainability in the field of radiology; however, obstacles such as inadequate leadership, support, authority, and facility limita tions impede the adoption of sustainable practices.

Limitations

The present study has several limitations. It relie d on voluntary participation from radiographers and employed snowb all sampling, which may introduce self-selection bias. Additionally, the on line survey format may limit geographical or technical access, thereby affecting the generalizability of study findings. Funding for this study: Not applicable Ethics committee - additional information: IRB from Gulf Medical University Author Disclosures: Mark F. Mcentee: Nothing to disclose Mohamed El Sayed Zakaria: Nothing to disclose Magdi Ali: Nothing to disclose Andrew England: Nothing to disclose Mohammad Ahmmad Rawashdeh: Nothing to disclose Considerations on the environmental sustainability of using ChatGPT in radiography *E. Scaramelli*, G. R. Bonfitto, A. Roletto, S. V. Fasulo, L. Bombelli, D. Catania; Milan/IT ([email protected]) Purpose or Learning Objective: ChatGPT has gained credit among radiographers, given its contribution to training a nd clinical activity. However, its environmental impact is overlooked. The increas ing use of AI is contributing to rising Greenhouse gas emissions, water waste and energy consumption. This research aims to explore the environmental imp act of using ChatGPT in radiography and raise awareness of its sustainabili ty in daily clinical practice, underscoring the need for radiographers to engage, given the gap in AI's environmental impact studies within radiology.

Methods

or Background: A literature review across databases (Google Scholar, PubMed) was conducted according to the PRI SMA statement to find recent evidence on environmental impact of ChatGPT in radiography and, more generally, in healthcare.

Results

or Findings: The literature shows limited evidence on the sustainability of AI in radiology, leading to an an alysis of the broader healthcare sector to estimate its impact and develo p insights for radiology. From this search, only 16 articles were found to be relevant to healthcare, and just one focused on radiology. The results reveal a dual nature of AI: while it's incredibly useful in healthcare, it has a significa nt environmental impact. Indeed, companies like Google, Microsoft, and Meta consume 2.2 billion cubic meters of water to cool their servers and produce e lectricity, marking a 48% increase in water usage. GPT-3 produces 8.4 metric tons of CO2 annually and requires 700,000 Liters of water for training. AI i s also responsible for 1-2% of global electricity consumption, raising important q uestions about its long-term sustainability.

Conclusion

Although tools such ChatGPT show usefulness in clin ical practice, including radiography, it’s essential to use them more responsibly, given their substantial environmental impact on pla net's resources.

Limitations

The lack of evidence in the literature, limits the strength of the

Conclusions

of this study. Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Luca Bombelli: Nothing to disclose Simone Vito Fasulo: Nothing to disclose Diego Catania: Nothing to disclose Andrea Roletto: Nothing to disclose Elena Scaramelli: Nothing to disclose Giuseppe Roberto Bonfitto: Nothing to disclose Saturday Abstract-based Programme 245 Green Imaging Revolution: Pioneering Sustainable Pr actices in Radiology for a Healthier Planet *A. A. Bherwani*¹, M. Sudds², L. Townsend-Sanders³, A. Prieto Valero⁴; ¹Orlando, FL/US, ²Wymondham/UK, ³Tampa, FL/US, ⁴Madrid/ES ([email protected]) Purpose or Learning Objective: This study highlights the critical need to incorporate sustainability in radiology, reducing t he environmental impact without compromising diagnostic accuracy or patient care. Focusing on energy-efficient practices, waste reduction, and te chnological innovations in modalities like CT, MRI, and general radiography, t he objective is to integrate eco-friendly strategies into radiological departmen ts as part of broader healthcare sustainability goals.

Methods

or Background: A comprehensive literature review of over 90 sources, including peer-reviewed articles, books, a nd case studies, was conducted. This research explores innovations such as energy-saving modes in MRI and CT, digital radiography's reduction of c hemical waste, and sustainable contrast media practices. Case studies from several healthcare institutes globally are analyzed, illustrating real -world applications of green radiology practices.

Results

or Findings: Adopting energy-efficient imaging systems and optimizing resource use led to significant environm ental and financial benefits. For example, Singapore General Hospital reduced ene rgy consumption by 30%, saving $1.2 million annually, while Mayo Clini c enhanced wastewater treatment, minimizing environmental contamination f rom contrast media. These practices and many others from various other instit utions not only reduced greenhouse gas emissions but also operational costs .

Conclusion

Incorporating sustainable practices in radiology is vital for reducing healthcare’s environmental footprint while maintaining high standards of patient care. This research provides a scalable framework for radiology departments worldwide, illustrating how integrating sustainability can enhance operational efficiency and contribute to global hea lth equity and environmental conservation.

Limitations

This study is based on case studies from select ins titutions, which may not reflect the diverse resources and regulator y environments of all healthcare systems. Additionally, the long-term imp acts of implementing these sustainable radiology practices require further inv estigation. Funding for this study: Not Applicable Ethics committee - additional information: Not Applicable Author Disclosures: Laurie Townsend-Sanders: Nothing to disclose Alicia Prieto Valero: Nothing to disclose Anand Ashok Bherwani: Employee: GE Healthcare Pharm aceutical Diagnostics Michael Sudds: Nothing to disclose Centargo: Insights from the CIMROD Experiment on Co ntrast Consumption, Efficiency, Patient Care and environme ntal impact Y. Anquetil¹, *T. Leturgez*², J. E. Jacquin¹, F. Kr uta³, F. Jambon¹; ¹Perigueux/FR, ²La Garenne Colombes/FR, ³Paris/FR ([email protected]) Purpose or Learning Objective: With the transition in France's contrast product supply model on March 1, 2024, the function alities of CT injectors have been re-evaluated. This study assesses the impact o f Centargo injectors on contrast media volume, patient care, preparation ti me, and environmental impacts at the CIMROD center, which acquired two Ce ntargo injectors.

Methods

or Background: To collect pertinent data, we conducted three measurements and utilized five data sources. The pr ivate hospital, equipped with two CT machines, transitioned from a single-us e Stellant injector to a multi-use system, ultimately adopting the Centargo injector with enhanced features. Data sources included the volume of contr ast media administered per patient, the time radiographers allocated to variou s tasks in their routines, protocol preparation and traceability times, and wa ste weight measurements. The evaluation period spanned from April to Septemb er and incorporated different injector configurations as well as a tagg ing system, Ubudu, to monitor workflow efficiency among radiographers.

Results

or Findings: CIMROD aimed for an average contrast volume of 80 mL using Ultravist 370. The Centargo injector succe ssfully achieved an average of 72.08 mL per injection across 1,099 proc edures, and after further optimization, this volume decreased to 68.31 mL, re sulting in a 15% reduction compared to the initial target. Workflow improvemen ts included a 50% reduction in injector preparation time, a 14% incre ase in patient care time, and a 6% increase in image preparation time. Time savin gs of 16% per exam were noted, with waste production decreasing by 66% comp ared to the single-use Stellant, equating to an annual reduction of 1.4 to nnes.

Conclusion

The findings confirm the benefits of the Centargo i njector, especially when combined with Smart Protocol and Na utilus software, RIS- Injector connectivity, and kVp optimization, leadin g to enhanced operational efficiency and reduced environmental impact.

Limitations

Single center evaluation. Funding for this study: Bayer funded Ethics committee - additional information: The experimentation isn't collecting patient level data so we didn't submitte d it to an ethical committee. Author Disclosures: Jerome Elisabeth Jacquin: Nothing to disclose François Kruta: Equipment Support Recipient: Ubudu Yohan Anquetil: Equipment Support Recipient: CIMROD Thibaut Leturgez: Employee: Bayer François Jambon: Equipment Support Recipient: CIMRO D Success factors for implementing an intervention us ing return letters for low-value MRIs *I. Ø. Brandsæter*, E. Kjelle, E. R. Andersen, B. M . Hofmann; Gjøvik/NO ([email protected]) Purpose or Learning Objective: The study aimed to investigate key stakeholders’ experiences with and reflections on s uccess factors for implementing an intervention using return letters t o reduce the use of three low-value magnetic resonance imaging (MRI) examinat ions.

Methods

or Background: An intervention to reduce low-value MRI was designed and implemented in private imaging centres in Norway in October 2022. The intervention used return letters based on Choosing Wisely recommendations for poor referrals for MRI of the l ower back, brain and knee sent to imaging centres. Two semi-structured indivi dual interviews were conducted with the medical directors of the two inc luded imaging providers (radiologists) and two focus group interviews with nine managers from the various private imaging centres (radiographers) ope rated by the two imaging providers were conducted. Inductive content analysi s in three steps was used to analyse the data.

Results

or Findings: The analysis resulted in five categories: general experience, anchoring, organisation, return letter procedure and outcome. In general, the intervention was well received. Suffic ient information, anchoring and support from the organisation’s leaders were id entified as crucial success factors. However, there were some barriers to the i mplementation, e.g. the medical directors in charge of the implementation f ound it hard to be hands-on and distribute information to the radiogpraphers at the imaging centres. Additionally, some Choosing Wisely recommendations were found vague and difficult to use by the radiographers doing the ref erral assessment.

Conclusion

This study provides insights into the practical and crucial details of implementing interventions to reduce low-value i maging using Choosing Wisely recommendations. The intervention was genera lly well received, and several key success factors were identified.

Limitations

The limitations of the study are that only managers and medical directors were included in the study, interviewing referrers and patients would have provided other important perspectives. Funding for this study: Funding was provided by the Research Council of Norway (Project number 302503). Ethics committee - additional information: Ethical approval is unnecessary according to national regulations in Norway (LOV-20 08–06-20–44), and this study was not submitted to the regional committees for medical and health research ethics. The Norwegian Agency for Shared Se rvices in Education and Research approved the processing and storage of per sonal information in this study (Ref. 974188) Author Disclosures: Ingrid Øfsti Brandsæter: Nothing to disclose Bjørn Morten Hofmann: Nothing to disclose Elin Kjelle: Nothing to disclose Eivind R. Andersen: Nothing to disclose Quantifying Energy Savings in Radiology: A Simple A pproach to Make the Radiology Department More Sustainable *K. Iaccarino*, D. Fazzini, S. Papa, M. Ali; Milan/ IT ([email protected]) Purpose or Learning Objective: To quantify the energy consumption of radiology reporting stations and explore a hypothet ical scenario to mitigate energy waste in a healthcare setting.

Methods

or Background: We monitored 10 reporting stations over a period of 90 days at the Centro Diagnostico Italiano (Milan, Italy). An energy logger was installed on each station to measure real-time powe r consumption, capturing data on both active and idle states. Stations were configured to enter stand-by mode after 4 hours of inactivity. We conducted a si mulation to assess the impact of shutting down the stations after 1 hour o f inactivity instead of allowing them to remain in stand-by mode.

Results

or Findings: The overall power consumption of the 10 reporting stations was approximately 16,615.65 kWh, which is equivalent to about 6 households in Italy (average annual consumption of 2,700 kWh per household). We identified three main power consumpt ion patterns: mainly-off, mainly-on, and always-off. The estimated on-mode co nsumption was 12,738.44 kWh per year, while stand-by consumption was 3,128.13 kWh and Saturday Abstract-based Programme 246 off-mode consumption was 749.06 kWh. By implementin g the hypothetical scenario of shutting down after 1 hour of inactivit y, we estimated potential energy savings of around 2,346.12 kWh/year. Conside ring that CDI has a total of 53 reporting stations across all sites, the estimated total annual energy savings would be approximately 12,434.44 kWh.

Conclusion

Optimizing energy usage in radiology departments is essential for promoting sustainability. Simple configuration chan ges can lead to significant reductions in energy waste, enhancing the environme ntal responsibility of healthcare facilities.

Limitations

The sample size of 10 reporting stations may not be representative of the entire radiology department, and the 90-day monitoring period might not account for seasonal variations in usage. The hypothetical scenario assumes uniform behavior across stations, which may not reflect actual workflows. Funding for this study: None Ethics committee - additional information: The ethics commettee approval is not applicable for this study Author Disclosures: Marco Ali: Consultant: Bracco Imaging S.p.A. Sergio Papa: Nothing to disclose Deborah Fazzini: Nothing to disclose Katia Iaccarino: Nothing to disclose 16:00-17:30 Research Stage 1 Research Presentation Session: Head and Neck RPS 2108 A journey through thyroid imaging Moderator E. Gotsiridze; Tbilisi/GE ([email protected]) Additional Value of Pertechnetate Scintigraphy to A CR-TIRADS and EU- TIRADS for Thyroid Nodule Classification in Euthyro id Patients L. Sollmann, M. Eveslage, M. Danzer, M. Schäfers, B . Heitplatz, D. Hescheler, B. Riemann, *B. Noto*; Münster/DE ([email protected]) Purpose or Learning Objective: Thyroid nodules are a highly prevalent, predominantly benign finding, yet their accurate ev aluation remains challenging. While ultrasound and TIRADS are now wi dely accepted as standard in evaluation, the utility of thyroid scin tigraphy in euthyroid patients remains debated. Previous studies have investigated the diagnostic potential of TIRADS or radionuclide scanning in isolation, but a n integrated approach has not been explored so far. This study aimed to evalu ate if pertechnetate scintigraphy enhances the diagnostic value of TIRAD S in a multimodal framework.

Methods

or Background: The diagnostic capabilities of ACR-TIRADS, EU- TIRADS, pertechnetate scintigraphy, and multimodal models were retrospectively analyzed for 322 nodules (231 benig n, 91 malignant) in 208 euthyroid patients undergoing thyroidectomy. Statis tical analysis employed generalized estimating equations.

Results

or Findings: Thyroid scintigraphy demonstrated an AUC of 0.6 (95%CI:0.55-0.66), ACR-TIRADS of 0.84 (95%CI:0.79-0 .89) and EU-TIRADS of 0.78 (95%CI: 0.72-0.83). Integrating thyroid sci ntigraphy with ACR-TIRADS enhanced diagnostic accuracy, yielding an AUC of 0. 86 (p=0.039). Similarly, combining thyroid scintigraphy with EU-TIRADS resul ted in an AUC of 0.80 (p =0.008), surpassing the individual TIRADS performan ces. Furthermore, the integration of thyroid scintigraphy adjusted the ma lignancy probability among TIRADS categories. Iso- or hyperfunctioning nodules in ACR-TIRADS TR4 and hypofunctioning nodules in TR3 exhibited comparable probabilities of malignancy. Similarly, iso- or hyperfunctioning nod ules in EU-TIRADS 4 showed similar malignancy probabilities to hypofunc tional nodules in EU- TIRADS 3, indicating the potential for more refined risk stratification.

Conclusion

This study demonstrates enhanced diagnostic perform ance achieved by integrating thyroid scintigraphy with A CR—and EU-TIRADS for classifying thyroid nodules in euthyroid patients. Such a multimodal approach could improve risk stratification and management de cisions, particularly in complex scenarios like multinodular goiter. Further research is warranted to validate these findings and explore their clinical implications.

Limitations

Retrospective design Funding for this study: The Medical Faculty, University of Münster, Germany supported B.N. as a clinician scientist. Ethics committee - additional information: The study protocol was approved by the ethics committee of the University of Munste r and performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments. Author Disclosures: Michael Schäfers: Nothing to disclose Moritz Danzer: Nothing to disclose Barbara Heitplatz: Nothing to disclose Lea Sollmann: Nothing to disclose Burkhard Riemann: Nothing to disclose Daniel Hescheler: Nothing to disclose Benjamin Noto: Nothing to disclose Maria Eveslage: Nothing to disclose Subtraction ultrasound microangiography for assessm ent of microvascularity patterns in diffuse thyroid diseas e *A. Borlea*, D. I. Stoian; Timisoara/RO ([email protected]) Purpose or Learning Objective: To evaluate the use of Subtraction Ultrasound Microangiography (SUMA) in the quantitat ive assessment of microvascularity patterns in diffuse thyroid diseas es and to compare these findings with normal thyroid tissue.

Methods

or Background: Conventional color Doppler ultrasound is limited by its subjective nature in assessing thyroid vascular ity. SUMA provides a quantitative approach through color pixel percentag e (CPP) measurement. A cohort of 220 subjects was studied, including 90 wi th autoimmune thyroiditis, 20 with Graves' disease, and 110 normal controls. S UMA was used to quantify CPP, and results were analyzed to identify differen ces between the groups. The relationship between TSH levels and CPP was als o explored.

Results

or Findings: Significant differences in CPP were observed across the groups. Normal controls (euthyroid) demonstrated a median CPP of 26% (IQR 18-37%), Graves' disease patients with clinical and subclinical hyperthyroidism had a median CPP of 75% (IQR 55-82%), and those wit h untreated hypothyroid Hashimoto thyroiditis (hypothyroid) had a median CPP of 63% (IQR 49-75%). A U-shaped relationship was found bet ween TSH levels and CPP, indicating alterations in vascularity in both hypo- and hyperthyroid states.

Conclusion

SUMA allows for a more objective and quantitative a ssessment of thyroid microvascularity, revealing distinct dif ferences between normal, autoimmune thyroiditis, and Graves' disease tissues . The technique may improve diagnostic accuracy compared to conventiona l Doppler ultrasound.

Limitations

Further studies are needed to validate SUMA’s diagn ostic potential, including its reproducibility across dif ferent ultrasound systems and settings, as well as its sensitivity in detecting disease progression or treatment response. Funding for this study: The study received no external funding Ethics committee - additional information: Victor BAbes University of Medicine and Pharmacy Author Disclosures: Dana I Stoian: Nothing to disclose Andreea Borlea: Nothing to disclose Ultrasound-guided microwave ablation versus thyroid ectomy for the treatment of solitary nodular retrosternal goiter *Y. Li*, Y. Luo, M. Zhang; Beijing/CN ([email protected]) Purpose or Learning Objective: The purpose of this study was to compare the clinical outcomes of microwave ablation versus thyroidectomy for patients with solitary nodular retrosternal goiter (RSG).

Methods

or Background: This retrospective study evaluated 243 patients wit h solitary nodular RSG treated with MWA (M group, n=1 11) or thyroidectomy (T group, n=132). Complications, thyroid function, and treatment variables, including procedure time, estimated blood loss, hos pitalization, and cost, were compared. The volume, volume reduction rate, sympto ms, and cosmetic score were also evaluated in the W group.

Results

or Findings: The rate of overall complications, transient RLN in jury and hypothyroidism was 35.6%, 9.1%, and 10.6% in th e T group, respectively, but these complications were 7.2%, 2,7%, and 0 in t he M group (all P<.05). Patients in the M group had a significantly shorter procedure time (median, 6.48 versus 95.0 minutes, P<.0001), less estimated blood loss (0 versus 50 ml, P<.0001), and lower cost (US ＄1541.25 versus ＄2839.40, P<.0001) than those treated by thyroidectomy. After MWA, the VRR was 76.31% and 89.37% at the 12 months and last follow-up time, respectiv ely. Of all the nodules treated by MWA, 50 (45%) received additional ablati on. The symptom and cosmetic scores were both significantly reduced at the last follow-up. Saturday Abstract-based Programme 247

Conclusion

MWA is an effective and safe treatment for solitary nodular RSG. Moreover, MWA is associated with a faster recovery, fewer complications, and superior esthetic results relative to thyroidectomy and it may be a potential alternative to surgery in selected patients, especi ally for those who are ineligible or unwilling to receive surgical treatme nt.

Limitations

This study was carried out in a single center, and multicenter studies are warranted. Funding for this study: This study recieved no funding. Ethics committee - additional information: Ethical approval was obtained from the Institutional Ethics Committee of the Chin ese PLA General Hospital. Author Disclosures: Yingying Li: Nothing to disclose Mingbo Zhang: Nothing to disclose Yukun Luo: Nothing to disclose Mid-term thyroid function alterations as predictors of long-term outcomes in radiofrequency ablation of benign thyro id nodules *Y-H. Chen*¹, P-L. Chiang¹, Y-H. Chang², C-K. Chou² , W-C. Lin¹; ¹Kaohsiung/TW, ²Kaohsiung City/TW ([email protected]) Purpose or Learning Objective: Radiofrequency ablation (RFA) has gained recognition as a highly effective treatment for ben ign thyroid nodules. While thyroid function alterations have been observed dur ing post-RFA follow-up, this study specifically focuses on the potential link be tween changes in thyroid function and volume reduction ratio of treated nodu les. Additionally, it seeks to evaluate whether fluctuations in thyroid function a t mid-term follow-up can serve as early indicators for the development of lo ng-term hypothyroidism following RFA.

Methods

or Background: In this retrospective study, 50 euthyroid individua ls (mean age = 47.0 years; 43 females, 7 males) with a total of 72 benign thyroid nodules (median volume = 4.61 mL) undergoing RFA we re evaluated. Comprehensive assessments, including clinical exami nations, ultrasound imaging, and blood tests, were conducted at specifi c intervals (pre-RFA, and at 6 months, 12 months, and annually post-RFA).

Results

or Findings: The mean follow-up period was 22.3 months. Both medium-term and long-term follow-ups revealed signi ficant reductions in T3 levels (p<0.001, p=0.005) and elevations in TSH lev els (p<0.001, p<0.001) compared to baseline measurements. A negative corre lation was found between medium-term T3 levels and long-term volume reduction ratio (r=- 0.332, p=0.005). Furthermore, patients with lower T 3 levels during medium- term follow-up demonstrated a significantly higher long-term volume reduction ratio compared to those with higher T3 levels (0.92 vs. 0.77, p=0.017).

Conclusion

Following radiofrequency ablation, notable unexpect ed alterations in thyroid function were observed, with out meeting the criteria of hypothyroidism. Additionally, a lower mid-term T3 l evel may be indicative of a better volume reduction ratio during long-term foll ow-up.

Limitations

The study is limited by biases associated with its retrospective design and a lack of short-term data within a six-m onth period. To address these limitations, further prospective studies with increased focus on short-term data are necessary. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study received approval from the institutional review board under the refer ence number 202200189B0. Author Disclosures: Wei-Che Lin: Nothing to disclose Yen-Hsiang Chang: Nothing to disclose Pi-Ling Chiang: Nothing to disclose Yi-Han Chen: Nothing to disclose Chen-Kai Chou: Nothing to disclose Comparision of K-TIRADS, EU-TIRADS and ACR-TIRADS G uidelines for Malignancy Risk Determination of Thyroid Nodules *E. Tobcu*, E. Karavaş, G. Taşova Yılmaz, Z. Tobcu; Balıkesir/TR ([email protected]) Purpose or Learning Objective: To evaluate the performances of three internationally recognized thyroid imaging reportin g and data systems (TIRADS) for risk stratification of malignancy in c omparison to one another.

Methods

or Background: A total of 225 thyroid nodules with definitive FNAB cytology or histopathological diagnoses were includ ed in the study. Various ultrasound (US) features were classified into categ ories based on three TIRADS editions. The guidelines were assessed regar ding sensitivity, specificity, predictive values, and diagnostic accu racy to compare diagnostic value.

Results

or Findings: American College of Radiology (ACR)-TIRADS demonstrated the best diagnostic accuracy (63.1%), the highest specificity (58.7%), and positive predictive value (36.3%) amon g three different TIRADS systems. Korean (K)-TIRADS exhibited the highest se nsitivity (94.2%), negative predictive value (96.1%), and the most fav orable negative likelihood ratio (0.13). The European (EU)-TIRADS had a sensit ivity of 90.4%, specificity of 48.6%, and diagnostic accuracy of 58.2%, ranking between the other two guidelines across most parameters.

Conclusion

The rigorous use of the guidelines established by e ach of the three TIRADS systems would have markedly reduced th e number of FNABs performed. The comparison of the three guidelines i n our study indicated that they are effective screening methods for identifyin g malignant thyroid nodules. Among them, K-TIRADS showed the most effective diag nostic performance in sensitivity, while ACR-TIRADS yielded the best spec ificity.

Limitations

The main limitation of our study was its single-cen ter design. Another limitation of our study was the lack of his topathological diagnoses for all nodules. All nodules with malignant cytological results underwent surgery that enables us to reach histopathological diagnose s, but only two nodules with benign cytology underwent surgery due to the clinic ian's discretion and the patient's preference. Cytology results were used as a reference standard diagnosis in the remaining nodules. Funding for this study: No funding Ethics committee - additional information: The study was performed in accordance with the ethical guidelines of the Helsi nki Declaration and approved by the local ethics review committee (2024 -3). Author Disclosures: Eren Tobcu: Nothing to disclose Erdal Karavaş: Nothing to disclose Gülden Taşova Yılmaz: Nothing to disclose Zeynep Tobcu: Nothing to disclose Thyroid nodule characterization: interobeserver eva luation of different TIRADS with and without AI software *C. Di Bella*¹, E. David², C. Solito¹, V. Dolcetti¹ , P. Pacini¹, G. Del Gaudio¹, M. Renda¹, C. Catalano¹, V. Cantisani¹; ¹Rome/IT, ² Catania/IT ([email protected]) Purpose or Learning Objective: To evaluate the diagnostic performance of CAD compared with TI-RADS systems and to compare th e performance of TI- RADS when used by operators with different levels o f experience.

Methods

or Background: Three operators with different levels of experience evaluated 484 thyroid nodules and the diagnostic ac curacy of three risk stratification systems (ACR-, EU-, K-TIRADS) and CA D software (S-Detect) in characterizing the nodules. Nodules were characteri zed and stratified by using the three TIRADS systems; then S-detect software wa s applied and the data were compared with each other and with the gold sta ndard (citology).

Results

or Findings: The sensitivity of the human operator measurement a nd the negative predictive value (NPV) is 100%, for al l three types of TIRADS. The positive predictive value (PPV) is 50%. The spe cificity is 78.4% (EU), 85.7% (ACR), 89% (K); this implies a certain propor tion of "false positives", especially in the use of the tirads EU. The sensiti vity of the measurement of s- detect alone is 66.7%, for all TRAIDS. This estimat e implies a low certainty of the negative result. The NPV is 96.2% (ACR, K) and 96.3% (EU). The PPV is 50% (ACR-K) and 66.7% (EU). The specificity is high er than the sensitivity: 92.7% (ACR - K) and 96.3% (EU).

Conclusion

S-DETECT combined with EU-TIRADS has similar result s as S- DETECT with ACR- and K- TIRADS in terms of sensitiv ity, specificity and NPV. However, it has a slightly better PPV, suggesting g reater accuracy in correctly diagnosing positive cases than the ACR- and K-class ification systems. S- Detect cannot yet be considered a substitute for th e human operator but is a valuable tool for characterizing thyroid nodules, w hen integrated with radiologist evaluation and for support tool for les s experienced operators and in doubtful cases

Limitations

Reduced patient sample. Ultrasound's operator depen dence Funding for this study: None Ethics committee - additional information: Ethics committee "Sapienza" Author Disclosures: Giovanni Del Gaudio: Nothing to disclose Maurizio Renda: Nothing to disclose Patrizia Pacini: Nothing to disclose Chiara Di Bella: Nothing to disclose Vincenzo Dolcetti: Nothing to disclose Carmen Solito: Nothing to disclose Vito Cantisani: Nothing to disclose Emanuele David: Nothing to disclose Carlo Catalano: Nothing to disclose Saturday Abstract-based Programme 248 Is Shear-Wave Elastography an Accurate Tool for Eva luating Nodules in Patients with Autoimmune Thyroiditis? *D. I. Stoian*, A. Borlea; Timisoara/RO ([email protected]) Purpose or Learning Objective: To evaluate the diagnostic accuracy of Shear Wave Elastography (SWE) in differentiating be nign from malignant nodules in patients with chronic autoimmune thyroid itis.

Methods

or Background: SWE has been widely studied in assessing thyroid nodules, but the background of thyroid lymphocitic infiltration and fibrosis may impact diagnostic performance. This study enrolled 130 subjects aged 18-84 years with a previous diagnosis of autoimmune thyro iditis and thyroid nodules. SWE measurements were performed on both thyroid par enchyma and nodules, assessing elasticity indices (EIs), includ ing mean and maximum values, and the nodule-to-thyroid (N/T) ratio. Conv entional ultrasound risk assessment using TIRADS was also evaluated. The rel ationship between elasticity indices and biochemical parameters and t hyroid volume was examined.

Results

or Findings: There were no statistically significant differences in thyroid function or autoimmunity parameters between benign and malignant nodules. Significant differences were found in TIRA DS scores (p < 0.0001), mean nodule EI (p < 0.0001), and N/T shear wave rat io (p < 0.0001). The mean nodule EI (47.2 kPa for malignant vs. 18.1 kPa for benign nodules) had the highest diagnostic performance, significantly o utperforming both the maximum EI (p = 0.0360) and the N/T ratio (p = 0.01 30). The mean nodule EI also demonstrated superior diagnostic accuracy comp ared to TIRADS (p = 0.0025).

Conclusion

SWE demonstrates high diagnostic accuracy in evalua ting nodules also in context of autoimmune thyroiditis. The mean nodule EI is the most reliable elastographic parameter, outperformin g other elasticity indices and TIRADS in distinguishing malignant from benign nodules.

Limitations

This study did not include a control group of nodul es without autoimmune thyroiditis, limiting comparisons of SWE performance in different thyroid backgrounds. Further studies are needed to assess reproducibility across other clinical settings. Funding for this study: There was no funding for the study Ethics committee - additional information: The studies involving humans were approved by The Ethics Committee of the Victor Babes University of Medicine and Pharmacy. The studies were conducted i n accordance with the local legislation and institutional requirements. T he participants provided their written informed consent to participate in this stu dy. Author Disclosures: Dana I Stoian: Nothing to disclose Andreea Borlea: Nothing to disclose Follow up or FNAB?: Malignancy Rates of ACR-TIRADS 4 and 5 Thyroid Nodules <10mm in Diameter *E. Tobcu*, E. Karavaş, Z. Tobcu, G. Taşova Yılmaz; Balıkesir/TR ([email protected]) Purpose or Learning Objective: The purpose of this study is to ascertain the rate of malignancy in nodules that are <10 mm in di ameter in the TR-4 and TR- 5 categories, as defined by the American College of Radiology Thyroid Imaging and Reporting Data System (ACR-TIRADS) 2017 whitepaper.

Methods

or Background: Assessment of thyroid nodules was conducted in accordance with the 2017 whitepaper of the ACR. The fine-needle aspiration biopsy (FNAB) procedure was performed under the gui dance of ultrasound. Lesions designated as Bethesda group 3, 4, 5, and 6 were classified as "non- benign group," while nodules defined as Bethesda gr oup 2 were classified as "benign group”. All patients underwent surgery, exc ept for those with benign cytology.

Results

or Findings: A total of 60 nodules were included in the study. B ased on cytological analysis, 12 of the 15 nodules class ified in the TR-4 category were determined to be benign thyroid nodules, while 3 nodules were classified as "suspicious for malignancy" (Bethesda-V) (%20) a fter cytological assessment. In the assessment of 45 TR-5 nodules, 3 0 were identified as non- benign (66.6%), while the remaining 15 were classif ied as benign thyroid nodules after cytological evaluation.

Conclusion

Our investigation demonstrated that <10mm thyroid n odules classified in categories TR-4 and TR-5, according t o ACR-TIRADS 2017 guideline, have a malignancy rate of 55%. We think that FNAB should be performed prior to the decision of active surveilla nce, in order to determine the PMTCs that exhibit aggressive cytologic features an d to identify those with benign cytology to avoid an unnecessary active surv eillance process.

Limitations

The research was conducted with a small number of p atients and was completed at a single center. Second, we did no t exclude patients with thyroiditis that may have an impact on the accuracy of FNAB. Funding for this study: The authors state that this study has not received any funding Ethics committee - additional information: This prospective study was approved by the institutional review board of our i n¬stitution (decision number: 2024/3-40), and any requirement of informed consent was waived. Author Disclosures: Eren Tobcu: Nothing to disclose Erdal Karavaş: Nothing to disclose Gülden Taşova Yılmaz: Nothing to disclose Zeynep Tobcu: Nothing to disclose Evaluating the diagnostic value of arterial enhance ment fraction from dual-layer spectral detector CT in lymph node metas tasis of papillary thyroid carcinoma *L. L. Ye*¹, X. Zheng¹, Y. Liao²; ¹Dongguan/CN, ²Gu angzhou/CN ([email protected]) Purpose or Learning Objective: To assess the diagnostic value of the arterial enhancement fraction (AEF) derived from dual-layer spectral detector CT scans in detecting lymph node metastasis in papilla ry thyroid carcinoma (PTC).

Methods

or Background: Preoperative spectral CT images of 58 lymph nodes from 25 PTC patients (7 males, 18 females; ag ed 28-74) confirmed by surgery and pathology were analyzed. All patients u nderwent lymph node dissection. Lymph nodes were classified into metast atic (N = 24) and non- metastatic (N = 34) groups based on pathology. AEF (defined as iodine concentration in the arterial phase / iodine concen tration in the venous phase) was measured. The Mann-Whitney U test was used to c ompare AEF between groups. ROC analysis evaluated AEF's predictive per formance for lymph node metastasis.

Results

or Findings: AEF was significantly higher in metastatic lymph no des than in non-metastatic ones (p<0.05). In terms of d ifferentiating lymph node metastasis, the AUC of AEF is 0.737 (95%CI 0.582-0. 891), with an accuracy of 0.810, a sensitivity of 0.583, and a specificity of 0.971.

Conclusion

AEF from dual-layer spectral detector CT demonstrat es significant differences between metastatic and non- metastatic lymph nodes in PTC and has diagnostic potential for identifying me tastatic lymph nodes, providing a basis for precise preoperative treatmen t planning.

Limitations

Not applicable Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Xiaolin Zheng: Consultant: mentor Li Li Ye: Speaker: speaker, main person in charge o f abstract Yuting Liao: Other: statistical researcher, unpaid Diagnostic of intravoxel incoherent motion diffusio n-weighted imaging histogram parameters in distinguishing between beni gn and malignant thyroid nodules *X. Li*, Y. Yue, J. Ren; Beijing/CN ([email protected]) Purpose or Learning Objective: To explore the diagnostic efficacy of histogram parameters of IVIM in differentiating ben ign and malignant thyroid nodules.

Methods

or Background: A total of 51 patients with thyroid nodules were retrospectively included from March 2017 to Septemb er 2022, including the benign group 24 cases and malignant group 27 cases . All results were confirmed by surgical pathology. multiple b-value s mall field diffusion weighted images were collected to generate true diffusion co efficient(D), pseudo diffusion coefficient(D*), and perfusion fraction(f ) images. At the same time, an apparent diffusion coefficient(ADC) image with a b- value of 990s/mm2 was generated. Manually outline the volume of interest( VOI) of the entire lesion. Using a self-designed program in Matlab, calculate the histogram parameters of the D, D *, and f at VOI, and calculate the mean ADC at VOI. Compare the intergroup differences between IVIM histogram param eters and ADC mean. Using multiple logistic regression, further select the optimal parameters and establish a multi parameter joint model and verify and compare the diagnostic performance of different models.

Results

or Findings: The statistically significant differences between b enign and malignant thyroid nodules are as follows(P<0.05 ): gender, ADC mean, 5th, 15th, 85th, and 95th percentile, mean, skewness, an d root mean square difference in the D-plot, 5th and 15th percentile p ercentile, mean, skewness, and coefficient of variation in the D * plot. The A UC values for skewness, 15th percentile, and root mean square difference of the D-plot combined with gender were 0.94, with a sensitivity of 88.46% and a specificity of 90.91%. The AUC value of the mean ADC combined with gender is 0 .86, with a sensitivity of 77.78% and a specificity of 87.50%.

Conclusion

Histogram parameters of IVIM has good diagnostic va lue in distinguishing benign and malignant thyroid nodules .

Limitations

Not applicable. Saturday Abstract-based Programme 249 Funding for this study: Not applicable. Ethics committee - additional information: This is a retrospective study. Author Disclosures: Yunlong Yue: Nothing to disclose Xingpeng Li: Nothing to disclose Jie Ren: Nothing to disclose New thyroid imaging reporting and data system (TI-R ADS) based on ultrasonography features for follicular thyroid neo plasms: A multicenter study *Y. Zhang*; Shanghai/CN ([email protected]) Purpose or Learning Objective: This study aimed to establish a new risk stratification system for FTN and new methods for n on-invasive and practical preoperative evaluation of thyroid follicular tumor s to reduce missed diagnoses and unnecessary biopsies.

Methods

or Background: 535 FTNs of 535 patients from four hospitals were included in this retrospective study. All the nodul es were randomly divided into test and validation groups. FTN-TIRADS was establis hed based on the results of univariate analysis and logistic regression of u ltrasonography features in the test group. Each nodule was evaluated and classifie d by existing risk stratification systems (EU-RSS, ATA-RSS, ACR-TIRADS , Chinese TIRADS [C- TIRADS]) and FTN-TIRADS. The diagnostic value of FT N-TIRADS in the validation group was verified and compared with the test group and the other four risk stratification systems. The unnecessary r ates of fine needle aspiration (FNA) of FTN-TIRADS and the other four risk stratif ication systems were compared, too.

Results

or Findings: Test group and validation group included 370 patien ts and 165 patients. The following features were indep endent risk factors and included in FTN-TIRADS: nodule composition, echogen icity, calcifications, halo sign, and indistinct boundary with thyroid capsule. The AUC of FTN-TIRADS was 0.855, statistically higher than EU-RSS, ATA-RS S, ACR-TIRADS, and C- TIRADS (0.759, 0.759, 0.753, 0.677, respectively). The FTN-TIRADS of the validation group had a similar diagnostic performan ce. The unnecessary FNA rate of the FTN-TIRADS was 26.0%, which was signifi cantly lower than that of EU-RSS (79.9%), ATA-RSS (92.5%), ACR-TIRADS (55.8%) and C-TIRADS (62.2%).

Conclusion

FTN-TIRADS achieved better differential diagnosis o f FTN than current risk stratification systems and significant ly reduced the rate of unnecessary FNA.

Limitations

The study was performed retrospectively; a more con clusive validation could be achieved using a prospective de sign that better captured the diverse characteristics of actual clinical case s. Funding for this study: This work was supported by the National Natural Science Foundation of China (Grants No. 81927801, 8 1725008, 81772849, 82171942, and 82371971) Ethics committee - additional information: The institutional review board of the university-affiliated hospital approved this re trospective study (approval number 22K82) Author Disclosures: Yifeng Zhang: Nothing to disclose 16:00-17:30 Research Stage 2 Research Presentation Session: Oncologic Imaging RPS 2116 Advances in imaging genitourinary cancer Moderator T. Akbas; Istanbul/TR ([email protected]) Total bone diffusion volume on whole-body diffusion -weighted imaging is a strong prognostic marker of disease survival in m etastatic castration- resistant prostate cancer *L. D'Erme*¹, A. Candito², G. Avesani¹, S. Bottazzi ¹, R. Emsley², D. Meo², J. Carmichael², N. Tunariu², D-M. Koh²; ¹Rome/IT, ² London/UK ([email protected]) Purpose or Learning Objective: To investigate the relationship between total bone diffusion volume (tBDV) and global apparent di ffusion coefficient (gADC) derived from whole-body diffusion-weighted MRI (WBD WI); as well as automatic bone scan index (aBSI) derived from bone scintigraphy with disease overall survival (OS) in metastatic castration-resi stant prostate cancer (mCRPC).

Methods

or Background: In this IRB approved study, we retrospectively reviewed 302 mCRPC patients (Jan 2015 - Dec 2023) w ho underwent baseline WBDWI before systemic anticancer treatment. Segment ation masks of bone disease on b900 WBDWI images were generated by an a utomated tool, and refined by a 3-year experienced oncological radiolo gist, to derive the tBDV and gADC values. The aBSI was derived in 265 patients w ith available baseline bone scintigraphy. Kaplan-Meier survival curves and log-rank tests for 5-years OS, including their hazard ratios (HR) and 95% conf idence intervals (CI), were compared between the three groups, stratified by th e median values of tBDV (89 mL), gADC (0.83), and aBSI (0.021). Significanc e was set at p < 0.05.

Results

or Findings: Patients with tBDV < 89 mL demonstrated significant ly longer OS compared to those with tBDV ≥ 89 mL (40.8 vs 23.7 months; p < 0.0001; HR 2.28, 95%CI 1.76–2.94). No significant s urvival difference was observed between the gADC groups (32.0 vs 27.2 mont hs; p = 0.4293). OS was significantly longer in patients with aBSI < 0. 021 than those with aBSI ≥ 0.021 (40.8 vs 24.4 months; p < 0.0001; HR 2.18, 95 %CI 1.66–2.85).

Conclusion

WBDWI-derived tBDV is a strong independent prognost ic marker for OS in mCRPC patients. The tBDV measurement is c omparable if not superior to aBSI as a predictor of disease survival .

Limitations

Retrospective design and longitudinal imaging data were not considered. Funding for this study: This study represents independent research funded by the National Institute for Health and Care Resea rch (NIHR) Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, London, and by the Ro yal Marsden Cancer Charity, and Cancer Research UK (CRUK) National Can cer Imaging Trials Accelerator (NCITA) and Prostate Cancer UK. The vie ws expressed are those of the author(s) and not necessarily those of the N IHR or the Department of Health and Social Care. This work uses data provide d by patients and collected by the NHS as part of their care and supp ort. Ethics committee - additional information: The study was approved by the Institutional Ethics Committee (no. 21/LO/0605). Author Disclosures: Giacomo Avesani: Nothing to disclose Dow-Mu Koh: Nothing to disclose Davide Meo: Nothing to disclose Nina Tunariu: Nothing to disclose Robby Emsley: Nothing to disclose Silvia Bottazzi: Nothing to disclose Antonio Candito: Nothing to disclose Luca D'Erme: Nothing to disclose Juliet Carmichael: Nothing to disclose Why we shouldn’t trust CT in the evaluation of bone metastases in patients with metastatic prostate cancer: a compari son between pattern of changes on CT and bone metastases MET-RADS-P cla ssification *S. Bottazzi*¹, L. Russo¹, G. Avesani¹, L. D'Erme¹, C. Messiou², D-M. Koh², E. Sala¹, N. Tunariu²; ¹Rome/IT, ²Sutton/UK ([email protected]) Purpose or Learning Objective: To compare changes on computed tomography (CT) with the MET-RADS-P response assess ment categories (RAC) of bone metastases in advanced prostate cance r patients (APCb) during treatment.

Methods

or Background: 102 patients (median age 68years, range 51-83) with APCb who underwent both CT and whole-body magn etic resonance imaging (WBMRI) within 30 days at baseline and duri ng treatment were included. Up to five focal lesions > 10 mm per pati ent were selected based on one or more of the following: [1] Sclerotic, lytic or mixed lesion on CT; [2] active bone marrow lesion on WBMRI [3] newly developed CT or MR lesion. Each lesion was assigned a CT pattern of change - based on changes in size and Hounsfield Unit (HU) – and a RAC according to METRA DS-P criteria. The CT patterns were corroborated with the RAC, grouped as response (RAC1-2), stable (RAC3), and progression (RAC4-5).

Results

or Findings: 358 lesions were identified. Of these, 70% (252/358 ) were sclerotic (SL), 6% (21/358) lytic (LL) and 4% (13/258) mixed lesions (ML). 20% (72/358) showing MR characteristics of active b one metastases were undetectable on CT. The most frequent CT patterns o f change on treatment were: stable SL (no changes in density or size) in 35.8% (128/358), increasing in size SL (>5 mm) in 11.4% (41/358), new SL (appea red during treatment) in 10.6% (38/358). Stable SL corresponded to the RAC c lassifications as follows: 22.3% (29/128) stable treated disease, 25.8% (33/12 8) responding, 23.4% progressing (30/128) and 28.1% (36/128) stable acti ve disease. New SLs corresponded to responding disease in 28.9% (11/38) .

Conclusion

20% of bone metastases are occult on CT. A stable S L on CT is a poor predictor of disease status. Hence, CT appea rs unreliable in the assessment of bone disease response inAPCb

Limitations

Na Saturday Abstract-based Programme 250 Funding for this study: This study represents independent research funded by the National Institute for Health and Care Resea rch (NIHR) Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, London, and by the Ro yal Marsden Cancer Charity, and Cancer Research UK (CRUK) National Can cer Imaging Trials Accelerator (NCITA) and Prostate Cancer UK. The vie ws expressed are those of the author(s) and not necessarily those of the N IHR or the Department of Health and Social Care. This work uses data provide d by patients and collected by the NHS as part of their care and supp ort. Ethics committee - additional information: Reference no. 21/LO/0605 Author Disclosures: Giacomo Avesani: Nothing to disclose Christina Messiou: Nothing to disclose Dow-Mu Koh: Nothing to disclose Nina Tunariu: Nothing to disclose Silvia Bottazzi: Nothing to disclose Evis Sala: Nothing to disclose Luca D'Erme: Nothing to disclose Luca Russo: Nothing to disclose Cancer Detection Rates in DWI-upgraded Transition Z one lesions align with risk assessment categories in PI-RADS v2.1: a Systematic Review and Meta-analysis *G. Agrotis*, E. H. P. Pooch, R. G. H. Beets-Tan, I . G. Schoots; Amsterdam/NL ([email protected]) Purpose or Learning Objective: To assess and compare cancer detection rates (CDRs) of transition zone (TZ) lesions that w ere upgraded from PI- RADSv2.1 score 2 to 3("2+1") or from score 3 to 4(" 3+1") using diffusion- weighted imaging (DWI) and evaluate their clinical impact.

Methods

or Background: A systematic literature search was conducted in Embase, Medline and Web of Science for studies eval uating TZ lesions with the use of DWI, with histology proven Grade Group ≥2 cancer (GG≥2) as primary outcome. Pooled estimates for sensitivity, specificity, CDRs, and Odds Ratio (OR) were derived from extracted data at lesi on level and quantitatively pooled using a bivariate binomial and random effect s model.

Results

or Findings: A total of 7 studies included 1,437 TZ lesions. GG ≥2 CDRs for PI-RADSv2.1 scores of 1, 2, 2+1, 3, 3+1, 4 , and 5 were respectively 2%[95% CI:0%-12%], 7%[4%-11%], 12%[6%-24%], 21%[18% -25%], 37%[23%-53%], 53%[33%-72%], and 86%[40%-98%]. GG ≥2 CDRs of TZ scores '2+1' and '2' were statistically different, with OR 3.13[1.31-7.48],p=0.01, while '2+1' and '3' scores were not, with an OR of 0.76[0.42-1.33],p=0.34. GG≥2 CDRs of TZ score '3+1' and '3' were statistically different, with an OR of 2.3[1.07-4.95],p=0.03, while scores '3+1' and '4' w ere not with an OR of 0.63[0.28-1.38],p=0.25. Still, false positive rates were substantial in both subcategories ('2+1': 76%[73.8%-78.2%] and '3+1': 4 5%[42.4%-47.6%]).

Conclusion

The risk of having significant prostate cancer in ‘ 2+1’ and ‘3+1’ Transition Zone lesions, with an upgrading based on DWI images, is appropriately categorized within the PI-RADS v2.1 s coring system, as shown by this meta-analysis. Especially TZ lesions with s core ‘3+1’ may impact individualized biopsy-decisions, as 2-in-5 harbor s ignificant disease, similar to score ‘4’ lesions. Still, the high false positive r ate in this sub-category emphasizes the need for strategies to minimize over diagnosis.

Limitations

Data availability and population differences Funding for this study: None Ethics committee - additional information: Not applicable Author Disclosures: Georgios Agrotis: Nothing to disclose Eduardo H. P. Pooch: Nothing to disclose Regina G. H. Beets-Tan: Nothing to disclose Ivo Gerardus Schoots: Nothing to disclose Evaluation of Arterial Enhancement Fraction and Ext racellular Volume Fraction from Dual-Layer Spectral CT for Typing and Grading Renal Cell Carcinoma *X. Zhang*¹, G. Zhang¹, H. Sun¹, Z. Jin¹, X. Lu², S -H. Yu¹, L. Xu³, J. Zhang¹, X. Bai¹; ¹Beijing/CN, ²Shenyang/CN, ³Hangzhou/CN ([email protected]) Purpose or Learning Objective: To explore the value of arterial enhancement fraction (AEF) and extracellular volume fraction (E CV) obtained from dual-layer spectral CT in the typing and grading of renal cell carcinoma (RCC).

Methods

or Background: In this retrospective study, patients with pathologically confirmed RCC who has undergone dual -layer spectral CT were included. RCC was classified into non-clear cell (n on-ccRCC) and clear cell (ccRCC). The ccRCC cases were further categorized a s high-grade or low- grade based on the WHO/ISUP grading system. AEF and ECV parameter maps were generated from both contrast-enhanced and iodine concentration (IC) images, producing quantitative parameters AEFH U, ECVHU, AEFIC, and ECVIC. Receiver operating characteristic curves wer e used to evaluate the ability of these parameters in RCC typing and gradi ng.

Results

or Findings: The study included 68 patients, comprising 13 with non- ccRCC and 55 with ccRCC. CcRCC showed higher values of AEFHU, ECVHU, AEFIC, and ECVIC compared with non-ccRCC. The multi variate model comprising AEFIC, and ECVIC demonstrated the highes t diagnostic accuracy for ccRCC, with an area under curve (AUC) of 0.822, sensitivity of 83.6%, and specificity of 76.9%. Among the ccRCC cases, 34 wer e low-grade and 21 were high-grade. High-grade ccRCCs exhibited significant ly higher ECVHU and ECVIC than low-grade tumors. The multivariate model with tumor diameter, and ECVIC achieved the highest diagnostic accuracy in identifying high-grade ccRCC, with an AUC of 0.909, sensitivity of 90.5%, and specificity of 76.5%.

Conclusion

AEF and ECV derived from dual-layer spectral CT can help distinguish ccRCC from non-ccRCC. Additionally, ECV can accurately identify high-grade ccRCC, offering valuable insights for RC C.

Limitations

Given the retrospective design and relatively small sample size of the present study, further studies should aim to in clude larger cohorts and consider prospective data collection to validate th ese findings. Funding for this study: This study has received funding by the National Hig h Level Hospital Clinical Research Funding [2022-PUMC H-A-033]; the Natural Science Foundation of Beijing Municipality [L232133 ]; the Chinese Academy of Medical Sciences Initiative for Innovative Medicine [2022-I2M-C&T-B-019]; National High Level Hospital Clinical Research Fund ing [2022-PUMCH-A-035]; National High Level Hospital Clinical Research Fund ing [2022-PUMCH-B-069]. Ethics committee - additional information: The study was conducted in accordance with the principles of the Declaration o f Helsinki and approved by the institutional Research Ethics Committee Author Disclosures: Jiahui Zhang: Nothing to disclose Gumuyang Zhang: Nothing to disclose Xiaomei Lu: Nothing to disclose Hao Sun: Nothing to disclose Lili Xu: Nothing to disclose Xin Bai: Nothing to disclose Zhengyu Jin: Nothing to disclose Xiaoxiao Zhang: Nothing to disclose Sheng-Hui Yu: Nothing to disclose Prognostic role of Whole-body MRI (WB-MRI) in patie nts with metastatic prostate cancer receiving systemic anti-cancer ther apy *C. Sattin*¹, C. Pizzi¹, F. Arnone¹, P. Hoxha¹, D. Berloco¹, F. Zugni¹, P. Summers¹, A. R. R. Padhani², G. Petralia¹; ¹Mila n/IT, ²Northwood/UK ([email protected]) Purpose or Learning Objective: To investigate the potential of the response assessment category (RAC) from MET-RADS-P guideline s as prognostic biomarker in metastatic castrate resistant prostate cancer (mCRPC) patients.

Methods

or Background: We enrolled mCRPC patients who underwent whole-body MRI at baseline and at each time point ( every 12 weeks disease until progression) after systemic anti-cancer thera py (SACT). We correlated the maximum RAC at time point 1 (TP1) with overall surv ival (OS). Patients were divided in two groups: those with a maximum RAC 1-2 (highly likely or likely to be responding, respectively) and those with a maxim um RAC 3-4-5 (stable disease, likely or highly likely to be progressing) at TP1. Survival curves were depicted in Kaplan-Meier plots and compared via a l og-rank test and hazard ratio (HR) using Cox regression model, with point c omparisons of three-year survival and median survival duration, using R.

Results

or Findings: Out of 31 mCRPC patients enrolled, a higher OS was observed in patients with a maximum RAC 1-2 (N=11) than in those with a maximum RAC 3-4-5 (N=21) at TP1 (log-rank test p=0. 005): median 34 months (lower bound 95%CI = 27 months) vs median 12 months (95%CI 11-28 months). The HR for the RAC 3-4-5 patients was 1.34 (95%CI 0.83 – 1.85, p= 0.009). Three-year OS was 30.3% for RAC1-2 vs 5.3% for RAC 3-4-5, for a difference of 25.0% (95%CI -11.1% - 61.2%, p=0.175) .

Conclusion

Our observations support the potential of RAC after TP1 as a prognostic biomarker in mCRPC undergoing SACT.

Limitations

Retrospective and monocentric study. Funding for this study: No fundings Ethics committee - additional information: Not applicable Author Disclosures: Fabio Zugni: Nothing to disclose Francesca Arnone: Nothing to disclose Giuseppe Petralia: Nothing to disclose Caterina Pizzi: Nothing to disclose Paolo Hoxha: Nothing to disclose Paul Summers: Nothing to disclose Professor Anwar R. R Padhani: Nothing to disclose Caterina Sattin: Nothing to disclose Donatello Berloco: Nothing to disclose Saturday Abstract-based Programme 251 Prognostic value of WB-MRI derived Bone Marrow Adip ose Tissue (BMAT) in bone metastatic prostate cancer patients treated with androgen deprivation + enzalutamide +/- zoledronic acid *N. Di Meo*, C. Buizza, P. Rondi, A. Dalla Volta, A . Borghesi, M. Ravanelli, A. Berruti, D. Farina; Brescia, BS/IT ([email protected]) Purpose or Learning Objective: To assess the prognostic significance of bone marrow adipose tissue (BMAT) in prostate cance r patients with hormone- sensitive bone metastases undergoing whole-body MRI (WB-MRI) and receiving enzalutamide treatment.

Methods

or Background: Imaging was conducted on a 1.5T MRI scanner using a MET-RADS-P-compliant protocol. Manual singl e-slice segmentation of fat fraction (FF%) sequences was performed by one o perator (R1) at the L3 vertebral level and across three contiguous slices at the femoral head. WB- MRI was performed at baseline and at 6 and 12 month s following the initiation of therapy. Absolute BMAT values and temporal chang es were recorded and correlated with survival outcomes.

Results

or Findings: Of the 126 patients enrolled in this prospective ph ase 2 clinical trial, 100 were available for analysis. No correlation was found between BMAT measurements at the L3 vertebra and the femora l head, with the latter showing significantly higher values (90.7% vs. 63.9 %, respectively). A significant positive correlation was identified bet ween baseline L3 BMAT and both progression-free survival (PFS) and overall su rvival (OS), with hazard ratios (HR) of 0.37 and 0.33, respectively, after a median split. Additionally, early changes in L3 BMAT were inversely associated with PFS and OS, with HRs of 1.89 and 2.96, respectively. BMAT at the fem oral head was not associated with survival outcomes.

Conclusion

L3 BMAT is a valuable prognostic and predictive bio marker that can be easily derived from WB-MRI. It may contribut e to more personalized treatment strategies for patients with metastatic p rostate cancer.

Limitations

No external validation. Funding for this study: No Funding Ethics committee - additional information: No Author Disclosures: Andrea Borghesi: Nothing to disclose Chiara Buizza: Nothing to disclose Davide Farina: Nothing to disclose Paolo Rondi: Nothing to disclose Marco Ravanelli: Nothing to disclose Alberto Dalla Volta: Nothing to disclose Nunzia Di Meo: Nothing to disclose Alfredo Berruti: Nothing to disclose Conventional Parameters of Periprostatic Fat on 18F -PSMA-1007 PET/CT: A Novel Biomarker for Predicting High ISUP Grade an d Short-Term Prognosis in Prostate Cancer *L. Chen*, Y. Yang, F. Yao; Wenzhou/CN ([email protected]) Purpose or Learning Objective: This study aimed to investigate the value of periprostatic fat area and 18F-PSMA-1007 uptake in predicting high ISUP grade and postoperative PSA persistence in prostate cancer patients using 18F-PSMA-1007 PET/CT.

Methods

or Background: A retrospective analysis was conducted on clinical data and 18F-PSMA-1007 PET/CT data of 350 prostate cancer patients. 3D- Slicer and Lifex software were utilized for delinea ting the region of interest for periprostatic fat and measuring periprostatic fat a rea and 18F-PSMA-1007 uptake. The primary outcome of this study was the I SUP grade greater than 3 based on surgical pathological results of radical p rostatectomy. The secondary outcome was postoperative PSA persistence, defined as routine follow-up tPSA > 0.1 ng/ml. Logistic regression analyses were performed to assess the association between characteristics and outcomes an d construct predictive models. Receiver operating characteristic curves we re utilized to determine optimal cutoff values and evaluate model performanc e.

Results

or Findings: Larger periprostatic fat area emerged as an indepen dent risk factor for higher ISUP grade (p < 0.001) and p ostoperative PSA persistence (p = 0.009) in prostate cancer patients . Higher 18F-PSMA-1007 uptake was also closely associated with higher ISUP grade (p < 0.001) and postoperative PSA persistence (p < 0.001). Models r espectively established to predict higher ISUP grade and postoperative PSA per sistence showed good predictive performance, with AUC values of 0.736 an d 0.745.

Conclusion

Larger periprostatic fat area and higher 18F-PSMA-1 007 uptake are independent risk factors for high ISUP grade an d postoperative PSA persistence, which can be used to predict high ISUP grade and the persistence of PSA.

Limitations

This is a small sample study. The generalizability of the results requires further consideration. Funding for this study: This study was supported by the Wenzhou Major Program of Science and Technology Innovation (Grant No. ZY2020012). Ethics committee - additional information: This retrospective study has been reviewed and approved by the the first affilia ted hospital of Wenzhou Medical University ethics committee. Author Disclosures: Yunjun Yang: Nothing to disclose Fei Yao: Nothing to disclose Lixuan Chen: Nothing to disclose Decrease in kidney volume predicts loss of renal fu nction in prostate cancer patients receiving LuPSMA treatment *F. Jungmann*, L. Steinhelfer, M. R. Makowski, M. E iber, R. Braren; Munich/DE ([email protected]) Purpose or Learning Objective: Lutetium-177 (177Lu) prostate-specific membrane antigen (PSMA) radioligand therapy (RLT) i s a novel treatment option for metastatic, castration-resistant prostat e cancer (mCRPC). Evidence is increasing that nephrotoxicity is a delayed side effect in a considerable fraction of patients. The purpose of this study was to identify prognostic markers for clinically significant deterioration of kidney function in patients undergoing 177Lu-PSMA RLT.

Methods

or Background: Total kidney volume (TKV) at 3 and 6 months following 177Lu-PSMA RLT was extracted from routine clinical CT scans using deep learning. A cut-off at ≥30% eGFR decline was defined as clinically significant deterioration of kidney function, given its indication as a substantial risk of end-stage renal disease. Differences betwee n patients developing an eGFR decline of ≥30% after 12 months and those who did not consideri ng baseline renal parameters, their relative changes ( ∆%), nephrotoxic risk factors, and the number of 177Lu-PSMA cycles were a nalyzed. Furthermore, distinct threshold values of significant features t o differentiate between the two patient groups were identified based on ROC analysi s using the Youden-Index.

Results

or Findings: A ≥10% decrease in TKV at six months predicted a severe eGFR decline of ≥30% at 12 months with high diagnostic accuracy (ROC-AUC of 0.90), surpassing all other parameters. Baseline risk factors, the number of prior treatment regimens and 177Lu-PSMA c ycles did not correlate with a higher eGFR decrease at 12 months.

Conclusion

Our retrospective analysis demonstrates the feasibi lity of fully automated kidney volume assessment from routine cli nical imaging data to predicting significant deterioration of kidney func tion at 12-month after 177Lu- PSMA RLT in mCRPC. It is more accurate than early r elative eGFR change and might contribute as a non-invasive biomarker wh en treatment decisions are pending including determining whether to contin ue/discontinue or adapt 177LuPSMA treatment.

Limitations

Retrospective, single-center Funding for this study: None Ethics committee - additional information: Ethical approval for this retrospective, HIPAA-compliant analysis was obtaine d from the local institutional review boards. The requirement for in formed consent was waived because of its retrospective design. Author Disclosures: Rickmer Braren: Nothing to disclose Marcus R. Makowski: Nothing to disclose Matthias Eiber: Nothing to disclose Friederike Jungmann: Nothing to disclose Lisa Steinhelfer: Nothing to disclose Deep learning-accelerated MRI Imaging in Patients w ith Prostate Cancer and Benign Prostatic Hyperplasia *V. Koch*, T. Vogl, R. Strecker, C. Booz, S. Mahmou di, L. D. Grünewald; Frankfurt/DE ([email protected]) Purpose or Learning Objective: The purpose of this study was to investigate the impact of deep learning-accelerated T2-weighted MRI imaging of prostate cancer and benign prostatic hyperplasia (BPH).

Methods

or Background: In this prospective study, adults who underwent 3- Tesla MRI of the prostate due to suspicion of prost ate cancer or benign prostatic hyperplasia were included. Standard seque nces were acquired according to a dedicated protocol compromising T1-, T2-, and diffusion- weighted imaging sequences. Additionally, T2-weight ed imaging sequences using the deep learning algorithm (T2DL) were acqui red in axial, coronal, and sagittal planes. Quantitative analysis encompassed time efficiency and

Objective

imaging parameters, including signal-to-n oise ratio (SNR) and contrast-to-noise ratio (CNR). Qualitative evaluati on was independently performed by three blinded radiologists to assess d iagnostic confidence, image quality, and lesion sharpness subjectively. Interre ader agreement was calculated using Fleiss κ. Saturday Abstract-based Programme 252

Results

or Findings: A total of 46 male patients (mean age, 70 ± 9 years ) were included. The study cohort encompassed 22 pati ents (48%) with prostatic cancer and 24 patients (52%) with BPH. Subjective e valuation of T2DL- sequences among all three readers revealed slightly superior diagnostic confidence, image quality, and lesion sharpness whe n compared to standard T2w sequences. Especially regarding focal lesions, T2DL-sequences allowed for significantly sharper demarcation with higher d iagnostic confidence in cancer diagnosis. Objective image analysis of T2DL revealed significantly higher SNR and CNR values when compared to conventi onal T2w-sequences. Acquisition times of T2w-sequences (axial, coronal, and sagittal plane) could be reduced by an average of 50 % using T2DL-sequenc es.

Conclusion

Our findings suggest that deep learning-accelerated T2w- sequences in MRI imaging of the prostate allow a re levant reduction in acquisition time while maintaining both subjective and objective image quality.

Limitations

Single-center study. Funding for this study: No funding. Ethics committee - additional information: Approval obtained. Author Disclosures: Christian Booz: Nothing to disclose Thomas Vogl: Nothing to disclose Vitali Koch: Nothing to disclose Scherwin Mahmoudi: Nothing to disclose Ralph Strecker: Nothing to disclose Leon David Grünewald: Nothing to disclose Feasibility of arterial spin labelling in MRI evalu ation of adnexal lesions: A comparative study with dynamic contrast enhanceme nt imaging *C. Meinzer*¹, K. Zhang¹, R. Gnirs¹, O. Zivanovic¹, H-U. Kauczor¹, H-P. Schlemmer¹, F. Kurz², T. Mokry¹; ¹Heidelberg/D E, ²Geneva/CH ([email protected]) Purpose or Learning Objective: To evaluate feasibility of arterial spin labeling (ASL) as a non-contrast MRI technique for assessing solid tissue of adnexal lesions and to compare its performance with dynamic contrast enhanced (DCE) MRI.

Methods

or Background: We prospectively included 11 adnexal lesions with solid tissue in nine females. Regions of interest ( ROIs) were annotated on DCE images, and then transferred anatomically to corres ponding sites on ASL images. From these ROIs, we extracted semi-quantita tive DCE parameters: area under the curve (AUC), relative area under the curve (relAUC), peak enhancement, time to peak, mean residence time, are a under the first moment curve, and wash-in-rate (WiR). From ASL perfusion m aps, we obtained mean adnexal blood flow (ABF). Correlation between ABF a nd DCE parameters was assessed using Pearson’s correlation coefficient. F or those parameters showing significant correlation, Bland-Altman plots were generated to evaluate agreement.

Results

or Findings: Pearson's correlation revealed significant correlat ions between ABF and two DCE parameters: relAUC (r=-0.75 , p=0.008), WiR (r=0.65, p=0.031). Other analysed DCE parameters sh owed no statistically significant correlations with ABF (p>0.05). Bland-A ltman analysis was performed for the significantly correlating paramet ers. For relAUC, mean difference was 135.85 (SD=18.04). No data points fe ll outside the limits of agreement, indicating good agreement between ASL an d DCE. Similarly, WiR showed a mean difference of -17.02 (SD=11.24), and no points outside the limits of agreement.

Conclusion

ABF demonstrated significant correlations with relA UC and WiR, indicating that ASL can provide comparable perfusio n information to DCE for these metrics. The Bland-Altman analysis further su ggests reasonable agreement between ASL-derived perfusion and DCE par ameters for relAUC and WiR.

Limitations

The small sample size limits the generalisability o f the findings. ASL MRI is susceptible to lower signal-to-noise rat ios and variability in perfusion measurements, which can impact accuracy a nd reproducibility. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study received institutional review board approval and written informed consent was obtained from all participants (S-337/2016). Author Disclosures: Oliver Zivanovic: Nothing to disclose Clara Meinzer: Nothing to disclose Ke Zhang: Nothing to disclose Theresa Mokry: Nothing to disclose Regula Gnirs: Nothing to disclose Hans-Ulrich Kauczor: Nothing to disclose Felix Kurz: Nothing to disclose Heinz-Peter Schlemmer: Nothing to disclose 16:00-17:30 Research Stage 3 Research Presentation Session: Breast RPS 2102 How to optimise and use breast MRI Moderator N. Sharma; Leeds/UK ([email protected]) Diagnostic accuracy of abbreviated magnetic resonan ce imaging for breast cancer screening: a multi-reader study *S. V. Grinsven*¹, R. Mann², K. M. Duvivier³, M. De Jong⁴, P. K. De Koekkoek-Doll³, C. Loo³, J. Veltman ⁵, W. B. Veldhuis¹, For The Dense Trial Study Group¹; ¹Utrecht/NL, ²Nij megen/NL, ³Amsterdam/NL, ⁴Den Bosch/NL, ⁵Almelo/NL ([email protected]) Purpose or Learning Objective: Costs and time of a full multi-parametric MRI protocol may be reduced by using an abbreviated MRI (AB-MRI) protocol. The DENSE trial’s multiparametric protocol provided the unique opportunity to study the accuracy of various AB-MRI protocols, to identify the minimal protocol necessary to maintain high diagnostic accu racy.

Methods

or Background: Seven radiologists performed incremental reads of a subset of 518 MRI examinations from the DENSE tri al (women with extremely dense breasts and negative mammography). Different sequences were added in four incremental steps, starting with : 1) both high resolution (hi- res) and ultra-fast T1-weighted images (T1WI), up t o 120 seconds after contrast-injection only, 2) complemented by diffusi on-weighted images (DWI), 3) T2-weighted images (T2WI), and 4) finally adding all remaining full protocol sequences: non-fatsat-T1-weighted pre-contrast imag es, all remaining dynamic phases, and curve-kinetics. Each radiologist assess ed the same MRI examinations and provided BI-RADS scores for all fo ur steps. We calculated the pooled sensitivity and specificity per incremen tal step by using a generalized estimating equation model, and the pool ed reading time per incremental step by using a linear mixed model.

Results

or Findings: The sensitivity and specificity of the most abbrevi ated MRI protocol (step 1) were not significantly differ ent from that of the full multiparametric MRI protocol (step 4) (p=0.68, p=0. 39). The pooled reading time of step 1 was almost 50% shorter than that of the full multiparametric MRI protocol (p<0.01), and the MR acquisition time was 70-80% shorter, depending on the hospital and scanner vendor.

Conclusion

In a screening setting, a full multiparametric MRI protocol, including pre-contrast DWI and T2WI, and delayed po st-contrast T1WI, did not provide significant additional diagnostic informati on for making a recall/no- recall decision compared to an ultrafast bi-dynamic T1WI-only protocol.

Limitations

A prospective screening study should confirm these results. Funding for this study: The DENSE trial is financially supported by the University Medical Center Utrecht (UMC Utrecht, Pro ject number: UMCU DENSE), the Netherlands Organization for Health Res earch and Development (ZonMw, Project numbers: ZONMW-200320002-UMCU and Z onMW Preventie 50-53125-98-014), the Dutch Cancer Society (KWF Kan kerbestrijding, Project numbers: DCS-UU-2009-4348, UU-2014-6859 and UU-2014 -7151), the Dutch Pink Ribbon / a Sister’s hope (Project number: Pink Ribbon-10074), Bayer AG Pharmaceuticals, Radiology (Project number: BSP-DEN SE), and Stichting Kankerpreventie Midden-West. For research purposes, Volpara Health Technologies (Wellington, New Zealand) has provided Volpara Imaging Software 1.5 for installation on servers in the scr eening units of the Dutch screening program. Ethics committee - additional information: On November 11, 2011, the trial was approved by the Dutch Minister of Health, Welfa re, and Sport, under advisement from the Health Council of the Netherlan ds. Author Disclosures: Mathijn De Jong: Nothing to disclose Petra Katharina De Koekkoek-Doll: Nothing to disclo se Wouter B. Veldhuis: Nothing to disclose Katya M. Duvivier: Nothing to disclose Jeroen Veltman: Nothing to disclose For The Dense Trial Study Group: Nothing to disclos e Sophie Van Grinsven: Nothing to disclose Claudette Loo: Nothing to disclose Ritse Mann: Nothing to disclose Saturday Abstract-based Programme 253 Breast MRI protocol strategies to reduce energy con sumption and carbon emissions: phantom and patient tests *J. T. Lee*¹, B. K. Seo¹, M. S. Bae¹, H. Choi², K. R. Cho², O. Woo², S. E. Song², S-Y. Kim², S. Cheon¹; ¹Ansan/KR, ²Seoul/KR ([email protected]) Purpose or Learning Objective: Environmental sustainability in healthcare is crucial, and MRI is a major energy-intensive device in radiology. We aimed to identify optimal energy-saving breast MRI protocols by comparing energy consumption and carbon emissions of abbreviated MRI , ultrafast dynamic contrast-enhanced MRI (Ultrafast-DCE), and artifici al intelligence (AI)-assisted protocols against multiparametric MRI, conventional dynamic contrast- enhanced MRI (Conventional-DCE), and non-AI-assiste d protocols using patient and phantom tests.

Methods

or Background: A 3-T MRI scanner equipped with a dedicated breast coil and a power meter providing a 1-Hz samp ling rate was used. We compared scan time (seconds), total energy (kW), en ergy consumption (kWh), and carbon emissions per scan (kgCO2e) between abbr eviated (n=74) and multiparametric (n=81) protocols, Ultrafast-DCE (n= 81) and Conventional-DCE (n=81), and AI-assisted (n=76) and non-AI-assisted (n=76) protocols in 307 patients. Additionally, the signal-to-noise ratio w as compared between AI- assisted and non-AI-assisted protocols using a brea st MRI phantom. The abbreviated protocol included T2-weighted imaging ( T2), diffusion-weighted imaging (DWI) (b values: 0 and 800 s/mm²), four-pha se DCE T1-weighted imaging (T1), and axillary T1. The multiparametric protocol consisted of T2, DWI (b values: 0, 800, and 1400 s/mm²), Ultrafast-D CE, five-phase DCE T1, and axillary T1.

Results

or Findings: Abbreviated MRI reduced scan time by 40%, total energy by 36%, and energy/carbon emissions by 62% c ompared to multiparametric MRI. Ultrafast-DCE reduced scan tim e by 81%, total energy by 83%, and energy/carbon emissions by 97% compared to Conventional-DCE. AI-assisted MRI reduced scan time by 29%, total ene rgy by 30%, and energy/carbon emissions by 52% compared to non-AI-a ssisted MRI, while increasing signal-to-noise ratio by 16% (all p < 0. 001).

Conclusion

Abbreviated, ultrafast, and AI-assisted MRI protoco ls significantly reduce energy consumption and carbon emissions, sup porting eco-friendly MRI practices.

Limitations

Not applicable. Funding for this study: National Research Foundation of Korea funded by the Korea government (RS-2024-00347290). Ethics committee - additional information: No Author Disclosures: Min Sun Bae: Nothing to disclose Okhee Woo: Nothing to disclose Soo-Yeon Kim: Nothing to disclose Bo Kyoung Seo: Nothing to disclose Sung Eun Song: Nothing to disclose Jeong Taek Lee: Nothing to disclose Sewon Cheon: Nothing to disclose Kyu Ran Cho: Nothing to disclose Hangseok Choi: Nothing to disclose Use of Diffusion-Weighted MRI in Screening High-Ris k Women Under 40 for Breast Cancer *C. C. Arıkan*, M. A. Arıkan, M. A. Nazli; Istanbul /TR ([email protected]) Purpose or Learning Objective: Diffusion-weighted imaging(DWI) presents a rapid, cost-effective, and non-contrast alternative to contrast-enhanced MRI(CE-MRI).This study aims to evaluate the diagnos tic performance of DWI compared to CE-MRI in breast cancer screening for h igh-risk women under 40 while assessing inter-rater agreement and correlati ng findings with biopsy outcomes and ultrasound BI-RADS categories.

Methods

or Background: We retrospectively analyzed the DWI and CE-MR images of 112 women under the age of 40 who had bre ast MRI screening due to high risk. Two radiologists independently review ed the images without knowledge of ultrasound or pathological results. Pa tients were classified as having "diffusion restriction present or absent" an d "pathological contrast enhancement present or absent." Inter-rater agreeme nt was assessed using Cohen’s kappa coefficient.MRI findings were compare d to ultrasound-based BI-RADS reports and biopsy results to determine sen sitivity and specificity.

Results

or Findings: Lesion detection inter-rater agreement was high for DWI(kappa=0.83), and moderate for CE-MRI(kappa=0.57 ). Of the 112 patients, 43 underwent biopsy, with 37 benign and 6 malignant diagnoses. DWI detected 5 out of 6 malignant lesions (sensitivity 83.3%).DWI exhibited 10 false positives (specificity 73%).CE-MRI detected all mal ignant lesions (sensitivity 100%), but showed 30 false positives (specificity 1 8.9%). Among the 79 patients classified by BI-RADS with no biopsy, 57 w ere BI-RADS1-2, with 5(%8.7) false positives on DWI and 14(%24,5) on CE- MRI. Of the remaining 12 BI-RADS3 cases were recommended for follow-up, DWI was positive in 2(%16.6) and CE-MRI in 7(%58.3).

Conclusion

DWI shows high sensitivity and specificity, particu larly in breast cancer screening for women at high risk below 40 ye ars.The high inter-rater agreement shows its reliability, and the reduced re call rate suggests that DWI has the potential to reduce unnecessary treatments compared to CE-MRI.

Limitations

The study is limited by loss to follow-up patients and those with incomplete diagnostic tests Funding for this study: No funding was provided for this study. Ethics committee - additional information: Başakşehir Çam and Sakura City Hospital Ethics Committee Author Disclosures: Mehmet Ali Nazli: Nothing to disclose Ceyda Ceren Arıkan: Nothing to disclose Mehmet Ali Arıkan: Nothing to disclose Patient comfort in supine breast MRI using a wearab le coil - preliminary questionnaire results *L. Nohava*, R. Czerny, M. Tik, E. Laistler, R. Fra ss-Kriegl; Vienna/AT Purpose or Learning Objective: Supine positioning during breast MRI using a wearable coil at 3 T has the potential to improve p atient comfort and to extend the inclusion criteria for breast MR examinations ( obesity, pregnancy). While the technical performance of the wearable coil for supine breast MRI is being assessed in ongoing studies, the aim of this study was the evaluation of patient perception with the goal of ensuring comfortable pa tient-oriented breast MR examinations.

Methods

or Background: A questionnaire evaluating the impact of radiofrequency coils on patient comfort in MRI as a n add-on to clinical studies comparing the performance of flexible coils with st andard rigid coils was developed. In an IRB-approved breast MRI study, pat ients underwent one exam in supine using a wearable flexible coil and o ne prone reference exam. After each exam, patients filled in the questionnai re with 18 items using a 7- point Likert scale. So far, 10 questionnaire sets w ere collected. The study population had a range of different bra sizes (70B- 95D), ages (20-64 years), and BMIs (19-30 kg/m2).

Results

or Findings: Significant improvement in patient comfort during s upine breast MRI was found for the following items: “I fe lt comfortable before the exam.” (p<0.009); “I found it cumbersome or physica lly demanding to take the lying position.” (p<0.026); ”I felt comfortable dur ing the exam.” (p<0.043); “I found the lying position comfortable.” (p<0.047). 9 patients commented on pain or discomfort in prone whereas only 1 patient comme nted on discomfort due to peripheral nerve stimulation in supine breast MRI.

Conclusion

Patients perceived supine BraCoil MRI as significan tly more comfortable than prone MRI, in anticipation of and during the exam.

Limitations

The limitations of the study are the small sample s ize, and mono- center character. Funding for this study: Funding was provided by by the Austrian Science Fund (FWF)/Agence Nationale de Recherche (ANR) gran t FWF I-3618/ANR- 17-CE19-0022 “BraCoil” and FWF grant P37189 “OPTIMA L”, the Horizon Europe Grants No. 101078393 “MRITwins” and No. 1010 71008 “CITRUS”, and the Austrian Society for Senology (ÖGS) support gra nt. Ethics committee - additional information: The study was approved by the Ethics Committee of the Medical University of Vienn a (EK No. 2137/2021). Author Disclosures: Lena Nohava: Nothing to disclose Roberta Frass-Kriegl: Nothing to disclose Elmar Laistler: Shareholder: ALSIX GmbH Martin Tik: Nothing to disclose Raphaela Czerny: Nothing to disclose A breast MRI image quality score (BreastMRI-QUAL): preliminary results S. Marziali, *L. Corradini*, M. Zanardo, C. Deprett o, G. Della Pepa, G. Irmici, G. P. Scaperrotta, F. Sardanelli; Milan/IT ([email protected]) Purpose or Learning Objective: Breast MRI is an established technique for diagnosing breast cancer using a multiparametric pr otocol, including sequences before/after contrast administration. The diagnostic performance depends on image quality, limited by misregistratio n artefacts due to patient movement. We propose a standardized image quality s core (BreastMRI- QUAL).

Methods

or Background: Two independent readers with 3 years of experience assigned a 4-level score to each sequenc e of 50 consecutive 1.5-T examinations at a tertiary cancer centre, as follow s: 0 = not diagnostic for any cause; 1 = relevant artefacts/malpositioning with c onserved diagnostic value for the specific case; 2 = slight artefacts/malposi tioning with conserved diagnostic value; 3 = excellent image quality with full diagnostic value. The score per sequence was summed as follows: (T2-weigh ted*1) + (DWI-b=0*0.5) + (ADCmaps*0.5) + (T1-weighted-precontrast*1) + (T1 -weighted- postcontrast*2) + (T1-weighted-subtracted*3). To ob tain a global score (GS) Saturday Abstract-based Programme 254 from 0 to 10, the sum was divided by 2.4, with scor es below 6 considered as insufficient.

Results

or Findings: The average GS between the two readers was 8.0 ± 1. 0 (mean ± standard deviation), with 42 cases (84%%) r eceiving a score ≥ 7. Only 1 case (2%) was scored 1 but ≤ 2 for 17 cases (34%), and > 2 for 1 case (2%). The Bland-Altman an alysis showed a mean difference (bias) of 0.28, with the limits of agree ment ranging from -1.82 to 2.38, indicating the level of agreement between rea ders. The average assessment time/examination was 3 min.

Conclusion

BreastMRI-QUAL is a reproducible quality score syst em. Breast MRI image quality at a tertiary cancer center was g ood-to-excellent in over 80% of cases. Multicenter-multivendor validation st udies are needed.

Limitations

Monocentric study, limited sample size. Funding for this study: No funding. Ethics committee - additional information: Use of anonymized datasets outside clinical workflow. Author Disclosures: Gianmarco Della Pepa: Nothing to disclose Sara Marziali: Nothing to disclose Francesco Sardanelli: Research/Grant Support: Bayer AG, Bracco imaging, GE healthcare Speaker: Bayer AG, Siemens Healthinee rs, Esaote Advisory Board: Bayer AG, Bracco imaging, GE healthcare Gianfranco Paride Scaperrotta: Nothing to disclose Moreno Zanardo: Nothing to disclose Catherine Depretto: Nothing to disclose Lisa Corradini: Nothing to disclose Giovanni Irmici: Nothing to disclose Detection of residual fibroglandular tissue on brea st MRI in women treated with mastectomy and DIEP flap breast recons truction *N. Smeins*, J. Rooij, Van, E. Heuts, J. B. Houwers , S. Tuinder, T. Van Nijnatten; Maastricht/NL ([email protected]) Purpose or Learning Objective: After breast amputation, and especially skin sparing mastectomy, there can be residual fibroglan dular tissue (RFGT). RFGT can influence the risk of breast cancer recurrence. However, women are not screened for RFGT after mastectomy. This study exam ines the frequency in which RFGT can be detected on breast MRI after mast ectomy and DIEP flap reconstruction and the influence of RFGT on breast cancer recurrence risk.

Methods

or Background: This retrospective, single-centre study included female patients who underwent mastectomy and DIEP f lap reconstruction. Post-reconstruction breast MRI exams from 2007-2022 were reassessed by a breast radiologist to detect potential presence of RFGT. The presence of RFGT was rated according to a confidence scale (1-5 ), with a score of 1 indicating ‘definitely no breast tissue’ and 5 indi cating ‘definitely breast tissue’. Locations suspected of RFGT rated with a score of 4 or more were considered RFGT. RFGT prevalence was correlated with disease r ecurrence.

Results

or Findings: A total of 73 patients (85 breasts) were included. RFGT was found in 15 (20.5%) patients and 16 (18.8%) bre asts. Ten (13.7%) local recurrences had occurred after a mean follow-up per iod of 164.3 months (range: 27.0-381.0 months). Presence of RFGT result ed in a relative risk of 2.58 (95% CI 0.83-7.98) for recurring disease. Most breast MRI exams were assessed with a score of 3 on the confidence scale (39.2%).

Conclusion

RFGT is frequently detected on breast MRI after mas tectomy and DIEP flap breast reconstruction and might have an a ssociation with disease recurrence. Future studies should focus on the clin ical consequences of visualization of RFGT on breast MRI and whether the re is a role for breast MRI in post-mastectomy patients.

Limitations

No limitations were identified. Funding for this study: N. Smeins received a salary from Kankeronderzoekfonds Limburg. Ethics committee - additional information: The study was approved by METC azM/UM (reference number METC 2022-3122). Author Disclosures: Joep Rooij, Van: Nothing to disclose Thiemo Van Nijnatten: Nothing to disclose Nieke Smeins: Nothing to disclose Estherm. Heuts: Nothing to disclose Stefania Tuinder: Nothing to disclose Janneke B. Houwers: Nothing to disclose Breast cancer on post bilateral mastectomy surveill ance MRI *T. Arazi Kleinman*¹, J. Lvovski¹, D. Walchok¹, G. Michal², T. Sella³; ¹Beer Yakov/IL, ²Tel Aviv/IL, ³Jerusalem/IL ([email protected]) Purpose or Learning Objective: Current guidelines regarding post bilateral mastectomy (BMx) follow-up indicate no need for ima ging. Regardless many patients are referred for breast MRI, though this p ractice is not evidence based. The aim of this study to evaluate the role o f MRI in detection of cancer in post BMx women.

Methods

or Background: Retrospective analysis of surveillance breast MRI i n women s/p BMx between the years 2017-2020, at a sin gle institution. Data collected included demographic information, persona l and family history of breast cancer, indication for mastectomy (prophylac tic vs. therapeutic) and reconstruction type. Suspicious MRI findings underw ent biopsy and were correlated with pathology. Malignancy or benignity were determined by either pathology or stability on imaging for at least 12 m onths. Descriptive statistics applied with p<0.05 considered significant.

Results

or Findings: 229 asymptomatic women s/p BMx aged 29-76±8.7 years underwent 709 surveillance studies for a tota l of 1418 breasts examined. Reconstructions included 1324 (93.3%) silicone, 47 (3.3%) autologous flaps and 47 (3.3%) with no reconstruction. 158 (69%) wom en underwent risk- reducing prophylactic Mx (rr-Mx) on one side and th erapeutic Mx (t-Mx) for cancer on the other, 45 (20%) underwent bilateral r r-Mx and 26 (11%) underwent bilateral t-Mx for bilateral breast cance r. Overall, 782/1418 breasts underwent rr-Mx and 184/1418 breasts t-Mx. Cancer w as detected in six breasts, five post t-Mx and one post rr-Mx. Overall cancer detection rate (CDR) was 0.4 (6/1418), higher in post t-Mx (CDR= 0.78, 5 /636) than post rr-Mx (CDR= 0.12, 1/782), p<0.05. No cancers were detecte d in women post rr-BMx.

Conclusion

Cancer risk in women undergoing bilateral rr-Mx is negligible and likely does not warrant MRI surveillance. In contra st, CDR in post t-Mx women was 0.78 on the side of prior cancer and surveillan ce MRI may be considered.

Limitations

Single-institution, retrospective Funding for this study: No Funding for this study Ethics committee - additional information: Retrospective study Author Disclosures: Guindy Michal: Nothing to disclose Tal Arazi Kleinman: Nothing to disclose Joana Lvovski: Nothing to disclose Daria Walchok: Nothing to disclose Tamar Sella: Nothing to disclose Percentage functional tumor volume on pre-treatment MRI within HER2+ breast cancer predicts pathologic complete response to combination neoadjuvant immunotherapy and chemotherapy *R. J. Weinfurtner*, S. Falcon, D. Ataya, M. Abdala h, O. Stringfield, N. Raghunand, B. Czerniecki, H. Soliman, H. Han; Ta mpa, FL/US ([email protected]) Purpose or Learning Objective: To determine if functional tumor volume (FTV) analysis of human epidermal growth factor 2 p ositive (HER2+) breast cancer on pre-treatment MRI can help predict pathol ogic complete response (pCR) in patients treated with dendritic cell vacci ne (DC1) neoadjuvant immunotherapy (NAI) followed by neoadjuvant chemoth erapy (NAC).

Methods

or Background: Patients with HER2+ breast cancer in this pilot tri al underwent pre-treatment MRI, followed by ultrasound -guided intratumoral and intranodal DC1 injections, and then NAC prior to po st-treatment MRI and surgery. FTV was calculated on pre-treatment post-c ontrast T1-weighted MRI images using a percent enhancement threshold of 70% and signal enhancement ratio set to 0. The %FTV was calculated as %FTV = FTV / segmented tumor volume. These were correlated with pathologic response at surgery using unpaired t-tests where p<0.05 was con sidered significant. FTV analysis was also compared to post-NAI/NAC pre-surg ical MRI reports for diagnostic test accuracy comparison.

Results

or Findings: Nineteen patients aged 29-74 (average 54) were included in the study, and 11 (57%) achieved pCR. M RI complete response (mCR) was seen in 11/19 (57%). However, accuracy fo r mCR predicting pCR was only 38%. For %FTV, median was 71%, and patient s achieving pCR had higher %FTV (78% vs 57%, respectively, p=0.007). As a diagnostic test, %FTV above median accurately predicted pCR in 79% (95% c onfidence interval of 54-94%) with sensitivity 75%, specificity 86%, PPV 90%, and NPV 67%.

Conclusion

In this pilot study of combination NAI/NAC treatmen t for HER2+ breast cancer, tumors with above median %FTV on pre -treatment MRI demonstrated more favorable response to treatment, achieving pCR in 78%. Given that post-treatment MRI evaluation demonstrat ed low accuracy in predicting pCR, pre-treatment FTV analysis may prov e a more accurate predictor in future studies. Saturday Abstract-based Programme 255

Limitations

This study was a pilot study with limited sample si ze. Funding for this study: Internal institution grant Ethics committee - additional information: Institutional Review Board (IRB) Author Disclosures: Robert Jared Weinfurtner: Nothing to disclose Mahmoud Abdalah: Nothing to disclose Hatem Soliman: Nothing to disclose Shannon Falcon: Nothing to disclose Dana Ataya: Nothing to disclose Brian Czerniecki: Nothing to disclose Hyo Han: Nothing to disclose Olya Stringfield: Nothing to disclose Natarajan Raghunand: Nothing to disclose Breast MRI: assessment of the Kaiser score in diffe rentiation of non- mass lesions *M. Vukojevic*, M. M. Nadrljanski, I. B. Krušac, D. Dimitrijevic, L. J. Raspopović, A. Djajic, M. Mihajlović; Belgrade/RS ([email protected]) Purpose or Learning Objective: To examine the performance of the Kaiser score in the diagnosis of nonmass breast lesions on MRI.

Methods

or Background: There were 39 female patients with pathologically confirmed nonmass lesions on breast MRI retrospecti vely analyzed. For each patient, the Kaiser score was determined and the BI -RADS category assigned. All patients were examined with full diagnostic pro tocol (T2W-STIR, T2W-TSE, T1W-TSE, DWI-ADC, 3D-FLASH) on 1.5T and 3T. Specifi city and sensitivity of the Kaiser score were computed and the correlation between the Kaiser score and the BI-RADS classification was determined.

Results

or Findings: In the group of patients with nonmass lesions (N=39 ), there were 10 patients (n1) with benign lesions (25 .64%) and 29 patients with malignant lesions (n2). The mean Kaiser score value in n1=3 and in n2=7. There was significant correlation between the Kaise r score and BI-RADS category: n1=0.92; n2=0.66; N=0.79. Sensitivity of the Kaiser score was 89.7% and specificity equaled 70.0%. ROC curve value reac hed 0.89.

Conclusion

Kaiser score represents a reproducible, sensitive a nd specific diagnostic tool for assessment of nonmass lesions o n breast MRI and may contribute to the adequate BI-RADS categorization a nd appropriate further steps in diagnostic algorithm of the patients with nonmass lesions.

Limitations

A single center retrospective analysis with the lim ited number of patients. Funding for this study: None Ethics committee - additional information: No decision was required for the retrospective analysis without the patient interven tion. Author Disclosures: Iva B. Krušac: Nothing to disclose Dejan Dimitrijevic: Nothing to disclose Luka Josif Raspopović: Nothing to disclose Marko Mihajlović: Nothing to disclose Andjela Djajic: Nothing to disclose Milos Vukojevic: Nothing to disclose Mirjan M. Nadrljanski: Nothing to disclose Distinguish HER2-low expression level in breast can cer: insights from qualitative and quantitative MRI analysis *Y. Shen*, C. You, Y. Gu; Shanghai/CN Purpose or Learning Objective: To investigate whether qualitative and quantitative MRI features can reflect HER2-low expr ession breast cancer.

Methods

or Background: The benefit of novel antibody-drug conjugates in HER2-low expression breast cancer suggests that the conventional binary classification HER2 status is insufficient to meet the needs of clinical diagnosis and treatment.The pre-treatment breast MRI images o f 232 patients with pathologically confirmed breast cancer were retrosp ectively analyzed. Clinicopathologic features and MRI features were re corded. The qualitative MRI features included BI-RADS descriptors in DCE-MR I, and intratumoral T2 hyperintensity and peritumoral edema in T2WI. The q uantitative features were generated by multi-b-value DKI, including mean, med ian, 5th, 95th percentile, skewness, kurtosis and entropy of ADC, Dapp and Kap p histogram from the mono-b and multi-b value models.

Results

or Findings: HER2 status was categorized into HER2-zero (n=60), HER2-low (n=91) and HER2-over expression (n=81). Fo r MRI features, the proportion of intratumoral T2 hyperintensity was hi gher in HER2-low than in other groups (p=0.009, p=0.008). For the lesion typ e, the mass lesions were more common in HER2-zero group than in HER2-low gro up (p=0.038). For mass lesions, mass shape (p<0.001) and margin(p<0.0 01) were significantly different between HER2-low and other groups, and ma ss shape is the independent predictive factor (HER2-low vs. HER2-ze ro: p=0.010, HER2-low vs. HER2-over: p=0.012). The area under the ROC cur ve (AUC) of qualitative features to distinguish HER2-low and -zero was 0.76 3 (95% CI: 0.667-0.859). Quantitative features differed between HER2-low and -overexpression groups, especially in NME-related lesions. All combined var iables (Combinedall) had the best performance in predicting HER2-low, with a n AUC of 0.802 (95% CI: 0.701 - 0.903).

Conclusion

Qualitative and quantitative MRI features are valua ble for noninvasively distinguishing HER2-low expression br east cancer, and have their advantages in mass and NME lesions, respectiv ely.

Limitations

Single-center retrospective study with limited samp les. Funding for this study: Not applicable. Ethics committee - additional information: Fudan University, Shanghai Cancer Center Author Disclosures: Yajia Gu: Nothing to disclose Chao You: Nothing to disclose Yiyuan Shen: Nothing to disclose Pre- and post-contrast assessment of apparent diffu sion coefficient in early tumor response assessment in patients on neoa djuvant chemotherapy *M. M. Nadrljanski*, I. B. Krušac, D. Dimitrijevic, L. J. Raspopović, A. Djajic, M. Mihajlović; Belgrade/RS ([email protected]) Purpose or Learning Objective: To assess the difference in pre- and post- contrast assessment of apparent diffusion coefficie nt (ADC) in early tumor response to neoadjuvant chemotherapy (NACT) after t he 2nd cycle.

Methods

or Background: There were 43 patients (N=43) included in retrospective analysis of ADC (b50, b850) in assess ment of early tumor response in responders (R, n1=19) and non-responder s (NR, n2=24), defined pathologically . In all patients, diffusion-weighte d imaging (EPI sequence) was performed before and after application of contrast medium (gadobutrol, 1 mmol/L; 0.1 mL/kg). All patients were examined on e ither 1.5T or 3T unit in same institution with full diagnostic protocol (T2W -STIR, T2W-TSE, T1W-TSE, DWI b50, b850, 3D-FLASH, DWI b50, b850).

Results

or Findings: In R, mean pre-contrast ADC: 1.17+/-0.07 mm2/s x 10 - 3, significantly differs from pre-contrast ADC in N R: 0.98+/-0.09 mm2/s x 10-3; p<0.0001. In R, mean post-contrast ADC value: 1.09+ /-0.08 mm2/s x 10-3, significantly differs from post-contrast ADC in NR: 0.91+/-0.09 mm2/s x 10-3; p<0.0001. Significant difference was obtained for R between mean value of ADC on pre- and post-contrast DWI: 1.17+/-0.07 mm2/ s x 10-3 vs. 1.09+/-0.08 mm2/s x 10-3; p<0.001. Significant difference was o btained for NR between mean value of ADC on pre- and post-contrast DWI: 0. 98+/-0.97 mm2/s x 10-3 vs. 0.91+/-0.09 mm2/s x 10-3; p<0.001.

Conclusion

DWI before and after the application of contrast me dia did not clinically influence the differentiation between R and NR, although ADC was generally lower after the application of contrast m edium. Although significantly different, ADC values for R before and after the ap plication of the contrast medium did not show clinical relevance and did not interfere with the interpretation of the findings. The same applied fo r the ADC values for NR.

Limitations

Small number of patients in a single center retrosp ective study. Funding for this study: None, Ethics committee - additional information: Institutional Review Board waived the need for decision for the collection and analysis of medical record information, with no participant interaction. Author Disclosures: Iva B. Krušac: Nothing to disclose Dejan Dimitrijevic: Nothing to disclose Luka Josif Raspopović: Nothing to disclose Marko Mihajlović: Nothing to disclose Andjela Djajic: Nothing to disclose Mirjan M. Nadrljanski: Nothing to disclose Saturday Abstract-based Programme 256 16:00-17:30 Research Stage 4 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 2105 Artificial intelligence in chest imaging Moderator C. M. Schaefer-Prokop; Amersfoort/NL ([email protected]) Standardized platform to evaluate, compare, and ana lyze AI-based software for detection and classification of lung n odules for the purpose of CT lung cancer screening implementation *X. Ouyang*¹, K. Togka², D. Han², H. L. Lancaster², I. Schuldink², A. N. Walstra², C. Van Der Aalst¹, H. J. De Koning¹ , M. Oudkerk²; ¹Rotterdam/NL, ²Groningen/NL ([email protected]) Purpose or Learning Objective: Low-dose CT detects lung nodules and consequently lung cancer (LC) at an early stage, pr oven to reduce LC mortality. To aid radiologists, commercially availa ble AI-based software have been developed to analyze lung nodules. Self-report ed performance metrics appear promising, however, there remains no indepen dent, standardized platform for external validation. We aimed to devel op a standardized, independent, trustworthy platform to assess and com pare the performance of commercially available AI-lung nodule analysis soft ware.

Methods

or Background: We developed a platform using a sequestered dataset of 560 scans from the EU-funded 4-IN-THE-LU NG-RUN (4ITLR) lung cancer screening implementation trial. The platform is based on systematic Structured Query Language (SQL) database architectu re. Output of AI software in different data formats was reformatted and stored to standardized SQL records, eliminating manual errors, and allowin g AI software results to be compared using uniformed data analysis algorithms t o the final consensus

Result

of an expert radiologist panel. Performance is evaluated on two levels: nodule level and participant level. Nodule level co mpares the detection/ classification of the reference nodule per particip ant and participant level was based on the largest-detected solid nodule.

Results

or Findings: Performance of AI software at nodule level was repo rted using frequencies of agreement and discrepancies wi th the 4ITLR consensus

Result

on reference nodule. At the participant leve l, Cohen's kappa coefficient is used to measure the agreement level with reference.

Conclusion

The standardized platform developed provides an ind ependent assessment of AI software performance. Clinical use rs benefit from reliable comparison of outcomes for lung nodule analysis and transparency of commercial AI in radiology.

Limitations

No limitations have been identified yet. Funding for this study: The 4-IN-THE-LUNG-RUN trial is funded by the European Union (grant number:848294) Ethics committee - additional information: Not applicable Author Disclosures: Matthijs Oudkerk: Nothing to disclose Anna N.H. Walstra: Nothing to disclose Harriet Louise Lancaster: Nothing to disclose Harry J. De Koning: Nothing to disclose Katerina Togka: Nothing to disclose Daiwei Han: Nothing to disclose Xiaotong Ouyang: Nothing to disclose Ilona Schuldink: Nothing to disclose Carlijn Van Der Aalst: Nothing to disclose Benchmarking of Artificial Intelligence and Radiolo gists for Lung Cancer Screening in CT: The LUNA25 Challenge *D. Peeters*¹, B. Obreja¹, N. Antonissen¹, R. Dinne ssen¹, Z. Saghir², E. Scholten¹, R. Vliegenthart³, M. Prokop¹, C. Jaco bs¹; ¹Nijmegen/NL, ²Hellerup/DK, ³Groningen/NL ([email protected]) Purpose or Learning Objective: The imminent implementation of lung cancer screening and growing workload for radiologists dem onstrates the need for safe and validated AI algorithms. At present, it is challenging to adequately validate and benchmark the increasing amount of AI algorithms being developed. In this study, we present the LUNA25 cha llenge, a public competition aiming to evaluate the diagnostic perfo rmance of AI algorithms and radiologists in lung nodule malignancy risk estimat ion at screening CT.

Methods

or Background: The LUNA25 dataset will include 5051 screening CT scans from the National Lung Cancer Screening Tr ial (NLST), with 624 malignant and 7414 benign nodules. Participating te ams can access this dataset to develop AI algorithms. For algorithm val idation, a separate set of 65 malignant and 818 benign nodules from the Danish Lu ng Cancer Screening Trial (DLCST) will serve as a hidden test set. Addi tionally, a subset from DLCST with indeterminate nodules measuring 5-15mm i n diameter will be assessed by a panel of radiologists with varying ex perience levels to benchmark radiologists’ performance against AI algo rithms. Performance will be measured using area under the ROC curve (AUC) an d at different operating points in terms of sensitivity and specificity.

Results

or Findings: With the NLST and DLCST cohorts collected, the challenge is ready to be introduced to the ECR audi ence. Preliminary results with an in-house developed AI algorithm demonstrate d a mean AUC of 0.91 [0.87, 0.95] on DLCST.

Conclusion

The LUNA25 challenge expects to establish a worldwi de benchmark for AI algorithms in estimating lung nodu le malignancy risk at screening CTs and offer insights into how AI compar es to radiologists across different experience levels and operating points.

Limitations

LUNA25 only benchmarks AI’s stand-alone performance , and does not address workflow integration or radiologis t-AI interaction, which are important for clinical adoption. Funding for this study: Funding was provided by the Dutch Cancer Society Ethics committee - additional information: The institutional review board waived the need for informed consent because of the retrospective design and data pseudonymization. Author Disclosures: Renate Dinnessen: Nothing to disclose Dre Peeters: Nothing to disclose Mathias Prokop: Nothing to disclose Zaigham Saghir: Nothing to disclose Ernst Scholten: Nothing to disclose Rozemarijn Vliegenthart: Nothing to disclose Noa Antonissen: Nothing to disclose Bogdan Obreja: Nothing to disclose Colin Jacobs: Nothing to disclose Systematic prioritisation of ai-detected chest x-ra y abnormalities for optimised lung cancer detection: a multicentre stud y *R. Bramley*¹, A. Sharman¹, R. Duerden², S. Lyon¹, M. Ryan³, E. Weber⁴, L. Brown¹, M. Evison¹; ¹Manchester/UK, ²Stockport/U K, ³Sydney/AU, ⁴Linköping/SE ([email protected]) Purpose or Learning Objective: This multicentre study aimed to establish a reproducible and data-driven method for selecting A I-detected chest X-ray (CXR) abnormalities to be prioritised for urgent re porting, supporting faster lung cancer diagnosis. By analysing cancer prevalen ce and clinical significance across two distinct cohorts from seven acute trusts, the study sought to maximise lung cancer detection while main taining a high negative predictive value (NPV).

Methods

or Background: The study involved two cohorts: a retrospective cohort of 1,282 CXR from primary care with detectab le lung cancer (Cohort 1), and a prospective cohort of 13,802 consecutive prim ary care adult CXR (Cohort 2), with AI deployed in shadow mode. The An nalise-AI platform identified 124 distinct findings. An interactive to ol was developed to assess prioritisation strategies based on the cancer preva lence ratio of each AI finding individually and in combination, combined with clin ical judgement.

Results

or Findings: The final prioritisation strategy flagged 41 AI fin dings which included 95.9% of cancers in Cohort 1 and 21. 6% of CXR in Cohort 2 (sensitivity 95.87%, specificity 79.11%, PPV 4.43%, NPV 99.95%). A further 15 AI findings were prioritised based on clinical judg ement as findings not associated with cancer, but requiring prioritisatio n as potentially needing prompt intervention.

Conclusion

This study demonstrates a reproducible and data-dri ven method for prioritising AI-detected CXR abnormalities, bal ancing the need for high sensitivity and NPV while reducing unnecessary prio ritisation of low-risk cases. The shadow mode approach ensured clinical safety be fore deployment, and the interactive tool provided a systematic means to assess prioritisation strategies, offering a practical alternative to tra ditional judgement-based

Methods

and supporting more efficient lung cancer d iagnosis.

Limitations

The tool is designed to support assessment of AI pe rformance in shadow mode in the referral population. Performance metrics should be validated before deployment in other populations. Funding for this study: Funding was provided by the NHS England National AI Diagnostics fund (AIDF). Ethics committee - additional information: The study was performed in shadow mode and did not impact on patient care. Saturday Abstract-based Programme 257 Author Disclosures: Rebecca Duerden: Nothing to disclose Matthew Evison: Nothing to disclose Rhidian Bramley: Nothing to disclose Sarah Lyon: Nothing to disclose Elodie Weber: Employee: Sectra Melissa Ryan: Employee: Annalise.ai Louise Brown: Nothing to disclose Anna Sharman: Nothing to disclose Beyond Nodules: A Deep Learning Approach for Compre hensive Lung Tumour Segmentation on CT *L. Petrychenko*, V. Pugliese, R. G. H. Beets-Tan, L. Topff, K. Groot Lipman; Amsterdam/NL ([email protected]) Purpose or Learning Objective: Several commercially available AI applications for lung nodule analysis on chest CT a re limited to the detection and segmentation of nodules up to 30 mm. There is c linical potential for AI- assisted volumetric analysis and treatment monitori ng of lung tumours of any size, including masses. We aim to develop a Deep Le arning model to detect and segment lung lesions, including primary cancers of all T-stages.

Methods

or Background: In this retrospective study, we collected 1001 ches t CT scans from 504 patients (mean age 66.4±10.3 years; 52% female) with histopathologically confirmed primary lung cancer, treated at the Netherlands Cancer Institute. Both the baseline and first follo w-up scans after treatment were included. Patients were randomly assigned to 9 0% training and 10% testing sets. Two radiologists (4-7 years of experi ence) performed manual segmentation of all lung nodules ≥ 3mm and masses. The deep learning model development utilized a Residual Encoder nnUNet back bone. SGD was selected as the optimizer, 10⁻² was set as the initial learning rate, 2 was set a s Batch Size, and a nnUNet ResEnc XL architecture was selected.

Results

or Findings: The dataset represented all T-stages (Tis/T1/T2/T3/ T4: 1.6/34/21/16/28%) and major histopathological types , with lesion sizes ranging from 3 to 135 mm. DL model achieved a median Dice S imilarity Coefficient (DSC) of 90.0% across all lung lesions, with a medi an of 1 false positive detection per scan. For primary lung tumors, detect ion sensitivity was 77.2%, and median DSC was 90.4%.

Conclusion

The DL model demonstrated very good segmentation performance for primary lung tumors of all sizes, i ncluding masses. The model has the potential to assist physicians in treatment monitoring and planning, though further improvements in detection sensitivit y could enhance its clinical utility.

Limitations

The model requires both external and clinical valid ation. Funding for this study: No additional funding was received; the study was conducted entirely at the Netherlands Cancer Instit ute. Ethics committee - additional information: The study received Institutional Review Board (IRB) approval. Author Disclosures: Valerio Pugliese: Nothing to disclose Kevin Groot Lipman: Nothing to disclose Regina G. H. Beets-Tan: Nothing to disclose Liliana Petrychenko: Nothing to disclose Laurens Topff: Nothing to disclose Foundation Model-based Unsupervised CT Kernel Conve rsion for Standardizing Emphysema Quantification *D. Park*, J-H. Kang, J. Jeong; Seoul, Republic of Korea/KR ([email protected]) Purpose or Learning Objective: Emphysema quantification is crucial for evaluation and management of chronic obstructive pu lmonary disease (COPD). Typically, emphysema is identified in compu ted tomography (CT) images reconstructed with smooth kernels. However, CT reconstruction kernels vary, and raw data are often deleted after reconstruction, making it hard to adjust the kernel retrospectively. Therefor e, this study aims to develop and validate a method for kernel conversion to stan dardize emphysema quantification using a foundational deep learning m odel.

Methods

or Background: Paired CT images from nine cases reconstructed with different kernels were used. Automated lung se gmentation was performed using TotalSegmentator, a foundational deep learnin g model. An unsupervised kernel conversion method was then applied to transf orm the images to a pre- defined kernel. The kernel conversion was evaluated by comparing the emphysema score (ES), defined as the ratio of regio ns with HU below -950 within the lung area, before and after the conversi on.

Results

or Findings: Before kernel conversion, the mean ES difference between images reconstructed with smoother kernels (ex: B30f and STANDARD) and those with sharper kernels (ex: B60f and LUNG) was 11.00±6.85%. After conversion to the target smooth kernel, the mean ES difference was reduced to 2.30±2.65%. Although the sample size was small, this reduction was statistically significant based on a paired t-test (p=0.011).

Conclusion

The foundational model enables the conversion of CT images reconstructed with different kernels to a target sm ooth kernel, allowing for standardized emphysema quantification without the n eed for additional datasets for model development. This result suggest s that the approach can be easily used by anyone with the appropriate software .

Limitations

For more rigorous validation, it is necessary to no t only compare the difference of ES before and after kernel conver sion but also comparative evaluation on ground-truth emphysema masks. Funding for this study: Not applicable. Ethics committee - additional information: We used a dataset from the Korea Testing Laboratory (KTL) challenge. Author Disclosures: Doohyun Park: Employee: VUNO Inc. Jung-Hyun Kang: Employee: VUNO Inc. Jonghun Jeong: Employee: VUNO Inc. Scientific Evidence of AI in Lung Nodule Evaluation on CT-examinations: A Systematic Review *J. Paramasamy*, J-W. Groen, A. Leliveld, B. Willem s, J. Aerts, A. Odink, J. J. Visser; Rotterdam/NL ([email protected]) Purpose or Learning Objective: The purpose of this study was to systematically review the scientific evidence demon strating the efficacy of CE- marked and/or FDA-cleared AI-applications for pulmo nary nodule evaluation on CT examinations.

Methods

or Background: Following the PRISMA guidelines, Medline, Embase, Web of Science, Cochrane, and Google Schola r databases were searched (Jan 1, 2012–Sep 30, 2024) for studies on AI-based evaluation of pulmonary nodules on CT-scans. Included articles we re classified according to a hierarchical model of AI-efficacy: Radiology AI D eployment and Assessment Rubric (RADAR) framework. Additionally, the evoluti on of evidence over time was examined.

Results

or Findings: A total of 98 articles encompassing AI-applications for lung nodule evaluation from 16 vendors were include d, with approximately 90% of clinical questions addressed through cross-s ectional studies. These publications primarily focused on automatic lung no dule detection, accounting for 61.8% of the studies. All included articles wer e classified based on their highest level of efficacy using RADAR, with the maj ority (41/98) at level 2 (diagnostic accuracy). Standalone nodule detection sensitivities in these studies ranged from 50% to 99%. No studies were ide ntified at efficacy levels 5 (patient outcomes), 6 (cost-effectiveness), or 7 (l ocal impact). The number of articles at levels 3 (diagnostic thinking) and 4 (t herapeutic impact) was 1 between 2012-2016, and increased to 40 between 2020 -2024.

Conclusion

Current scientific evidence for AI-applications in lung nodule evaluation primarily emphasizes diagnostic accuracy . However, there is a noticeable shift in research towards exploring the potential clinical impact of this technology.

Limitations

No meta-analysis was conducted due to significant h eterogeneity in methods and reporting. Moreover, vendor involvem ent in most studies could potentially influence outcomes and introduce bias. Furthermore, as AI for lung nodule evaluation rapidly evolves, the included art icles since 2012 may reflect variations in AI-application performance over time. Funding for this study: Unrestricted institutional grant Ethics committee - additional information: Systematic reviews of existing published literature Author Disclosures: Jasika Paramasamy: Nothing to disclose Arlette Odink: Nothing to disclose Joachim Aerts: Nothing to disclose Asabi Leliveld: Nothing to disclose Bo Willems: Nothing to disclose Jacob Johannes Visser: Grant Recipient: Grant to in stitution from Qure.ai; consulting fees from Tegus; payment to institution for lectures from Roche; travel grant from Qure.ai; participation on a data safety monitoring board or advisory board from Quibim, Contextflow, Noaber Fou ndation, and NLC Ventures; leadership or fiduciary role on the steer ing committee of the PINPOINT Project (payment to institution from Astra Zeneca) and RSNA Common Data Elements Steering Committee (unpaid); p hantom shares in Contextflow and Quibim. Jan-Willem Groen: Nothing to disclose Saturday Abstract-based Programme 258 On the effect of lesion number on the FROC performa nce in AI-based lung nodule detection *T. Escobar*, E. Oubel; Montpellier/FR ([email protected]) Purpose or Learning Objective: In the lung nodule detection context, we aimed to determine weather the number of lesions pe r patient acts as a confounding variable in performance evaluation, pot entially affecting metrics like sensitivity and FROC, and if this factor shoul d thus be rigorously controlled during testing.

Methods

or Background: Two experiments were conducted using the LIDC- IDRI dataset. In both experiments, a trained model was evaluated on sub- groups of patients sorted by deacreasing lesion num ber. The first experiment formed cumulative sub-groups by adding and discardi ng 10 patients at a time, creating groups with varying patient numbers. To en sure no effect nor spurious correlation related to patient number, the second e mployed a sliding window of 100 patients with a step of 10. For each sub-group, FROC were computed based on 5-fold cross-validation predictions for th e whole dataset. Additionally, false positives (FP/s), false negatives (FN/s), tru e positives per scan (TP/s), and sensitivity (Se) were evaluated to identify whi ch parts of the FROC were affected.

Results

or Findings: A clear inverse relationship between the number of lesions and FROC scores was observed. Pearson and S pearman correlation coefficients were significant and equal to -0.9 for both experiments. As lesion number increased, TP/s, FP/s, and FN/s increased, w hile Se decreased (i.e. TP/s increase did not compensate for FP/s and FN/s ones.).

Conclusion

The number of lesions per patient inversly affects the FROC in lung nodule detection models. The number of lesions per patient should thus be controlled and documented during model evaluatio n to ensure accurate performance assessments and to clarify under which conditions they are guaranteed. Further studies are required to rigorou sly examine these effects and validate the hypotheses.

Limitations

Limitations include no specific investigation of th e sources of FP/s and FN/s increase with number of lesions. Funding for this study: This study was totally funded by the compagny Intrasense SA as part of its research and developme nt activity. Ethics committee - additional information: Not applicable Author Disclosures: Thibault Escobar: Employee: Intrasense SA Estanislao Oubel: Employee: Intrasense SA Optimizing Healthcare Sustainability through AI-Ass isted Lung Cancer Detection at the time of initial CXR *J. Packer*, M. Storey, A. Chung, S. J. Rickaby, G. Dean, S. C. Shelmerdine, C. Malamateniou; London/UK ([email protected]) Purpose or Learning Objective: Lung cancer is the leading cause of cancer mortality in the UK. Early diagnosis is essential b ut hindered by workforce shortages and limited CT access. This study evaluat es whether the 'Artificial Intelligence triage to same-day CT' (AI-CT) pathway , using the Annalise CXR v2.3 model, can enhance healthcare sustainability b y reducing patient visits, travel emissions, and administrative workload while improving CT access for suspected cancer.

Methods

or Background: Sustainability indices for 26,660 patients (January 2022–October 2023) were assessed across five NHS ce ntres in London. Key metrics included time from chest radiograph to CT r eport (Time to CT), AI accuracy, and cancer suspicion on CXR and CT, pre- and post-AI-CT implementation. Time to CT was measured using survi val analysis, and diagnostic performance (AUC-ROC, F1 scores) was cal culated based on CT- confirmed cancer. Time saved for patients and admin teams was estimated, and carbon reduction was calculated using the Carbo n Trust online calculator.

Results

or Findings: From 26,660 chest radiographs and 573 CT scans, 75 of 10,833 patients received same-day CT post-AI, compa red to 13 of 8,434 pre- AI, eliminating 150 appointments. Each appointment saved ~1.5 hours per patient, totalling 225 hours, with ~37.5 additional hours saved for admin teams. With travel emissions estimated at 1 kgCO2e per pat ient, this potentially resulted in a 150 kgCO2e reduction. The AI model sh owed high sensitivity (91%) but low specificity (22%, F1 score 0.56). The re was a significant increase in CT within 1 and 3 days post-suspicious CXR (HR 1.93, 1.34; p < 0.001).

Conclusion

The AI-CT pathway improved same-day CT access and r educed patient visits and emissions. However, the model's low specificity suggests a need for supervised triage to optimize performance.

Limitations

The AI’s low specificity and dependence on co-locat ed facilities, limit generalizability. Funding for this study: None Ethics committee - additional information: Local trust clinical audit and QI registration forms approved. Author Disclosures: Mathew Storey: Nothing to disclose Simon Joseph Rickaby: Nothing to disclose Susan Cheng Shelmerdine: Nothing to disclose Jack Packer: Nothing to disclose Anthony Chung: Nothing to disclose Geraldine Dean: Nothing to disclose Christina Malamateniou: Nothing to disclose An artificial intelligence software for the detecti on of benign and non- typically benign pulmonary nodules on chest CT scan s *S. Bennani*¹, N-E. Regnard², M. Durteste¹, V. Mart y¹, R. Quilliet¹, A. Pourchot¹, L. Clovis¹, J. Ventre¹, G. Chassagnon ¹; ¹Paris/FR, ²Lieusaint/FR Purpose or Learning Objective: Detecting lung nodules on chest computed tomography (CT) is an important task that extends b eyond the realm of lung cancer screening. This study aimed to compare the p erformance of radiologists to an AI software in identifying both non-typically benign and benign nodules on CT scans.

Methods

or Background: We retrospectively collected thin-section chest CT scans from private practices across France focusing on patients aged 15 or older. The dataset included patients with non-typic ally benign (solid and sub- solid nodules), benign (granulomas and intrapulmona ry lymph nodes), or no nodules. An expert thoracic radiologist defined the ground truth using past and follow-up scans as well as radiologist reports. We compared the performance of four radiologists who had access to limited clin ical information with that of an AI software, LungCT (Gleamer, Paris, France). We co nducted patient-wise ROC and lesion-wise FROC analyses.

Results

or Findings: The final dataset included 250 chest CT scans (age = 66 ± 23 y, 117 women, 133 men). Among these, 128 scans contained at least one non-typically benign nodule, 40 displayed only beni gn nodules and 82 were nodule-free. The analysis focused on nodules with a diameter >6mm. The patient-wise AUC of the AI was 0.97 [0.93,1.00] and that of radiologists was 0.88 [0.83,0.92]. On a lesion-wise basis, the AI ac hieved a sensitivity of 79% [75%,83%] for 0.30 false positive (FP) per scan. Ra diologists exhibited an average sensitivity of 72% [67%,76%] with a mean FP rate of 0.34 [0.25,0.45].

Conclusion

The AI solution demonstrated robust patient-wise pe rformance and comparable lesion-wise detection of non-typical ly benign and benign nodules on CT scans to radiologists.

Limitations

The study’s retrospective design and limited sample size could affect the generalisability of results. Future rese arch should evaluate the performance of AI-assisted radiologists. Funding for this study: Gleamer (Paris, France) funded this study. Ethics committee - additional information: Not applicable Author Disclosures: Nor-Eddine Regnard: Founder: Gleamer Guillaume Chassagnon: Nothing to disclose Vincent Marty: Employee: Gleamer Lauryane Clovis: Employee: Gleamer Rémi Quilliet: Employee: Gleamer Souhail Bennani: Employee: Gleamer Marion Durteste: Employee: Gleamer Jeanne Ventre: Employee: Gleamer Aloïs Pourchot: Employee: Gleamer Improving the generalisation of radiographic AI usi ng automated data curation to mitigate shortcut learning *I. A. Selby*¹, E. González Solares¹, A. Breger², M . Roberts¹, J. Babar¹, F. J. Gilbert¹, N. Walton¹, C-B. Schönlieb¹, J. R. Weir-Mccall³; ¹Cambridge/UK, ²Vienna/AT, ³London/UK ([email protected]) Purpose or Learning Objective: To investigate whether automated data curation pipelines for chest radiographs can improv e deep-learning model performance on unseen data.

Methods

or Background: Two public datasets, MIDRC-1A and MIDRC-R1, were used to develop diagnostic COVID-19 models usi ng four architectures (DenseNet121/ResNet152V2/VGG16/EfficientNetB3). Eac h was trained four times using a different data curation workflow: WF1 . Raw pixel data with partitioning stratified on dataset and COVID-19 sta tus; WF2. DICOM-cleaned data with look-up tables applied, lateral projectio ns and non-chest radiographs excluded, classes balanced on Manufacturer and Proj ection tags, and partitioning additionally stratified on the same me tadata; WF3. Cases excluded using an open-source data-cleaning pipeline (AutoQC , https://gitlab.developers.cam.ac.uk/maths/cia/covid -19-projects/autoqc). Partitioning was stratified on projection and the p resence of a pacemaker using AutoQC annotations; and WF4. The previous two workf lows combined. COVID-19 diagnosis was inferred from laboratory tes ts, and model performance was assessed using five other public da tasets. Generalisation from internal-to-external data was quantified using ΔAUCs. Saturday Abstract-based Programme 259

Results

or Findings: 43,176 radiographs were included in WF1, with 33.2% (14,328) being COVID-19-positive. The development s ets of the other workflows were up to 60% smaller. Similarly, the ex ternal test sets ranged from 24,563-to-38,417 patients, depending on workflow. T he WF1 models experienced the largest fall in generalisation (mea n ΔAUC = -0.15 [95%CI:- 0.17,-0.14]), while models trained utilising AutoQC (WF3-4) demonstrated the most consistent performance with mean ΔAUCs = -0.04 [95%CI:-0.06,-0.02] and -0.02 [95%CI:-0.04,0.00] for WF3 and WF4 (p<0.0 5). The WF2 models had a mean ΔAUC = -0.07 [95%CI:-0.09,-0.05].

Conclusion

Automated data curation can improve the generalisat ion of deep learning models for chest radiographs, facilitating more consistent performance on data from new locations and equipment.

Limitations

Future work should evaluate the tool in multiclassi fication tasks and non-COVID-19 datasets. In addition to the curre nt pacemaker detection, tools for a broader range of support apparatus are necessary. Funding for this study: The authors wish to acknowledge support from the EU/EFPIA Innovative Medicines Initiative 2 Joint Un dertaking - DRAGON (101005122) (I.S., A.B., M.R., L.E.S., J.B., C.-B.S ., E.S., J.W.M., AIX- COVNET); the National Institute for Health and Care Research (NIHR) Cambridge Biomedical Research Centre (BRC-1215-2001 4) (I.S., L.E.S., J.H.F.R., E.S., J.W.M.); Wellcome Trust (J.H.F.R.), British Heart Foundation (J.H.F.R.); the EPSRC Cambridge Mathematics of Info rmation in Healthcare Hub EP/T017961/1 (M.R., J.H.F.R., C.-B.S.); Cancer Research UK (CRUK) National Cancer Imaging Translational Accelerator ( NCITA) [C42780/A27066] (L.E.S.); Cambridge Mathematics of Information in H ealthcare (CMIH) Hub EP/T017961/1; Austrian Science Fund (FWF, project T -1307) (A.B.); and the Trinity Challenge BloodCounts! project (M.R., C.-B. S.). The AIX-COVNET collaboration is also grateful to Intel for financi al support. C.B.S. additionally acknowledges support from the P hilip Leverhulme Prize, the Royal Society Wolfson Fellowship, the EPSRC adv anced career fellowship EP/V029428/1, EPSRC grants EP/ S026045/1 and EP/T00 3553/1, EP/N014588/1, EP/T017961/1, the Wellcome Innovator Awards 215733/Z/ 19/Z and 221633/Z/20/Z, the European Union Horizon 2020 research and innovation program under the Marie Skodowska-Curie grant agreement No. 777826 NoMADS, the Cantab Capital Institute for the Mathematics of Information and the Alan Turing Institute. Please note that the content of this publication re flects the authors’ views and that neither IMI nor the European Union, EFPIA, or the DRAGON consortium are responsible for any use that may be made of the information contained therein. Ethics committee - additional information: The Brent Research Ethics Committee, the Health Research Authority (HRA), and Health and Care Research Wales (HCRW) provided ethical approval for our retrospective study (IRAS ID: 282705, REC No.: 20/HRA/2504, R&D No.: A0 95585). Informed consent was not required as data was pseudonymised. Author Disclosures: Carola-Bibiane Schönlieb: Nothing to disclose Eduardo González Solares: Nothing to disclose Michael Roberts: Nothing to disclose Fiona J. Gilbert: Nothing to disclose Jonathan R. Weir-Mccall: Nothing to disclose Anna Breger: Nothing to disclose Ian Andrew Selby: Nothing to disclose Judith Babar: Nothing to disclose Nicholas Walton: Nothing to disclose 260 Sunday, March 2 Sunday Abstract-based Programme 261 08:00-09:00 Research Stage 1 Research Presentation Session: Breast RPS 2202 Personalised risk prediction of breast cancer Moderator P. Clauser; Vienna/AT ([email protected]) Author Disclosures: Paola Clauser: Board Member: EUSOBI; Grant Recipien t: Siemens Associations of automatically measured breast densi ty with breast cancer risk and duration of the pre-clinical detect able phase in a Dutch screening cohort *J. Peters*¹, D. Van Der Waal¹, M. Smid-Geirnaerdt¹ , C. Van Gils², M. Broeders¹; ¹Nijmegen/NL, ²Utrecht/NL ([email protected]) Purpose or Learning Objective: Breast density could impact screening strategies. Women with dense breasts face higher br east cancer risk. Furthermore, lesion masking may shorten the pre-cli nical detectable phase (PCDP), increasing interval cancer rates. This stud y examines the associations of automated breast density measures with breast ca ncer risk and PCDP duration.

Methods

or Background: Digital mammograms were used from 60,739 participants in a prospective Dutch screening cohor t (PRISMA study, 2014- 2019). Dense volume (DV,cm3), volumetric breast den sity (VBD,%) and Volpara Density Grade (VDG1-4) were assessed using Volpara version 1.5.0. Breast cancer diagnoses were ascertained through li nkage with the Netherlands Cancer Registry. Participants with prio r breast cancer (N=73) or screen-detected breast cancer at study entry (N=401 ), were excluded. Information on time to screen-detected and interval cancers was used to fit a three-state (1:cancer-free, 2:pre-clinical detectab le cancer, 3:clinical cancer) Markov regression model. Hazard ratios (HRs) were c alculated for the effects of breast density on state transition intensities 1 ->2 (=breast cancer risk) and 2->3 (=1/PCDP duration).

Results

or Findings: After a median 4.2 years (IQR 3.9–4.6) we observed 430 screen-detected and 316 interval cancers. Log-trans formed VBD and DV were positively associated with increased breast cancer risk (HR 1.12 [95%CI 1.05- 1.21] and HR 1.32 [95%CI 1.05-1.35] per one-standar d-deviation increase, respectively). Both measures were associated with s horter PCDP (HR 1.48 [95% CI 1.30-1.69] and HR 1.19 [95%CI 1.05-1.35]). Mean PCDP duration was 1.63 years [95%CI 1.11–2.41] for women with highest density (VDG4), compared to 3.41 years [95%CI 3.31-3.52] for VDG1-3 .

Conclusion

Breast density is associated with breast cancer ris k and PCDP duration. VBD has the strongest association with PC DP, indicating a reduced sensitivity of biennial mammography, while DV conta ins more information on breast cancer risk.

Limitations

Other confounders than age, e.g. BMI, were not yet included in the model. Funding for this study: The PRISMA study is funded by ZonMw and KWF. Ethics committee - additional information: CMO Arnhem-Nijmegen Author Disclosures: Jim Peters: Nothing to disclose Carla Van Gils: Nothing to disclose Maartje Smid-Geirnaerdt: Nothing to disclose Mireille Broeders: Nothing to disclose Danielle Van Der Waal: Nothing to disclose Mammographic biomarkers of cardiovascular risk: the BAKER study *D. Capra*, O. Hoda, C. B. Monti, M. Zanardo, F. Sa rdanelli; Milan/IT ([email protected]) Purpose or Learning Objective: Mammography could offer two sex-specific biomarkers to spotlight cardiovascular risk in wome n: breast arterial calcifications (BAC) and breast density. We conduct ed a prospective case- control study evaluating the association between BA C and gynaecological and cardiovascular risk factors.

Methods

or Background: Consecutive women showing BAC and age- and breast density-matched controls referred for annual mammography were prospectively enrolled. We recorded anthropometric variables, traditional cardiovascular risk factors and gynaecological risk factors. Breast density was classified as low breast density (BI-RADS categorie s A and B) or high breast density (BI-RADS C and D).

Results

or Findings: 72 BAC patients and 72 controls were enrolled (medi an age 70.0 years, IQR 62.5 to 78 years). Women with B AC had a younger age at menopause (50 vs 52 years, p=0.008), and showed ass ociations with breastfeeding (p=0.041) and parity (p=0.038). Women with BAC show a borderline significant trend towards the use of ant i-ipertensive medications (p=0.092). No other differences between cases and c ontrols were observed (p>0.101). Higher breast density was significantly associated with younger age (p<0.001), lower body weight (p<0.001), lower systo lic blood pressure (p=0.003), and higher HDL cholesterol (p=0.017), wh ereas lower breast density was associated with longer time since menop ause (p=0.003) and use of anti-ipertensive medications (p<0.001).

Conclusion

Women with BAC have a younger menopausal age, which represents a precocious shift towards a less favour able cardiometabolic hormonal balance. Similarly, women with low breast density show an unfavourable cardiovascular risk profile, using mor e often anti-ipertensive medications, having a higher systolic blood pressur e and lower levels of HDL cholesterol.

Limitations

Age and breast density matching reduces the statist ical power to observe associations between breast density and car diovascular risk factors among women of similar age. Absence of follow up to record cardiovascular events. Funding for this study: General Electric Healthcare supported this study. Ethics committee - additional information: Ethics committe approval number 90/INT/2020, 08/09/2020. All participants si gned informed consent. Author Disclosures: Francesco Sardanelli: Advisory Board: Bayer AG Advi sory Board: Siemens Healthineers Advisory Board: Bracco Imaging Advisor y Board: Esaote Advisory Board: GE Healthcare Omar Hoda: Nothing to disclose Caterina Beatrice Monti: Nothing to disclose Moreno Zanardo: Nothing to disclose Davide Capra: Nothing to disclose Short-term risk prediction of breast cancer compari ng risk tools for digital mammography and digital breast tomosynthesi s in U.S. screening populations E. F. Conant¹, C. Parghi², P. Hall³, *M. Eriksson*³ ; ¹Philadelphia, PA/US, ²Addison, TX/US, ³Stockholm/SE ([email protected]) Purpose or Learning Objective: Image-derived AI-based risk models demonstrate ability to predict breast cancer risk u sing digital mammography (DM) and digital breast tomosynthesis (DBT) imaging data. However, a direct comparison of performances within the same screenin g population has yet to be conducted.

Methods

or Background: We conducted a nested case-control study including women aged 35-98 from Solis and UPenn scr eening cohorts, between 2014 and 2021. Participants were followed f or two screens, with cancer diagnosed before August 2022. Two image-base d ProFound AI Risk models, one for DM and one for DBT, estimated absol ute 1-year breast cancer risks at study-entry. We assessed models’ discrimin atory performance (AUC) controlling for screening site and classified risks according to U.S. Preventive Services Task Force (USPSTF) thresholds.

Results

or Findings: Study included 780 women with incident breast cance r (mean age 63.4±11.2) and 7,481 controls (mean age 5 7.0±10.6). Cancers were diagnosed on average 1.3±0.5 years (range 4 mo nths to 4 years) after study-entry. At study entry, AUCs of DM and DBT mod els were 0.71 (95% CI: 0.69-0.73) and 0.75 (95% CI: 0.73-0.77), respective ly (p<0.01). Comparing UPenn and Solis data, similar estimates were observ ed for respective models. Based on USPSTF guidelines, 14% of women were class ified as high-risk. Among this group, DM model predicted 40% (95% CI: 3 6-43%) of future breast cancers, compared to 48% (95% CI: 44-52%) by DBT mo del (p<0.01). Non- significant differences in proportions of future br east cancers were observed comparing sites.

Conclusion

The image-derived DM and DBT AI-risk models predict ed 40- 48% of future breast cancers at study-entry in two U.S. screening populations. The DBT model predicted a significantly higher prop ortion of future cancers compared to the DM model emphasizing the need for s ome women to obtain supplemental screening.

Limitations

Study is limited to only 2 sites in U.S. Funding for this study: iCAD, Inc. Swedish Research Council Swedish Breast Cancer Association Ethics committee - additional information: The study has been reviewed and approved. Author Disclosures: Mikael Eriksson: Patent Holder: iCAD, Inc. Chirag Parghi: Nothing to disclose Per Hall: Patent Holder: iCAD, Inc. Emily F. Conant: Advisory Board: iCAD, Inc. Sunday Abstract-based Programme 262 Impact of Breast Density Metrics on Personalized Br east Cancer Screening Protocols *G. Gennaro*¹, L. Bucchi², A. Ravaioli², F. Caumo¹; ¹Padova/IT, ²Meldola/IT ([email protected]) Purpose or Learning Objective: To evaluate the impact of different breast density metrics on the personalization of breast ca ncer screening.

Methods

or Background: The RIBBS study (ClinicalTrials.gov NCT05675085) is a personalized breast screening study targeting young women. The protocol used digital breast tomosynthesis (DBT) and double reading to stratify participants based on individual breast cancer risk and breast density. In this protocol, breast density determined the need for su pplemental ultrasound (US), while breast cancer risk guided the frequency of sc reening. A quantitative software tool provided volumetric breast density, w ith a 25% threshold used to identify women who needed supplemental US. This stu dy compares stratification based on this volumetric approach wi th stratification using categorical breast density metrics, both objective and human.

Results

or Findings: A total of 10,269 women, all aged 45 years, were enrolled in the RIBBS study. Of these, 1,904 women (18.5%) had ≥25% breast volumetric density and underwent additional US. Usi ng the categorical breast density provided by the same software, 41.1% of par ticipants would have been categorized as having a BIRADS “d” category density , 2.2 times higher than that identified through the volumetric metric. If B IRADS categorization had been performed by human readers, considering every DBT classified as “d” by at least one reader, the percentage would have drop ped to 32.3%, still 1.7 times higher than the current protocol.

Conclusion

This study shows that the choice of breast density metrics can significantly influence the stratification process in personalized breast cancer screening. Quantitative metrics allow more precise stratification than categorical approaches, improving the feasibility o f supplemental imaging in clinical practice.

Limitations

Changes in screening performance due to the use of categorical breast density for supplemental US remain unassesse d, as the current results are based solely on the applied protocol. Funding for this study: This specific subanalysis had no funding Ethics committee - additional information: The RIBBS study was approved by the Ethics Committee with the following code "RI BBS 2019/37" Author Disclosures: Francesca Caumo: Nothing to disclose Lauro Bucchi: Nothing to disclose Alessandra Ravaioli: Nothing to disclose Gisella Gennaro: Nothing to disclose Changes in Mammographic Density for Breasts Develop ing and not Developing Breast Cancer *J. Gjesvik*, N. Moshina, S. Sagstad, M. Larsen, Å. S. Holen, M. B. Bergan, S. Hofvind; Oslo/NO ([email protected]) Purpose or Learning Objective: The evidence on longitudinal changes in mammographic density in breasts developing cancer i s limited. We aimed to analyse mammographic density among women developing and not developing breast cancer over three consecutive screening roun ds in BreastScreen Norway.

Methods

or Background: In this retrospective cohort study, 66,696 women aged 50-69 years with three consecutive screening e xaminations performed in Rogaland and Hordaland counties, 2007-2020, were in cluded. A total of 909 women were diagnosed with screen-detected and 287 w ith interval cancer. Mammographic density data was obtained from an auto mated software (Volpara 1.5.0 and 1.5.4.0) and included absolute ( cm3) and percent (%) dense volume per woman and breast. A linear mixed-e ffects model with a fixed effect for each woman was applied on a breast level to define the changes in absolute and percent dense volume. The model was ad justed for age at entry, breast volume, and history of benign breast disease .

Results

or Findings: Mean age for women not developing breast cancer was 62.5 years (standard deviation, SD: 5.1), while it was 62.3 (SD: 4.4) for women with screen-detected cancer and 61.9 (SD: 4.8) for interval cancer. Absolute and percent dense volume decreased over time in all women. In breasts developing cancer the rate of decrease was lower fo r absolute dense volume, estimate=0.004 (95% CI: 0.002-0.007, p=0.041), comp ared to breasts not developing cancer. The rate of decrease was also lo wer for percent dense volume, estimate=0.003 (95%CI 0.000-0.007, p=0.053) , in breasts developing versus not developing cancer.

Conclusion

Absolute dense volume decreased to a lower degree i n breasts developing versus not developing cancer. Longitudin al changes in absolute dense volume could be used for more precise breast cancer risk prediction and screening personalization.

Limitations

The study is retrospective and the population is fa irly homogenous. Funding for this study: No funding Ethics committee - additional information: Regional Commitee for Medical and Health Research Ethics Author Disclosures: Nataliia Moshina: Nothing to disclose Jonas Gjesvik: Nothing to disclose Marie Burns Bergan: Nothing to disclose Åsne Sørlien Holen: Nothing to disclose Silje Sagstad: Nothing to disclose Solveig Hofvind: Nothing to disclose Marthe Larsen: Nothing to disclose Using Artificial Intelligence to Detect Subclinical Breast Cancer *J. Gjesvik*¹, N. Moshina¹, C. Lee², D. L. Migliore tti³, S. Hofvind¹; ¹Oslo/NO, ²Seattle, WA/US, ³Davis, CA/US ([email protected]) Purpose or Learning Objective: Investigate whether an artificial intelligence algorithm (AI) trained for detecting breast cancer scored the breast developing breast cancer and the breast not developing breast cancer differently years before diagnosis.

Methods

or Background: In this retrospective cohort study, we included women aged 50-69 who attended three consecutive bie nnial screening rounds between 2004 and 2018, as part of BreastScreen Norw ay. A total of 116 495 women were included in the final study population, 1265 with screen-detected breast cancer detected at, and 342 with interval ca ncer diagnosed within two years after, the third screening round. We used a c ommercial AI algorithm to score each breast with a risk score between 0 and 1 00.

Results

or Findings: For women developing screening-detected breast cancer the mean AI-score at the first screening rou nd for the breast developing breast cancer was 19.2 (SD: 28.6), and 82.7 (SD: 26 .7) after the third screening round. The score was 9.5 (SD: 19.0) in th e first and 5.0 (SD: 15.7) in the third screening round for the breast not develo ping breast cancer. For interval cancer, the mean scores for breasts develo ping cancer were 17.8 (SD:26.3) and 33.1 (SD: 33.8), respectively, and me an scores for breasts not developing cancer were 10.5 (SD: 19.9) and 8.4 (SD: 18.7), respectively. For women not developing breast cancer, the mean AI sco re was 7.1 (SD: 15.2) in the first and 6.4 (SD: 14.5) in the third screening rounds, respectively.

Conclusion

AI-scores were higher in breasts developing cancer up to 6 years before it was diagnosed. The findings suggests that commercial AI algorithms for breast cancer detection might be considered for identifying women at higher risk of developing breast cancer.

Limitations

This is a retrospective study, and the population i s mostly homogenous. Funding for this study: Funded by the Norwegian Cancer Society (Pink Ribbon) Ethics committee - additional information: Regional Committee for Medical and Health Research Ethics, Norway Author Disclosures: Christopher Lee: Nothing to disclose Nataliia Moshina: Nothing to disclose Jonas Gjesvik: Nothing to disclose Diana L Miglioretti: Nothing to disclose Solveig Hofvind: Nothing to disclose External validation of a mammographic masking predi ction model in the Dutch Breast Cancer Screening Program *S. D. Verboom*¹, J. G. Mainprize², J. Peters¹, M. Broeders¹, M. Yaffe², I. Sechopoulos¹; ¹Nijmegen/NL, ²Toronto, ON/CA ([email protected]) Purpose or Learning Objective: To externally validate a lesion masking prediction model for mammograms, Mammatus, develope d on a North American cohort, in a larger retrospective breast c ancer screening cohort from one screening center in The Netherlands.

Methods

or Background: A total of 935 digital mammograms from the Dutch Breast Cancer Screening Program with a unilateral i nvasive breast cancer that was either screen detected or diagnosed within 24 m onths after a negative screening (interval cancer) were included. All mamm ograms were retrospectively evaluated for the visibility of mal ignant masses using all available diagnostic imaging and clinical informati on. Mammatus was applied on the contralateral mammogram to eliminate the inf luence of the lesion. The area under the receiver operator characteristics (R OC) curve (AUC) when using Mammatus to distinguish examinations with scr een-detected cancers (assumed low masking risk) from interval cancers (a ssumed high masking risk) was computed. The AUC was compared to that of the o riginal cohort and to that obtained using volumetric breast density (VBD) as a predictor. A second three-category ROC analysis was performed, with int erval cancers that were retrospectively visible classified as intermediate lesion masking. Sunday Abstract-based Programme 263

Results

or Findings: Mammatus achieved an AUC of 0.70 (95%CI 0.67-0.74) for distinguishing between screen-detected- (n=632) and interval-cancer exams (n=303). This performance did not differ from the original study (AUC=0.75 (0.68-0.82), p=0.20), and outperformed VB D (AUC=0.66 (0.62- 0.70, p<0.002). The three-category ROC analysis sho wed that Mammatus outperformed VBD at identifying low risk of lesion masking (AUC=0.74 (0.70- 0.77)), however, not for identifying high risk (AUC =0.69 (0.65-0.74)).

Conclusion

Mammatus performed well in predicting breast cancer -masking risk in a Dutch screening cohort. This suggests tha t adding information other than density improves prediction of lesion masking.

Limitations

There is no ground truth of lesion masking risk, th erefore the best possible approximation is used. Funding for this study: aiREAD financed by the Dutch Research Council (NWO), Dutch Cancer Society (KWF), and Health Holla nd (HH) Ethics committee - additional information: The Radboudumc ethics committee declared that this study falls outside th e scope of the Dutch Medical Research involving Human Subjects Act and could be carried out without approval of an Institutional Review Board. Author Disclosures: Jim Peters: Nothing to disclose Mireille Broeders: Research/Grant Support: Hologic, Screenpoint Medical, Sectra Benelux, Volpara Healthcare, Lunit, and iCAD Speaker: Hologic and Siemens Healthcare Sarah Delaja Verboom: Nothing to disclose James G. Mainprize: Founder: Calavera Surgical Desi gn, Inc. Research/Grant Support: GE Healthcare Martin Yaffe: Research/Grant Support: GE Healthcare Ioannis Sechopoulos: Speaker: Canon, Siemens Health care Research/Grant Support: Siemens Healthcare, Canon Medical Systems, ScreenPoint Medical, Sectra Benelux, Volpara Healthcare, Lunit Advisory Board: Koning Corp. 08:00-09:00 Research Stage 2 Research Presentation Session: Neuro RPS 2211 Contrast agents and energy sustainability in neuroimaging Moderator Y. Özsunar; Aydın/TR ([email protected]) Efficacy of gadoquatrane, a novel low-dose high rel axivity macrocyclic gadolinium-based contrast agent, at 5-, 10- and 15- min post injection in comparison to gadobutrol for CNS CE-MRI *C. Deuschl*¹, K. Kudo², A. Liu³, M. A. Klemens³, B . M. Hofmann³, P. Palkowitsch³, B. P. Liu⁴; ¹Essen/DE, ²Sapporo/JP, ³Berlin/DE, ⁴Chicago, IL/US ([email protected]) Purpose or Learning Objective: This dose-finding Phase 2 study for the novel tetrameric macrocyclic GBCA gadoquatrane inve stigated the efficacy of 0.04 mmol Gd/kg body weight (bw) gadoquatrane in co mparison to 0.1 mmol Gd/kg bw gadobutrol for CE-MRI in patients with CNS lesions at 5-, 10- and 15-min post injection (pi).

Methods

or Background: Adult patients with CNS lesions were included in the study and underwent two CE-MRIs within an inter val of 3 to 14 days, first gadobutrol, then gadoquatrane. Images were acquired at 5- (3D IR-GRE), 10- (2D-SE) and 15-min pi (3D IR-GRE). The 5-, 10- and 15-min CE-MRIs were assessed in the Blinded Independent Central Review using visualization parameters (contrast enhancement, border delineatio n, internal morphology) to investigate the time course of enhancement. Quantit ative signal intensity measurements were also performed. Safety parameters were assessed and pharmacokinetic (PK) samples were taken to determin e plasma concentrations for both GBCAs.

Results

or Findings: The sum of the lesion visualization parameters at e ach point for the average reader was very similar for g abobutrol (9; 8.86; 9.09) and gadoquatrane (8.94; 8.8; 8.97), with differences cl ose to zero (-0.06; -0.06; - 0.12). These results were representative for the th ree individual readers and each visualization parameter. The results of the qu antitative measurements supported the qualitative evaluations. The safety a nd concentration-time profile of gadoquatrane were very similar to gadobutrol, bu t dose-proportionally lower.

Conclusion

Gadoquatrane at a dose of 0.04 mmol Gd/kg bw shows similar efficacy as gadobutrol at 0.1 mmol Gd/kg bw on 5-, 10- and 15-min pi CE-MRI images.

Limitations

The limitation of the study is the small sample siz e of 50 evaluable patients. Funding for this study: This Phase 2 study was sponsored by Bayer AG. Ethics committee - additional information: Ethics vote from all sites available. Author Disclosures: Alex Liu: Employee: Bayer Benjamin P. Liu: Nothing to disclose Cornelius Deuschl: Nothing to disclose Petra Palkowitsch: Employee: Bayer Kohsuke Kudo: Nothing to disclose Birte Maria Hofmann: Employee: Bayer Mark Alexander Klemens: Employee: Bayer Clinical experience with gadopiclenol in patients wi th brain metastases: Single and double doses, comparison of standard and delayed scanning *J. Vymazal*, Z. Ryznarova, R. Liščák, A. Rulseh; Prague/CZ ([email protected]) Purpose or Learning Objective: To demonstrate the role of gadopiclenol in MR imaging of brain metastases with at standard (0. 05 mmol/kg) and double dose (0.1 mmol/kg), including delayed examination ( ca 15 minutes after contrast administration).

Methods

or Background: Ninety-one subjects with known brain metastases underwent MRI prior to radiosurgery (44 female, mea n age 64.2 ±12 years). A standard T1W 3D SPACE sequence was acquired in all subjects (3 subjects at 1.5T, 88 at 3T) post-contrast, followed by a delaye d acquisition. In roughly half of subjects (n=47), the delayed acquisition was pre ceded by additional contrast administration (cumulative double dose). After rand omization, 3 blinded readers evaluated the number of lesions for each pa tient in 2 sessions, 1 month apart. A paired t-test was used to compare ea rly and late evaluations in the two groups (subjects with cumulative single dos e or double dose), the Welch t-test was used to compare differences in (la te-early) single versus (late- early) double doses.

Results

or Findings: Agreement between all three blinded readers was use d for evaluation. The late single dose exam revealed 170 metastatic lesions compared to 161 metastatic lesions on the early exa m. Late double-dose exam revealed 138 metastatic lesions compared to 121 met astatic lesions on the single dose early exam. The number of lesions on ea rly versus late (single/double dose) exams did not differ significa ntly (p=0.4 and p=0.11, respectively). No significant difference was detect ed between double dose and delayed single dose.

Conclusion

Although relatively more lesions were detected on d elayed exam (single/double dose), the difference did not reach significance, likely reflecting small effect size. Furthermore, no significant diff erence was found between late single and double dose.

Limitations

The study was performed at a single institution. On ly the SPACE sequence was used for blinded reading. Funding for this study: Supported by MH CZ - DRO (NNH, 00023884) No. 244302 Ethics committee - additional information: Approved on June, 5, 2024 Author Disclosures: Roman Liščák: Author: co-author Zuzana Ryznarova: Author: co-author Josef Vymazal: Consultant: Bracco Consultant: GE He althcare Aaron Rulseh: Author: corresponding author Global body exposure to gadolinium after administra tion of the Human Equivalent Dose (HED) of gadopiclenol, in compariso n to the HED of gadobutrol and gadoterate in Healthy Rats *M. Rasschaert*, E. Couloumy, C. Hollenbeck, E. Ren ard, I. Janot, N. Decout, M. Lefebvre, C. Factor, P. Robert; Roissy CdG Cedex /FR Purpose or Learning Objective: Gadopiclenol is a high relaxivity macrocyclic Gadolinium-Based Contrast Agent (mGBCA) approved in 2022 in the USA and December 2023 in Europe. The aim of this study was to evaluate the time- dependent gadolinium Gd exposure in main body organ s over 5 months in rats after a single injection of gadopiclenol or two mGB CAs.

Methods

or Background: Healthy female rats were allocated to 3 groups: gadopiclenol (Elucirem®), gadobutrol (Gadovist®) an d gadoterate (Dotarem®). One single administration of GBCA at the Human Equi valent Dose was realized, blinded. Selected tissues (including cent ral (CNS) and peripheric nervous system (PNS) organs, excretion organs, bone ) were collected for total Gd determination by Inductive Coupled Plasma-Mass S pectrometry at 1 day, 1 week, 1 or 5 months after injection (n=10/group and timepoint). The global exposure to Gd in all investigated organs was estim ated by calculating the area under the curve (AUC). Sunday Abstract-based Programme 264

Results

or Findings: After gadopiclenol administration, the Gd exposure over the 5 months compared to gadoterate and gadobutrol was respectively -43% and -41% in CNS, -28% and -32% in PNS (sciatic nerv e, spinal nodes, footpads, spinal cord), -25% and -68% in skin, -1% and -26% in liver, -19% to - 44% in kidney and -31% to -39% in spleen. In bone, the Gd exposure after gadopiclenol administration was +80% and +16% highe r in femur diaphysis, - 1% and +48% in epiphysis. In bone marrow, the AUC w as -22% and -35% lower compared to gadobutrol and gadoterate respect ively. Globally, the overall Gd exposition was -25% and -40% after gadop iclenol compared to gadoterate and gadobutrol, respectively.

Conclusion

In our experimental conditions, the measured exposu re to Gd after gadopiclenol injection is significantly reduc ed compared to the other marketed GBCAs in most of the investigated organs.

Limitations

Animal Study Funding for this study: None Ethics committee - additional information: Research project 2015080615264829 approved by French MESR Ministry Author Disclosures: Claire Hollenbeck: Employee: Guerbet Philippe Robert: Employee: Guerbet Cecile Factor: Employee: Guerbet Ilona Janot: Employee: Guerbet Nathalie Decout: Employee: Guerbet Mylène Lefebvre: Employee: Guerbet Emilie Couloumy: Employee: Guerbet Elisabeth Renard: Employee: Guerbet Marlène Rasschaert: Employee: Guerbet Use of gadolinium-based contrast agents (GBCA) over the years: impact on monitoring multiple sclerosis disease activity S. M. Sceppacuercia¹, C. Tur¹, *A. H. M. E. Hammam* ², W. H. E. Hamed², O. Sarwani², D. Deborah¹, C. Auger Acosta¹, T. A. Y ousry², A. Rovira Cañellas¹; ¹Barcelona/ES, ²London/UK ([email protected]) Purpose or Learning Objective: To describe the evolution of GBCA use, the detection of Gd-enhancing, and new T2 lesions over time in two independent cohorts of MS patients. We also examined the associ ation between new T2 lesions and clinical features, focusing on the role of GBCA use.

Methods

or Background: We included all MS patients clinical MRIs at the Multiple Sclerosis Centre of Catalonia, Spain, and the Queen Square MS Centre, UK, during May from 2015 to 2022. Clinical and demographic data included age, sex, disease duration, clinical pheno type, progression, relapses since the last MRI, and DMT variables. Brain MRI da ta included GBCA use, the number of Gd+ lesions, and new T2 lesions. Statisti cal analysis used logistic regression models adjusted for confounders.

Results

or Findings: 479 patients(cohort1), and 794 (cohort2). Both coho rts showed a significant decrease in GBCA use over time , an increase in DMT exposure (p=0.032), and a reduction in relapses (p< 0.001) and progression events (p=0.006). A significant reduction in Gd-enh ancing lesions was observed in Barcelona. The detection of new lesions also decreased over time, likely due to an older population (p<0.001) and inc reased use of high-efficacy DMTs (p<0.001). GBCA administration was independent ly associated with a higher likelihood of detecting new lesions (p=0.030 ).

Conclusion

Over time, there was a decrease in GBCA use, Gd-enh ancing lesion detection, and new/enlarging T2 lesions, sug gesting a trend towards a less aggressive disease course, possibly due to mor e effective DMTs and an aging population. GBCA use was independently associ ated with higher detection of new T2 lesions, possibly reflecting he ightened neuroradiologist vigilance when new inflammatory activity is suspect ed and the facilitation of T2 lesion detection by Gd-enhancing lesions.

Limitations

This includes the retrospective nature, a single mo nth/year representation and unequal MRIs' frequencies among years and centres. Funding for this study: A retrospective study with no fund needed. Ethics committee - additional information: A retrospective study. Author Disclosures: Sofia Maria Sceppacuercia: Nothing to disclose Tarek A. Yousry: Nothing to disclose Weaam Hamed Elsayed Hamed: Nothing to disclose Alejandro Rovira Cañellas: Other: Alex Rovira serv es/ed on scientific advisory boards for Novartis, Sanofi, Synthetic MR, BMS, Roc he, and Biogen, and has received speaker honoraria from Bayer, Sanofi, Merc k-Serono, Teva Pharmaceutical Industries Ltd, Novartis, Roche, BMS and Biogen, and is CMO and co-founder of TensorMedical. Ahmed Hassan Mohamed Elzeki Hammam: Nothing to disc lose Carmen Tur: Other: Carmen Tur has received honorari a from Roche and Novartis and is a steering committee member of the O’HAND trial and of the Consensus group on Follow-on DMTs. Cristina Auger Acosta: Nothing to disclose Omran Sarwani: Nothing to disclose Deborah Deborah: Nothing to disclose Different rates and patterns of symptoms associated with gadolinium exposure (SAGE) between linear and macrocyclic gado linium based contrast agents *I. Shahid*, E. Lancelot; Paris/FR ([email protected]) Purpose or Learning Objective: Some patients who received multiple administrations of gadolinium-based contrast agents (GBCAs) have been reported to develop “symptoms associated with gadol inium exposure” (SAGE). The aim of this study was to analyze pharmacovigila nce data and to explore the various SAGE patterns of linear and macrocyclic GBCAs among patients exhibiting three or more SAGE symptoms.

Methods

or Background: SAGE symptoms were searched by preferred terms (PTs) in different system organ class (SOC) categor ies in the FDA Adverse Event Reporting System (FAERS) database, over a 6-y ear period ranging from 2014 to 2019, for 3 linear and 3 macrocyclic GBCAs.

Results

or Findings: The analysis of FAERS data revealed a significantly higher SAGE weight for the linear GBCAs (20-24%) th an for the macrocyclic GBCAs (5-9%). For the linear GBCAs, the most preval ent combinations of 3 SAGE symptoms were reported in 152-164 occurrences, while for the macrocyclic GBCAs this range was significantly lowe r (1-13 occurrences). Moreover, the patterns of SAGE combinations differe d significantly between both categories of GBCAs.

Conclusion

Linear GBCAs are associated with higher SAGE report ing rates than macrocyclic GBCAs. They also present different patterns of SAGE combinations.

Limitations

This study did not provide any demonstration of a c ausal relationship between the reported events and GBCA a dministration to the patients. Funding for this study: None Ethics committee - additional information: Not Applicable (No patients were involved) Author Disclosures: Eric Lancelot: Employee: Guerbet Imran Shahid: Employee: Guerbet How Compressed-Sense acceleration technique impacts the Central Nervous System Magnetic Resonance Imaging energy co nsumption *S. Kalari*¹, I. Seimenis¹, E. Psatha², D. Apostolo u¹, I. Loulakas¹, E. Karavasilis²; ¹Athens/GR, ²Alexandroupolis/GR ([email protected]) Purpose or Learning Objective: Compressed-sensing is a new acquisition technique that enables efficient signal acquisition and reconstruction reducing the scanning time. The aim of our study was to esti mate the impact of CS on energy consumption during brain imaging.

Methods

or Background: Thirty individuals underwent brain MRI using the same exam protocol including 3DT1, 3DT2 and 3D Τ2 Flair with and without CS on a 3.0TPhilips MRI system. Energy consumption was recorded using kilowatt-hour energy measurement sensors (0.017 Hz sampling rate) for all the above sequences. The 3DT1, 3DT2 and 3DT2 Flair with and without CS image quality was assessed quantitatively using VolBrain free available online software. Images were also reviewed from an experie nced neuroradiologist who evaluated their total image quality and 3DT2 Fl air lesion detection sensitivity. Sequences’ signal to noise ratio (SNR) , image quality indices derived from VolBrain, qualitative indices and thei r energy consumption were introduced to SPSS to perform paired t-test statist ical analysis (p<0.05).

Results

or Findings: The mean acquisition time of 3DT2 FLAIR, 3DT2 and 3DT1 with and without CS were 210sec±0.8sec vs 390s ec±1.1sec, 200sec±0.7sec vs 370sec±1sec and 230sec±0.1sec vs 400sec±1.2sec respectively. We didn’t find any statistical differ ences in both image qualitative and quantitative quality metrics. There was no subs tantial difference of the SNR. Concerning the energy consumption/min CS incre ased by 42% the energy consumption rate in 3DT2 FLAIR (0.57kWh/min versus 0.40kWh/min) therefore the total energy consumption was decrease d by 10% (1,99 versus 2.2 kWh). 3DT2 and 3DT1 energy with and without CS consumption rate were 0.60kWh/min, 0.56kWh/min and 0.58kwh/min and 0.53kw h/min. Thus, the 3DT2 and 3DT1 total energy consumption were decreas ed 45,9% and 42,5%, respectively.

Conclusion

CS image acceleration technique has the potential t o reduce MRI scanning time and energy required while maintaining high image quality.

Limitations

Not applicable Funding for this study: No funding was provided for this study. Ethics committee - additional information: The study received institutional review board approval and written consent was obtai ned from all participants. Sunday Abstract-based Programme 265 Author Disclosures: Evlampia Psatha: Nothing to disclose Selmina Kalari: Nothing to disclose Efstratios Karavasilis: Nothing to disclose Ioannis Seimenis: Nothing to disclose Ioannis Loulakas: Nothing to disclose Dimitrios Apostolou: Nothing to disclose A methodology for reducing the environmental impact of the energy consumption of a neuroradiology MRI system using da ily electricity emission data *A. Roletto*¹, M. Verga², G. L. Viganò², S. Zanoni² ; ¹Milan/IT, ²Brescia/IT ([email protected]) Purpose or Learning Objective: Magnetic resonance imaging (MRI) scanners contribute significantly to the carbon footprint of healthcare sector, and efforts to reduce environmental impact are based on reducin g energy consumption. The aim of this work was to develop a methodology t o reduce carbon emissions by reorganising an MRI neuroradiology sys tem's diagnostic activities based on daily electricity emission data.

Methods

or Background: In July 2024, at a large public hospital of Brescia (Italy), the energy profiles of two weeks of diagno stic activity of a neuroradiology MRI system were recorded using HT GS C60 Power Meter/Analyzer (HT-Italia, Italy), calculating the total energy consumption and its distribution over days. Finally, the amount of carbon dioxide equivalent (CO2e) emitted by electricity consumption was calcu lated using open-source daily emission data (www.electricitymaps.com) and t he amount that can be saved by rescheduling diagnostic activities to lowe r emission day periods.

Results

or Findings: The total weekly energy consumption was 2305.1 kWh, spread over 5 working days, from 7am to 7pm. The mo st performed examinations were brain MRI (25 kWh/scan) and lumba r spine MRI (18 kWh/scan). Total emissions were 373.5 kgCO2e, equiv alent to emissions from one homes' energy use for six months. Examinations performed between 4pm and 7pm have the highest carbon emission index (213 .4-252.3 gCO₂eq/kWh). Rescheduling them earlier in the working day or at the weekend would reduce emissions by 18%, without any decrease of energy co nsumed.

Conclusion

This study shows how clinicians and managers can re duce carbon emissions from MRI diagnostic activity, not only by reducing energy consumption through cutting unnecessary examination s or optimising the acquisition technique, but also by rescheduling wor k activities according to daily electricity emission data, scheduling examina tions during low-carbon emission times.

Limitations

Short period of data collection limited the strengt h of the

Conclusions

of this study. Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Matteo Verga: Nothing to disclose Andrea Roletto: Nothing to disclose Gian Luca Viganò: Nothing to disclose Simone Zanoni: Nothing to disclose 08:00-09:00 Research Stage 3 Research Presentation Session: Oncologic Imaging RPS 2216 Structured reporting, radiomics and deep learning Moderator M. E. Mayerhöfer; Vienna/AT Evaluating the Impact of Structured Radiology Repor ting on Clinical Practice and Decision-Making: A Survey in a Large T ertiary University Hospital – STAR Study *M. Mancino*, G. Avesani, A. Infante, S. Gaudino, B . Merlino, L. Natale, E. Sala; Rome/IT ([email protected]) Purpose or Learning Objective: To evaluate the impact of structured radiology reports (SRRs) on clinical decision-makin g and patient management, specifically focusing on improvements in diagnostic accuracy, treatment planning, patient outcomes, and data standardizatio n. Additionally, we aim to measure clinicians' satisfaction with the clarity, comprehensiveness, and utility of SRRs in comparison to standard narrative reports .

Methods

or Background: SRRs were introduced one and a half years ago at Fondazione Policlinico Universitario A.Gemelli, a l arge tertiary university italian hospital , and have since been implemented in nearl y all radiological procedures . This transition was the result of a co mprehensive collaboration among radiologists, surgeons, and clinicians to ens ure that the structured templates met clinical needs while adhering to guid elines from major radiological societies. An extensive and anonymous survey was conducted among non-radiologist clinicians, surgeons, and res idents from various specialties to gather feedback on SRRs' clarity, cl inical impact, adaptability, research value, and efficiency compared to traditio nal reports.

Results

or Findings: Survey responses indicate increased clinician satisfaction, improved communication, and greater w orkflow efficiency with SRRs. Preliminary findings suggest better data inte rpretation, ease of retrieval, and enhanced multidisciplinary discussions, particu larly in oncology. SRRs were perceived as more effective than traditional r eports in supporting clinical decisions and improving collaboration.

Conclusion

SRRs significantly improve clinical practice by del ivering clearer, more consistent interpretations that directly enhan ce patient outcomes and support effective decision-making. The increased cl arity and standardization foster better collaboration among clinicians, ultim ately benefiting patient care. Continuous feedback from users is essential to refi ne SRRs and ensure they remain adaptable and impactful. Furthermore, SRRs p rovide a foundation for consistent data collection, crucial for advancing r esearch and supporting evidence-based practices.

Limitations

The single-center survey limits the generalizabilit y of the results. Future studies should include multiple centers to v alidate these findings. Funding for this study: No specific funding was obtained for this study. Ethics committee - additional information: This study has been notified to the Ethics Committee of Fondazione Policlinico Univ ersitario A. Gemelli IRCCS. Author Disclosures: Giacomo Avesani: Nothing to disclose Amato Infante: Nothing to disclose Simona Gaudino: Nothing to disclose Luigi Natale: Nothing to disclose Evis Sala: Nothing to disclose Matteo Mancino: Nothing to disclose Biagio Merlino: Nothing to disclose Unraveling tumour heterogeneity with radiogenomics: comparing single instance and multiple instances learning AI approac hes *D. I. Rodríguez Sánchez *, R. Spaans, S. Rostami, O. Maxouri, Z. Bodalal, R. G. H. Beets-Tan; Amsterdam/NL ([email protected]) Purpose or Learning Objective: Tumour genetic heterogeneity is a fact in cancer research. While biopsying every lesion is cl inically infeasible, imaging- based methods (such as radiogenomics) promise clini cians non-invasive insights into the underlying tumour biology. This s tudy assesses the impact of accounting for tumour heterogeneity by comparing Si ngle Instance Learning (SIL) and Multiple Instance Learning (MIL) AI appro aches.

Methods

or Background: A cohort of 1666 routine contrast-enhanced CT scans, including over 11.000 segmented lesions, was retrospectively collected along with matched next-generation sequencing data. The morphological phenotype was quantified from each lesion by radiom ic features, with subsequent feature selection using orthogonal princ ipal feature selection (OPFA) and five-fold cross-validation. SIL and MIL- based machine learning

Methods

were compared to predict the mutational sta tus of TP53, KRAS, and EGFR. Interlesional morphological heterogeneity was measured using spatial distance metrics.

Results

or Findings: For genes with established low biological variabili ty between lesions (TP53 and KRAS), accounting for tum our heterogeneity did not improve the radiogenomic predictive performance . However, for genes with high discordance (EGFR), MIL-based machine learning methods (AUC range=0.72-0.78) performed significantly better tha n SIL (AUC=0.56). These findings are also supported by interlesional hetero geneity scores, which did not differ between wild-type and mutated TP53 or KRAS c ases. Conversely, EGFR-mutated patients demonstrated significantly hi gher interlesional morphological heterogeneity than their wild-type co unterparts (p<0.0001).

Conclusion

MIL models may better reflect tumour heterogeneity, particularly in cases with high biological variability. Incorpor ating MIL into radiogenomic models may enhance their predictive accuracy by acc ounting for real-world tumour heterogeneity.

Limitations

The study's limitations include external validation . A large-scale multicenter radiogenomics dataset is currently bein g finalised to validate these results. Sunday Abstract-based Programme 266 Funding for this study: Funding was provided by the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska- Curie grant agreement number 101034290 (EMERALD Int ernational PhD Program for Medical Doctors). Ethics committee - additional information: IRB approval: IRBd19-147 Author Disclosures: Zuhir Bodalal: Nothing to disclose Robert Spaans: Nothing to disclose Diana Ivonne Rodríguez Sánchez : Nothing to disclos e Sajjad Rostami: Nothing to disclose Regina G. H. Beets-Tan: Nothing to disclose Olga Maxouri: Nothing to disclose Enhancing speed and precision of lesion tracking in follow-up lung CT using deep-learning-based registration *S. Kuckertz*¹, S. Heldmann¹, F. Peisen², J. H. Mol tz³; ¹Lübeck/DE, ²Tübingen/DE, ³Bremen/DE Purpose or Learning Objective: Continuous lesion assessment in cancer patients is integral to radiologists’ work. Part of this process is the tedious and time-consuming (re-)localisation and measurement of lesions. Fast and precise image registration facilitates the workflow by auto matically matching previous and current observations. In this study we evaluate our deep-learning-based lung registration using a longitudinal dataset incl uding expert lesion segmentations, comparing it to a state-of-the-art c onventional non-learning approach.

Methods

or Background: We follow a 3-level coarse-to-fine deep-learning registration approach. At each level, we input the baseline and follow-up CT scan at a different resolution into a U-Net, result ing in a deformation field that maps corresponding locations from baseline to follo w-up. Combining the deformations from all levels allows accurate tracki ng of anatomical points. Our

Method

is trained on 681 follow-up image pairs and evaluated on a distinct dataset consisting of 90 image pairs with 307 manua lly segmented lung lesions. We compare our approach to a non-learning GPU-accelerated registration. Each lesion centre in the baseline sc an is propagated onto the follow-up, where we check whether it maps inside th e given corresponding lesion.

Results

or Findings: With our learning-based approach, 81.1% of the baseline lesion centres were correctly mapped to th e corresponding lesion in the follow-up (conventional approach: 73.6%). The m edian distance between the propagated and the given lesion centre was 1.9 mm (conventional approach: 3.2 mm) and the average calculation time was 0.92 s (conventional approach: 14.56 s).

Conclusion

Our learning-based registration approach enhances b oth speed and accuracy, enabling precise relocation of all lu ng lesions in follow-up scans in less than a second. This facilitates radiologist s’ workflow, also enabling cursor synchronisation and change highlighting in v iewers.

Limitations

CTs were cropped to the thorax area and resampled t o an isotropic resolution of 1.5 mm. Funding for this study: Funding was provided by the Federal Ministry of Education and Research of Germany (BMBF) as part of SPIRABENE (project number 13GW0561B). Ethics committee - additional information: Only retrospective data was used for this work. The outcome had no effect on patient treatment. Author Disclosures: Sven Kuckertz: Nothing to disclose Felix Peisen: Nothing to disclose Stefan Heldmann: Nothing to disclose Jan Hendrik Moltz: Nothing to disclose Radiomic gradient in the peritumoral tissue of live r metastases: A biomarker to drive clinical practice? *A. Ammirabile*¹, F. Fiz², E. M. Ragaini¹, S. Sirch ia³, S. Viganò¹, M. Francone¹, L. Cavinato¹, E. Lanzarone³, L. Viganò¹; ¹Milan/IT, ²Genova/IT, ³Bergamo/IT ([email protected]) Purpose or Learning Objective: To investigate the variation of three textural features (mean HU, entropy, and uniformity) in the peritumoral tissue around colorectal liver metastases (CRLM) as distance from the tumor increases.

Methods

or Background: This retrospective study included all consecutive patients with histologically proven CRLM who underw ent locoregional treatment (resection/ablation) between January 2010 and December 2022. Inclusion criteria were high-quality CT with an ade quate portal phase and identifiable hypodense CRLM (>10 mm). Multiple VOIs were generated: 1) manual tumor segmentation (Tumor-VOI); 2) multiple automatic concentric rims at increasing distance from CRLM (1 to 10 millimete rs); 3) manual segmentation of a virtual biopsy of non-tumoral par enchyma (Liver-VOI). Radiomic features were extracted by the LifeX softw are. The percentage variation of index values across different VOIs was calculated, using Liver-VOI as reference. Subgroup analyses were based on tumor size (10-30 vs. >30 mm) and chemotherapy administration (no chemotherap y vs. responders).

Results

or Findings: 63 CRLM in 51 patients (median age 67 years, 14 females) were analyzed. Median peritumoral HU value s were similar to Liver- VOI, except within the 1-mm VOI around CRLM (p=0.00 2). Entropy progressively decreased (from 3.11 of CRLM to 2.54 of Liver-VOI, p<0.001) while uniformity increased (from 0.135 of CRLM to 0 .199 of Liver-VOI, p<0.001). At 10 mm from CRLM, entropy was similar t o Liver-VOI in 62% of cases and uniformity in 46%. Smaller CRLM and respo nders to chemotherapy showed higher and earlier normalization of entropy and uniformity values.

Conclusion

The radiomic analysis of peritumoral tissue in CRLM demonstrated a peculiar gradient of decreasing entr opy and increasing uniformity despite a normal radiological appearance , representing a potential biomarker for personalized clinical decision-making .

Limitations

Retrospective analysis; small cohort; heterogeneous CT data; missing correlation with pathologic/surgical data. Funding for this study: AIRC grant #2019−23822 Ethics committee - additional information: Protocol 988/22 Author Disclosures: Sara Sirchia: Nothing to disclose Angela Ammirabile: Nothing to disclose Ettore Lanzarone: Nothing to disclose Francesco Fiz: Nothing to disclose Samuele Viganò: Nothing to disclose Elisa Maria Ragaini: Nothing to disclose Marco Francone: Nothing to disclose Lara Cavinato: Nothing to disclose Luca Viganò: Nothing to disclose Integrating MRI and PET/CT Radiomics for Enhanced S urvival Prediction in Esophageal Cancer *C. Noirot*, D. Abler, L. Haefliger, S. Mantziari, M. Schäfer, N. Vietti Violi, A. Depeursinge, C. Dromain, M. Jreige; Lausanne/CH ([email protected]) Purpose or Learning Objective: Prognosis evaluation in esophageal cancer remains challenging. Accurate survival prediction i s crucial for treatment planning and follow-up strategies. Although MRI and 18F-FDG PET/CT provide valuable information, they have limitations in accu rately predicting patient outcomes. This study aimed to develop radiomics mod els based on MRI and PET/CT to predict overall survival in esophageal ca ncer patients using baseline and follow-up imaging.

Methods

or Background: Sixty patients (M/F: 50/10, mean age 66±9 years) with newly diagnosed esophageal cancer were prospec tively included (2017- 2022). Patients underwent staging with 18F-FDG PET/ CT and MRI, with follow- up MRI after neoadjuvant treatment. Tumors were man ually segmented using Mint™ Software, and radiomics features were extract ed via QuantImage v2 platform. The dataset, including 645 features from MRI and PET/CT, was split into training (80%) and test (20%) sets. Various su rvival prediction algorithms were compared. Model performance was assessed with the concordance index (C-index) using bootstrapping for confidence interv al (CI) estimation.

Results

or Findings: Radiomics features were analyzed at baseline from b oth PET/CT and MRI for 52 patients, and at follow-up MR I for 49 patients. Mean survival was of 37 months (range: 1.8 to 78.1). The MRI model (14 features) achieved a C-index of 0.733 (95% CI: 0.718–0.756), and the PET/CT model (5 features) achieved 0.724 (95% CI: 0.707–0.746) for predicting OS. A combined model with 19 features improved the C-index to 0.86 8 (95% CI: 0.853–0.881), while a follow-up MRI model (16 features) reached 0 .807 (95% CI: 0.790– 0.827).

Conclusion

The radiomics model based on MRI and PET/CT demonst rated robust performance in predicting survival for esoph ageal cancer patients. Integrating multi-modal baseline and follow-up imag ing radiomics features into survival models could enhance prognostic accuracy, improving personalized management strategies in esophageal cancer.

Limitations

The limitations of the study are the number of pati ents. Funding for this study: No funding was received. Ethics committee - additional information: The study was approved by "Commission Cantonale d'Ethique de la Recherche sur l'être humain" à Lausanne (CER-VD 2017-00388) Author Disclosures: Markus Schäfer: Nothing to disclose Camille Noirot: Nothing to disclose Adrien Depeursinge: Nothing to disclose Laura Haefliger: Nothing to disclose Daniel Abler: Nothing to disclose Styliani Mantziari: Nothing to disclose Naïk Vietti Violi: Nothing to disclose Mario Jreige: Nothing to disclose Clarisse Dromain: Nothing to disclose Sunday Abstract-based Programme 267 AI-Assisted Annotation for Improving Federated Lear ning in Automated RCC Image Segmentation K. S. Younis¹, *J. Garrett*², A. Elhanashi³, A. Gen tili⁴, S. Faghani⁵, S. Kuanar⁵, Y. Singh⁵, Y. Huo⁶, G-M. Conte⁵, J. Yacoub⁷, O. Unal²; ¹Cleveland, OH/US, ²Madison, WI/US, ³Pisa/IT, ⁴San DIego, CA/US, ⁵Rochester, MN/US, ⁶Nashville, TN/US, ⁷Washington, DC/US ([email protected]) Purpose or Learning Objective: To evaluate the impact of AI-assisted annotation tools on improving consistency in datase t labeling for federated learning, focusing on image segmentation tasks for renal cell carcinoma (RCC) in CT and MR images.

Methods

or Background: One of the major challenges in distributed learning is variability in data labels across sites, especia lly in image segmentation tasks where mask generation methods can differ significan tly. This study leverages a publicly available dataset from The Cancer Imaging Archive (TCGA-KIRC) for RCC, which contains rich imaging and clinical data. AI-assisted annotation tools standardize data labeling before training mod els such as Unet and Swin UNETr. The primary goal is to assess if these tools enhance model performance or reduce the number of cases needed fo r effective training in federated learning environments. Radiologists from various institutions across multiple states in the US manually annotated the im ages and federated learning was conducted using NVIDIA nvFLARE.

Results

or Findings: Preliminary results suggest that AI-assisted annota tion improves model consistency, with segmentation accur acy increasing by approximately 12% when compared to non-standardized data labels. The model efficiency was also reflected in reduced data redundancy and higher annotation agreement between annotators. This enhan cement allowed for more precise radiomic analysis across datasets. Add itionally, the number of samples required for training decreased by 20%, ind icating the efficiency of AI- assisted annotation in generating reliable training datasets.

Conclusion

AI-assisted annotation holds promise for improving performance and efficiency in federated learning environments, particularly for automated image segmentation tasks like RCC detection. The en hanced consistency in label generation helps to reduce the variability in troduced by multiple sites, thereby improving model generalizability.

Limitations

The study does not include external validation acro ss diverse imaging platforms, and annotation tools were not ev aluated for real-time performance during federated learning. Funding for this study: Nvidia Education grant Ethics committee - additional information: N/A Author Disclosures: Yashbir Singh: Nothing to disclose Abdussalam Elhanashi: Nothing to disclose Orhan Unal: Nothing to disclose Amilcare Gentili: Nothing to disclose Shahriar Faghani: Nothing to disclose Joseph Yacoub: Consultant: Microsoft Nuance Khaled Salem Younis: Nothing to disclose Yuankai Huo: Nothing to disclose Shiba Kuanar: Nothing to disclose Gian-Marco Conte: Nothing to disclose John Garrett: Grant Recipient: Research Support fro m the NIH, GE Healthcare, and the American Cancer Society Advisory Board: Rad Unity Corp. Shareholder: NVIDIA, Inc. and RadUnity Corp. 08:00-09:00 Research Stage 4 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 2205 Imaging informatics, quality and new techniques Moderator L. Lofino; Milan/IT ([email protected]) Universal Medical Imaging Encoding Datasets - A new standard for creating large-scale diagnostic imaging datasets *B. O. Klaudel*¹, A. Obuchowski², A. Komor¹, P. Fr ąckowski¹, K. Rogala¹, K. Knitter¹; ¹Gdańsk/PL, ²Banino/PL ([email protected]) Purpose or Learning Objective: We present a new standard for creating and unifying large-scale diagnostic imaging datasets. I t aims to address the challenges in medical AI, particularly the lack of standardized, comprehensive data for training foundation models in medical imag ing. By combining multiple open-source datasets, unifying them to a common ont ology and providing standardized preprocessing pipelines, we seek to ac celerate the development of more generalized and robust medical AI models.

Methods

or Background: We combined over 20 open-source datasets, resulting in more than one million annotated medica l images, including CT, MRI, and X-ray modalities. We created reusable prep rocessing pipelines to transform diverse source datasets into a unified fo rmat. A key innovation is the adoption of the RadLex ontology, developed by the R adiological Society of North America, to standardize labels and masks acro ss all included datasets. It addresses issues of inconsistent labeling, regional variations, and differing granularity in existing medical imaging datasets.

Results

or Findings: The project has created the largest publicly availa ble dataset of annotated radiological imaging to date, with over one million images, 40+ labels, and 15 annotation masks. The datasets a re accompanied by ready- to-use preprocessing pipelines that can be easily a dapted to incorporate new data sources. The unified ontology based on RadLex enables consistent labeling, facilitating more effective model trainin g and cross-dataset compatibility.

Conclusion

UMIE datasets represent a significant advancement i n medical imaging AI, providing a standardized foundation for developing more generalized models. By addressing the challenges of data scarcity, inconsistent formatting, and labeling discrepancies , it paves the way for more robust and widely applicable AI solutions The open- source nature of the project encourages collaboration and further expansion of t he datasets.

Limitations

UMIE datasets inherits potential biases and incompl ete labelling coverage from source datasets, and limitations of t he RadLex ontology. Funding for this study: No funding was required. Ethics committee - additional information: All data used in this study come from existing opensource datasets. Author Disclosures: Andrzej Komor: Nothing to disclose Kacper Rogala: Nothing to disclose Barbara Olga Klaudel: Nothing to disclose Aleksander Obuchowski: Nothing to disclose Kacper Knitter: Nothing to disclose Piotr Frąckowski: Nothing to disclose Analysis of key principles for improving the effici ency of medical data annotation processes for machine learning *P. Pilius*¹, N. Smirnov²; ¹Almaty/KZ, ²Haar/DE ([email protected]) Purpose or Learning Objective: The aim of this work is to analyze the fundamental principles of project management and ev aluate potential opportunities for its implementation for improving efficiency in the preparation of medical data for machine learning: for data coll ection, working with annotators, and results validation.

Methods

or Background: We analyzed personal experiences from 9 projects conducted during 2022-2024 years, aimed at creating annotated medical datasets with various levels of complexity and acro ss different fields. The clients were private companies developing artificia l intelligence technologies, and the annotators included certified radiologists, residents, and radiographers. Sunday Abstract-based Programme 268 To enhance efficiency, we studied and applied princ iples from an advanced training course on the foundations of project manag ement.

Results

or Findings: The duration of the projects ranged from 3 weeks to 3 months. The number of annotated datasets included: 75 MRI, 3720 CT, 55,500 X-rays, 6000 angiograms, 1700 mammograms, and 3200 radiological text report annotations. All projects were completed on time and met quality requirements. Based on practical experience and the implementation of fundamental project management concepts, we've form ulated recommendations and standards for conducting simila r projects.

Conclusion

The organization of medical data annotation process es for machine learning presents numerous specific challen ges. To optimize the workflow and achieve desired outcomes, it is essent ial to have not only sufficient experience in handling medical images bu t also a solid understanding of fundamental project management principles.

Limitations

The work is based on personal experience and observ ations and is primarily of a recommendatory nature. It is not possible to objectively compare the outcomes of projects where basic manage ment principles were and were not applied, as the projects were not repe ated under identical conditions. However, the variability and uniqueness of each project allowed for a comprehensive analysis, leading to the formulatio n of universal standards. Funding for this study: No funding Ethics committee - additional information: Kazakh National Scientific and Research Center of Oncology and Radiology Author Disclosures: Polina Pilius: Nothing to disclose Nikolai Smirnov: Nothing to disclose Data Interoperability in a Clinical Pathway From Fr ee-Text Reports *K. Nairz*, N. Cihoric, F. Dennstädt, M. Schmerder, H. Bonel, H. Von Tengg-Kobligk; Bern/CH ([email protected]) Purpose or Learning Objective: Structured reporting (SR) in radiology has been shown to be the most favorable scheme to provi de imaging information to referrers. Implementation of SR is associated with additional effort for the radiologists, but there are recent advances with La rge Language Models (LLMs) that are prompted to convert free text into a structured format. We aimed at killing two birds with a stone by leveragi ng LLMs to generate structured data that enhances interoperability as w ell. As a proof of concept we we selected breast cancer patient pathways for docu mentation. We structured the free-text information from various sources, inc luding health interviews, mammography reports, biopsy results, pathology find ings, and tumor board discussions, to ensure that the data could be effec tively transmitted and support therapeutic decision-making.

Methods

or Background: Our approach is based on the use of Common Data Elements (CDEs), which are minimal information unit s or precisely defined questions associated with a set of standardized ans wers, each having explicitly defined values. Building on in-depth analyses of re ports and guidelines such as BI-RADS, we defined corresponding sets of Common Da ta Elements and created templates to establish a structured format. To ensure data protection we utilized a locally installed LLM (Llama 3), whic h was prompted to answer CDE-based questions. This process enabled the mappi ng of free-text content to a structured format, which was ultimately stored in FHIR compatible JSON format.

Results

or Findings: By defining key CDE values, or specific combination s of these values, clinicians can identify critical insi ghts that may suggest a particular therapeutic course or provide predictive indicators for patient outcomes.

Conclusion

Prompting LLMs to answer CDE-based structures prove s to be a viable approach to promote data interoperability in complex medical settings.

Limitations

The study focuses specifically on breast cancer pat hways. Funding for this study: Innosuisse 59228.1 Ethics committee - additional information: Kantonale Ethikkommission Bern Author Disclosures: Harald Bonel: Nothing to disclose Max Schmerder: Nothing to disclose Nikola Cihoric: Nothing to disclose Knud Nairz: Nothing to disclose Fabio Dennstädt: Nothing to disclose Hendrik Von Tengg-Kobligk: Nothing to disclose Deep learning based automated field of view positio ning for prostate magnetic resonance imaging *A. S. Quinsten*¹, A. Wetter², M. Raczkowski³, L. T rembecki³, R. Buchkremer⁴, D. Matusiewicz¹, K. Nassenstein¹, M. Forsting¹, A. Demircioglu¹; ¹Essen/DE, ²Hamburg/DE, ³Wrocław/PL, ⁴Düsseldorf/DE Purpose or Learning Objective: Prostate magnetic resonance imaging (MRI) is typically conducted according to manual prescrip tions by radiographers. This approach is time-consuming, error-prone, inconsiste nt due to rater variability, and has low reproducibility. The aim of the study w as to develop a deep learning-based framework for the automatic planning of the field of view (FoV) in the oblique coronal and axial planes in prostate MRI according to Prostate Imaging Reporting and Data System (PI-RADS) guideli nes.

Methods

or Background: The retrospective multicentre study included 2109 patients from diagnostic (Sites I and III) and radi otherapy (Site II) centres. The variability within and between raters was evaluated by three assessors. Three distinct deep neural networks were developed with t he objective of predicting the oblique coronal and axial FoV. The optimal netw ork was evaluated on three external cohorts using a non-inferiority test, and its clinical utility was assessed.

Results

or Findings: The optimal model demonstrated non-inferior performance, with slice position differences rangin g from 0.21 ± 0.99 and 0.37 ± 0.48. At Sites I and III, the predictions were predominantly non-inferior, with FoV overlaps of 86.6 ± 5.8% and 88.7 ± 6.0% and angle differences of 4.66 ± 4.89° (Site I) as well as 3.46 ± 2.80° (Site III). In contrast, the predictions for Site II demonstrated inferior overlap (67.0 ± 9.7% and 63.6 ± 8.8%) and higher angle differences (9.18 ± 9.49°). Consequently, the clinical utility was excellent for Sites I and III (97.9–100%) but lower for Site II (85.3–89.0%).

Conclusion

The utilisation of a deep learning-based framework for the automated positioning of the FoV in oblique coronal and axial planes for prostate MRI is a viable approach, exhibiting high clinical utility.

Limitations

The present study did not include images acquired w ith the endorectal coil. Funding for this study: No funding was provided for this study Ethics committee - additional information: The ethics committee notification can be found under the number 22-10740-BO. Author Disclosures: Michael Forsting: Nothing to disclose Kai Nassenstein: Nothing to disclose Anton Sheahan Quinsten: Nothing to disclose Rüdiger Buchkremer: Nothing to disclose David Matusiewicz: Nothing to disclose Axel Wetter: Nothing to disclose Lukasz Trembecki: Nothing to disclose Aydin Demircioglu: Nothing to disclose Maciej Raczkowski: Nothing to disclose RADAR - real-time automated detection and analysis of radiopaque devices using CT topograms *C. S. Schmidt*, M. Walter, J. Haubold, F. Nensa, R . Hosch; Essen/DE Purpose or Learning Objective: The aim of this study was to develop a deep learning (DL) model for the automatic detection and localisation of medical devices known to cause metal artefacts in CT images , utilising their corresponding topograms.

Methods

or Background: A dataset of 943 CT topograms with radiopaque medical devices was manually annotated via box labe lling by a radiology resident with three years of experience in CT imagi ng. The following classes were defined: cochlear implant, cardiac conduction device (pacemaker, defibrillator, stimulator), implanted port, prosthe tic heart valve, (embolisation) coil, osteosynthesis (nail-, plate-, screw-, and wi re-fixation, spinal instrumentation hardware), sternal wires, external fixation hardware, hip prosthesis, shoulder prosthesis, knee prosthesis, d enture (prosthesis, implant). An 80/10/10% split for training, validation and tes ting was performed and the YOLO11X model was trained for 100 epochs. The model was evaluated using mAP50 scores, precision (P) and recall (R).

Results

or Findings: The model achieved an average mAP50 score of 0.83, Precision of 0.86 and Recall of 0.79 over all class es and the following (mAP50/P/R) scores for the respective classes: coch lear implant (0.92/0.93/0.85), cardiac conduction device (0.90/0 .82/0.93), implanted port (0.91/0.97/0.79), prosthetic heart valve (0.86/0.82 /0.78), coil (0.99/0.98/1), osteosynthesis (0.63/1/0.55), sternal wires (0.51/0 .74/0.57), external fixation hardware (0.52/0.51/0.33), hip prosthesis (0.99/0.8 9/1), shoulder prosthesis (0.88/0.88/0.86), knee prosthesis (0.99/0.95/1), de nture (0.88/0.81/0.76).

Conclusion

The presented model demonstrates an accurate detect ion of most radiopaque medical devices in CT scout images. It could thus be utilised as an efficient orchestration tool for selecting a cohort of high quality imaging studies without interfering artefacts.

Limitations

The limitations of the study are its small sample s ize and that scout images were annotated by a single observer. A dditionally, certain medical devices can be challenging to identify and localise on topograms, which could cause relevant features to go undetecte d. Funding for this study: No funding was received for this study. Ethics committee - additional information: Informed consent was waived by the ethics committee due to the retrospective setti ng. Author Disclosures: Johannes Haubold: Nothing to disclose Cynthia Sabrina Schmidt: Nothing to disclose Marie Walter: Nothing to disclose René Hosch: Nothing to disclose Felix Nensa: Nothing to disclose Sunday Abstract-based Programme 269 Evaluating the Impact of Quantum Technology on Radi omics: A Comparative Study of Classical and Quantum Random F orest Models *F. Mariotti*, A. Agostini, A. Borgheresi, L. Pierp aoli, F. Ricciardiello, A. Zannotti, D. Nicolini, A. B. Galosi, A. Giovagno ni; Ancona/IT ([email protected]) Purpose or Learning Objective: This study aims to evaluate the impact of quantum technology on the performance of radiomics random forest (RF) models for medical imaging. We simulated a semi-qua ntum approach, involving quantum embedding followed by classical R F, and a fully quantum approach using a quantum random forest (QRF) model.

Methods

or Background: We used three radiomic datasets: 1. Perineural infiltration of peripancreatic fat in pancreatic ad enocarcinoma on CT, 2. Characterization of renal nodules in CT, 3. Predict ion of LI-RADS category on abbreviated MRI protocols. For the quantum approach es, we compared the original random forest (RF) models with simulated q uantum-embedded RF and quantum random forest (QRF) algorithms, implemented in Python using an 8- qubit configuration. The comparison involved analyz ing the accuracy and the receiver operating characteristic (ROC) curves usin g statistical significance set at p-values < 0,05

Results

or Findings: The classical RF achieved the highest accuracy for the pancreas (0.9167) and kidney (0.8571) datasets. For the liver dataset, both the quantum embedding RF and QRF outperformed the class ical approach (0.8462 vs. 0.7692), with the ROC curves showing st atistically significant improvement (p < 0.01). In the pancreas dataset, qu antum methods showed slightly lower accuracy (0.8333), and for the kidne y, they also performed worse (0.7857). This indicates that the benefits of quant um approaches may be data- dependent, providing advantages in some cases but n ot yielding consistent improvements across all datasets.

Conclusion

Quantum machine learning is a feasible approach for radiomic datasets, showing variable results and the potentia l to outperform classical methods. However, the variability in performance su ggests that fine-tuning of quantum algorithms may be necessary depending on th e specific characteristics of each dataset.

Limitations

Small datasets used and simulation of quantum proce sses with a 8-qubit setup. Further research should involve larg er datasets and physical quantum devices. Funding for this study: This study did not receive any specific funding fro m public, commercial, or not-for-profit sectors. The research was conducted without external financial support. Ethics committee - additional information: Not Applicable Author Disclosures: Daniele Nicolini: Nothing to disclose Alice Zannotti: Nothing to disclose Luca Pierpaoli: Nothing to disclose Andrea Agostini: Nothing to disclose Francesco Ricciardiello: Nothing to disclose Francesco Mariotti: Nothing to disclose Andrea Benedetto Galosi: Nothing to disclose Alessandra Borgheresi: Nothing to disclose Andrea Giovagnoni: Nothing to disclose A new framework for 3D data representation in Exten ded Reality (XR) on iPhone, iPad and Apple Vision Pro *A. M. C. Boehner*, A. Jacob, A. Isaak, C. C. Piepe r, J. A. Luetkens, D. Kütting; Bonn/DE ([email protected]) Purpose or Learning Objective: 3D data is rarely spatially displayed in routine. However, patient-clinician and clinician-c linician interaction may benefit from such representation in Extended Reality (XR). Additionally, radiologists may aid surgeons during surgery via audiovisual com munication to demonstrate 3D data if needed.

Methods

or Background: We developed and tested a workflow integrating different software platforms (e.g.‘Medical Imaging XR’, ‘Fiji’) to display DICOM images on iPhone and iPad (n=35) and Apple Vision P ro (AVP, n=10). The system enables fused XR visualization of CT, MRI, P ET. Handheld devices were utilized to aid sonographic correlations of he patic lesions (n=11); by surgeons during surgery preparations (n=10); and fo r patient information (n=14). Integrated systems were tested in a mock au diovisual call from the operating room via the AVP to the other devices loc ated on and off campus.

Results

or Findings: Our framework allowed for fast integration of 3D datasets across devices with low computational burd en. XR during sonographic correlation of hepatic lesions signific antly reduced the time needed to identify lesions from 4:50min to 2:45min (P<0.05). Patient reported full acceptance of XR usage. AVP allowed real-time image-data and view sharing between the radiologist and surgeon.

Conclusion

Integration of XR across smartphones, tablets and A VP enhanced medical imaging communication between all parties, reducing time to locate lesions and improving patient-physician i nteractions. AVP further facilitates sterile audiovisual communication betwe en surgeons and radiologists during procedures, allowing for remote and swift consultation without leaving the sterile field.

Limitations

Our method was tested exclusively on Apple products , limiting its generalizability to other platforms. Funding for this study: This project was part of the ISMC, funded by the Ministry of Economic Affairs, Innovation, Digitaliz ation and Energy of the state of North Rhine-Westphalia Ethics committee - additional information: Ethics University Hospital Bonn, Germany (2024-228-BO) Author Disclosures: Julian Alexander Luetkens: Nothing to disclose Alexander Isaak: Nothing to disclose Alexander Marc Christian Boehner: Nothing to disclo se Alice Jacob: Nothing to disclose Daniel Kütting: Nothing to disclose Claus Christian Pieper: Nothing to disclose 08:00-09:00 Room G1 Research Presentation Session: Radiographers RPS 2214 Advancing radiography through education and research: innovations, challenges, and future directions Moderators B. Horehledova; Heerlen/NL ([email protected]) J. Santos; Coimbra/PT ([email protected]) The College of Radiographers’ Education and Career Framework (fourth edition): Exploring the guideline implementation ga p across England using Normalisation Process Theory *H. L. Spencer*¹, K. Williamson², A. Robertson², M. N. K. Anudjo¹, C. Burton³; ¹Bournemouth/UK, ²London/UK, ³Norwich/UK ([email protected]) Purpose or Learning Objective: In 2022, the College of Radiographers (CoR) published the fourth edition of their Education and Career Framework (ECF). This essential document provides a professional blu eprint for the radiography career trajectory with the overarching aim of impro ving patient outcomes. However, publication does not guarantee implementat ion; there often exists a dissonance between policy intent and policy in-acti on. Therefore, if we are to access the full benefits of the ECF, its implementa tion requires careful consideration.

Methods

or Background: To advance our understanding of the translational gap between policy and practice, this observational mixed-methods study employed Normalisation Process Theory (NPT) as a th eoretical frame. Focusing on the diagnostic radiography profession i n England, a national consultation survey was deployed, alongside four co nsultation workshops. The data collection methods were underpinned by NPT. Th e framework approach was adapted for the qualitative data analysis. The quantitative survey data, meanwhile, was analysed using descriptive and infer ential statistics.

Results

or Findings: The data collection took place between April-June 2 023. In total, 142 survey responses were returned. Each workshop was comprised of 7-11 participants. The findings were deductively interpreted through the lens of NPT, from which five core themes emerged: making sense of complexity (coherence); bringing people together (cognitive pa rticipation); being strategic (collective action); evaluating complexity (reflexi ve monitoring); implementation in the ‘real world’ (barriers and enablers).

Conclusion

By furthering our understanding of the guideline im plementation gap, it was then possible to propose recommendation s to enhance the ECF’s adoption. The recommendations were study-derived, l inked to responsible stakeholders, and grouped into four strategic prior ities, aligned with the NPT domains. Through these evidence-based recommendatio ns, it is hoped the ECF can be translated more fully from page to the ‘ real world’ for the benefit of the profession and its service users.

Limitations

Nonapplicable. Funding for this study: This study was undertaken as part of a Clinical Education Improvement Fellowship secondment, suppor ted by NHS England (South East) Workforce, Training, and Education, Ca nterbury Christ Church Sunday Abstract-based Programme 270 University, and the Florence Nightingale Foundation . However, no direct funding was received for this study. Ethics committee - additional information: Canterbury Christ Church University Faculty of Medicine, Health, and Social Care Ethics Panel (Reference: ETH2223-0262). Author Disclosures: Christopher Burton: Nothing to disclose Messiah Narh Kwame Anudjo: Nothing to disclose Holly Louise Spencer: Nothing to disclose Kathryn Williamson: Employee: Society and College o f Radiographers Amy Robertson: Employee: Society and College of Rad iographers Newly qualified radiographers' perception of the in duction programme in a radiology department: a survey study *L. Bombelli*, G. R. Bonfitto, A. Roletto, E. Scara melli, S. V. Fasulo, D. Catania; Milan/IT ([email protected]) Purpose or Learning Objective: The role of radiographers is rapidly evolving, putting them in a key role in a contest of increasi ng complexities in patient care. With the rising demand for specialized skills , it is crucial to implement structured induction programs for newly qualified r adiographers (NQR), also to avoid poor workplace performance and even the dismi ssal of workers. This study aims to explore the perceptions of NQR who ha ve already experienced an induction programme.

Methods

or Background: A survey was distributed to NQR in a large university hospital in Italy. The questionnaire gat hered demographic data and by using a 5-point Likert scale assessed 29 sentenc es concerning issues in their work, including patient management, decision- making, work organization, self-confidence development and relationships with other team members.

Results

or Findings: Twenty-two NQR participated in the survey. Among these, 32% (n=7) graduated within the last year and 46% (n=10) reported that they had their first experience in a healthcare set ting. Regarding Likert scale evaluation, only 10 out of 29 sentences received sc ores of 4 or higher, indicating “Agreement”. Participants with prior wor k experiences reported greater self-confidence in their skills once the in duction program has been completed. Conversely, radiographers with no work e xperience indicated feeling more supported by management.

Conclusion

In conclusion, NQR involved in this study felt adeq uately prepared for clinical practice after completing the induction program, despite some differences between radiographers with differe nt levels of experiences. A successful induction program for NQR is essential t o foster a proactive mindset, promote appropriate work methods, enhance collaboration among team members, reduce radiographers’ stress, turnove r and ensure a high quality of patient care.

Limitations

Quantitative study design and limited sample size m ay have caused limitations. Funding for this study: No funding for the study. Ethics committee - additional information: No ethic committe Author Disclosures: Luca Bombelli: Nothing to disclose Simone Vito Fasulo: Nothing to disclose Diego Catania: Nothing to disclose Andrea Roletto: Nothing to disclose Elena Scaramelli: Nothing to disclose Giuseppe Roberto Bonfitto: Nothing to disclose Patient, Public and Practitioner Partnership within Imaging and Radiotherapy: An exploration of the implementation and use of the College of Radiographers Guiding Principles *R. M. Strudwick*¹, A. Ramlaul², P. Shuttleworth³, C. Fiyebor¹; ¹Ipswich/UK, ²High Wycombe/UK, ³Leeds/UK ([email protected]) Purpose or Learning Objective: In 2014 the National Health Service (NHS) released the Five Year Forward plan1, envisioning a shift in power from health professionals to patients and the public. In respon se the Society and College of Radiographers (SCoR) produced the “Patient, Public and Practitioner Partnership within Imaging and Radiotherapy: Guidin g Principles” (P4) document which was implemented within four domains of radiography practice; service delivery, service development, education an d research2. This project explored how these guidelines were implemented; and whether improvement to the quality and scope were needed, leading to ma king recommendations for updating the document.

Methods

or Background: A qualitative methodological framework was adopted with two phases. Phase 1 – a survey explori ng use of the P4 document’s guiding principles. There was no maximum number of participants to ensure inclusivity. Phase 2 - six focus groups f rom the four domains3.

Results

or Findings: 626 participants completed the phase 1 survey. 18.8 5% (n=118) of participants were aware of the document and used it as a reference tool for practice, teaching, and research. 81.15% ( n=508) of participants stated they were unaware of the document. Themes from phas e 2; importance of service user involvement in service delivery and ev aluation, resources to ensure service user involvement, suggestions to upd ate the P4 document and use of the P4 document in radiographer education. P articipants acknowledged guidance in the document was best practice. They re ported more awareness of patients’ needs and the effect this has on radiogra phers in supporting their needs.

Conclusion

Participants recommended the document be given grea ter visibility. The voices of patients and the public m ust be heard within radiography practice.

Limitations

Small sample size Funding for this study: Feedback from this study can be used for the future development of the P4 document. Ethics committee - additional information: University of Suffolk Ethics committee approval Author Disclosures: Pamela Shuttleworth: Nothing to disclose Aarthi Ramlaul: Nothing to disclose Ruth Mary Strudwick: Grant Recipient: CoRIPS funded project Chioma Fiyebor: Nothing to disclose Patients’ perception of Radiographers’ communicatio n skills during planar X-ray imaging: a single centre study *F. Zarb*, P. Bezzina, D. Ciantar; Msida/MT ([email protected]) Purpose or Learning Objective: To explore patients’ perception of radiographers’ communication skills during planar x -ray imaging examinations at a general hospital in Malta

Methods

or Background: A random sample of patients attending for planar x- ray examinations at a medical imaging department at a general hospital in Malta filled out a self-designed questionnaire cons isting of demographic data and scored a series of statements on a Likert scale of 1-4, with 1 being ‘Very unsatisfied’, 2 being ‘Unsatisfied’, 3 being ‘Satis fied’ and 4 being ‘Very satisfied’. Data was collected and analysed to iden tify trends and understand patients’ perception of radiographers’ communicatio n skills. The Friedman test and Kruskal Wallis test were used to analyse the ga thered data.

Results

or Findings: Mean scores were either 3-Satisfied or 4-Very Satis fied. There were no significant (p>0.05) discrepancies wh en comparing the different demographics. Females provided significantly (p<0.0 5) higher mean rating scores than males. Patients’ scores varied the most in the statement ‘Radiographers informed me of what they had to do n ext’.

Conclusion

This is the first study of its kind performed local ly evaluating patients’ perceptions of Radiographers communicatio n skills. The mechanism in place locally for training radiographers in pati ent communication appears to be effective, as patients are given instructions cl early and without undue discomfort. To make it simpler for patients once th ey have completed the x-ray and put their minds at ease during their entire sta y at the hospital, a clearer pathway should be implemented. Having radiographers with good communication skills improves the patients’ experie nce when attending for imaging services.

Limitations

Limited sample size, honesty of participants fillin g the questionnaire and a number of radiographers being f oreigners making communication a challenge. Funding for this study: No funding was received for this study. Ethics committee - additional information: Ethical permission for this study was sought and obtained from the Faculty of Health Science Research Ethics Committee (FREC), University of Malta. Author Disclosures: Francis Zarb: Nothing to disclose Paul Bezzina: Nothing to disclose Dean Ciantar: Nothing to disclose A post-graduation expectation analysis of Italian R adiographers. The OPEN project *A. Masperi*; Abbiategrasso/IT ([email protected]) Purpose or Learning Objective: The aim of this study was to explore student satisfaction with the OPEN project via a survey and to identify factors influencing radiographers' post-graduation decision s.

Methods

or Background: Radiography undergraduates face changing career paths, influenced by technology, aspirations, finan cial gain and post-graduate education choices. In January 2024 the Radiographer bachelor’s degree course at the University of Milan launched the OPEN project aimed at creating a new postgraduate orientation programme. Following AMEE guidelines, a semi-structured survey was sent to 17 undergraduate students in the OPEN project to identify factors influencing their post- graduation decisions. University of Milan protocols ensured consent, anonymity, and confidentiality. The survey, Sunday Abstract-based Programme 271 in three parts with Likert scale questions, showed good internal consistency via Cronbach's alpha.

Results

or Findings: Out of 17 participants, 16 consented to participate (94%), with all attending more than 50% of meetings . The internal consistency of the survey was excellent (αC = 0.83551) and results were expressed in terms of median and interquartile range. Students p rioritise careers in research-oriented hospitals with opportunities for research and academic advancement. They seek facilities that offer perman ent contracts, incentivise clinical activities for salary growth and a balance d working environment.

Conclusion

The Radiography degree programme at the University of Milan has demonstrated the effectiveness of postgraduate orientation programmes that bridge academic and career gaps.

Limitations

It should be noted that the results of this new pro ject are based on a limited cohort of subjects from a single unive rsity. Furthermore, gender preferences were not explored, which could have pro vided valuable insights to the survey by broadening the field of interview. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Andrea Masperi: Nothing to disclose Radiation Awareness and Occupational Concerns Among Radiographers and Students K. Brage¹, J. Jensen¹, O. Brage¹, *M. W. Kusk*², P. L. Hansen¹, M. Roland Pedersen³, H. Precht⁴; ¹Odense/DK, ²Esbjerg/DK, ³Vejle/DK, ⁴Kolding/DK ([email protected]) Purpose or Learning Objective: To assess the knowledge, perceptions, and concerns of professionals working with ionizing rad iation (IR) regarding their occupational exposure and its implications on their health, fertility, and offspring.

Methods

or Background: This cross-sectional survey included European radiographers and students and was distributed via social media and professional bodies (EFRS and Danish Society of Rad iographers) from March to July 2024. The questionnaire covered demographic s, knowledge of IR, and concerns, using a five-point Likert scale.

Results

or Findings: A total of 629 participants from 32 countries were included: 414 women (mean age 34.34 ± 12.18), 208 m en (mean age 38.27 ± 11.81), and 7 non-binary or undisclosed individuals . Of these 28.30% were students. Mean exposure time was 10.69 years ± 10.6 8. Overall, 29.77% of respondents agreed that their radiation exposure co uld negatively impact their health, 23.75% expressed concerns about fertility, and 18.59% about their children's health. No significant differences were observed between sex in these responses. Of the radiographers 0.44% of did not fully understand the occupational risks of IR, 5.77% were dissatisfied w ith their radiation protection education, and 4.66% felt unqualified to inform pat ients. Additionally, 1.99% were unsure about self-protection, and 4.00% felt t hey lacked the necessary protection means. Regarding the International Basic Safety Standard, a total of 28.95% felt not up to date while the number was 12. 42% for the National Legislation.

Conclusion

This study highlights concerns regarding IR and rel ated health with up to 30% expressing concerns. While only a mi nority of respondents felt they needed more education or resources to protect themselves, a third lacked knowledge on the International Basic Safety Standar ds.

Limitations

Selection bias could be present as this topic may h ave attracted respondents with greater concerns. Funding for this study: None Ethics committee - additional information: The University of Southern Denmark Research Ethics Committee (23/70920) approv ed this project on 8 December 2023. Before accessing the questionnaire, participants were informed about the study's purpose and were assured of the confidentiality of their responses. Only those who provided their info rmed consent proceeded to complete the questionnaire. Author Disclosures: Helle Precht: Nothing to disclose Karen Brage: Nothing to disclose Martin Weber Kusk: Nothing to disclose Pernille Lund Hansen: Nothing to disclose Janni Jensen: Nothing to disclose Malene Roland Pedersen: Nothing to disclose Oliver Brage: Nothing to disclose Abdominal Ultrasound Simulation based on CT examina tions as an educational tool for enhancing Ultrasound acquisiti on competences of radiography students R. S. T. Ribeiro, C. Schiesser, *C. Campeanu*, C. S . D. Reis; Lausanne/CH ([email protected]) Purpose or Learning Objective: To assess the effectiveness of simulate d CT examinations as an educational intervention for enh ancing radiography students' competencies in ultrasound(US).

Methods

or Background: A pilot study was conducted with third-year students enrolled in a four-week US module. Partici pants had prior knowledge of physics but no US practical experience. The educ ational intervention utilised abdominal CT examinations to enhance competences in US image acquisition, analysis and transducer positioning. A 20 images se t comprising abdominal anatomy was administered before and after the modul e. Assessments were made using a Likert-scale across 4 categories: anat omical identification, anatomical topographical correlation, sectional pla nes recognition, transducer positioning. Data were analysed using measures of c entral tendency and dispersion to assess improvements.

Results

or Findings: The simulated US based on CT examinations improved students' US competencies. Anatomical Identificatio n scores increased by 49.8%, mainly in the gallbladder (2.30-point increa se), liver (1.85-points), and pancreas (1.55-points). Anatomical Topographical Co rrelation improved by 34.2%, particularly for the pancreas (1.80-point) a nd gallbladder (1.50-points), reflecting enhanced spatial understanding. Sectiona l Planes Recognition was enhanced by 46.5%, for gallbladder (2.40-point) and kidneys (1.86 ), demonstrating better comprehension of cross-section al anatomy. Transducer Positioning presented a a gain of 66.1%, with impro vements for gallbladder (2.60) and kidneys (2.27). Overall, the interventio n effectively addressed initial skill gaps, leading to advancements in both specifi c anatomical structures analysis and hands-on practical competencies in US.

Conclusion

The use of CT examinations proved to be a valuable tool to improve not only anatomical identification but also students’ spatial understanding and practical application skills in U S. The greatest improvements were observed in areas that were initi ally the most challenging, such as pancreas and gallbladder identification and transducer positioning.

Limitations

Only abdominal anatomy was tested. Reduced student panel (5 students) Funding for this study: Not applicable Ethics committee - additional information: Not applicable Author Disclosures: Cosmin Campeanu: Nothing to disclose Claudia Sa Dos Reis: Nothing to disclose Ricardo Silva Teresa Ribeiro: Nothing to disclose Claire Schiesser: Nothing to disclose Evaluating internship guides: Can student opinion c reate a virtuous circle? *A. Devetti*¹, S. Da Dalt¹, F. R. Fabris², L. Ceres er¹, M. G. Belgrano², R. Girometti¹, C. Zuiani¹; ¹Udine/IT, ²Trieste/IT ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the efficacy of a peer-assessment tool designed to enhance the qual ity of clinical tutoring experiences for undergraduate students of a Radiogr apher Bachelor Degree internship program. By collecting student feedback on their tutors' performance, the institution sought to identify str engths, weaknesses, and implement targeted improvement strategies.

Methods

or Background: Over three academic years, 940 student evaluations were collected from 180 clinical tutors across four departments (Radiotherapy, Nuclear Medicine, Medical Physics, a nd Diagnostics). Tutors were evaluated on seven dimensions using a 10-point Likert scale. The reliability of the evaluation tool was assessed usi ng Cronbach's alpha, which consistently yielded values between 0.92 and 0.94.

Results

or Findings: The results indicate a high overall level of studen t satisfaction with the clinical tutoring experience. The mean overall rating was 8.87 out of 10, with a standard deviation of 0.97. While slight variations were observed across departments, these differences were not statistically significant. The high Cronbach's alpha values sugge st that the evaluation tool is reliable and internally consistent.

Conclusion

The implementation of a peer-assessment tool has pr oven to be an effective method for gathering valuable feedback on the quality of clinical tutoring. By providing tutors with individualized f eedback, the institution has fostered a culture of continuous improvement. The a ggregated data has also allowed for department-wide analysis, enabling the identification of areas where additional training or support may be needed.

Limitations

The evaluation tool was designed specifically for t he Radiographer Bachelor Degree internship program and may not be generalizable to other healthcare settings. Additio nally, the study relied solely on student perceptions of tutor performance. Funding for this study: None Sunday Abstract-based Programme 272 Ethics committee - additional information: None Author Disclosures: Stefano Da Dalt: Nothing to disclose Angie Devetti: Nothing to disclose Chiara Zuiani: Nothing to disclose Manuel Gianvalerio Belgrano: Nothing to disclose Rossano Girometti: Nothing to disclose Lorenzo Cereser: Nothing to disclose Francesca Romana Fabris: Nothing to disclose 09:30-11:00 Research Stage 1 Research Presentation Session: Musculoskeletal RPS 2310 Imaging of the various pathologies of the spine Moderator C. Loupatatzis; Männedorf/CH ([email protected]) 3D Ultrashort Echo Time MRI for Assessing the Carti laginous Endplate of the lumbar intervertebral discs: Correlation with D isc Degeneration and Modic Changes in Conventional Fast Spin Echo Sequen ces *Y. Kim*¹, J. G. Cha², S. Lee¹; ¹Seoul/KR, ²Bucheon /KR ([email protected]) Purpose or Learning Objective: To investigate the association between cartilaginous endplate (CEP) abnormalities on 3-dim ensional ultrashort echo time MRI with cone trajectory technique (3D UTE) an d disc degeneration and endplate Modic change on conventional MRI.

Methods

or Background: Ninety one patients (44 men, 47 women, mean age: 55.75 years, range: 19-85 years) underwent MRI of t he lumbar spine with conventional sagittal T1, T2-weighted and fat-suppr essed T2 weighted fast spin echo sequence and sagittal 3D UTE cone traject ory technique (TR = 16.1 ms, TE = 0.032 ms and 6.6 ms) in 3T MRI. Two muscul oskeletal radiologists assessed CEP abnormalities (irregularity, thickenin g, thinning and defects) of the superior and inferior endplates of the L3-4, L4 -5 and L5-S1 discs on 3D UTE and disc degeneration with the Pfirrmann gradin g system, and presence of Modic change of the endplate on conventional MRI by consensus. The relationship of CEP abnormalities with the disc deg eneration and Modic change was tested using Pearson's chi-square test a nd Spearman's correlation analysis.

Results

or Findings: All CEP abnormalities were positively correlated wi th Pfirrmann grading system (Spearman ρ, 0.31-0.47) and Modic change (Spearman ρ, 0.24-0.50) with statistical significance (p < 0.0 01 for all Pearson's chi-square and Spearman's correlation ana lysis).

Conclusion

The CEP abnormality in 3D UTE MRI may be associated with the severity of disc degeneration and the presence of M odic change.

Limitations

The number of patients in the study is relatively s mall. MRI findings did not correlate with pathological findin gs. The MRI findings did not correlate with the patient's clinical symptoms or p rognosis. This is a cross- sectional study that does not allow for a longitudi nal study of the patient. Funding for this study: None Ethics committee - additional information: This study was designed prospectively, and was approved by the Inha Univers ity hospital's review board, and informed consent was obtained from all t he participating patients. Author Disclosures: Seunghun Lee: Nothing to disclose Yeoju Kim: Nothing to disclose Jang Gyu Cha: Nothing to disclose Added Value of Color-Coded Fat-Calcium Dual-Energy CT in the Detection of Spine Occult Bone Metastasis – a Pilot Study *J. Li*¹, J. Liu²; ¹Fujian/CN, ²Xiamen/CN ([email protected]) Purpose or Learning Objective: To assess the capability of color-coded Fat- Calcium dual-energy CT (DECT) in identifying spinal occult bone metastases (S-OBMs).

Methods

or Background: DECT images of a consecutive series of lung cancer patients were retrospectively analyzed. Two radiologists reviewed conventional CT images and color-coded Fat-Calcium images, recording the locations (diffuse infiltration, focal involvement of vertebral cancellous, vertebral edges, basivertebral venous plexus, and a ppendages) and number of occult bone metastases (OBMs) identified on the col or-coded Fat-Calcium images. Diagnostic performance measures (sensitivit y, specificity, positive predictive value (PPV), negative predictive value ( NPV), and accuracy) were then assessed.

Results

or Findings: A total of 24 patients were included, comprising 80 spinal occult bone metastases (S-OBMs). Color-coded DECT images show 100% sensitivity, PPV, and accuracy in diagnosing d iffusely invasive S-OBMs. The overall sensitivity of color-coded DECT images for focal occult metastases was measured at 96.1%. However, the PPV and accurac y of DECT for focal OBMs were influenced by the lesion’s location. The PPV and accuracy of OBMs in vertebral trabecular regions were higher th an those in vertebral edges, basivertebral venous plexus, and attachments (PPV: 81%, 7.7%, 2.0%, and 6.0%, respectively; accuracy: 95.2%, 67.2%, 47. 4%, and 14%, respectively). The diagnostic performance for verte bral cancellous regions was the highest, with sensitivity, specificity, PPV, NP V, and accuracy of 94%, 95.5%, 81%, 98.7%, and 95.2%, respectively.

Conclusion

Color-coded Fat-Calcium DECT significantly improves the detection of OBM in the spine.

Limitations

Firstly, it was a retrospective study with a small sample size. Secondly, this study was based on non-enhanced DECT , and whether enhanced DECT could improve diagnostic efficiency r equires further study. Funding for this study: Natural Science Foundation of Fujian Province, Chin a (grant numbers: 2023J01181) Ethics committee - additional information: Fujian Cancer Hospital Ethics Committee (K2023-198-01) Author Disclosures: Jianfang Liu: Nothing to disclose Jie Li: Nothing to disclose Ex vivo and in vivo validation of dual-layer detect or spectral-CT fat quantification of vertebrae bone marrow *Y. F. Melzer*, G. Campbell, N. F. Schubert, I. Fie dler, B. Busse, I. Molwitz; Hamburg/DE ([email protected]) Purpose or Learning Objective: To evaluate and validate dual-layer detector spectral-CT fat quantification (dlCT) of the verteb rae bone marrow.

Methods

or Background: Isolated human cadaver vertebrae (n=14) of 10 body donors were scanned within 72-96 hours after d eath using dlCT (CT7500) at 120 kV and a 3T MRI (Ingenia) (Philips Healthcare, the Netherlands). Spherical volumes of interest (VOIs, 11 mm diameter) were placed in the center of all vertebral bodies. Addit ionally, n=13 patients were prospectively included (mean age 57±9 years; three females) who underwent prior to liver transplantation multiphase dlCT and MR imaging. VOIs were defined at the third lumbar vertebrae. Within the V OIs fat was quantified in dlCT scans without contrast agent using three-mater ial decomposition for hydroxyapatite, red bone marrow, and fat. Reference values for red bone marrow were generated from the blood pool. MRI fat quantification was performed using mDIXONquant sequences (TE shortest, TR shortest, FA 3°). For statistics, Pearson’s correlations and Bland Al tman analysis were employed.

Results

or Findings: Ex vivo correlation between dlCT and MRI was high (r2=0.94, p<0.001) with a mean difference of -0.55 [95% intervals of agreement -11.0, 9.9]. In vivo, correlation between dlCT and MRI was moderate (r2=0.47, p=0.01). The mean difference amounted to 14.4 [95% intervals of agreement - 3.9, 32.7].

Conclusion

Ex vivo dlCT fat quantification of the vertebral bo ne marrow delivers reproducable results. In vivo measurements require further calibration of dlCT using MRI and - due to challenges of MRI fa t quantification in the presences of bone - preferably histology as a refer ence.

Limitations

Small sample size due to ongoing recruitment and ne cessary further calibration of dlCT fat quantification with histological analyses, for which the vertebra are currently prepared by formalin fix ation. Funding for this study: None. Ethics committee - additional information: 2023-300414-WF (Ärztekammer Hamburg) Author Disclosures: Björn Busse: Nothing to disclose Isabel Molwitz: Nothing to disclose Graeme Campbell: Nothing to disclose Imke Fiedler: Nothing to disclose Niklas Ferdinand Schubert: Nothing to disclose Yasmin Fede Melzer: Nothing to disclose Sunday Abstract-based Programme 273 VERIFACT: Revealing the Hidden Epidemic of Undiagno sed Vertebral Fractures in Routine CT Scans H. P. Dimai, J. Igrec, *J. Steiner*, R. Riedl, M. F uchsjäger; Graz/AT ([email protected]) Purpose or Learning Objective: Osteoporosis is a condition characterized by low bone mass and increased fracture risk. Vertebra l fractures are the most common, often undetected despite serious health con sequences. Studies reveal that many fractures are missed in radiograph ic and CT reports, with false-negative rates ranging from 30% to 84%, highl ighting significant underreporting in clinical practice. The study aims to assess recognition rates in chest and abdominal CTs.

Methods

or Background: This retrospective analysis included 1,500 CT images from 1,380 patients. Two independent board-c ertified radiologist with multi-year-experience reviewed each scan for verteb ral fractures, noting the number, location, and type of fracture, and classif ying them based on the Genant classification system (Grades 2 and 3). Rate r agreement was assessed using the Kappa coefficient and AC1 statis tic. Fracture documentation rates were calculated based on whethe r fractures were noted in the formal radiology report summary or only in the free-text narrative. Stratified analysis was performed by patient gender, fracture location, and the presence of multiple fractures.

Results

or Findings: Vertebral fractures were found in 11.5% of patients . Agreement between the raters was excellent, with a Kappa value of 0.94 (95% CI: 0.92–0.97) and AC1 of 0.99 (95% CI: 0.98–0.99). However, fractures were documented in the summary in only 14.7% of cases (9 5% CI: 9.8%–20.9%), while 35.3% were mentioned only in the narrative po rtion (95% CI: 28.1%– 43.0%). Overall, 50% of the fractures were reported in either the summary or the narrative (95% CI: 42.3%–57.8%).

Conclusion

While interrater agreement on fracture detection wa s high, the rate of documentation, particularly in the structur ed summary, was low. These findings suggest a need for better reporting protoc ols to ensure vertebral fractures are clearly communicated, improving patie nt outcomes and clinical decision-making.

Limitations

n/a Funding for this study: n/a Ethics committee - additional information: Approved by local ethics committee Author Disclosures: Jakob Steiner: Nothing to disclose Regina Riedl: Nothing to disclose Michael Fuchsjäger: Nothing to disclose Hans Peter Dimai: Nothing to disclose Jasminka Igrec: Nothing to disclose Spectral Collagen Imaging: Assessment of Thoracic D isk Herniation and Degeneration *M. Dimitrova*, C. Booz, S. Mahmoudi, A. Gökduman, L. D. Grünewald, S. Bernatz, E. Höhne, T. Vogl, I. Yel; Frankfurt/DE Purpose or Learning Objective: This study investigates the diagnostic efficacy of Dual-Energy CT (DECT)-derived collagen maps in evaluating thoracic disk herniation and degeneration.

Methods

or Background: A retrospective analysis was conducted involving 51 patients who underwent dual-source DECT (Somatom Force; Siemens Healthineers) and MRI of the thoracic spine within a two-week timeframe. Two blinded radiologists assessed the presence and type of herniation using the North American Spine Society's classification for i ntervertebral disk pathology, evaluating both conventional grayscale CT and DECT collagen maps. Diagnostic accuracy, sensitivity, and specificity w ere calculated with MRI as the

Reference

standard. Additionally, subjective assess ments of diagnostic confidence and image quality were performed, and in ter-reader reliability was evaluated using the intraclass correlation coeffici ent.

Results

or Findings: Analysis of 612 intervertebral disks showed that DE CT collagen maps achieved significantly higher sensiti vity (98.0%), specificity (97.7%), and diagnostic accuracy (97.7%) compared t o conventional CT (sensitivity: 72.0%, specificity: 97.0%, diagnostic accuracy: 96.0%; p < 0.001). Substantial inter-reader reliability was noted (κ=0.76, p < 0.001), with DECT collagen maps providing enhanced diagnostic confide nce and image quality (p < 0.001). For assessing disk degeneration, DECT col lagen maps demonstrated high sensitivity (83.0%), specificity (81.9%), and diagnostic accuracy (83.4%) in distinguishing non/mild from moderate/severe degene ration, with inter-reader reliability also showing strong agreement (κ=0.82, p < 0.001). Subjective evaluations reported moderate to high diagnostic co nfidence (median 3.5) and moderate to good image quality (median 3.5).

Conclusion

DECT-derived collagen maps significantly enhance th e detection of thoracic disk herniation and degeneration, offer ing improved diagnostic accuracy, reliability, confidence, and image qualit y over conventional CT. This imaging technique serves as a valuable alternative for patients who cannot undergo MRI.

Limitations

retrospective study single-centre Study small patie nt group CT system from a specific vendor Funding for this study: No funding. Ethics committee - additional information: Approval by local ethics committee. Author Disclosures: Christian Booz: Nothing to disclose Ibrahim Yel: Nothing to disclose Mirela Dimitrova: Nothing to disclose Thomas Vogl: Nothing to disclose Scherwin Mahmoudi: Nothing to disclose Aynur Gökduman: Nothing to disclose Leon David Grünewald: Nothing to disclose Simon Bernatz: Nothing to disclose Elena Höhne: Nothing to disclose Cervical spine motion in dynamic X-rays – approach, results, conclusions *M. J. Łubiński*, A. Majos, P. Kowalski; Lodz/PL ([email protected]) Purpose or Learning Objective: Dynamic X-ray is very usefull tool to evaluate cevical spine motion. The aim of this stud y is to find reference ranges, correlations and practical tips which radiologists can use in cervical spine mobility assesment.

Methods

or Background: 288 patients aged 19-78 years old without cervical spine osteoarthritis or suffering from first degree osteoarthritis in Kellgren- Lawrance classtification were examined. We performe d lateral X-rays in three projections – neutral, flexion and extension. In th e functional X-ray examinations of the cervical spine, the following p arameters were assessed: the Cobb angle C2-C7, angular and horizontal segmen tal mobility and segmental cervical curvature in flexion and extensi on.

Results

or Findings: Reference ranges for extension and flexion C2-C7 Co bb angle, angular and horizontal segmental mobility we re found. Analysis showed the biggest mobility of C4-C5 segment. Correlations suggest that horizontal displacement index is the most universal parameter in cervical spine mobility assessment.

Conclusion

A large group of patients and multitude measurement s made it possible to find reliable and clinically useful ref erence ranges and parameters which can be used in routine evaluation of cervical spine functional tests.

Limitations

Not applicable. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Not applicable Author Disclosures: Agata Majos: Author: Co-researcher Marcin Janusz Łubiński: Author: Main researcher Piotr Kowalski: Author: Co-researcher Longitudinal assessment of structural abnormalities in the lumbar spine of adolescent competitive alpine skiers over 48 mon ths *G. C. Feuerriegel*, D. Meyer, D. Fitze, J. Haniman n, C. Stern, S. Fröhlich, J. Scherr, J. Spörri, R. Sutter; Zurich/CH Purpose or Learning Objective: To longitudinally assess and compare spinal structural abnormalities in adolescent competitive alpine skiers over 48 months and to compare MRI findings in asymptomatic and sym ptomatic skiers and explore their clinical relevance.

Methods

or Background: Adolescent competitive alpine skiers recruited for a cross-sectional MRI investigation underwent a 3T MR imaging of the lumbar spine at baseline and after 48 months. All MR image s were assessed for structural changes occurring in the intervertebral disc, vertebral body and facet joints. At both baseline and follow-up, athletes' l ow back pain (LBP) symptoms were assessed and Athletes were classified as sympt omatic if at least one 'substantial' episode of health problems related to back overuse had occurred in the 12 months prior to the MRI examination. The Wilcoxon signed-rank test and Pearson's chi-squared test were used to compare the measurements.

Results

or Findings: A total of 63 athletes (mean age at follow-up 19.6± 1.2 years, 25 female) were included in the study. A sig nificant increase in LBP affecting training and competition was observed at follow-up (P = 0.04). Of the athletes with LBP, 63% (n=16) reported recurrent LB P, 14% (n=4) reported permanent backpain, and 26% (n=7) reported one-time LBP since baseline. Assessment of structural changes revealed a signifi cant increase in the number of athletes with disc dehydration (P 0.05).

Conclusion

Overuse related structural changes progress during adolescence and are not self-limiting. However, structural chan ges are not directly correlated with LBP. This finding may facilitate th e development of appropriate treatment and prevention strategies that do not foc us solely on spinal changes.

Limitations

Structural abnormalities were only assessed by MRI and not confirmed by other modalities. Sunday Abstract-based Programme 274 Funding for this study: This study was generously supported by the Balgrist Foundation. Ethics committee - additional information: Cantonal Ethics Committee Zurich Author Disclosures: Johannes Scherr: Nothing to disclose Jörg Spörri: Nothing to disclose Stefan Fröhlich: Nothing to disclose Christoph Stern: Nothing to disclose Jonas Hanimann: Nothing to disclose Reto Sutter: Nothing to disclose Georg Constantin Feuerriegel: Nothing to disclose Daniela Meyer: Nothing to disclose Daniel Fitze: Nothing to disclose Multiparametric quantitative MRI in Charcot-Marie-T ooth 1A inherited neuropathy: correlation with motor function and bal ance performance *D. Bianco*, F. Zaottini, S. Rinaldi, M. Pansecchi, M. Hamedani, S. Massucco, E. Rovetta, C. Martinoli; Genova/IT ([email protected]) Purpose or Learning Objective: We aimed to evaluate the feasibility of a quantitative multiparametric MRI protocol of lumbo- sacral plexus and proximal sciatic nerve to differentiate patients affected by Charcot-Marie-Tooth type 1A (CMT1A) neuropathy from controls and to correlate t hese imaging parameters with clinical grading scale of disease's severity.

Methods

or Background: Patients with clinical, electrophysiological and genetical proven CMT1a were prospectively enrolled. The 3T MRI protocol included the following sequences: Diffusion Tensor Imaging, 2 points T2 DIXON, T1 mapping and T2 mapping. The MR parameters were independently measured by two radiologists. The same day of MRI e xamination, the CMT1a patients were clinically assessed using CMTNS score and Berg Balance Score (BBS). An age and sex matched control group without clinical signs of neuropathy (NN) was enrolled.

Results

or Findings: n=11 patients (7 f,4 m), 47.57 yo ± 14.39 and n= 8 NN controls (5 f, 3 m) 46.5 yo ± 14.39 underwent MRI. The interobserver reliability of measurements was good (ICC=0,65). Lumbosacral pl exus roots and sciatic nerve cross sectional area, Fractional Anisotropy ( FA), T1 and T2 relaxation time were significantly different between the two g roups (p<0,05). Bilateral L5 and S1 T1 relaxation values and FA significantly co rrelated with CMTNS (respectively R=0.86, p=0.013 and R=0,77, p=0,04) a nd BBS (respectively R=0.62, p=0.041 and R= 0.75, p=0,042). Sciatic Nerv e FA demonstrated strong correlations with both CMTNS (R=0.92, p=0.01 0) and BBS (R=0.89, p=0.019).

Conclusion

These findings suggest that FA and T1 relaxation ti me of the lumbosacral plexus and sciatic nerve are the MRI pa rameters that better correlate with balance performance and overall func tional disability in CMT1A patients, representing potential biomarker for dise ase severity and longitudinal evaluation.

Limitations

Small sample size. Funding for this study: The study was funded by the Italian Ministry of hea lth trough the public grant BANDO RICERCA FINALIZZATA 2 021. Ethics committee - additional information: Comitato Etico Territoriale - Regione Liguria Author Disclosures: Deborah Bianco: Nothing to disclose Simone Rinaldi: Nothing to disclose Federico Zaottini: Nothing to disclose Carlo Martinoli: Nothing to disclose Mehrnaz Hamedani: Nothing to disclose Edoardo Rovetta: Nothing to disclose Michelle Pansecchi: Nothing to disclose Sara Massucco: Nothing to disclose T2 relaxation times of the pubic symphysis in ostei tis pubis *N. Andjelic*¹, N. Holl², B. Klaan², M-A. Weber²; ¹ Sremska Kamenica/RS, ²Rostock/DE ([email protected]) Purpose or Learning Objective: This study aimed to evaluate the T2 relaxation times of the cartilage layer of the pubi c symphysis in male athletes with osteitis pubis and examine the correlation bet ween these values, pubic bone marrow edema (BME), the pubic symphysis width, and the presence of cleft injuries.

Methods

or Background: Sixty-two male athletes (median age, 28 years) presenting with groin pain were examined using a 3- Tesla MRI system. T2 mapping was applied using a T2W sequence to assess the pubic symphysis in three ways: interpubic disc, articular cartilage, a nd the entire hyaline- fibrocartilage complex. T2 relaxation times were me asured, and BME and cleft injuries were identified. Correlations between T2 v alues, symphyseal width, and the presence of BME and pubic cleft injuries we re explored.

Results

or Findings: The median T2 relaxation times for the hyaline- fibrocartilage complex, interpubic disc, and articu lar cartilage were 49.7 ms, 54.6 ms, and 46.2 ms, respectively, with significan t differences between (p- value < 0.001), and a notable distinction between t he right and left sides of the articular cartilage. Athletes with BME had higher T 2 relaxation times for the hyaline-fibrocartilage complex and interpubic disc (p-value < 0.01) but not for the articular cartilage. A moderate positive correl ation (r = 0.4) was found between symphysis width and T2 relaxation times.

Conclusion

T2 mapping provides valuable insights into the stru ctural changes in the pubic symphysis in athletes with ost eitis pubis. Higher T2 relaxation times in the interpubic disc and whole s ymphyseal hyaline/fibrocartilage complex are associated with BME, suggesting their potential use in evaluating osteitis pubis.

Limitations

The study was limited to male patients, and the abs ence of an asymptomatic control group is a notable limitation. Additionally, the pre- screening of all participants introduces a potentia l selection bias due to the study design. Funding for this study: This research project was part of the ESOR Bracco Research Fellowship 2024 Ethics committee - additional information: The study was approved by the Ethical Committee of Rostock University (approval N o. A 2020-0040) Author Disclosures: Norman Holl: Nothing to disclose Marc-André Weber: Nothing to disclose Bastian Klaan: Nothing to disclose Nikola Andjelic: Nothing to disclose Ultrasound-guided navigation system for spine surge ry *A. Lubansu*, P. Pandin; Brussels/BE ([email protected]) Purpose or Learning Objective: Current image-guided navigation systems in spine surgery rely on ionizing radiation from intra operative fluoroscopy or CT scans. This study evaluates the feasibility and uti lity of fusing intraoperative ultrasound (US) imaging with preoperative lumbar CT scans to create a novel US-guided spinal navigation system, aiming to reduc e radiation exposure.

Methods

or Background: Over one year, 25 patients undergoing lumbar spine surgery participated in this study. Cortical borders of spinal structures were co-registered and fused with preoperative CT s cans. The accuracy and time required for co-registration were assessed. Wh en navigation accuracy was within 2mm, various spinal procedures, includin g screw removal or placement and canal or foraminal recalibration, wer e performed using the system.

Results

or Findings: Accurate co-registration was achieved in under 10 minutes for all cases. Anatomical landmarks for co- registration varied depending on the target region (sacrum, sacroiliac joint, lumbosacral junction, or lumbar vertebrae). The system facilitated percut aneous screw placement, intradural lesion localization, and optimal neural structure decompression. No complications related to US-guided navigation occur red. The total radiation dose was reduced compared to conventional non-navig ated procedures.

Conclusion

This preliminary experience suggests that US-guided navigation for spinal procedures is feasible, accurate, safe, and potentially beneficial in reducing radiation exposure. The system demonstrate d versatility across various spinal regions and procedures. Further rese arch with larger cohorts and more indications is needed to fully understand the potential advantages and limitations of this innovative navigation syste m.

Limitations

This study is limited by its small sample size and single-center design. A direct comparison with conventional navig ation techniques was not assessed. The learning curve for implementing this new technique was not evaluated. Funding for this study: No Ethics committee - additional information: Not applicable Author Disclosures: Alphonse Lubansu: Nothing to disclose Pierre Pandin: Nothing to disclose Sunday Abstract-based Programme 275 09:30-11:00 Research Stage 2 Research Presentation Session: Imaging Informatics and Artificial Intelligence RPS 2305 Healthy aging, body composition and prevention: the true potential of AI? Moderator M. Pop; Tg.Mures/RO Automated coronary calcification assessment on unga ted unenhanced chest CT using an optimised nnUNet framework for pa tient prognostication in non-small-cell lung cancer *J. Y. Anifowose*, Z. Li, G. Agarwal, E. Aboagye, B . Ariff, S. Copley, M. Chen; London/UK ([email protected]) Purpose or Learning Objective: To develop an automated software for assessing coronary calcification in non-small cell lung cancer (NSCLC) patients using an optimised deep learning nnUNet framework f or disease prognostication.

Methods

or Background: Cardiovascular risk is higher in NSCLC patients than in the general population, but is often underd iagnosed in clinical practice. Attenuation correction CTs from routinely acquired PET-CT staging scans are ungated unenhanced studies which offer an opportuni ty to assess this risk without incurring additional radiation exposure or radiology workload. nnUNet is a state-of-the-art deep learning architecture de monstrating superior performance in medical image segmentation applicati ons. We trained nnUNet models for coronary calcification on ungated unenha nced chest CTs (n = 100) from a public domain dataset (Stanford AIMI) and te sted them on independent data: attenuation correction CTs of PET-CT scans of NSCLC patients acquired at our multi-centre institution between 2012 and 20 18 (n = 287, age: 66.8 ± 10.1, male: female 174:113). The reference truth se gmentations were drawn and verified by two radiologists of 8 and 2 years o f experience. Models with varying batch sizes and convolutional filters were developed and benchmarked; with the best performing one selected to develop a composite prognostic predictor, based on model-derived corona ry calcification score and significant NSCLC features.

Results

or Findings: The best performing nnUNet has a 3D_fullres configuration with batch size of 4 and patch size 2 8x224x224. All cases of coronary calcifications were successfully detected. Multivariable Cox showed statistical significance of disease histology and s tage on patient survival. The composite predictor achieved statistically signific ant prognostic risk stratification (p-value < 0.05).

Conclusion

An optimised nnUNet framework can facilitate automa ted coronary calcification assessment on ungated unenha nced CT to support a composite prognostic predictor in NSCLC patients.

Limitations

Retrospective study. Single external validation coh ort. Funding for this study: Academy of Medical Sciences award SGL026 ∖1024. Ethics committee - additional information: Retrospective observational study IRAS: 243592 REC: 18HH4616 Author Disclosures: Zechen Li: Nothing to disclose Susan Copley: Nothing to disclose Ben Ariff: Nothing to disclose Mitchell Chen: Nothing to disclose Eric Aboagye: Nothing to disclose Girija Agarwal: Nothing to disclose Jubril Yinka Anifowose: Nothing to disclose Deep learning-based biological age estimation from MRI predicts cardiometabolic events in the general population *M. Jung*¹, M. Reisert², H. Rieder², S. Rospleszcz² , M. T. Lu¹, F. Bamberg², V. Raghu¹, J. Weiß²; ¹Boston MA/US, ²Freiburg/DE ([email protected]) Purpose or Learning Objective: Chronological age is one of the cornerstones of medical decision-making, but it's an imperfect m easure of health. We propose a deep learning framework (MRI-Age) for est imating biological age from MRI and investigated its value in predicting c ardiometabolic outcomes in the general population beyond chronological age.

Methods

or Background: We used 30,389 individuals from the German National Cohort (NAKO) to develop MRI-Age, which ta kes MRI-derived volumetric body composition, including subcutaneous (SAT), visceral (VAT), intramuscular adipose tissue (IMAT), and skeletal m uscle (SM) from the 1st to 5th lumbar vertebra as input and outputs an age est imate in years. For downstream analyses, we calculated MRI-Age accelera tion, defined as an age- specific z-score of the age estimate. We then valid ated this framework in an external testing set of 36,317 individuals from the UK Biobank (UKBB). Incident outcomes were diabetes, MACE, and all-cause mortali ty. Multivariable Cox regression assessed the association between MRI-Age acceleration categories “negative” (MRI-Age acceleration 1) and ou tcomes adjusted for traditional cardiometabolic risk factors in the UKB B.

Results

or Findings: In 36,317 UKBB participants (65.1±7.8 years, 51.7% female; median follow-up 4.8 years), we found a hig her incidence of diabetes, MACE, and death in individuals with positive MRI-ac celeration. In multivariable- adjusted Cox regression, we observed a significant positive association between positive MRI-Age acceleration and diabetes (aHR: 1.87, 95% CI [1.56-2.25], p<0.001), MACE (aHR: 1.26, 95% CI [1.0 1-1.57], p=0.038), and all-cause mortality (aHR: 1.37, 95% CI [1.09-1.72], p=0.007).

Conclusion

Deep learning-derived biological age from MRI predi cts cardiometabolic outcomes in the general population beyond chronological age and cardiometabolic risk factors. Individuals at hi gh MRI-Age could benefit from personalized prevention strategies, lifestyle interventions, and treatment planning.

Limitations

Limited age-range. Predominantly white population. Funding for this study: This project was conducted with data from the German National Cohort (NAKO) (www.nako.de). The NA KO is funded by the Federal Ministry of Education and Research (BMBF) [ project funding reference numbers: 01ER1301A/B/C, 01ER1511D, and 01ER1801A/B/ C/D], federal states of Germany, and the Helmholtz Association, t he participating universities and the institutes of the Leibniz Asso ciation. This research has been conducted using the UK Biobank Resource under Application Number 80337. We thank all participants who took part in t he NAKO and UKBB study and the staff of these research initiatives. MJ was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Founda tion) - 518480401. VKR was funded by Norn Group Longevity Impetus Gran t, NHLBI K01HL168231, and AHA Career Development Award 93517 6. Ethics committee - additional information: Informed consent was obtained from all participants in the UK Biobank and the Ger man National Cohort study. In addition, we received local IRB approval (IRB of the University of Freiburg: 23-1316-S1-retro and 24-1099-S1-retro). Author Disclosures: Susanne Rospleszcz: Nothing to disclose Marco Reisert: Nothing to disclose Jakob Weiß: Nothing to disclose Matthias Jung: Nothing to disclose Fabian Bamberg: Nothing to disclose Hanna Rieder: Nothing to disclose Vineet Raghu: Nothing to disclose Michael T. Lu: Nothing to disclose Body Composition in the general population: MRI-der ived reference curves from over 66,000 individuals and their assoc iation with cardiometabolic outcomes *M. Jung*¹, M. Reisert², H. Rieder², S. Rospleszcz² , M. Russe², M. T. Lu¹, F. Bamberg², V. Raghu¹, J. Weiß²; ¹Boston MA/US, ²F reiburg/DE ([email protected]) Purpose or Learning Objective: Body composition (BC) plays an important role in risk estimation in patients with cardiometa bolic disease and cancer, but

Reference

curves are missing to place individual me asurements in context. We developed a deep learning framework to quantify BC from MRI to calculate

Reference

curves and investigated its value for pre dicting cardiometabolic outcomes.

Methods

or Background: BC extracted from MRI data of the UK Biobank (UKBB) and German National Cohort included 1) subcu taneous (SAT), 2) visceral (VAT), 3) intramuscular adipose tissue (IM AT), 4) skeletal muscle (SM), and 5) SM fat fraction (SMFF). Reference curv es were generated using generalized additive models for each BC metric to c alculate age, sex, and height-specific z-scores. Multivariable Cox regress ion assessed the association between z-score categories (low: z1) and outcomes (incident diabetes; major adverse cardiova scular events [MACE]; and all-cause mortality) adjusted for traditional c ardiometabolic risk factors in the UKBB.

Results

or Findings: Among 66,608 individuals (57.7±12.9 years; BMI: 26.2±4.5 kg/m2, 48.3% female), we observed sex diff erences in BC volumes and distributions with SAT, VAT, SMFF, and IMAT pos itively and SM negatively associated with age. We found graded ass ociations between BC z- score categories and health outcomes in the UKBB. I n multivariable adjusted Cox regression, z-score risk categories had hazard ratios of up to 2.69 for incident diabetes (high VAT), 1.41 for incident MAC E (high IMAT), and 1.49 for all-cause mortality (low SM) compared to middle cat egories. Sunday Abstract-based Programme 276

Conclusion

BC measures normalized for age, sex, and height are associated with cardiometabolic outcomes beyond traditional ri sk factors in the general population. We will provide open-source BC referenc e curves, which may accelerate the clinical translation of BC-based ris k assessment for cardiometabolic disease and support future BC resea rch.

Limitations

Study population is predominantly white Western Eur opean adults. Funding for this study: This project was conducted with data from the German National Cohort (NAKO) (www.nako.de). The NA KO is funded by the Federal Ministry of Education and Research (BMBF) [ project funding reference numbers: 01ER1301A/B/C, 01ER1511D, and 01ER1801A/B/ C/D], federal states of Germany, and the Helmholtz Association, t he participating universities and the institutes of the Leibniz Asso ciation. This research has been conducted using the UK Biobank Resource under Application Number 80337. We thank all participants who took part in t he NAKO and UKBB study and the staff of these research initiatives. MJ was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Founda tion) - 518480401. VKR was funded by Norn Group Longevity Impetus Gran t, NHLBI K01HL168231, and AHA Career Development Award 93517 6. Ethics committee - additional information: Informed consent was obtained from all participants in the UK Biobank and the Ger man National Cohort study. In addition, we received local IRB approval (IRB of the University of Freiburg: 23-1316-S1-retro and 24-1099-S1-retro). Author Disclosures: Susanne Rospleszcz: Nothing to disclose Marco Reisert: Nothing to disclose Jakob Weiß: Nothing to disclose Matthias Jung: Nothing to disclose Maximilian Russe: Nothing to disclose Fabian Bamberg: Nothing to disclose Hanna Rieder: Nothing to disclose Vineet Raghu: Nothing to disclose Michael T. Lu: Nothing to disclose AI-Driven MRI Biomarker Extraction and Machine Lear ning Analysis of Their Association with Diabetes: A UK Biobank Study *S. Kim*¹, D. W. Kim¹, C. Han¹, D. Kim², D. Yoon¹; ¹Seoul/KR, ²Daegu/KR ([email protected]) Purpose or Learning Objective: To evaluate AI-derived imaging biomarkers from whole-body MRI in detecting and predicting dia betes mellitus (DM).

Methods

or Background: An open-source multi-label segmentation model was applied to Dixon whole-body MRIs from the UK Bi obank to segment organs and body compositions. Volume indices (volum e/m³) and fat fractions of each structure were calculated automatically. For D M detection at the time of MRI, logistic regression was performed. Excluding b aseline DM, random survival forest analysis was performed for predicti ng future DM. Area under curve (AUC) and Harrell’s C-index was used. Perform ance of imaging biomarkers was compared to the Leicester Diabetes R isk Score.

Results

or Findings: Among the 2,924 participants, 149 had DM at baselin e. Of the 2,775 participants without baseline DM, 28 d eveloped DM and were included in the survival analysis (median follow-up 4.1 years, up to 8.9 years). For DM detection, adrenal gland volume index, kidne y volume index, and pancreatic fat fraction (AUC 0.748, 0.716, and 0.71 0 respectively) were the top classifiers. The multivariate model, using 10 selec ted imaging features, achieved AUC of 0.802. In survival analysis, pancre atic fat fraction, adrenal gland volume index, and torso fat volume index (C-i ndex 0.713, 0.685, and 0.678 respectively) were the top predictors. The mu ltivariate model with six selected imaging features achieved C-index 0.780, o utperforming the Leicester Diabetes Risk Score (C-index 0.651). When imaging f eatures were combined with clinical features, performance further improve d (C-index 0.794).

Conclusion

AI-derived MRI biomarkers demonstrated strong perfo rmance in detecting current DM and predicting future onset, h ighlighting their potential utility in opportunistic screening.

Limitations

Further validation of the open-source segmentation model is necessary to assess its quantitative and qualitativ e performance. Funding for this study: MD-PhD/Medical Scientist Training Program through the Korea Health Industry Development Institute, fu nded by the Ministry of Health & Welfare, Republic of Korea Ethics committee - additional information: Our institution has received IRB approval for UK Biobank-related research, and any a dditional ethical considerations are adhered to in accordance with th is approval. Author Disclosures: Songsoo Kim: Nothing to disclose Donghyun Kim: Nothing to disclose Dong Won Kim: Nothing to disclose Changho Han: Nothing to disclose Dukyong Yoon: Nothing to disclose CompositIA: an open-source pipeline for automated q uantification of body composition scores from thoraco-abdominal CT s cans *R. F. Cabini*, A. Cozzi, S. Leu, B. Thelen, R. Kra use, F. Del Grande, S. Rizzo, D. U. Pizzagalli; Lugano/CH ([email protected]) Purpose or Learning Objective: This study aims to develop and validate CompositIA, an automated, open-source pipeline for quantifying body composition scores from thoraco-abdominal CT scans.

Methods

or Background: CompositIA consists of three main steps: automatic identification of the L1 and L3 vertebrae, segmenta tion of image slices at these spinal levels, and quantification of body compositi on indices. Two Deep Learning models were used: MultiResUNet for detecti ng CT slices intersecting the L1 and L3 vertebrae, and UNet for segmenting th e corresponding axial slices. The pipeline was trained on 205 contrast-en hanced thoraco-abdominal CT scans and tested on an independent dataset of 54 scans. Manual segmentation was performed by two radiology residen ts, who identified the centers of the L1 and L3 vertebrae and segmented th e corresponding axial slices. Performance was evaluated via mean absolute error (MAE) for L1/L3 detection, volumetric Dice similarity coefficient ( vDSC) for segmentation, and mean percentage relative error (PRE), regression an alysis, and Bland–Altman plots for body composition indices estimation.

Results

or Findings: On the independent dataset CompositIA achieved a MAE of about 5 mm in detecting slices intersecting the L1 and L3 vertebrae, with a MAE < 10 mm in at least 85% of cases, and a vDSC greater than 0.85 in segmenting axial slices. Regression and Bland–Altma n analyses demonstrated a strong linear relationship and good agreement bet ween automated and manual scores (p values < 0.001 for all indices), w ith mean PREs ranging from 5.13% to 15.18%.

Conclusion

CompositIA facilitated automated quantification of body composition scores, achieving high precision in ind ependent testing.

Limitations

The main limitation of the study is the small size of the training set. Funding for this study: Raffaella F. Cabini, Benedikt Thelen, Rolf Krause a nd Diego U. Pizzagalli were supported financially by t he grants ExaTrain (to Rolf Krause), and FIR (to Diego U. Pizzagalli). Ethics committee - additional information: This study was approved by the local Ethics Committee (Comitato Etico Cantonale, R epubblica e Cantone Ticino, Switzerland; protocol code 2021-00943). All patients whose CT scans were included in the training set provided informed consent for the participation in the study. Author Disclosures: Andrea Cozzi: Nothing to disclose Rolf Krause: Nothing to disclose Stefania Rizzo: Nothing to disclose Filippo Del Grande: Nothing to disclose Diego Ulisse Pizzagalli: Nothing to disclose Raffaella Fiamma Cabini: Nothing to disclose Benedikt Thelen: Nothing to disclose Svenja Leu: Nothing to disclose A novel CT-based biological age model, based on aut omated abdominal CT biomarkers for accurate longevity prediction *J. Garrett*¹, M. Lee¹, A. Pyrros², R. Summers³, M. Kattan⁴, P. J. Pickhardt¹; ¹Madison, WI/US, ²Downers Grove, IL/US, ³Bethesda, MD/US, ⁴Cleveland, OH/US ([email protected]) Purpose or Learning Objective: To derive and test a CT-biological age (CTBA) model using explainable fully automated abdo minal CT-based tissue biomarkers predictive of survival in a large adult population.

Methods

or Background: In this retrospective cohort study, an automated suite of explainable CT-based AI algorithms quantif ying skeletal muscle (L3 level), fat (L3 level), aortic calcification, bone density, and solid organs (liver/spleen/kidney volume) was applied to a large adult cohort undergoing abdominal CT between January 2001-January 2021. Mul tivariable Cox proportional hazards regression survival analysis w as performed to determine final CT biomarker selection based on index of pred iction accuracy (IPA). Using all-cause mortality as a primary outcome, the CTBA model informed only by CT biomarkers and blinded to demographics was co mpared to a model based on demographic data (chronological age/sex/ra ce). The model was also applied to an external validation cohort of 40,718 adults.

Results

or Findings: 123,281 adults (mean age, 53.6 years [SD 17.4]; 47% women) underwent abdominal CT during the study inte rval. Median post-CT follow-up was 5.3 years (IQR,1.9-10.4 years). CT bi omarkers of greatest importance to the model were (in descending order): muscle attenuation, aortic calcification, visceral fat attenuation, and bone d ensity. The CTBA model significantly outperformed demographic data for pre dicting longevity (IPA=29.2 vs. 21.7; 10-year AUC=0.880 vs. 0.779; p<0.001). Ag e- and sex-corrected survival HR for highest-vs-lowest risk CTBA quartil e was 8.73 (95% CI,8.14- 9.36); HR for highest-risk vs remaining quartiles w as 3.13 (95% CI,3.04-3.23). Sunday Abstract-based Programme 277 CTBA model performed well in the external validatio n cohort (IPA=28.6; AUC=0.888).

Conclusion

A novel CTBA model informed only by objective fully automated "opportunistically” derived abdominal CT biomarkers outperformed a demographics (CA/sex/race) based model and improves survival prediction.

Limitations

Data from a single large academic medical center we re used for model training. Funding for this study: None Ethics committee - additional information: IRB Waiver of consent; retrospective analysis. Author Disclosures: Michael Kattan: Nothing to disclose Perry J. Pickhardt: Advisory Board: Nanox AI Matthew Lee: Nothing to disclose John Garrett: Shareholder: NVIDIA Advisory Board: R adUnity Ayis Pyrros: Nothing to disclose Ronald Summers: Nothing to disclose Enabling Exchange of Quantitative CT-Assessed Body Composition Data using FHIR: A First Step into Interoperable Body Co mposition Profiling *Y. Wen*, J. H. Eil, K. A. Borys, J. Kohnke, K. Arz ideh, J. Haubold, F. Nensa, O. Pelka, R. Hosch; Essen/DE Purpose or Learning Objective: This study aims to demonstrate the integration of AI-generated body composition and or gan measurements from CT images with Fast Healthcare Interoperability Res ources (FHIR) to standardise and enhance CT-derived biomarker intero perability.

Methods

or Background: FHIR is a widely used interoperability standard tha t enables health information exchange across differen t healthcare systems. With the development of AI models, modern AI application s cannot only analyse the data but also generate relevant data for patient mo nitoring and assessment, such as models for body composition analysis. The m issing step in advancing personalised medicine is combining AI-generated hea lthcare results with an interoperable and standardised format. Therefore, t his study integrated the

Results

of the Body and Organ Analysis (BOA) model into FHIR profiles, including measurements of 11 semantic body regions, seven tissues, and 104 landmarks, to streamline and provide interoperabili ty of CT-derived biomarkers in radiology.

Results

or Findings: Two FHIR profiles, Body Composition Analysis Observation and Body Structure Volume Observation p rofiles, have been developed to capture body composition measurements and record the volume of body structure generated from the BOA model, inc orporating terminology coding and references to related FHIR resources.

Conclusion

The presented FHIR profiles provide an interoperabl e format for AI-generated body composition data, standardising t he storage and exchange of AI-generated biomarkers derived from CT images. The contributed profiles can also be extended in future work to support othe r radiological modalities (e.g. MRI) or other image-based AI model biomarkers (e.g. CT-based bone mineral density).

Limitations

The created profiles focus on tissue and organ volu metrics and should be enhanced to include other available image -based markers and imaging modalities. Funding for this study: No funding was received for this study. Ethics committee - additional information: This study does not require ethics committee approval, since no identifiable or sensitive patient data was used. Author Disclosures: Katarzyna Anna Borys: Nothing to disclose Jan Horst Eil: Nothing to disclose Johannes Haubold: Nothing to disclose Judith Kohnke: Nothing to disclose Yutong Wen: Nothing to disclose Kamyar Arzideh: Nothing to disclose René Hosch: Nothing to disclose Felix Nensa: Nothing to disclose Obioma Pelka: Nothing to disclose Transfer of a CT-based 3D body composition analysis segmentation model to MRI T2-weighted sequences using a generati ve adversarial network *C. Bojahr*¹, J. Haubold¹, O. Pollok¹, C. S. Schmid t¹, K. A. Borys¹, M. Mancino², L. Umutlu¹, F. Nensa¹, R. Hosch¹; ¹Ess en/DE, ²Rome/IT Purpose or Learning Objective: This study aims to adapt CT-based deep learning (DL) models using CycleGAN-based style tra nsfer, enabling accurate body composition analysis (BCA) without extensive m anual annotation on T2- weighted MRI sequences.

Methods

or Background: This study analyzed data from 120 patients (96 train, 24 test) who underwent whole-body CT and MRI within 48 hours. A CycleGAN was trained to convert CT images to T2-wei ghted MRIs, producing synthetic MRIs that preserve CT structures with MRI styling. BCA was assessed on the corresponding CT scans using the Bo dy and Organ Analysis (BOA) framework, and 10 body-region class segmentat ions were transferred to synthetic MRIs to train an initial nnU-Netv2 3D seg mentation model. This model was used to generate proposals for all 120 MR Is, which two trainees under guidance of a radiologists (with 8 years of e xperience) refined. A second model was then trained on the refined segmentations , and evaluated by comparing both models to expert annotations using t he Sørensen-DICE score.

Results

or Findings: The comparison between the two models (style transf er vs. expert refined) revealed the following DICE-sco res: subcutaneous tissue (0.835 vs. 0.978), muscle (0.845 vs. 0.965), abdomi nal cavity (0.943 vs. 0.988), thoracic cavity (0.895 vs. 0.977), bone (0. 774 vs. 0.919), glands (0.576 vs. 0. 899), pericardium (0.697 vs. 0.945), mediast inum (0.731 vs. 0.914), brain (0.894 vs. 0.965), spinal canal (0.886 vs. 0.970) a nd the average of all classes (0.808 vs. 0.952).

Conclusion

The presented approach shows rapid adaptation of CT BCA models to MRI without manual annotation, achieving notable segmentation performance. When refined by experts, these metrics are further enhanced, enabling precise body composition analysis with red uced annotation effort.

Limitations

Validation of different MRI scanners is necessary t o ensure the generalizability and robustness of the proposed met hod. Funding for this study: No funding was provided for this study. Ethics committee - additional information: Informed consent was waived by the ethics committee due to the retrospective setti ng. Author Disclosures: Christian Bojahr: Nothing to disclose Katarzyna Anna Borys: Nothing to disclose Olivia Pollok: Nothing to disclose Johannes Haubold: Nothing to disclose Lale Umutlu: Nothing to disclose Cynthia Sabrina Schmidt: Nothing to disclose Matteo Mancino: Nothing to disclose René Hosch: Nothing to disclose Felix Nensa: Nothing to disclose Deep Learning-Based Fully Automated Body Compositio n Analysis as a Prognostic Factor in ARDS Patients using CT-Based O pportunistic Biomarkers *J. Kohnke*, K. Schmidt, F. Espeter, K. Pattberg, J . Haubold, F. Nensa, R. Hosch; Essen/DE Purpose or Learning Objective: Acute Respiratory Distress Syndrome (ARDS) is a severe condition with high morbidity an d mortality. Early risk assessment is crucial for improving outcomes and gu iding treatment. While body composition parameters have recently emerged a s prognostic factors, they are not commonly considered. However, image-ba sed Body Composition Analysis (BCA) can help extract relevant informatio n about patients. By leveraging deep learning, these features can be eff ectively used for enhanced risk stratification using detailed body information .

Methods

or Background: Thoracic CT scans from 960 ARDS patients (37.4 % female; median age = 54.7; interquartile range 43 .0 - 64.6), were analyzed. The scans were obtained within two days before or a fter ICU admission. Extracted BCA features include lung volume and sarc openia marker (muscle volume / bone volume). Based on the features, terti les were determined separately for both genders (lower tertile vs. othe rs). Kaplan-Meier, Log-Rank, and Cox-regression analyses compared 30-days surviv al between the tertiles.

Results

or Findings: Kaplan-Meier analysis revealed significant differen ces in survival based on lung volume markers (p = 0.02 for male; p = 0.09 female) and sarcopenia (p = 0.01 male; p = 0.52 female). Co x regression indicated that Lung volume (p = 0.02) and gender (p = 0.01) had si gnificant effects on survival, while sarcopenia (p = 0.07) was slightly not statistically significant for survival.

Conclusion

The results suggest that image based BCA from routi ne CT imaging could improve risk predictions in ARDS pati ents by using additional information of the patient's body.

Limitations

Although BCA parameters show promise, generalizabil ity is limited as all data were from a single center, high lighting the need for validation in broader clinical settings. Furthermore, the diff erences between the results depending on gender require further investigation. Funding for this study: None Ethics committee - additional information: This study adhered to all guidelines defined by the approving institutional r eview board of the investigating hospital. The Institutional Review Bo ard waived written informed consent due to the study's retrospective nature. Co mplete anonymization of all data was performed before inclusion in the study. Sunday Abstract-based Programme 278 Author Disclosures: Karsten Schmidt: Nothing to disclose Johannes Haubold: Nothing to disclose Judith Kohnke: Nothing to disclose René Hosch: Nothing to disclose Kevin Pattberg: Nothing to disclose Felix Nensa: Nothing to disclose Florian Espeter: Nothing to disclose Enhancing Autopsy Evaluations with AI-Driven Body C omposition Biomarkers from Post-Mortem CT Scans *J. Garrett*¹, M. Golden¹, M. Lee¹, S. Berry², N. A ppel³, H. Edgar³, P. J. Pickhardt¹; ¹Madison, WI/US, ²Kalamazoo, MI/U S, ³Albuquerque, NM/US ([email protected]) Purpose or Learning Objective: To correlate fully automated post-mortem CT (PMCT)-based measures of aortic calcification, skel etal muscle, and intra- abdominal fat of decedents with causes of death and comorbidities.

Methods

or Background: Retrospective study of the New Mexico Decedent Image Database (NMDID) with non-contrast PMCT scans between 2010-2017. An automated pipeline of AI-driven algorithms for q uantifying skeletal muscle, subcutaneous fat, visceral fat, and aortic calcific ation (Agatston score) from the abdominal component of PMCT scans was used. Scans w ith more than minimal decomposition were excluded. Cause of death was categorized as acute or chronic. CT-based biological age was deriv ed using a predetermined model.

Results

or Findings: The final cohort included 6638 decedents (mean age 50 ± 18 years; 74% male). Deaths were classified as 80% acute, 10% chronic, and 10% uncertain. Muscle density and area at the L 3 level were higher in the acute group compared to the chronic group (26 HU vs . 18 HU, p<0.001; 192 cm² vs. 183 cm², p<0.001) and higher in those witho ut cancer (25 HU vs. 16 HU, p<0.001; 190 cm² vs. 169 cm², p<0.01). Aortic A gatston scores were higher in heart disease deaths (5120 vs. 2098, p<0. 001). Diabetic patients had higher L3 visceral fat area (227 cm² vs. 175 cm², p <0.001) and lower muscle density (17 HU vs. 25 HU, p<0.001). The chronic gro up had a larger biological- chronological age gap than the acute group (median age gap, 19 years vs. 10 years; p<0.001).

Conclusion

Fully automated quantitative CT-based tissue biomar kers from PMCT scans match expectations based on previous stu dies on living patients and correlate with acuity of death and chronic co-m orbidities.

Limitations

The process imperfect of categorizing decedents int o “acute” or “chronic” causes of death based on death certificat es is imperfect without accounting for all potential medical confounders. Funding for this study: None Ethics committee - additional information: IRB Exempt study; non-human subjects per HIPAA Author Disclosures: Perry J. Pickhardt: Advisory Board: Nanox AI Heather Edgar: Nothing to disclose Matthew Lee: Nothing to disclose Shamsi Berry: Nothing to disclose John Garrett: Shareholder: NVIDIA Advisory Board: R adUnity Nicollette Appel: Nothing to disclose Max Golden: Nothing to disclose Deep Learning Models for Cardiomegaly Detection Ena bles Assessment of Cardiomegaly Prevalence in an International CT D ata Repository: Insights from AICT Consortium *U. Zidan*¹, N. Bi¹, A. Chandrashekar¹, M. Bown², E . Joviliano³, V. Grau¹, E. R. Ranschaert⁴, R. Lee¹; ¹Oxford/UK, ²leicester/UK, ³São Paulo/BR , ⁴Ghent/BE ([email protected]) Purpose or Learning Objective: To develop high-performance ML/DL pipelines for the detection and characterization of cardiomegaly in a diverse international repository of CT scans.

Methods

or Background: The AICT consortium (www.aict.ai) consists of 10 clinical sites across 3 continents, contributing CT scans in an agnostic fashion to a common research repository. The ultimate goal is to collect 1 million CT studies, enabling ML/DL training at an unprecedente d scale. . This pilot analysis includes the first 5487 unique individuals encompassing 1978 chest CT scans performed from March 2017 to September 202 4. Two published models (Superem Total Segmentor) were used to detec t cardiomegaly.

Results

or Findings: Here we report the findings on cardiomegaly (define d as a cardiothoracic ratio [CTR] > 0.50). Of the 1978 i ndividuals, 1577 did not exhibit cardiomegaly (784 males and 793 females), a nd 401 had cardiomegaly (178 males and 223 females). The overall prevalence of cardiomegaly is 20%. The prevalence is higher among females (22%) compar ed to males (19%) (p<0.05). The average age of those with cardiomegal y is on average 70 years old (range: 21-96) [m:65,(21-96); f:75,(36-95); p<0 .05]. The mean CTR in those with cardiomegaly is 0.55 (±0.05) [m: 0.55 (±0.05); f: 0.56 (±0.06); p=ns].

Conclusion

The AICT Consortium repository, combined with high- throughput ML/DL analytic pipelines, provides novel insights i nto the prevalence and demographic distribution of cardiomegaly in a conte mporary international cohort. This data enhances our understanding of car diomegaly epidemiology and supports the development of advanced detection methods.

Limitations

[To be added based on study outcomes] Funding for this study: Horizon Europe and UK Research Innovation Ethics committee - additional information: The study was approved by HRA (22/HRA/2302) Author Disclosures: Vicente Grau: Nothing to disclose Anirudh Chandrashekar: Nothing to disclose Edwaldo Joviliano: Nothing to disclose Regent Lee: Nothing to disclose Usama Zidan: Nothing to disclose Matt Bown: Nothing to disclose Ning Bi: Nothing to disclose Erik R. Ranschaert: Nothing to disclose Detection of osteoporotic vertebral body compressio n fracture in computed tomography scans of the chest and abdomen using artificial intelligence Nanox.AI *V. Mathew*, D. Pearce, N. Kate Rose, S. Saini, E. Bogoch; Toronto, ON/CA ([email protected]) Purpose or Learning Objective: The detection of undiagnosed vertebral compression fractures (VCFs) is critical due to the ir association with increased risk of future fragility fractures. Primary objecti ve is to evaluate the performance of Nanox.AI HealthOST in identifying incidental VCF s on outpatient chest and abdomen CT scans by assessing sensitivity, specific ity, PPV, and NPV. Secondary objective is to quantify missed VCFs on b y initial reporting radiologist.

Methods

or Background: HealthOST is an AI solution from Nanox.AI, providing automatic image analysis of the spine fro m CT images to support clinicians in the evaluation and assessment of indi cators of osteoporosis. Retrospective analysis on 590 outpatient CT cases f rom St. Michael’s Hospital at Unity Health Toronto. Two radiologists, includin g a senior musculoskeletal radiologist established a consensus “gold standard” for comparison with AI results. Two AI thresholds for vertebral height red uction were examined: mild (>20%) and moderate (>25%). Original radiologist re ports were reviewed to quantify missed VCFs on these scans.

Results

or Findings: At the 20% threshold, AI showed a sensitivity of 91 .1%, specificity of 52.7%, PPV of 17.1%, and NPV of 98.2 %. At the 25% threshold, sensitivity decreased to 79.9%, while specificity i mproved to 94.2%, with a PPV of 50.7% and NPV of 98.4%. AI increased fracture de tection by 88% compared to initial radiologist findings at the 20% threshol d and 92% at the 25% threshold.

Conclusion

Nanox.AI HealthOST shows potential as an effective tool for VCF screening, with high sensitivity at the 20% thresho ld and improved specificity at 25%. Given the variable specificity and substantial rate of false positives, a secondary review by radiologists is recommended for accuracy. Increased detection rate by the AI in comparison to the initi al radiologist report highlights the AI's capability to assist in fracture detection and enhancing diagnostic accuracy.

Limitations

None Funding for this study: AMGEN Inc. Ethics committee - additional information: Ethics committee approval REB# 21-183 Author Disclosures: Noah Kate Rose: Nothing to disclose Vinu Mathew: Nothing to disclose Dawn Pearce: Nothing to disclose Earl Bogoch: Research/Grant Support: Amgen Sidharth Saini: Nothing to disclose Sunday Abstract-based Programme 279 09:30-11:00 Research Stage 3 Research Presentation Session: Chest RPS 2304 Lung cancer imaging: characterisation and prognosis Moderator R.-I. Milos; Vienna/AT ([email protected]) Pulmonary adenocarcinoma: Correlation of Pathologic al Growth Pattern and Radiological Morphology on Computed Tomography L. Biggemann, H. Bohnenberger, J. Vincke, P. Kraus, T. Overbeck, A. Hammerstein-Eqourd, *J. Uhlig*; Göttingen/DE ([email protected]) Purpose or Learning Objective: Pulmonary adenocarcinoma (AC) can present with different pathological growth patterns . This study evaluates whether these growth patterns correlate with the ra diological morphology of the tumor on computed tomography (CT).

Methods

or Background: Patients with surgically resected pulmonary AC and preoperative CT imaging were retrospectively includ ed. Cases were sampled to distribute growth patterns approximately evenly. Pathological growth patterns were assessed on a representative patholog ical slice. The predominant growth pattern was defined as >=60%. Ra diological morphology was assessed using preoperative thoracic CT scans a nd compared across growth patterns using the chi-square test.

Results

or Findings: A total of n=345 patients were included (43.8% fema le; median age 68 years). Pathological growth patterns were acinar (n=68), lepidic (n=51), micropapillary (n=58), papillary (n=50), mu cinous (n=51), and solid (n=67). While age was balanced across AC growth pat terns, lepidic and micropapillary ACs were more common in women (56.9% / 51.7%); and solid and acinar ACs more likely in men (71.6% / 57.4%; o verall p <0.05). Pathological growth patterns demonstrated specific morphologies on CT regarding nodule type, margin, ground glass opaciti es, central low attenuation, air bronchogram, associated lymphadenopathy and loc ation of distant metastases (each variable p<0.05). For example, lep idic ACs commonly presented as ground-glass or partially solid nodule s (13.7% / 35%); acinar ACs with associated ground-glass opacities (77.9%); sol id ACs with contact to pulmonary fissures (50.7%) and central low attenuat ion (58.2%); and mucinous ACs with air bronchogram (41.2%).

Conclusion

Radiological morphology of pulmonary ACs on CT corr elates well with pathological growth patterns, which could aid guiding diagnostic and treatment patterns.

Limitations

Limitations include that only a representative path ological slice of the pulmonary AC was assessed, whereas growth patte rns might vary in the full tumor volume; and that CT-imaging was performe d on different CT- scanners, introducing heterogeneity. Funding for this study: This study has been supported by Siemens Healthineers. Ethics committee - additional information: Ehtics committee of the University Medical Center Goettingen Author Disclosures: Hanibal Bohnenberger: Nothing to disclose Tobias Overbeck: Nothing to disclose Paul Kraus: Nothing to disclose Johannes Uhlig: Investigator: Siemens Healthineers Advisory Board: Bayer Jan Vincke: Nothing to disclose Lorenz Biggemann: Nothing to disclose Alexander Hammerstein-Eqourd: Nothing to disclose AI-Based Computer-Aided Volumetry for Invasiveness Evaluation in Lung Adenocarcinoma: Influence of Radiation Dose Reducti on and Reconstruction Algorithms on High-Definition CT *Y. Ozawa*, D. Takenaka, H. Nagata, T. Ueda, M. Nom ura, T. Yoshikawa, Y. Ohno; Toyoake/JP ([email protected]) Purpose or Learning Objective: To determine the influence of radiation dose and reconstruction method on artificial intelligenc e (AI)-based computer-aided volumetry (CADv) for nodule component measurement a nd diagnostic performance to evaluate invasiveness in lung adenoc arcinoma on high- definition CT (HDCT).

Methods

or Background: 112 consecutive patients with 181 lung adenocarcinomas underwent thin-section HDCTs at sta ndard-dose (SDCT: 9.0±1.8 mGy), reduced-dose (RDCT: 1.7±0.2 mGy) and ultra-low-dose (ULDCT: 0.8±0.1 mGy) levels. All HDCT data were rec onstructed with hybrid- type iterative reconstruction (IR) and deep learnin g reconstruction (DLR). Then, standard references for solid and GGO compone nts and consolidation- to-tumor ratio (CTR) were computationally determine d with the simultaneous truth and performance level estimation (STAPLE) met hod from annotated CT data by three board-certified chest radiologists. T hen, each component volume and CTR on all HDCT data were measured by AI-based CADv software. Each component volume and consolidation-to-tumor ratio ( CTR) were correlated between CADv measurement on each CT data and standa rd reference. Then, Measurement differences of each index between stand ard reference and each CADv measurement were compared among all CT data by Tukey’s HSD test. Finally, diagnostic performance of invasiveness was compared among all CTR measurements by ROC analysis.

Results

or Findings: There were significant correlations for each compon ent and volume and CTR on all HDCTs (hybrid-type IR: 0. 71≤r≤0.88, p<0.0001; DLR: 0.71≤r≤0.88, p0.05). Area under the curve (AUC) of each HDCT with DLR (SDCT: AUC=0.98, RDCT: AUC=0.98, ULDCT: AUC=0.97) was sign ificantly larger than all HDCTs with hybrid-type IR (AUC=0.95, p<0.0 5).

Conclusion

Reconstruction method rather than radiation dose re duction had significantly affected to diagnostic performance of AI-based CADv for invasiveness evaluation in lung adenocarcinoma on H DCT.

Limitations

Number of subjects Funding for this study: Reconstruction method is more important than radiation dose reduction for keeping diagnostic per formance of AI-based CADv for invasiveness evaluation in lung ademocarcinoma on high-definition CT. Ethics committee - additional information: Research grants from Canon Medical Systems Corporation and Smoking Research Fo undation Author Disclosures: Yoshiyuki Ozawa: Research/Grant Support: Grants-in- Aid for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Smoking Rese arch Foundation Masahiko Nomura: Nothing to disclose Takahiro Ueda: Research/Grant Support: Grants-in-Ai d for Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Daisuke Takenaka: Nothing to disclose Hiroyuki Nagata: Research/Grant Support: Grants-in- Aid for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Canon Medica l Systems Corporation Takeshi Yoshikawa: Nothing to disclose Yoshiharu Ohno: Research/Grant Support: Smoking Res earch Foundation Research/Grant Support: Canon Medical Systems Corpo ration Variability of Enlarged Mediastinal Lymph Node Meas urement in CT: Manual vs. Automatic Assessment A. Olesinski, R. Lederman, Y. Azraq, J. Sosna, *L. Joskowicz*; Jerusalem/IL ([email protected]) Purpose or Learning Objective: Detection of enlarged mediastinal lymph nodes (LNs) in CECT chest scans is necessary for th e assessment of oncology patients. It is performed by manually measuring the short axis length (SAL) of the LNs. We aimed to quantify the interobserver agr eement and variability of the SAL measurements and to compare them to automat ically computed SALs from volumetric LN delineations.

Methods

or Background: We retrospectively analyzed 40 CECT chest scans from patients with mediastinal lymphadenopathy. Two radiologists with 25 years and 35 years of experience independently meas ured the SALs of 471 LNs by: identifying the scan slice in which the LN appears largest, drawing the line segment along its short axis; and classifying the LN as normal (SAL<10mm) or enlarged (SAL≥10mm). Four weeks later, one of the radiologists manually segmented the volume of the L N's along its contours. The LNs were automatically classified as normal/enl arged based on their SALs computed from the delineation (COMP). Confusion mat rices were computed, as well as the differences between the SALs.

Results

or Findings: The normal/enlarged LN overall agreement (371 norma l LNs, 49 enlarged LNs) between both radiologists was 94.8% (420/471). For agreement/disagreement, the SAL differences (std) w ere 1.2(1.1)mm and 3.5(3.2)mm. Note that the SALs difference is nearly twice as large for disagreement as for agreement. The normal/enlarged overall agreement between the manual and the automatically computed S ALs and both radiologists was 94.8%(421/471) and 92.4%(411/471).

Conclusion

Identification of enlarged mediastinum lymph nodes in chest CECT based on short axis measurements derived manua lly or computed from lymph nodes delineations has high accuracy. The agr eement of the computed SALs from LN contour delineations is within the man ual SALs interobserver Sunday Abstract-based Programme 280 variability. Accurate automatic LN segmentations ca n be obtained with existing methods.

Limitations

Single institution and single contour delineation. Funding for this study: None. Ethics committee - additional information: Yes, Helsinki Committee of the Hadassah University Medical Center Author Disclosures: Richard Lederman: Consultant: HighRAD Leo Joskowicz: Consultant: HighRAD Jacob Sosna: Consultant: HighRAD Alon Olesinski: Consultant: HighRAD Yusef Azraq: Nothing to disclose Multiparametric 18F-FDG PET/MRI based on restrictiv e spectrum imaging and amide proton transfer-weighted imaging facilita tes the assessment of lymph node metastases in non-small cell lung cancer *N. Meng*¹, X. Liu¹, J. Pan¹, X. Yu¹, Y. Wu¹, Y. Ya ng², Z. Wang³, M. Wang¹; ¹Zhengzhou/CN, ²Bei Jing/CN, ³Shanghai/CN ([email protected]) Purpose or Learning Objective: To investigate the value of multiparametric 18F-FDG PET/MRI based on tri-compartmental restrict ive spectrum imaging (RSI), amide proton transfer-weighted imaging (APTW I), and diffusion- weighted imaging (DWI) in the assessment of lymph n ode metastases (LNM) of non-small cell lung cancer (NSCLC)

Methods

or Background: A total of 152 NSCLC patients were enrolled. 18F- FDG PET- derived parameter (SUVmax), RSI-derived pa rameters (f1, f2, and f3), APTWI-derived parameter (MTRasym(3.5ppm)), DWI -derived parameter (ADC), and were calculated. Logistic regression ana lysis was used to identify independent predictors, and combined diagnostics. A UC, calibration curves and decision curve analysis (DCA) were employed to assess the performance of the combined diagnostics.

Results

or Findings: MTRasym(3.5ppm), SUVmax, f2, and f3 were higher and ADC and f1 were lower in LNM-positive group tha n in LNM-negative group (all P < 0.05). Maximum lesion diameter, f1, MTRasy m(3.5ppm) , SUVmax, and ADC were independent predictors of LNM status i n NSCLC patients, and the combination of them had an optimal diagnostic e fficacy (AUC = 0.978; sensitivity = 95.35 %; specificity = 90.91 %), whic h was significantly higher than maximum lesion diameter, f1, MTRasym(3.5ppm), SUVmax, and ADC (AUC = 0.774, 0.810, 0.832, 0.834, and 0.783, respe ctively, and all P < 0.01). The combined diagnosis showed a good performance (A UC = 0.968) in the bootstrap (1000 samples)-based internal validation. Calibration curves and DCA demonstrated that the combined diagnosis not on ly provided better stability, but also resulted in a higher net benefi t for the patients involved.

Conclusion

Multiparametric 18F-FDG PET/MRI based on RSI, APTWI , and DWI is beneficial for the non-invasive assessment o f LNM status in NSCLC, and the combination of maximum diameter, f1, MTRasy m(3.5ppm), SUVmax, and ADC may serve as a promising biomarker.

Limitations

This study was conducted at a single institution wi th a relatively small sample size Funding for this study: The National Key R&D Program of China (2023YFC2414200), the National Natural Science Foun dation of China (82371934), the Joint Fund of Henan Province Scienc e and Technology R&D Program (225200810062). Ethics committee - additional information: The current study received approval from the local ethics committee, and all p articipants provided written informed consent for participation. Author Disclosures: Xue Liu: Nothing to disclose Yang Yang: Nothing to disclose Jiayin Pan: Nothing to disclose Nan Meng: Nothing to disclose Meiyun Wang: Nothing to disclose Yaping Wu: Nothing to disclose Zhe Wang: Nothing to disclose Xuan Yu: Nothing to disclose Diagnostic Accuracy in NSCLC Lymph Node Staging wit h Total-Body and Conventional PET/CT *C. Mingels*¹, M. H. Madani¹, F. Sen¹, H. Nalbant¹, Y. G. Abdelhafez¹, M. Guindani², R. Badawi¹, B. A. Spencer¹, L. Nardo¹ ; ¹Sacramento/US, ²Los Angeles, CA/US Purpose or Learning Objective: To characterize diagnostic accuracy for nodal (N)-staging with [18F]FDG Total-Body (TB) and short-axial field-of-view (SAFOV) PET/CT in non-small cell lung cancer (NSCLC ) patients.

Methods

or Background: In this prospective, randomized, single center head - to-head comparative study 48 patients underwent TB and SAFOV PET/CT. 700 nodal levels (1R/L, 2R/L, 3a/p, 4R/L, 5, 6, 7, 8R/L, 9R/L, 10-14R/L) of 28 patients could be associated to histopathological f indings, imaging after localized or systemic treatment, which allowed calc ulation of sensitivity, specificity, positive (PPV) and negative predictive value (NPV). Thresholds for maximum standardized uptake value (SUVmax), tumor-t o-background ratio (TBR), metabolic tumor volume (MTV) and total-lesio n glycolysis (TLG) were calculated.

Results

or Findings: TB and SAFOV PET/CT showed high diagnostic accuracy indices for patient-based N-staging. Sensi tivity and specificity were 86.0% (CI: 77.0-95.0%) and 98.3% (CI: 97.3-99.3%) f or TB; 77.2% (CI: 66.3- 88.1%) and 97.4% (CI: 96.1-98.6%) for SAFOV PET. PP V was higher for TB (81.7%, CI: 71.9-91.5%) compared to SAFOV PET (72.1 %, CI: 60.9-83.4%), however, this finding was not statistically signifi cant (p=0.08). NPV for TB (98.6%, CI: 97.9-99.6%) and SAFOV PET/CT (98.0%, CI : 96.9-99.1%) were comparable (p=0.22). Overall, NSCLC N-staging was a ffected in six cases on SAFOV and only in one case on TB PET/CT. Semi-quant itative analysis revealed a SUVmax-threshold of 3.0 to detect TP les ions on both scanners. TBR, MTV and TLG thresholds were lower on TB compar ed to SAFOV PET (TBR: 1.2 vs. 1.7, MTV: 0.5 ml vs. 1.0 ml and TLG: 1.0 ml vs. 3.0 ml).

Conclusion

TB and SAFOV PET/CT showed high diagnostic accuracy for N- staging in NSCLC. Sensitivity and PPV on TB PET/CT were slightly higher compared to SAFOV PET/CT. TB PET/CT showed lower ra te of incorrect N- staging and lower semi-quantitative thresholds.

Limitations

Small sample size, composite reference standard wit h imaging Funding for this study: Research reported in this publication was supported by the National Institutes of Health under award nu mber R01CA249422. The work was also supported by the In Vivo Translationa l Imaging Shared Resources with funds from NCI P30CA093373 and by th e Fred and Julia Rusch Foundation for Nuclear Medicine Research and Education. Hande Nalbant’s funding is partially provided by United I maging Health’s UIH Fellowship Gift. Ethics committee - additional information: This study was approved by the UC Davis institutional review board (IRB 1506448). Written informed consent for inclusion was obtained. The study was performed in accordance with the Declaration of Helsinki. Author Disclosures: Fatma Sen: Nothing to disclose Hande Nalbant: Nothing to disclose Clemens Mingels: Nothing to disclose Michele Guindani: Nothing to disclose Mohammad H. Madani: Nothing to disclose Ramsey Badawi: Nothing to disclose Lorenzo Nardo: Nothing to disclose Benjamin A. Spencer: Nothing to disclose Yasser Gaber Abdelhafez: Nothing to disclose Whole-lesion iodine map histogram analysis versus s ingle-slice spectral CT parameters for determining of visceral Pleural I nvasion in NSCLC *K. Zhu*, J. Zhou; Lanzhou/CN ([email protected]) Purpose or Learning Objective: To evaluate and compare the performances of whole-lesion iodine map histogram analysis to th ose of single-slice spectral CT parameters in discriminating of visceral pleural invasion in NSCLC.

Methods

or Background: A total of 99 NSCLC patients underwent preoperative spectral CT and were divided into two groups: VPI and non- VPI.There were 65 men and 34 women with a mean age of 59.33 ± 8.62 (standard deviation) years(range:37-79 years) .The whole-lesion iodine map histogram parameterswere measured for each NSCLC pa tient. By placing regions of interest at representative levels of the tumor and normalizing them, spectral CT parameters IC and NIC were obtained. Di scriminating capabilities of spectral CT and histogram parameters were assess ed and compared using area under the ROC curve (AUC)and logistic regressi on models.

Results

or Findings: The SD, Variance and CV of the iodine map histogram analysis,and iodine concentration and normalized io dine concentration of single-slice spectral CT parameters were significan tly different of visceral pleural invasion in NSCLC (P < 0.001 to P = 0.03). The CV of histogram parameters (AUC=0.65; 95% [CI]: 0.54-0.76) and norm alized iodine concentration (AUC=0.75; 95% CI: 0.64-0.85) from sp ectral CT parameters had the best performance for distinguishing whether visceral pleural invasion occurred in NSCLC. At ROC curve analysis no signifi cant differences in AUC were found between histogram parameters (AUC = 0.84 ; 95% CI: 0.76-0.91) and spectral CT parameters (AUC = 0.66; 95% CI: 0.5 5-0.77) (P = 0.24).

Conclusion

Both whole-lesion iodine map histogram analysis and single-slice spectral CT parameters help discriminate whether of visceral pleural invasion in NSCLC, and the single-slice spectral CT paramete rs performed better in terms of diagnostic efficacy.

Limitations

This finding has not been validated in an independe nt population, limiting their generalizability. Future prospective studieswith larger and external patient cohorts are warranted. Funding for this study: This research has been supported by the National Natural Science Foundation of China (82371914). Sunday Abstract-based Programme 281 Ethics committee - additional information: The study was approved by the Institutional Ethics Committee (2021A-498) and exem pted from patient informed consent. Author Disclosures: Junlin Zhou: Nothing to disclose Kaibo Zhu: Nothing to disclose MRI-Based Molecular Imaging vs. FDG-PET/CT: Capabil ity for Postoperative Recurrence Prediction with FDG-PET/CT in Stage I NSCLC Patients *Y. Ozawa*, H. Nagata, T. Ueda, M. Nomura, T. Yoshi kawa, D. Takenaka, Y. Ohno; Toyoake/JP ([email protected]) Purpose or Learning Objective: To compare the prediction capability for postoperative recurrence among FDG-PET/CT and MRI-b ased molecular information from chemical exchange saturation trans fer (CEST) imaging and diffusion-weighted imaging (DWI) in stage I non-sma ll cell lung cancer (NSCLC) patients.

Methods

or Background: 79 pathologically diagnosed and surgically treated NSCLC patients who underwent CEST imaging, DWI and FDG-PET/CT, follow-up and pathological examinations were includ ed in this study. According to the follow-up and pathological examination resul ts, all patients were divided as recurrence (n=13) and non-recurrence (n=66) grou ps. In each lesion, magnetization transfer ratio asymmetry at 3.5ppm (M TRasym), apparent diffusion coefficient (ADC) and SUVmax of each nodu le were assessed by ROI measurements. To compare all indexes between two gr oups, Student’s t-test was performed. To determine the significant predict ors, multiple logistic regression analysis was performed. Then, ROC analys is was performed to compare distinguishing two groups among all indexes and combined significant predictors. Finally, sensitivity (SE), specificity (SP) and accuracy (AC) were compared among all methods by McNemar’s test.

Results

or Findings: There was significant difference of each index betw een two groups (p<0.05). Multiple logistic regression a nalyses determined MTRasym (Odds ratio [OR]: 1.31, p=0.03) and ADC (OR : 0.002, p=0.008) as significant predictors. When applied each threshold value, SPs and ACs of MTRasym (SP: 81.8%, AC: 82.2%), ADC (SP: 87.9%, AC: 86.1%) and combined predictors (SP: 89.4%, AC: 89.9%) were sig nificantly higher than those of SUVmax (SP: 69.7%, p<0.05; AC: 72. %, p<0. 05). Moreover, AC of combined predictors was significantly higher than t hat of MTRasym (p=0.03).

Conclusion

MRI-based molecular information has better predicti on capability for postoperative recurrence than FDG-PET/CT in sta ge I NSCLC patients.

Limitations

Limited study cohort number and follow-up periods i n some patients are considered as limitations in his study . Funding for this study: Canon Medical Systems Corporation Ethics committee - additional information: Fujita Health University Hospital Author Disclosures: Yoshiyuki Ozawa: Research/Grant Support: Grants-in- Aid for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Smoking Rese arch Foundation Masahiko Nomura: Nothing to disclose Takahiro Ueda: Research/Grant Support: Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Daisuke Takenaka: Nothing to disclose Hiroyuki Nagata: Research/Grant Support: Canon Medi cal Systems Corporation Research/Grant Support: Grants-in-Aid f or Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Takeshi Yoshikawa: Nothing to disclose Yoshiharu Ohno: Research/Grant Support: Canon Medic al Systems Corporation Research/Grant Support: Smoking Researc h Foundation Interstitial lung abnormalities are significant poo r prognostic factors in resected clinical stage Ⅰ non-small cell lung cancer *T. Hino*, T. Akamine, T. Hida, K. Sagiyama, Y. Yam asaki, K. Tabata, K. Ishigami; Fukuoka/JP ([email protected]) Purpose or Learning Objective: Interstitial lung abnormalities (ILA) are known to be associated with increased mortality; ho wever, the impact of ILA on post-operative prognosis of non-small cell lung can cer (NSCLC) remains unclear. The aim of study is to assess the associat ion between mortality in post-operative clinical stage Ⅰ NSCLC patients and the presence of ILA.

Methods

or Background: We retrospectively evaluated the patients who underwent chest CT, followed by radical resection f or clinical stage Ⅰ NSCLC from 2006 to 2018. The presence of ILA was evaluate d on high-resolution CT images by two radiologists and one thoracic surgeon . Five-year overall survival (OS), recurrence-free survival (RFS), and cumulativ e incidence of other cause of death stratified by the presence of ILA were ass essed.

Results

or Findings: Among 709 patients included in this study, 80 patie nts had ILA (11.2%). Five-year OS and RFS were signific antly lower in patients with ILA than those without ILA (43.6% vs. 90.1 %, log-rank test p<0.001; 44.7% vs. 80.6 %, log-rank test p<0.001). Multivari able analysis demonstrated that the presence of ILA was an independent poor pr ognostic factor in both OS and RFS (HR: 1.45, 95% CI: 1.14–1.84, p=0.003; HR: 1.37, 95% CI: 1.08– 1.74, p=0.010, respectively). Five-year cumulative incidence of other cause of death was significantly higher in patients with ILA than those with non-ILA (21.1% vs. 5.4%, Gray’s test p<0.001).

Conclusion

The presence of ILA affected the cause of other dea th and was an independent poor prognostic factor in clinical s tage Ⅰ NSCLC.

Limitations

All the participants were retrospectively collected . ILA was not conformed with pathological specimens. Funding for this study: N/A Ethics committee - additional information: Kyushu University Hospital IRB No. 23240-00 Author Disclosures: Kosuke Tabata: Nothing to disclose Kousei Ishigami: Nothing to disclose Yuzo Yamasaki: Nothing to disclose Akamine Takaki Akamine: Nothing to disclose Koji Sagiyama: Nothing to disclose Takuya Hino: Nothing to disclose Tomoyuki Hida: Nothing to disclose Intratumoral and peritumoral CT radiomics in predic ting anaplastic lymphoma kinase mutations and survival in patients with lung adenocarcinoma: a multicenter study *G. Lin*, W. Chen, J. Ji; Lishui/CN Purpose or Learning Objective: To explore the value of intratumoral and peritumoral radiomics in preoperative prediction of anaplastic lymphoma kinase (ALK) mutation status and survival in patients with lung adenocarcinoma.

Methods

or Background: We retrospectively collected data from 505 eligible patients with lung adenocarcinoma from four hospita ls (training and external validation sets 1–3). The CT-based radiomics featur es were extracted separately from the gross tumor volume (GTV) and GT V incorporating peritumoral 3-, 6-, 9-, 12-, and 15-mm regions (GPT V3, GPTV6, GPTV9, GPTV12, and GPTV15), and screened the most relevant features to construct radiomics models to predict ALK (+). The combined m odel incorporated radiomics scores (Rad-scores) of the best radiomics model and clinical predictors was constructed. Performance was evaluat ed using receiver operating characteristic (ROC) analysis. Survival o utcomes were examined using the Cox proportional hazards model.

Results

or Findings: The GPTV3 radiomics model using a support vector machine (SVM) algorithm achieved the best predictiv e performance, with the highest average AUC of 0.811 in the validation sets . Clinical TNM stage and pleural indentation were independent predictors. Th e combined model incorporating the GPTV3-Rad-score and clinical pred ictors achieved higher performance than the clinical model alone in predic ting ALK (+) in three validation sets (AUC: 0.855 vs. 0.648, 0.882 vs. 0. 634, 0.810 vs. 0.663). The prediction score of the combined model could strati fy survival outcome in patients receiving ALK-TKI therapy (P=0.026) and im munotherapy (P=0.012).

Conclusion

The presented combined model based on GPTV3 effecti vely mined tumor features to predict ALK mutation status and stratify survival in patients with lung adenocarcinoma.

Limitations

This is a retrospective study and may have varying degrees of selection bias. Funding for this study: This research was funded by the National Natural Science Foundation of China (Grant No.82072026 to J iansong Ji), Key Project of Joint Construction by Provincial and Ministerial Authorities (Grant No.WKJ- ZJ-2452 to Minjiang Chen), Medical and Health Gener al Project of Zhejiang Province (Grant No. 2024KY568 to Weiyue Chen, and N o. 2023KY425 to Guihan Lin). Ethics committee - additional information: This study was approved by the Institutional Review Boards, and the requirement fo r informed consent was waived due to retrospective nature. Author Disclosures: Jiansong Ji: Nothing to disclose Weiyue Chen: Nothing to disclose Guihan Lin: Nothing to disclose Sunday Abstract-based Programme 282 09:30-11:00 Research Stage 4 Research Presentation Session: Genitourinary RPS 2307 Malignant lesions of the female pelvis: advances in imaging techniques, diagnostic and follow-up Moderator A. M. Hötker; Zurich/CH A Machine Learning Model Based on Endometrium MRI R adiomics to Predict Histological Diagnosis From Biopsy in Subje cts at Risk of Endometrial Cancer: Pilot study *R. V. Ninkova*, M. Gennarini, V. Miceli, A. Cupert ino, F. Curti, C. Catalano, L. Manganaro; Rome/IT ([email protected]) Purpose or Learning Objective: Aim of this study was to develop a machine learning model based on Magnetic Resonance Imaging (MRI) to stratify the single-subject risk of endometrial cancer (EC).

Methods

or Background: From September 2023 to July 2024, we collected MRI images from 41 patients. Among these subjects, 15 patients (36.6%) belonged to class "microsatellite instability (MSI) " and 26 patients (63.4%) belonged to class "microsatellite stability (MSS)", according to histological diagnosis from biopsy. This image set was used for the training and cross- validation of different machine learning models. A robust radiomic approach was applied, under the hypothesis that radiomic fea ture could be able to capture the disease heterogeneity among the two gro ups.

Results

or Findings: Three models consisting of 3 ensembles of machine- learning classifiers (random forests, support vecto r machines and k-nearest neighbor classifiers) were developed for the binary classification task of interest (“MSI” vs. “MSS”), based on supervised lea rning, using histological diagnosis from biopsy as reference standard. The be st model showed ROC- AUC mean value of 84 % [78.4-89.8], accuracy mean v alue of 75.6% [69.6- 81.7], sensitivity mean value of 68.9 % [59.3-78.5] , specificity mean value of 79.5 % [74-85], PPV mean value of 71.66 %[58.1-73.9 ], and NPV mean value of 81.6% [76.3-86.9] (p <0.005 mean value).

Conclusion

The radiomics-based machine learning model achieved a high diagnostic performance in the molecular stratificat ion of patients with EC, which could improve risk stratification and support clinical therapeutic decision. This is a preliminary study which could provide a n ew perspective for patients with EC, allowing a complete and accurate identific ation of the disease and promoting personalized treatment.

Limitations

The primary limitation of this study is the small p atient cohort. Expanding the sample size and investigating additio nal molecular pathological profiles will be necessary for further validation. Funding for this study: None Ethics committee - additional information: The study was performed in line with the principles of the Declaration of Helsinki. Author Disclosures: Marco Gennarini: Nothing to disclose Valentina Miceli: Nothing to disclose Federica Curti: Nothing to disclose Angelica Cupertino: Nothing to disclose Roberta Valerieva Ninkova: Nothing to disclose Carlo Catalano: Nothing to disclose Lucia Manganaro: Nothing to disclose FIGO 2023 staging of endometrial cancer: is there s till a role for radiology? *A. Rame*¹, S. Bottazzi¹, G. Avesani¹, M. Bonatti², V. Celli¹, E. Perrone¹, T. Pasciuto¹, B. Gui¹, E. Sala¹; ¹Rome/IT, ²Bolzano /IT ([email protected]) Purpose or Learning Objective: To investigate the role of MRI in the preoperative evaluation of endometrial cancer (EC) considering the new FIGO 2023 staging system.

Methods

or Background: Patients diagnosed with EC from two institutions between 2019 and 2023 were retrospectively included . Inclusion criteria were the availability of preoperative MRI and biopsy, mo lecular data and definitive histopathological data obtained after surgery. Two radiologists retrieved MRI findings from reports and combined them with biopsy results (grading and histology) to determine the FIGO 2023 preoperative stage. Definitive histopathological and molecular data served as the gold standard (final staging). The preoperative evaluation was compared with final staging (FIGO stages I, II, III and IV as independent categories) and discrepancies were recorded.

Results

or Findings: 231 patients were included. The agreement between preoperative evaluation and final staging was 74% ( 171/231), while the FIGO stage was discordant at 26% (60/231). Causes of dis cordance were: lymph node (LN) involvement in 26.7% (16/60), changes in grading or histological subtype between biopsy and surgical specimen in 23. 3% (14/60), metastases and carcinosis in 15% (9/60: 5 in the upper abdomin al peritoneum, 1 in the pelvic peritoneum and 3 distant), the presence of s ubstantial lymphovascular space invasion (LVSI) in 18.2% (10/55). Less common causes of discrepancies were myometrial invasion (6.7%; 4/60) , cervical stroma invasion (5%; 3/60), ovarian involvement (3.3%; 2/60) and va ginal invasion (1.7%; 1/60). Overall, parameters not assessable preoperat ively (LVSI and definitive grading and histological subtype) accounted for sta ging discrepancies in 10.4% (24/231) of cases.

Conclusion

Our study showed that the inclusion of parameters n ot assessable preoperatively in the new FIGO 2023 stag ing system does not significantly diminish the role of MRI in the preop erative evaluation of EC, resulting in staging discrepancy in only approximat ely 10% of cases.

Limitations

No limitations. Funding for this study: No funding was received for this study. Ethics committee - additional information: ID 6194, approved on 7/12/2023 Author Disclosures: Matteo Bonatti: Nothing to disclose Tina Pasciuto: Nothing to disclose Giacomo Avesani: Nothing to disclose Emanuele Perrone: Nothing to disclose Benedetta Gui: Nothing to disclose Veronica Celli: Nothing to disclose Silvia Bottazzi: Nothing to disclose Evis Sala: Nothing to disclose Anna Rame: Nothing to disclose Value of enhanced t1 mapping MR imaging in the eval uation of the depth of myometrial invasion in endometrial cancer: compa red with dynamic contrast-enhanced MR imaging *X. Liu*, Z. Yuan, Y. Li, J. Ren, Y. He, H. Xue, Z. Jin; Beijing/CN ([email protected]) Purpose or Learning Objective: To compare the diagnostic efficiency of enhanced T1 mapping MR imaging and dynamic contrast -enhanced (DCE) imaging for assessing the depth of myometrial invas ion in patients with endometrial cancer.

Methods

or Background: 46 women diagnosed with endometrial cancer underwent preoperative MR imaging. Two radiologists independently assessed the depth of myometrial invasion, categorized as no myometrial invasion, superficial myometrial invasion, or deep myometrial invasion, on T2WI+DWI+DCE MR imaging, followed by T2WI+DWI+enhan ced T1 mapping MR imaging, four weeks later. The findings were the n compared to histopathological examinations. The diagnostic perf ormance comparison was conducted using the chi-square test.

Results

or Findings: The overall accuracy for accessing depth of myometr ial invasion on T2WI+DW+DCE and T2WI+DW+enhanced T1 map ping were 76.1%, 80.4% for reader 1, and 78.3%, 80.4% for rea der 2, respectively. The increment was not statistically significant for eit her reader. While assessing the absence of myometrial invasion, the precision, sens itivity, and specificity achieved by both radiologists using DCE were 100%, 16.7%, and 100%, whereas for enhanced T1 mapping the precisions were 100%, with sensitivities of 50% and 33%, and specificities of 100%. In evalu ating superficial myometrial invasion, these values using DCE were 82 .4%/82.8%, 84.8%/87.9%, and 53.8%; using enhanced T1 mapping, these values were 87.5%/85.3%, 84.8%/87.9%, and 69.2%/61.5%. Assessme nt of deep myometrial invasion achieved these values of 54.5%/ 60%, 85.7%, and 87.2%/89.7% with DCE; 54.5%/60%, 85.7%, 87.2%/89.7% with enhanced T1 mapping. Inter-reader agreement, measured with kapp a values, was 0.831 with DCE-MRI and 0.899 with enhanced T1 mapping. Bo th radiologists concurred that enhanced T1 mapping substantially im proved diagnostic confidence over DCE-MRI.

Conclusion

Enhanced T1 mapping demonstrates superior diagnosti c efficiency in the evaluation of myometrial invasion in endometrial cancer compared with DCE MR imaging.

Limitations

This was a pilot study with a small sample size. Funding for this study: This work was supported by grants from Natural Science Foundation of China (grant No. 82271886). Ethics committee - additional information: This retrospective study was approved by the institutional review board, which w aived the requirement of informed consent. Sunday Abstract-based Programme 283 Author Disclosures: Jing Ren: Nothing to disclose Yuan Li: Nothing to disclose Zhilin Yuan: Nothing to disclose Zhengyu Jin: Nothing to disclose Huadan Xue: Nothing to disclose Yonglan He: Nothing to disclose Xinyu Liu: Nothing to disclose MRI in early-stage cervical cancer after cone biops y: can it predict the presence of residual tumor? *A. Amerighi*, M. Dolciami, A. Napoletano, I. Isufi , G. Avesani, E. Sala, B. Gui; Rome/IT Purpose or Learning Objective: Cervical cancer (CC) is increasingly diagnosed at earlier stages due to cancer screening programs. Cone biopsy is often necessary to rule out invasive disease and ca n sometimes suffice for treatment; however, determining residual disease is essential for treatment planning. Despite its crucial role in managing CC, the role of Magnetic Resonance Imaging (MRI) in determining residual dis ease in patients after conization remains poorly understood, with only a f ew studies focusing on this topic. We aimed to assess MRI accuracy in detecting residual disease after cervical conization for early-stage cancer and comp are the diagnostic performance of MRI with and without a contrast agen t.

Methods

or Background: We retrospectively enrolled all patients with early - stage CC who received conization before MRI and the n surgically treated (hysterectomy, trachelectomy, or re-conization). Tw o radiologists evaluated MRI scans for residual disease in the cervix, blind ed to surgical outcomes.

Results

or Findings: 119 patients were included in the study. MRI showed an accuracy of 78,18%, sensitivity of 65.45%, specific ity of 90.91%, positive predictive value (PPV) of 87.80%, and negative pred ictive value (NPV) of 72.46%. There was no significant change in the MRI performance with and without contrast medium, in accuracy (81,05% vs 75, 80%, respectively), sensitivity (76,92% vs 55,17%), specificity (85,19% vs 96,43%), PPV (83,33% vs 94,12%), and NPV (79,31% vs 67,50%).

Conclusion

MRI showed good accuracy in assessing residual tumo r after conization with high specificity and PPV; however, the main problem remains the high number of false negatives. The use of cont rast medium did not significantly affect the MRI's performance, as it l ed to a slight increase in true negatives but also an increase in false positives.

Limitations

Retrospective study Funding for this study: None Ethics committee - additional information: ID prot. 7073 prot. 0022234/24 Author Disclosures: Andrea Amerighi: Nothing to disclose Giacomo Avesani: Nothing to disclose Miriam Dolciami: Nothing to disclose Ina Isufi: Nothing to disclose Anna Napoletano: Nothing to disclose Benedetta Gui: Nothing to disclose Evis Sala: Nothing to disclose Diagnostic performance of DWI and ADC in characteri zing the N parameter classified according to the Node-RADS sco re in patients with cervical cancer *M. Gennarini*, R. V. Ninkova, V. Miceli, A. Cupert ino, S. Riccardi, F. Curti, C. Cutonilli, C. Catalano, L. Manganaro; Rome/IT ([email protected]) Purpose or Learning Objective: This study evaluates the diagnostic accuracy of the Node-RADS score in Magnetic Resonance Imagin g (MRI) and assesses the significance of the Apparent Diffusion Coeffici ent (ADC) in identifying actual lymph node neoplastic involvement in patients with cervical cancer (CC).

Methods

or Background: A retrospective analysis was performed on cervical cancer patients who underwent preoperative MRI and radical surgery with lymphadenectomy from February 2018 to July 2024. Ly mph node involvement risk was evaluated for the primary pelvic lymph nod es, assigning scores ranging from 1 to 5: 1 (very low), 2 (low), 3 (uncl ear), 4 (high), and 5 (very high). The mean ADC, relative ADC (rADC), and corre cted ADC (cADC) for lymph nodes rated as Node-RADS 3, 4, and 5 were mea sured and compared against lymph node histology results.

Results

or Findings: In this study, 156 lymph nodes from 54 patients, wi th a Node-RADS score greater than 2, were included. Of t hese, 108/156 (69.2%) were histologically confirmed as positive, while 48 /156 (30.8%) were negative. The mean ADC value proved most statistically signif icant, showing a sensitivity of 87.0%, specificity of 82.6%, a positive predicti ve value (PPV) of 92.2%, and a negative predictive value (NPV) of 73.1%. The are a under the curve (AUC) values for Node-RADS >2 and Node-RADS 3 were 0.892 and 0.677, respectively, with ADC thresholds of <0.963×10^-3 m m^2/s and <0.983×10^-3 mm^2/s.

Conclusion

The ADC measurement of lymph nodes provides crucial data that aids in the accurate classification of patients wit h cervical cancer. Utilizing an ADC cut-off of 0.963×10^-3 mm^2/s, the MRI demonstr ated high diagnostic sensitivity. Thus, ADC serves as a valuable tool fo r enhancing the diagnostic process in cervical cancer management.

Limitations

Single center retrospective design study. Funding for this study: None Ethics committee - additional information: N/A Author Disclosures: Marco Gennarini: Nothing to disclose Claudia Cutonilli: Nothing to disclose Valentina Miceli: Nothing to disclose Federica Curti: Nothing to disclose Sandrine Riccardi: Nothing to disclose Angelica Cupertino: Nothing to disclose Roberta Valerieva Ninkova: Nothing to disclose Carlo Catalano: Nothing to disclose Lucia Manganaro: Nothing to disclose Assessing the feasibility of Magnetic Resonance Ima ging Compilation for determining the treatment strategies and predicting the recurrence risk factors and short-term efficacy in cervical cancer *X. Ou*, Y. Li, Y. Pei, W. Li; Hunan/CN ([email protected]) Purpose or Learning Objective: To investigate the feasibility of MAGiC (sy- T2WI; sy-T1, sy-T2, and sy-PD maps) to determine tr eatment plan and predict recurrence risk factors (RRF) ，short-term treatment efficacy (STE) in cervical cancer (CC) patients using high-resolution T2-weigh ted (hr-T2WI）and diffusion-weighted imaging (DWI) as reference stand ards.

Methods

or Background: 194 patients suspected of CC were enrolled. For CC underwent CCRT, hrT2WI performed in 2 months for evaluating STE (completed response and no-CR ). MAGiC can generate synthetic morphologic images (syT2WI) and quantitative synthetic images ( sy-T1, T2 and PD maps). For syT2WI, The evaluation of image quality and sta ging using sy-T2WI and hr-T2WI was conducted. The accuracy, sensitivity an d specificity of sy-T2WI were analyzed for making treatment strategies (IB-I IA: surgery; IIB-IVA: CCRT). The AUC was used to predict RRF and STE, use quantitative sy-T1, T2, PD maps.

Results

or Findings: 69 out of 119 CC(IIB－IVA) received CCRT . 50 out of 119 CC(IA-IIA) received surgery. SyT2WI was no sign ificant differences with hrT2WI in four aspects (P＞0.05). The accuracy, sensitivity and specificity of sy-T2WI was 0.908, 0.908 ,0.999 for differentiating IB-IIA from IIB-IVA, and an excellent agreement between them (k = 0.935; p < 0. 001). T2, T1 and ADC values had a significant differences to identified CR from no-CR and identified RRF from no-RRF (P < 0.05) but PD.The diagnostic pe rformance of ADC was inferior to T2 for STE, which was similar to T2 for RRF. Furthermore, T1 ＋T2 was superior to ADC for predicting RRF (AUC: 0.980 vs. 0.776; p = 0.005) and forecasting STE (AUC: 0.982 vs. 0.737; p < 0.001) .

Conclusion

MAGiC is a promising technique for deciding therape utic planning and predicting RRF and STE in CC , which is similar and even superior to hr- T2WI and DWI.

Limitations

Not applicable Funding for this study: Not applicable Ethics committee - additional information: the local Research Ethics Committee Author Disclosures: Wenzheng Li: Nothing to disclose Yigang Pei: Nothing to disclose Yue Li: Nothing to disclose Xiaorong Ou: Nothing to disclose A retrospective comparative study of diagnostic alg orithms for the differential diagnosis of uterine leiomyoma and sar coma: preliminary

Results

*C. Vercelli*¹, F. Rosa², C. Martinetti¹, D. Schett ini¹, A. Gastaldo², N. Gandolfo¹; ¹Genova/IT, ²Savona/IT ([email protected]) Purpose or Learning Objective: In 2022, an important Consensus Statement and diagnostic algorithm for assessing the risk of uterine leiomyosarcoma (LMS) using MRI was published (Hindman et al.). The aim of our study is to compare the diagnostic performance of the recently proposed algorithms available in the literature for the differential di agnosis of uterine mesenchymal tumors on MRI.

Methods

or Background: After a literature review, three diagnostic algorit hms were identified: Hindman et al. (2022), Rosa et al. (2023), and Wahab et al. (2020). A retrospective evaluation was conducted on 50 MRIs performed for suspected uterine mesenchymal masses. Each lesion w as categorized in a Sunday Abstract-based Programme 284 double-blind manner by an expert Radiologist in gyn ecological imaging according to the three algorithms. Histological exa mination or appropriate follow-up was used as reference standard for the fi nal diagnosis.

Results

or Findings: The study included 52 uterine masses: 37 benign lesions (leiomyomas) and 15 malignant lesions (7 LM S, 2 STUMP, 2 endometrial stromal sarcomas, 1 adenosarcoma, and 3 non-sarcomatous malignant lesions). The three algorithms demonstrat ed equal specificity in diagnosing LMS (97.3%). However, when analyzing the performance in the differential diagnosis between malignant and benign lesions (including rarer histotypes), the Rosa et al. algorithm showed super ior sensitivity (93.33% vs 73.33%) and diagnostic accuracy (96.15% vs 90.38%).

Conclusion

The differential diagnosis of uterine mesenchymal l esions presents a diagnostic challenge with significant im plications for management and outcome. The three algorithms demonstrated high diagnostic accuracy; in particular, the Rosa et al. algorithm proved more e ffective in identifying also rarer malignant histotypes

Limitations

Retrospective study; Small cohort with high prevale nce of sarcomas. Funding for this study: No funds Ethics committee - additional information: N. Registro CER Liguria: 78/2023 - DB id 12883 N. CET - Liguria: 104/2024 - DB id 13756 Author Disclosures: Nicoletta Gandolfo: Nothing to disclose Daria Schettini: Nothing to disclose Carola Martinetti: Nothing to disclose Alessandro Gastaldo: Nothing to disclose Francesca Rosa: Nothing to disclose Caterina Vercelli: Nothing to disclose Clues for a new MR scoring system of uterine mesenc hymal tumors E. Zlotykamien Taieb, *D. Gherman*; Vincennes/FR ([email protected]) Purpose or Learning Objective: To externally validate a previous MRI-based expert consensus algorithm and evaluate the potenti al improvement of an MR- scoring system's accuracy in diagnosing uterine mes enchymal tumors.

Methods

or Background: A bicentric retrospective observational cohort stud y was conducted from January 2018 to December 2023 in cluding women with a pathological diagnosis of myometrial tumor followin g a pelvic MRI within six months. Clinical and MR criteria were recorded blin dly by two radiologists. Continuous variables were analyzed using a Mann–Whi tney test, and categorical variables using Fisher’s exact test. Od ds ratios for predicting malignancy were calculated with 95% confidence inte rvals and p-values.

Results

or Findings: The cohort included 455 women with mesenchymal tumors: 437 leiomyomas, 2 STUMPs, and 16 leiomyosar comas. Using initial criteria (pelvic hypertrophic lymph nodes, T2W sign al intensity, DW signal intensity compared to endometrium, and ADC cutoff v alue of 0.9 × 10⁻³ mm²/sec), the model accurately classified 420 out o f 455 cases (Accuracy: 80.9%, sensitivity was 61.1% and specificity 93.6%. A refined approach added “irregular tumor margins” and menopausal status, mo dified DW signal compared to bladder, and an ADC cutoff value of 1.2 3 × 10⁻³ mm²/s, improving classification to 445 out of 455 cases (Accuracy: 9 2.5%; sensitivity: 83.3% ; specificity: 98.4%. The refined algorithm significa ntly improved accuracy, allowing the development of a 5-category scoring sy stem.

Conclusion

MR imaging effectively differentiates leiomyosarcom a from other uterine tumors. The new algorithm increases diagnos tic accuracy, helping prevent morcellation risks in women with uterine le iomyosarcoma

Limitations

Our study is retrospective, which does not permit a voidance of all biases. Even though the cohort was bicentric, the p revalence of malignancy remained low. Due to the sample size, differentiati on between STUMP and frankly invasive UMT was not possible. No external validation of our modified score is yet available, necessitating further studi es. Funding for this study: None Ethics committee - additional information: Institutional ethics committee approval and waiver of informed consent (CRM-2405-4 10) Author Disclosures: Diana Gherman: Nothing to disclose Eva Zlotykamien Taieb: Author: conceiving and writi ng this study Magnetic resonance spectroscopy integration with mu ltiparametric MRI: Enhancing diagnostic precision in sonographically i ndeterminate adnexal masses *D. Garg*, R. Kaur, R. Bedi, R. Gupta, B. Goel, U. Handa; Chandigarh/IN ([email protected]) Purpose or Learning Objective: Proton magnetic resonance spectroscopy (¹H-MRS) is a non-invasive imaging technique that o ffers insights into biochemical metabolism. While its role in brain and prostate malignancies is well-established, its application in evaluating adn exal masses is still in its early stages. This study investigates the role of ¹H-MRS in characterising adnexal masses and its value, in conjunction with dynamic c ontrast-enhanced MRI (DCE-MRI) and diffusion-weighted imaging (DWI), in enhancing the diagnostic accuracy of conventional MRI for differentiating ad nexal masses.

Methods

or Background: We conducted a prospective study including 62 histologically confirmed adnexal masses (19 benign and 43 malignant), which were indeterminate on ultrasound. Patients underwen t conventional MRI, DWI (apparent diffusion coefficient), DCE-MRI (time int ensity curves), and ¹H-MRS. Single-voxel spectroscopy analysed resonance peaks for choline, N-acetyl aspartate (NAA), creatine, lactate, and lipids. Cho line-to-creatine ratios were compared between benign and malignant tumours, and ROC curves were used to define optimal thresholds.

Results

or Findings: Conventional MRI showed sensitivity, specificity, p ositive predictive value (PPV), and negative predictive val ue (NPV) of 97.67%, 57.89%, 84%, and 91.67%, respectively. We detected choline peak in 100% of malignant and 47.4% of benign masses, NAA in 79.1% of malignant and 31.6% of benign masses, as well as lipid peaks in 3 6.8% of benign and 20.9% of malignant masses. The mean choline-to-creatine r atio is 1.05+0.55 in benign and 12.18+12.38 in malignant tumours, statistically significant (p<0.05). With a choline-to-creatine threshold of 2.01, sensitivity, specificity of 100% is achieved. The addition of ¹H-MRS, DWI, and DCE-MRI improved the diagnostic accuracy of conventional MRI, with 100% sensitivity, 94.74% specificity, 97.73% PPV, and 100% NPV.

Conclusion

1H-MRS has promising role in characterising adnexal masses and in conjunction with DWI and DCE-MRI, enhances t he diagnostic accuracy of conventional MRI.

Limitations

The sample size is limited. Funding for this study: No funding was provided for this study. Ethics committee - additional information: The ethics committee notification can be found under the number ECR/658/Inst/PB/2014/ RR-20 Author Disclosures: Uma Handa: Nothing to disclose Rekha Gupta: Nothing to disclose Raveena Bedi: Nothing to disclose Dollphy Garg: Nothing to disclose Ravinder Kaur: Nothing to disclose Bharti Goel: Nothing to disclose 11:30-12:30 Research Stage 1 Research Presentation Session: Abdominal and Gastrointestinal RPS 2401 Acute abdominal diseases and imaging of the bowel Moderator M. Zins; Paris/FR ([email protected]) 0.5-mSv ultra-low dose appendiceal CT using deep le arning–based denoising algorithm: a comparison with conventional 2.0-mSv low dose CT *B. J. Choi*, W. Chang, J. H. Hwang, J. Cho, Y. J. Lee, Y. H. Kim, S. H. Park, J. Y. Choi; Seongnam-si/KR Purpose or Learning Objective: To demonstrate that 0.5-mSv ultra-low dose CT using deep learning–based denoising algorithm (D LA) has non-inferiority in diagnosing acute appendicitis, compared to conventi onal 2.0-mSv low dose CT.

Methods

or Background: We used 2.0-mSv CT images of 30 patients with suspected appendicitis from the prior prospective s tudy. The original 2.0-mSv CT were reconstructed using iterative model reconst ruction (IMR). We simulated 0.5-mSv CT images from the original 2.0-m Sv CT. Then we applied IMR and DLA, resulting in three CT image groups per patient (IMR 2.0-mSv, IMR 0.5-mSv, and DLA 0.5-mSv groups). Six radiologi sts (three abdominal and three non-abdominal radiologists) rated the likelih ood of appendicitis on a five- point Likert scale. Primary end point was compariso n of the pooled area under the receiver operating characteristic curve (AUC) b etween DLA 0.5-mSv CT and IMR 2.0-mSv CT, with a non-inferiority margin o f 0.06. Secondary end points included comparison of AUC between DLA 0.5-m Sv CT and IMR 0.5- mSv CT, diagnostic sensitivity/specificity.

Results

or Findings: The AUC of DLA 0.5-mSv CT was non-inferior to that of IMR 2.0-mSv CT [AUC difference: 0.003 (95% CI: -0.0 11, 0.017)]. The AUC of Sunday Abstract-based Programme 285 DLA 0.5-mSv CT was slightly higher among the non-ab dominal radiologists compared to IMR 0.5-mSv CT [AUC difference: 0.034 ( 95% CI: -0.130, 0.198)]. Diagnostic sensitivity/specificity were 100% (9/9)/ 95% (20/21) for all readers with both DLA 0.5-mSv CT and IMR 2.0-mSv CT. Howeve r, the sensitivities of two non-abdominal radiologists with IMR 0.5-mSv CT were mildly compromised [78% (7/9) and 89% (8/9), respectively] .

Conclusion

0.5-mSv ultra-low dose CT using DLA was non-inferio r to conventional 2.0-mSv low dose CT using IMR in diagn osing acute appendicitis.

Limitations

The limitations of the study are 1) 0.5-mSv CT imag es were simulations, not real data and 2) small sample size , collected from a single tertiary hospital. Funding for this study: Funding was provided by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (NRF-2022R1F1A1072570). Ethics committee - additional information: The institutional review board approved this study, and the requirement for inform ed consent was waived (B- 2401-879-112). Author Disclosures: So Hyun Park: Nothing to disclose Jungheum Cho: Nothing to disclose Ji Young Choi: Nothing to disclose Yoon Jin Lee: Nothing to disclose Won Chang: Nothing to disclose Young Hoon Kim: Nothing to disclose Byung Jin Choi: Nothing to disclose Jin Hee Hwang: Nothing to disclose Upright CT vs. Supine CT: Diagnostic Capabilities f or Inguinal Hernias and Subtypes in Emergency Department *T. Yoshikawa*, H. Nagata, T. Ueda, M. Nomura, D. T akenaka, Y. Ozawa, Y. Ohno; Toyoake/JP ([email protected]) Purpose or Learning Objective: Upright CT is recently and clinically set in our institution, and the purpose of this study was to directly compare capabilities for diagnosis and subtype classificati on of inguinal herniation between upright CT (uCT) and conventional supine CT (sCT).

Methods

or Background: 258 consecutive patients who suspected inguinal hernia underwent sCT and uCT within a week, surgica l treatment or follow-up examination. From this cohort, 120 inguinal hernias and computationally selected 120 out of 396 non-inguinal hernia were vi sually assessed by two board certified general and abdominal radiologists by 5-point scales as well as subtypes of hernia. Inter-observer agreements for p robability of hernia and subtype were assessed by kappa statistics with χ2 test. Then, ROC analysis was performed to compare diagnostic performance bet ween two CTs. Then, sensitivity (SE), specificity (SP) and accuracy (AC ) were compared each other by McNemar’s test. Moreover, subtype classification accuracy (SAC) was also compared between uCT and sCT by McNemar’s test.

Results

or Findings: Inter-observer agreement for probability of hernia were determined as significant and almost perfect on uCT (κ=0.84, p<0.0001) and substantial on sCT (κ=0.77, p<0.0001), and that for subtype classificati on were also significant and almost perfect on both CTs (uC T: κ=0.83, p<0.0001; sCT: κ=0.81, p<0.0001). Area under the curve (AUC) of uCT (AUC=0.99) were significantly larger than that of sCT (AUC=0.97, p< 0.05). SE and AC of uCT (SE=92.5%, AC=96.3%) were significantly higher than those of sCT (SE: 80.8%, p<0.0001; AC: 90.4%, p<0.0001). SAC of uCT ( 87.5%) was significantly higher than that of sCT (73.3%, p<0.0 001).

Conclusion

Upright CT has better diagnostic performance for in guinal hernia and subtype classification than conventional supine CT in routine clinical practice.

Limitations

Lack of clinical outcome evaluation Funding for this study: Research grant from Canon Medical Systems Corporation Ethics committee - additional information: Fujita Health University Hospital Author Disclosures: Yoshiyuki Ozawa: Research/Grant Support: Grant-in-A id for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Smoking Rese arch Foundation Masahiko Nomura: Nothing to disclose Takahiro Ueda: Research/Grant Support: Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, S ports, Science and Technology Daisuke Takenaka: Nothing to disclose Hiroyuki Nagata: Research/Grant Support: Grants-in- Aid for Scientific Research from the Japanese Ministry of Education, C ulture, Sports, Science and Technology Research/Grant Support: Canon Medica l Systems Corporation Takeshi Yoshikawa: Nothing to disclose Yoshiharu Ohno: Research/Grant Support: Smoking Res earch Foundation Research/Grant Support: Canon Medical Systems Corpo ration Dual-Energy CT of Gastrointestinal Bleeding - Influ ence on diagnostic accuracy and reader confidence *M. Oberparleiter*, H-C. Breit, J. Vosshenrich, P. Hehenkamp, A. C. Seifert, A. Kobe, C. J. Zech, M. Obmann; Basel/CH Purpose or Learning Objective: Current guidelines suggest replacing unenhanced images with DECT-derived virtual non-con trast images (VNC) in suspected upper GI bleeding based on only two clini cal studies. Our study compares diagnostic accuracy, reader confidence, an d reading time of a conventional triphasic versus a dual-energy CT prot ocol in patients with upper and lower GI bleeding.

Methods

or Background: This retrospective study included 52 patients with active GI bleeding (22 upper, 30 lower) and 52 cont rols who underwent non- contrast, arterial, and portal-venous phase abdomin al CT. For each case, a triphasic conventional CT dataset and a DECT datase t with VNC, iodine images, and arterial and portal venous phase images were created. Two residents and two fellowship-trained abdominal radi ologists evaluated all cases for active GI bleeding. Radiation dose and reading time were recorded. Diagnostic confidence was rated on a 5-point Likert scale. Inter-reader agreement was assessed using Fleiss' kappa. Sensiti vity and specificity were compared using McNemar's test, reading time, and re ader confidence with the Wilcoxon signed-rank test.

Results

or Findings: Inter-reader agreement was substantial ( =0.80). Sensitivity and specificity for detecting GI bleedi ng using conventional CT did not differ from DECT (91% and 95%, vs. 93% and 96%, p=0.30 and p=0.77, respectively). Subgroup analysis of lower GI bleedi ng showed a sensitivity of 88% in conventional CT versus 93% in DECT (p=0.18). Diagnostic confidence increased from 4(IQR, 4-5) to 5(IQR, 4-5) when usin g DECT (p<0.01). Mean reading time per case was 102 s for both datasets ( p=0.62). Total DLP without true unenhanced images was 21% lower.

Conclusion

DECT-derived VNC and iodine images can replace true non- contrast images when searching for GI-bleeding. Gui delines should be extended to include lower-GI-bleeding.

Limitations

Sample size was moderate, the study had a single-ce nter design, and only dual-source and split-beam DECT scanners w ere used. Funding for this study: This research received no specific grant from any funding agency in the public, commercial, or not-fo r-profit sectors. Ethics committee - additional information: The need for informed consent was waived due to the retrospective nature of this study. Author Disclosures: Markus Obmann: Nothing to disclose Hanns-Christian Breit: Nothing to disclose Christoph Johannes Zech: Nothing to disclose Adrian Kobe: Nothing to disclose Paul Hehenkamp: Nothing to disclose Jan Vosshenrich: Nothing to disclose Alina Carolin Seifert: Nothing to disclose Moritz Oberparleiter: Nothing to disclose Correlation of CT-derived Quantitative Image Featur es and Inflammatory Laboratory in Pyelonephritis *A. W. Marka*, M. Graf, S. Ziegelmayer, M. R. Makow ski, A. Sauter, T. Huber; Munich/DE ([email protected]) Purpose or Learning Objective: To investigate the relationship between inflammatory laboratory markers and quantitative CT -derived imaging features in patients with acute pyelonephritis (APN).

Methods

or Background: In this single-center retrospective study, we evaluated patients with clinical symptoms of APN at our institution from December 2018 to April 2024. Inclusion criteria com prised APN symptoms, elevated inflammatory markers (CRP and/or WBC), and CT morphologic signs of APN. Exclusion criteria included concomitant acu te pathology and poor image quality. A total of 102 patients (mean age 60 .1±18.8) were initially identified; 14 were excluded due to acute pathology and 5 due to poor image quality, leaving 83 for final analysis. CT scans fo llowed a standardized protocol, and two radiologists blinded to clinical and lab data conducted image analysis. Inflammatory markers were collected on th e scan day. Statistical analyses included Spearman correlation, Mann-Whitne y-U tests, and linear regression.

Results

or Findings: Spearman correlation analysis revealed strong posit ive correlations between CRP levels and both total volu me (r=0.76, p<0.001) and percentage (r=0.71, p<0.001) of renal perfusion def icit. Multivariate linear regression showed total perfusion deficit volume ex plained 54.5% of CRP variability (p<0.001). WBC count also correlated si gnificantly with total volume (r=0.379, p<0.001) and percentage (r=0.374, p<0.001 ) of perfusion deficit. Procalcitonin levels moderately correlated with fat stranding area (r=0.482, p=0.0014) but not other CT features. Locoregional l ymphadenopathy was significantly associated with elevated CRP and WBC counts, but not procalcitonin levels. Sunday Abstract-based Programme 286

Conclusion

Quantitative CT-derived features, particularly rena l perfusion deficits, are significantly associated with inflamm atory markers in APN. These findings suggest CT imaging can serve as a surrogat e for inflammation severity, potentially guiding clinical management. Further research is needed to explore the clinical implications of these assoc iations.

Limitations

-Small cohort -Scarcity of procalcitonin levels -No correlation with duration of a patient's hospitalization Funding for this study: None Ethics committee - additional information: Data collection, processing, and analysis were approved by the institutional review board (protocol number 180/17S), and informed consent was waived. Author Disclosures: Andreas Sauter: Nothing to disclose Markus Graf: Nothing to disclose Thomas Huber: Nothing to disclose Marcus R. Makowski: Nothing to disclose Alexander Wolfgang Marka: Nothing to disclose Sebastian Ziegelmayer: Nothing to disclose Fast abdominopelvic T2-weighted imaging with deep l earning reconstruction for acute abdomen: feasibility study J. Xu, L. Zhu, W. Liu, *Y. Lu*, J. Liu, C. Ma, Y. Z hang, X. Wang, F. Feng; Beijing/CN ([email protected]) Purpose or Learning Objective: To evaluate the image quality and diagnostic performance of single-shot fast spin-echo T2 weight ed imaging with deep learning reconstruction (SSFSE-DL) in volunteers an d patients with acute abdomen, in comparison to SSFSE without deep learni ng reconstruction (SSFSE-nonDL) and conventional PROPELLER sequences.

Methods

or Background: Thirty-five healthy volunteers, as well as 35 patie nts with acute abdominal pain from emergency room were prospectively enrolled. Abdominopelvic MRI at 3T was performed using three T2-weighted imaging sequences: SSFSE-DL (acquisition time: 34s), SSFSE- nonDL, and conventional PROPELLER (acquisition time: 2-3min), in random order. Two blinded radiologists independently evaluated the ov erall image quality, noise, motion artifacts and clarity of major abdominopelvi c organs. Diagnostic confidence for the presence or absence of common ab dominopelvic diseases was rated on a 1-5 Likert scale. Signal-to-noise ra tio (SNR), contrast-to-noise ratio (CNR) as well as image noise for the liver, p ancreas and spleen were also quantified. Intra- and inter- observer agreement we re assessed, and comparisons of image quality and diagnostic perform ance between the three sequences were made.

Results

or Findings: Intra- and inter- observer agreement for the qualit ative analysis and diagnostic performance were good to ex cellent (0.776-0.967). SSFSE-DL yielded significant higher SNR and CNR, an d lower noise than SSFSE-nonDL and PROPELLER in both volunteers and pa tients (all P<0.05). SSFSE-DL obtained significantly higher image qualit y and lower noise than SSFSE-nonDL and PROPELLER (both P<0.05). SSFSE-DL a nd SSFSE- nonDL had significantly lower motion artifacts and better clarity of major abdominopelvic organs than PROPELLER (both P<0.05). The AUC for detecting common abdominopelvic diseases in SSFSE-D L (0.977-1) and SSFSE-nonDL (0.887-1) were significantly higher tha n PROPELLER (0.585- 0.953).

Conclusion

SSFSE-DL achieved superior image quality and diagno sis performance for volunteers and patients with acute abdomen.

Limitations

The number of patients with positive diagnosis for each specific disease was relatively small. Funding for this study: This study was funded by the National Natural Science Foundation of China (82371950) Ethics committee - additional information: This prospective single-center study was approved by the local institutional revie w board, and informed consent was obtained from all participants prior to inclusion in the study. Author Disclosures: Wei Liu: Nothing to disclose Feng Feng: Nothing to disclose Yitong Lu: Nothing to disclose Chenxue Ma: Nothing to disclose Xuan Wang: Nothing to disclose Jia Xu: Nothing to disclose Liang Zhu: Nothing to disclose Jingjuan Liu: Nothing to disclose Yifei Zhang: Nothing to disclose Diagnostic performance of low-dose abdominal CT wit h artificial intelligence iterative reconstruction for acute pan creatitis *X. Zhang*¹, S. Zhong², G. Zhang², X. Zhou¹; ¹Chong qing/CN, ²Shanghai/CN ([email protected]) Purpose or Learning Objective: To characterize the diagnostic performance of low-dose (LD) abdominal CT combined with artific ial intelligence iterative reconstruction (AIIR) for assessing acute pancreati tis based on CT severity index (CTSI).

Methods

or Background: A total of 30 patients with acute pancreatitis who underwent follow-up CT examination were prospective ly enrolled. All patients underwent standard-dose (SD) CT followed by LD-CT i n the same breath hold, where an immediate LD-scan was added in the portal venous phase. The SD- protocol was 120 kVp, ref. 141 mAs, and hybrid iter ative reconstruction (HIR), whereas the LD-protocol was 120kVp, ref. 50mAs and AIIR reconstruction. To obtain the CTSI, SD- and LD-CT images at portal ven ous phase were independently scored by two radiologists for assess ing pancreatic inflammation, necrosis, and extrapancreatic complic ations. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of he althy and inflamed parenchyma were measured and calculated.

Results

or Findings: Compared to SD-CT, LD-CT examination achieved a 65.18% reduction in effective radiation dose for th e portal venous phase (6.06 ± 1.28 mSv vs. 2.11 ± 0.45 mSv, p < 0.05). Based on CTSI scoring, LD-AIIR was found comparable to SD-HIR in evaluating the se verity of acute pancreatitis (5.02 ± 1.42 vs. 5.11 ± 1.78, p = 0.86). Inter-observer agreement for assessing the severity of acute pancreatitis wa s excellent (k = 0.89). LD- AIIR showed superior conspicuity compared to SD-HIR for both the inflamed (SNR: 1.96 ± 0.83 vs. 1.74 ± 0.98; CNR: 4.17 ± 1.58 vs. 2.11 ± 0.87; both p < 0.05) and the healthy parenchyma (SNR: 8.03 ± 1.71 vs. 4.91 ± 1.04; CNR: 3.06 ± 1.84 vs. 1.78 ± 1.03; both p < 0.05).

Conclusion

AIIR allows for significant radiation dose reductio n without compromising image quality or diagnostic performanc e for the evaluation of acute pancreatitis.

Limitations

n/a Funding for this study: n/a Ethics committee - additional information: Ethics committee of Chongqing University Jiangjin Hospital Author Disclosures: Xue Zhou: Nothing to disclose Xiufu Zhang: Nothing to disclose Guozhi Zhang: Nothing to disclose Sihua Zhong: Nothing to disclose SBOM-AI TRIAL: Set up and validation of AI-based au tomatic total Small Bowel length Measurement using CT and MRI in Obese patients candidates for metabolic surgery *M. Zerunian*¹, S. Nardacci¹, N. Petrucciani¹, I. T oniolo², D. De Santis¹, D. Caruso¹, C. G. Fontanella², G. Silecchia¹, A. La ghi¹; ¹Rome/IT, ²Padova/IT ([email protected]) Purpose or Learning Objective: Total small bowel length(TSBL) is crucial to achieve successful metabolic/bariatric surgery. A n on-invasive measurement of the TSBL will impact on surgical strategy to avoid short-bowel syndrome after surgery. Cross-sectional imaging(CSI) might play an important role by measuring TSBL non-invasively. We aimed to set up a reliable AI-based automatic method using preoperative CSI to measure the TSBL in candidate to bariatric/metabolic surgery.

Methods

or Background: This multicentre prospective TRIAL included patients eligible for bariatric surgery(BMI >35 kg/ m2 and at least one obesity- related comorbidity,BMI>40 kg/m2) underwent the sam e day MRI-and CT- enterography before bariatric surgery. TSLB assesse d right before the surgery at the operation table and, TSLB <250 cm considered the cut-off as risk for developing short-bowel syndrome. TSLB obtained on M RI and CT by manual segmentation(Slicer3D). A Convolutional Neural Netw ork with U-NET developed, consisting in a contracting path followe d by an expansive path, RELU activation and a softmax activation function t o reduce the feature map. The Adam optimization algorithm with a constant lea rning rate was used for the learning process. Training stopped after 100 epochs . DICE coefficient were calculated to quantify the accuracy of the predicti on compared to annotated-by- radiologist images.

Results

or Findings: Fifty patients enrolled (27 female,age range 26-54 years old,mean BMI 41.47). Patients underwent surgery sho wed a TSBL mean measured intraoperatively of 652.85±87.37cm. CSI showed good concordance with TSBL mean of 589.08±82.95cm(k= 0.70).DICE coef ficient of the training set showed a DICE score ranging between 35%-45%, co nfirmed in the validation set. All methods correctly categorized t he patients according to the cut-off considered as risk factor to develop short- bowel syndrome.

Conclusion

Automatic AI-based segmentation of small bowel on n on-invasive CSI might be a useful tool to assess obese patients to personalize the treatment and reduce complications.

Limitations

Small sample size Sunday Abstract-based Programme 287 Funding for this study: Italian Ministry of University and Research (MUR) Research Projects of Significant National Interest – PRIN (ID: MUR 2022MPAE29_003) Ethics committee - additional information: Multicenter interventional study Author Disclosures: Niccolò Petrucciani: Nothing to disclose Damiano Caruso: Nothing to disclose Marta Zerunian: Nothing to disclose Domenico De Santis: Nothing to disclose Ilaria Toniolo: Nothing to disclose Gianfranco Silecchia: Nothing to disclose Andrea Laghi: Nothing to disclose Chiara Giulia Fontanella: Nothing to disclose Stefano Nardacci: Nothing to disclose Vascular enhancement in single-pass abdominal CT: E ffects of a fixed injection duration in patients with non-traumatic a cute abdomen *A. Stanzione*, V. Arpaia, A. E. Antonini, R. Liuzz i, L. Sommella, L. Mannacio, A. Brunetti, L. Camera; Naples/IT ([email protected]) Purpose or Learning Objective: To evaluate the effects of a fixed injection duration (FID) on vascular enhancement in a Single- Pass (SP) abdominal CT performed in patients with nontraumatic acute abdom en (ANTA).

Methods

or Background: 100 patients (58M, 42F; aged 52±20 yrs ) with ANTA underwent a SP contrast-enhanced CT (Somatom D rive, Siemens) performed using a Single Source at either 80 kVp (G roup A; BMI 19±3), 100 Kvp (Group B; BMI 25±4) or 120 Kvp (Group C; BMI 30 ±2.5) . In all groups a non-ionic iodinated contrast media (370 mgI/ml) was administered as follows: Group A (0.37 grI/Kg); Group B (0.52 grI/Kg); Group C (0.63 grI/Kg). All patients underwent a SP protocol with a FID (50 sec ) and a tailored scan delay (SD). In all patients Signal- (SNR) and Contrast-to -Noise Ratios (CNR) were calculated for the abdominal aorta (AA) and the mai n portal vein (MPV) using the psoas muscles as reference tissue. Statistical analysis was performed with ANOVA (p < 0.05).

Results

or Findings: No significant differences were observed in the demographics of either Group A (20M/12F; 47±20 yrs) or B (36M/22F; 54±19 yrs) whereas a female preponderance was observed in Group C (2M/8F; 50±18 yrs). Despite significant differences (p < 0.001) were observed in both the volumes (53±11 vs 103±17 vs 141±19 ml) as well as the injection rates (1.1±0.2 vs 2.1±0.4 vs 2.8±0.4 ml/sec) of the contrast media for Group A, B and C, respectively, SNR and CNR were not significa ntly different for both AA and MPV.

Conclusion

SP performed with a FID results in a consistent vas cular enhancement.

Limitations

Unbalanced sample size Funding for this study: None Ethics committee - additional information: Local IRB approval Author Disclosures: Luigi Mannacio: Nothing to disclose Laura Sommella: Nothing to disclose Valerio Arpaia: Nothing to disclose Luigi Camera: Nothing to disclose Raffaele Liuzzi: Nothing to disclose Andrea Ennio Antonini: Nothing to disclose Arnaldo Stanzione: Nothing to disclose Arturo Brunetti: Nothing to disclose 11:30-12:30 Research Stage 2 Research Presentation Session: Cardiac RPS 2403 Cardiac imaging: insights from trials Moderator N. Lama; Athens/GR ([email protected]) Association between metabolic phenotypes and MRI-de rived cardiac function parameters, major adverse cardiovascular e vents, and mortality: Insights from the UK Biobank *B. Bogner*, M. Jung, M. Reisert, J. Maushagen, S. Rospleszcz, C. L. Schlett, F. Bamberg, J. Weiß, J. Taron; Freiburg/DE Purpose or Learning Objective: To evaluate the association between metabolic phenotypes, MRI-derived cardiac function, major adverse cardiovascular events (MACE), and all-cause mortali ty.

Methods

or Background: We analyzed 22,348 UK Biobank (UKBB) participants who underwent cardiac MRI. Obesity was defined as BMI ≥30kg/m²; metabolically unhealthy as the presence of >=1 metabolic syndrome components or diabetes resulting in four phenotypes : metabolically healthy non-obese (MHN), metabolically healthy obese (MHO), metabolically unhealthy non-obese (MUN), metabolically unhealthy obese (MUO ). Associations between MRI-derived left ventricular ejection fract ion (LVEF, %), cardiac index (cardiac output/body surface area [L/min/m2]), MACE , and all-cause mortality were investigated using uni-/multivariable linear a nd Cox regression analyses adjusted for age, sex, and smoking status.

Results

or Findings: Among 22,348 UKBB participants (mean age 64.1±7.5 years, 48.8% male), 45.4% were MHN, 6.0% MHO, 36.5% MUN and 12.1% MUO. Over a median follow-up of 5.2 years, 52 MACE events and 371 deaths occurred. Compared to MHN, both obese phenotypes sh owed a significant negative association with LVEF whereas a positive a ssociation was found for MUN (p5-fold higher risk of MACE for both obese phenotypes (MHO (HR 5.53 [95% CI 2.10- 14.58], p<0.001; MUO (HR 5.24 [95% CI 2.37-11.57], p<0.001) compared to MHN. A similar pattern was seen for all-cause morta lity (MHO (1.57 [1.04, 2.36], p=0.03; MUO (1.61 [1.20, 2.17]; p=0.001).

Conclusion

Alterations in metabolic health are linked to diffe rences in cardiac function, MACE and all-cause mortality risk indepen dent of age, sex, and smoking. This suggests distinct mechanisms affectin g cardiac health and emphasizes the need to consider both metabolic heal th and obesity for personalized risk assessment.

Limitations

Findings may not be generalizable to non-UK populat ions or ethnic groups underrepresented in the UKBB. Funding for this study: None Ethics committee - additional information: Approved by the UK Biobank. Author Disclosures: Susanne Rospleszcz: Nothing to disclose Christopher L. Schlett: Nothing to disclose Marco Reisert: Nothing to disclose Jakob Weiß: Nothing to disclose Balazs Bogner: Nothing to disclose Matthias Jung: Nothing to disclose Jana Taron: Nothing to disclose Juliane Maushagen: Nothing to disclose Fabian Bamberg: Nothing to disclose The aging heart: Associations between cardiac struc ture, function, and demographic factors in a population-based study *B. J. Kerber*¹, T. Küstner², S. Gatidis³; ¹Zurich/ CH, ²Tübingen/DE, ³Stanford, CA/US Purpose or Learning Objective: Exploring the relationship between demographic factors and quantitative imaging phenot ypes based on cardiac MR (CMR) in the NAKO population study to provide in sights into the aging heart and advance the understanding of cardiovascul ar disease.

Methods

or Background: Steady-state free precession short-axis CMR full- cycle sequences from 29,104 participants of the NAK O study, aged 19 to 74 years (16,201 male and 12,903 female, aged 47.9 +/- 12.4, resp. 48.7 +/- 12.2 years), were analyzed. A custom-trained nnU-Net mod el was used to segment Sunday Abstract-based Programme 288 images into left ventricle (LV), leftventricular my ocardium, and right ventricle (RV). From these segmentations, key cardiac metrics including LV mass (LVM), LV and RV end-diastolic and end-systolic vol umes (LVEDV/RVEDV/LVESV/RVESV), ejection fractions (LVEF /RVEF), stroke volumes (LVSV/RVSV) and cardiac output (LVCO) were calculated. Correlation analysis and group comparisons were performed to ex amine associations between the participants' demographic and quantitat ive imaging phenotypes.

Results

or Findings: In this study, LVEDV/RVEDV (r=-0.22/r=-0.23), LVESV/RVESV (r=-0.17/r=-0.19), LVSV/RVSV, (r=-0.21/ r=-0.20) and LVCO (r=-0.19) significantly decreased with age, with ma les showing a higher baseline and steeper decline. LVM and LVEF were nea rly stable (r=0.01/r=0.01). LVEF was consistently higher in fe males (p<0.001). RVEF increased with age for females and decreased for ma les (r=0.07, r=-0.02). Hypertension, diabetes, smoking and high blood lipi ds were associated with significantly higher LVM and lower LVEF/RVEF (p<0.0 01), while high HDL was negatively correlated with LVM (r=-0.36), LVEDV/RVE DV (r=-0.20/r=-0.22) and positively with LVEF/RVEF (r=0.08/r=0.14).

Conclusion

A CMR full-cycle segmentation model was developed a nd applied to a large population study spanning a broa d age range. The resulting imaging phenotypes revealed strong associations wit h age, disease, and demographic factors.

Limitations

The analysis was performed automatically using only basic quality control. Funding for this study: No specific funding. Ethics committee - additional information: The study protocol, participant information, and consent forms for the NAKO study w ere reviewed by all local ethics committees of the participating institutions . Author Disclosures: Sergios Gatidis: Nothing to disclose Bjarne Jonas Kerber: Nothing to disclose Thomas Küstner: Nothing to disclose Variability of coronary artery calcium score in the multicentre DISCHARGE trial: Agreement between readings at the core laboratory and the clinical centres *F. Biavati*¹, M. Mohamed¹, S. Tsogias¹, B. Föllmer ¹, M. Bosserdt¹, J. Dodd², M. Dewey¹; ¹Berlin/DE, ²Dublin/IE Purpose or Learning Objective: To assess the agreement between core laboratory and the clinical centre measurement of c oronary artery calcium (CAC) score using the Agatston method.

Methods

or Background: At each of the 26 clinical centres across 16 European countries [NCT02400229], radiologists meas ured the patients’ CAC score based on noncontrast coronary computed tomogr aphy (CT). Two readers measured the CAC score at the core laborato ry blinded to the clinical centre reading. Bland-Altman analysis of the CAC sc ores values and Cohen’s kappa of the CAC score risk categories (I: 0, II: 1 -400, III: >400) was performed.

Results

or Findings: 1550 patients (mean age, 59 years ± 10 [SD], 56.3% women) were included. The Bland-Altman analysis sho wed a mean absolute difference of 2.0 and limits of agreement of ± 93.3 between the core laboratory and the clinical centres. There was agreement in th e CAC score risk categories in 96.6% (1498 of 1550) of patients between the cor e laboratory and the clinical centre reading. Discrepancies between CAC score risk categories occurred mostly between categories I and II (88.5%, 46 of 52). Agreement according to Cohen’s kappa was excellent (0.94, 95% CI: 0.93, 0.96; p<.001). Most disagreements in the assignment of risk catego ries were between category I and II (88.5%).

Conclusion

CAC score measurements had good agreement between t he core laboratory and clinical centres in a Pan-Europ ean multicentre trial suggesting that CAC score measurements can be widel y implemented as part of cardiac CT based on its reproducibility.

Limitations

Our study has limitations. It included only stable chest pain patients. The population was from 26 European centr es, limiting global applicability. Some CT scans were excluded due to m issing data or non- matching reconstruction parameters, and only filter ed back projection reconstruction was used. Funding for this study: Funding was provided by grants from the EU-FP7 Framework Program (FP 2007-2013, EC-GA 603266). Ethics committee - additional information: The study was approved by The German Federal Office for Radiation Protection and the local or national authorities at each trial site. The reference numbe r is: EA1/294/13. Author Disclosures: Jonathan Dodd: Author: Receives royalties as a co-a uthor of book chapters in the Stat-Dx book Series Diagnostic Imaging – Cardio vascular and the textbook CT and MRI in Cardiology, Elsevier. Other: Associat e Editor for Radiology, a member of the Editorial Board for Radiology Cardiot horacic Imaging, and an Associate Editor for the Quarterly Journal of Medic ine. All non-paid. Grant Recipient: Received funding from EU-FP7 Framework P rogram (DISCHARGE EU FP EC-GA 603266). Mahmoud Mohamed: Nothing to disclose Marc Dewey: Board Member: M.D. is European Society of Radiology (ESR) Publications Chair (2022-2025); the opinions expres sed in this presentation are the author’s own and do not represent the view of E SR. Other: Hands-on cardiac CT courses (www.ct-kurs.de) Author: Cardiac CT (Springer Nature). Grant Recipient: EU (EC-GA 603266 in HEALTH.2013.2. 4.2-2) DFG (DE 1361/14-1, DE 1361/18-1, BIOQIC GRK 2260/1, Radiomi cs DE 1361/19-1 [428222922] and 20-1 [428223139] in SPP 2177/1), GU IDE-IT (DE 1361/24-1), Berlin University Alliance (GC_SC_PC 27), G-BA (01N VF23002), Berlin Institute of Health (Digital Health Accelerator). R esearch/Grant Support: Siemens, General Electric, Philips, Canon. Patent H older: Patent on fractal analysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10,991,109, approved). Sotirios Tsogias: Nothing to disclose Federico Biavati: Nothing to disclose Maria Bosserdt: Nothing to disclose Bernhard Föllmer: Nothing to disclose Patient Acceptance of Coronary CT Angiography Versu s Invasive Coronary Angiography in Patients with Stable Chest Pain M. Bosserdt, *K. Schulze*, M. Mohamed, A-M. Stantie n, M. Dewey, E. Schöneberger; Berlin/DE Purpose or Learning Objective: Patient preference between coronary computed tomography (CT) and invasive coronary angi ography (ICA) in a multicentre analysis in Europe is unknown. Therefor e, we compare patient preference for CT and ICA in a European multicentre randomised controlled trial.

Methods

or Background: A total of 3561 patients with a clinical indication for ICA with stable chest pain and an intermediate like lihood of obstructive coronary artery disease were analysed in this presp ecified secondary analysis from the randomised DISCHARGE trial (NCT02400229) c onducted between October 2015 and April 2019 in 26 European centres. Patient preference using a previously validated questionnaire completed at l east 24 hours after CT or ICA, including preparation for the tests, anxiety, comfort, level of helplessness, pain, willingness to undergo the tests again, overa ll satisfaction and preference.

Results

or Findings: The questionnaire was completed by 89.7% in the CT group (1622/1808) and 89.4% (1567/1753, P=.75) in t he ICA group. Patients reported significantly higher satisfaction with CT (mean (SD): CT: 1.37 (0.53) vs. 1.48 (0.60); score: 1-5; P<.0001), were more wi lling to undergo CT again (mean (SD): CT: 1404/1622 (86.6%) vs. ICA: 1127/156 7 (71.9%); P<.0001), and were better prepared to CT (mean (SD): CT: 1.40 (0.56) vs. 1.49 (0.64); score: 1-5; P=.0005). They were less anxious before CT (mean (SD): CT: 1.07 (1.17) vs. ICA: 1.62 (1.33); score: 1-4; P<.0001), felt less helpless during CT (mean (SD): CT: 0.54 (0.72) vs. ICA: 0.91 (0.90); s core: 0-4; P<.0001), and felt more comfortable (mean(SD): CT: 1.64 [0.66] vs. ICA : 1.90 [0.80]; score: 1-5; P<.0001).

Conclusion

In this multicentre, randomised trial of patients r eferred for ICA with stable chest pain and an intermediate likeliho od of obstructive coronary artery disease, patient preference was in favour of coronary CT angiography.

Limitations

Not applicable. Funding for this study: This study was funded by grants from the EU-FP7 Framework Program (FP 2007-2013, EC-GA 603266). Ethics committee - additional information: The study was approved by ethics committee at Charité (EA1/294/13). Author Disclosures: Kenrick Schulze: Nothing to disclose Mahmoud Mohamed: Nothing to disclose Eva Schöneberger: Nothing to disclose Anne-Marieke Stantien: Nothing to disclose Marc Dewey: Board Member: M.D. is European Society of Radiology (ESR) Publications Chair (2022-2025); the opinions expres sed in this presentation are the author’s own and do not represent the view of E SR. Grant Recipient: EU (EC-GA 603266 in HEALTH.2013.2.4.2-2) DFG (DE 1361/ 14-1, DE 1361/18-1, BIOQIC GRK 2260/1, Radiomics DE 1361/19-1 [42822292 2] and 20-1 [428223139] in SPP 2177/1), GUIDE-IT (DE 1361/24-1) , Berlin University Alliance (GC_SC_PC 27), G-BA (01NVF23002), Berlin I nstitute of Health (Digital Health Accelerator). Author: Cardiac CT (S pringer Nature) Patent Holder: Patent on fractal analysis of perfusion ima ging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10, 991,109, approved) Research/Grant Support: Siemens, General Electric, Philips, Canon. Other: Hands-on cardiac CT courses (www.ct-kurs.de) Instit utional research agreements: Siemens, General Electric, Philips, Can on. Patent on fractal analysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10,991,109, approved) Maria Bosserdt: Nothing to disclose Sunday Abstract-based Programme 289 CAD-Man EXTEND: Long-term clinical results of a sin gle centre randomised controlled trial comparing CT with ICA *A-M. Stantien*¹, F. Biavati¹, A-C. Stahl¹, S. Chim ed¹, M. Mohamed¹, L. M. Serna Higuita², M. Bosserdt¹, M. Dewey¹; ¹Ber lin/DE, ²Tübingen/DE Purpose or Learning Objective: To investigate major adverse cardiovascular events (MACE) after 10 years of follow-up in patien ts with an intermediate probability of coronary artery disease (CAD) underg oing computed tomography (CT) or invasive coronary angiography (ICA).

Methods

or Background: This is the 10-year long-term clinical follow-up of the single centre randomised CAD-Man (Coronary Arte ry Disease Management) trial comparing CT with ICA in patients with atypical angina or chest pain clinically referred for ICA. Clinical fo llow-up was done by interviewing patients using questionnaires asking a bout MACE (myocardial infarction, stroke, unstable angina, (surgical) re- /revascularization, cardiac death) in the past 10 years. Additionally, every pa tient was offered a cardiac CT scan and blood samples were taken. The associati on between randomisation group and MACE was assessed using a m ultivariate Cox proportional hazards model.

Results

or Findings: Out of 329 patients included in CAD-Man 106 patient s completed the clinical long-term follow-up resultin g in a total median follow-up of 4.9 years. Additional 18 MACE cases occurred, 7/ 49 (14%) of which in the CT group and 11/57 (19%) in the ICA group. There wa s no statistically significant difference in MACE between the two rand omisation arms. The HR was 0.86 (95% CI 0.42–1.74) in the CT group.

Conclusion

After 10 years of follow-up, the survival and the o ccurrence of MACE was similar in the CT group compared to the IC A group for patients referred for ICA because of atypical angina or stab le chest pain and an intermediate pretest probability of CAD.

Limitations

The limitations of the study are the single centre design and the low number of MACE, limiting the generalisability o f our findings. Funding for this study: This study was funded by a grant of the Heisenberg programme. Ethics committee - additional information: The study was approved by ethics committee at Charité (EA1/124/23). Author Disclosures: Mahmoud Mohamed: Nothing to disclose Anne-Marieke Stantien: Nothing to disclose Lina Maria Serna Higuita: Nothing to disclose Surenjav Chimed: Nothing to disclose Marc Dewey: Research/Grant Support: Institutional r esearch agreements with Siemens, General Electric, Philips, Canon. Grant Re cipient: EU (EC-GA 603266 in HEALTH.2013.2.4.2-2) DFG (DE 1361/14-1, D E 1361/18-1, BIOQIC GRK 2260/1, Radiomics DE 1361/19-1 [428222922] and 20-1 [428223139] in SPP 2177/1), GUIDE-IT (DE 1361/24-1), Berlin Univer sity Alliance (GC_SC_PC 27), G-BA (01NVF23002), Berlin Institute of Health (Digital Health Accelerator). Patent Holder: Patent on fractal anal ysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773 A1, and USPTO 2021 10,991,109, approved) Author: Cardiac CT (Springer Nature). Other: Hands-on cardiac CT-couses (www.ct-kurs.de) Board Member: Eu ropean Society of Radiology (ESR) Publications Chair (2022-2025) Ann-Christine Stahl: Nothing to disclose Federico Biavati: Nothing to disclose Maria Bosserdt: Nothing to disclose Improved Prediction of Obstructive Coronary Artery Disease by DISCHARGE Trial Pretest Calculator Combined with Ca rdiac CT *M. Mohamed*¹, M. Dewey¹, V. Wieske¹, P. Schlattman n², R. Haase¹, J. Dodd³; ¹Berlin/DE, ²Jena/DE, ³Dublin/IE Purpose or Learning Objective: To evaluate the accuracy of pretest probability (PTP) calculations alone and in combina tion with computed tomography angiography (CTA) results for the diagno sis of CAD in stable chest pain.

Methods

or Background: Individual patient data (IPD) meta-analysis of 65 prospective diagnostic accuracy studies of patients clinically referred to invasive coronary angiography (ICA) with stable che st pain in 22 countries. Three clinical probability models a PTP model based on age, sex and chest pain type (termed the updated DISCHARGE trial PTP c alculator), a CTA alone model and the updated DISCHARGE Trial PTP calculato r and CTA model combined were constructed. The models were built by multivariable logistic regressions with a dataset-specific random intercep t and were compared using the area under the receiver-operating-characteristi c curve (AUC) and the decision curve analysis (DCA).

Results

or Findings: 5332 stable chest pain patients with clinically ind icated ICA were included. The updated DISCHARGE pretest pr obability calculator was more accurate than the original Diamond-Forrest er model (AUC: 0.68, 95% CI: 0.66-0.69 versus 0.63, 95% CI: 0.62-0.65). The combination of DISCHARGE pretest probability calculator with CTA f indings significantly improved accuracy compared with either model alone (AUC: 0.86, 95% CI: 0.85-0.87 versus 0.81, 95% CI: 0.80-0.82). The impr oved prediction of CAD by combining CTA with the updated DISCHARGE Trial PTP calculator prediction model was consistent in the DCA with an increased n et benefit for all chest pain types and was almost equally seen in patients with typical or atypical angina (0.85, 95% CI: 0.84-0.86) and nonanginal or other chest discomfort (0.88, 95% CI: 0.86-0.89).

Conclusion

Combining the updated DISCHARGE trial PTP calculato r with CTA provides more accurate prediction than CTA alon e for the diagnosis of obstructive CAD.

Limitations

The ICA indication resulted in a relatively high CA D prevalence of 48.3%. Funding for this study: The COME-CCT Consortium is funded by a joint program of the German Research Foundation and the G erman Federal Ministry of Education and Research (01KG1110) and t he Digital Health Accelerator of the Berlin Institute of Health to Ma rc Dewey. All researchers are independent of the funding bodies. The funding bodi es had no role in the study design; in the collection, analysis and interpretat ion of data; in the writing of the report; and in the decision to submit the manuscrip t for publication. Ethics committee - additional information: This study was approved by the local research ethics committee of Charité (EA-1-08 0-08) and the German Federal Office for Radiation Protection (Z5-22462/2 -2008-048). All patients gave written informed consent before randomisation. Author Disclosures: Jonathan Dodd: Nothing to disclose Mahmoud Mohamed: Nothing to disclose Peter Schlattmann: Nothing to disclose Viktoria Wieske: Nothing to disclose Marc Dewey: Nothing to disclose Robert Haase: Nothing to disclose Pericoronary adipose tissue volume but not attenuat ion is associated with quantitative coronary plaque metrics on corona ry CT angiography: Insights from the PROMISE trial *M. C. Langenbach*¹, I. Hadzic¹, T. Mayrhofer¹, J. Karady¹, S. Shah², M. T. Lu¹, M. Ferencik¹, P. Douglas², B. Foldyna¹; ¹Boston, MA /US, ²Durham/US ([email protected]) Purpose or Learning Objective: Pericoronary adipose tissue (PCAT) is related to pericoronary inflammation and contribute s to atherogenesis and adverse outcomes. We investigated the association b etween PCAT and advanced plaque characteristics in patients with st able chest pain.

Methods

or Background: PCAT was quantified around all three major epicardial vessels on non-contrast CT images from t he PROMISE trial using a validated deep-learning algorithm. Quantitative cor onary plaque metrics included total plaque volume and burden (TPV, mm³; TPB, %), and plaque composition (calcified plaque (CP), non-calcified p laque (NCP), and low- density NCP (LD-NCP; <30HU). Multivariable linear r egression analyses related global PCAT density (per 10HU) and BSA-inde xed PCAT volume (per 10 cm³/m²) to plaque metrics (per 10mm³/1%), adjust ed for signal-to-noise ratio, tube voltage, risk score, and BMI.

Results

or Findings: In 3,620 participants (age: 60±8 years; women: 1,945(51.2%), mean total heart PCAT volume and dens ity were 12.9±3.1cm³/m² and -81.4±6.4HU. Greater PCAT volume related to higher CP volume (Coef. 2.24, 95%CI:0.40–4.08, p=0.017), LD-N CP volume (Coef. 0.28, 95%CI:0.05–0.52, p=0.018), and TPV (Coef. 3.84, 95% CI:-0.21–7.88, p=0.06). A 10 cm³/m² increase in PCAT volume was associated with a 7% higher TPB (Coef. 6.93, 95%CI:3.58–10.28, p<0.001), 4% higher CP burden (Coef. 3.97, 95%CI:2.42–5.51, p<0.001), 3% higher NCP burden (Co ef. 2.96, 95%CI:0.33– 5.59, p=0.027), and 1% higher LD-NCP burden (Coef. 0.68, 95%CI:0.34–1.01, p<0.001). PCAT density showed no significant associ ation with plaque volume and burden including composition.

Conclusion

PCAT volume is associated with plaque volume, burde n, and composition suggesting a relationship with both ath erogenesis and plaque architecture. In contrast, PCAT density, a known me asure of inflammation, was not associated with either. These findings emphasiz e the complex physiology of pericoronary fat and underscore the need to furt her investigate PCAT’s potential role as a target for treatment interventi ons.

Limitations

Secondary analysis Funding for this study: DFG (project number: 502109212) Ethics committee - additional information: Massachusetts General Hospital (2009P002231) Author Disclosures: Ibrahim Hadzic: Nothing to disclose Pamela Douglas: Nothing to disclose Thomas Mayrhofer: Nothing to disclose Borek Foldyna: Nothing to disclose Maros Ferencik: Nothing to disclose Marcel Christian Langenbach: Grant Recipient: DFG r esearch grant Julia Karady: Nothing to disclose Michael T. Lu: Nothing to disclose Svati Shah: Nothing to disclose Sunday Abstract-based Programme 290 Evaluating Radiation Exposure in Patients with Stab le chest Pain in the DISCHARGE trial M. Bosserdt¹, M. Mohamed¹, M. C. Williams², M. Dewe y¹, *J. Knape*¹; ¹Berlin/DE, ²Edinburgh/UK Purpose or Learning Objective: To assess 3.5 years of cumulative radiation doses of cardiovascular imaging to participants und ergoing computed tomography (CT) or invasive coronary angiography (I CA) for suspected coronary artery disease.

Methods

or Background: This is a prespecified analysis of a multicentre, randomised DISCHARGE trial involving 3561 participa nts with stable chest and who were referred for ICA, conducted between Octobe r 2015 and April 2019 in 26 European centres. Participants were randomised t o either CT (1808) or ICA (1753). Radiation dose from CT (dose-length product ), SPECT (injected activity), PET-CT (injected activity) and ICA (kern -area product) was assessed for 3.5 years after randomisation. Effective dose w as calculated using conversion factors appropriate for the imaging moda lity. Missing data were imputed using the mean. Wilcoxon rank sum test was used to assess group differences.

Results

or Findings: Over a median follow-up period of 3.5 years, a tota l of 1845 (CT: 1796 vs. ICA: 49) CT scans, 1939 (CT: 344 vs. ICA: 1595) ICA without PCI, 584 (CT: 269 vs. ICA: 315) ICA with PC I, 75 (CT: 46 vs. ICA: 29) SPECT and 81 (CT: 66 vs. ICA: 15) PET-CT were perfo rmed. Total per- participant cumulative dose was higher in the CT gr oup (median, 6.1 mSv; IQR, 3.9-10.3 mSv) compared with ICA group (median, 4.4 mSv; IQR, 2.2-9.1 mSv, P<0.001). The cumulative dose varied across pa rticipating centre, with a mean 9-fold for CT-group and 4-fold for ICA-group v ariation between the highest and lowest dose.

Conclusion

The cumulative radiation dose over 3.5 years was hi gher in the CT group compared to the ICA group. However, the to tal radiation dose for the CT and ICA group differed greatly among the partici pating centres.

Limitations

At a follow-up radiation dose was not available for all procedures Funding for this study: This study was funded by grants from the EU-FP7 Framework Program (FP 2007-2013, EC-GA 603266) Ethics committee - additional information: The study was approved by ethics committee at Charité (EA1/294/13)). Author Disclosures: Mahmoud Mohamed: Nothing to disclose Marc Dewey: Grant Recipient: EU (EC-GA 603266 in HE ALTH.2013.2.4.2-2) DFG (DE 1361/14-1, DE 1361/18-1, BIOQIC GRK 2260/1, Radiomics DE 1361/19-1 [428222922] and 20-1 [428223139] in SPP 2 177/1), GUIDE-IT (DE 1361/24-1), Berlin University Alliance (GC_SC_PC 27 ), G-BA (01NVF23002), Berlin Institute of Health (Digital Health Accelera tor). Other: Editor: Cardiac CT (Springer Nature). Hands-on cardiac CT courses (www .ct-kurs.de) Institutional research agreements: Siemens, General Electric, Phi lips, Canon. Patent on fractal analysis of perfusion imaging (jointly with Florian Michallek, EPO 2022 EP3350773A1, and USPTO 2021 10,991,109, approved) M .D. is European Society of Radiology (ESR) Publications Chair (2022 -2025); the opinions expressed in this presentation are the author’s own and do not represent the view of ESR. Michelle Claire Williams: Nothing to disclose Maria Bosserdt: Nothing to disclose Jakob Knape: Nothing to disclose 11:30-12:30 Research Stage 3 Research Presentation Session: Interventional Radiology RPS 2409 Hepato-biliary interventions in benign conditions Moderator G. Maleux; Leuven/BE ([email protected]) Multicenter outcomes analysis of self-expandable bi odegradable stents for the management of benign biliary strictures in 81 patients with pediatric liver transplantation *P. Marra*¹, D. Barnés Navarro², L. F. Fernández Ro dríguez³, G. Barbiero⁴, S. Mcguirk⁵, C. Gonzalez-Junyent², T. Hernández Cabrero³, M. C . Minà⁴, S. Sironi¹; ¹Bergamo/IT, ²Barcelona/ES, ³Madrid/ES, ⁴Padova/IT, ⁵Birmingham/UK ([email protected]) Purpose or Learning Objective: Percutaneous transhepatic cholangiography (PTC), billioplasty and biliary drainage are routin e treatments for benign strictures after pediatric liver transplantation (p LT). This multicentric study evaluated the efficacy and safety of biodegradable biliary stents.

Methods

or Background: ELLA (Ella-CS Ltd) is self-expandable stent made of polydioxanone which degrades in 3-6 months. We e valuated a total of 81 patients (39 females, median age 4 y/o; 42 males, m edian age 5 y/o) with benign biliary strictures developed after pLT in fi ve European centers. All the patients underwent percutaneous bilioplasty followe d by ELLA stent placement between October 2014 and March 2024. Stricture feat ures and treatment timing were assessed. Efficacy in terms of freedom from stricture recurrence and safety in terms of complications were analyzed.

Results

or Findings: Regarding stricture features, 42.7% of strictures h ad and extension <1 cm, 57.3% ≥1 cm; 69.7% of patients had strictures at the hepaticojejunostomy, 2.5% had intrahepatic strictur es and 28% had both intrahepatic and anastomotic strictures. The time b etween PTC and stent placement varied from 0 to 744 days (median of 36 d ays, IQR 61 days) with maintainance of a drainage; 53/81 (65.4%) patients underwent additional bilioplasty sessions before ELLA placement. Success ful stent placement was achieved in 100% of cases and complications occurre d in 23.5% of cases, all minor, mostly infectious cholangitis. Eighteen pati ents had a stricture recurrence (22.2%) during a median follow-up of 784 days (IQR 1200 days); of these, 55.6% were retreated with ELLA.

Conclusion

Biodegradable self-expandable biliary stent are be safe and effective for the treatment of benign biliary strictures after pLT. Further studies investigating factor predisposing to stent failure with standardized protocols are required to define the ideal candidates and the bes t timing for stenting.

Limitations

Retrospective design; variable protocols among cent ers Funding for this study: None Ethics committee - additional information: Comitato Etico of Bergamo - Multicenter ELLA Author Disclosures: Sandro Sironi: Nothing to disclose Teresa Hernández Cabrero: Nothing to disclose Giulio Barbiero: Nothing to disclose Paolo Marra: Nothing to disclose Simon Mcguirk: Nothing to disclose Lucia Fernández Fernández Rodríguez: Nothing to dis close Carla Gonzalez-Junyent: Nothing to disclose Maria Carla Minà: Nothing to disclose Daniel Barnés Navarro: Nothing to disclose Sunday Abstract-based Programme 291 Percutaneous portal vein recanalization of non-cirr hotic extrahepatic portal vein obstruction: technical considerations a nd clinical outcomes in 15 children *P. Marra*, R. Muglia, F. S. Carbone, L. Dulcetta, L. D'Antiga, S. Sironi; Bergamo/IT ([email protected]) Purpose or Learning Objective: Portal hypertension resulting from extrahepatic portal vein obstruction (EHPVO) in chi ldren has been managed primarily through Meso-Rex bypass. The aim of the s tudy is to report a preliminary series of patients who underwent attemp ts at portal vein recanalization (PVR) prior to other types of interv ention.

Methods

or Background: A cohort of consecutive patients with EHPVO from 2021-2024 was retrospectively collected. After tran sjugular wedge hepatic venography for the study of the native intrahepatic portal system, percutaneous PVR was attempted via transhepatic and/or transplen ic access. Clinical and procedural data, technical and clinical success, co mplications and follow up data were recorded. Technical success was considere d at least the partial revascularization of the native portal system.

Results

or Findings: Fifteen patients (7 males; median age 8 years) with severe portal hypertension due to EHPVO underwent 2 0 percutaneous transhepatic (n=1), transplenic (n=12) or simultane ous transhepatic/transplenic (n=7) attempts at portal vein recanalization. Rex v ein was judged patent ad wedge hepatic venography in 2/15 (13%). Successful recanalization was achieved in 9/15 patients (60%). No major adverse e vents were observed. After successful angioplasty, 8/9 patients required metal stenting to obtain sustained patency. None of the failed patients was considered eligible for Meso-Rex bypass and underwent TIPS (n=2), splenecto my (n=1), surgical shunt (n=1). Two patients were followed-up without further interventions. After a median follow-up of 6 months patency of the main portal vein was demonstrated in all the patients who achieved PVR, with clinical and laboratory improvement of portal hypertension.

Conclusion

Our preliminary experience suggests that 60% of chi ldren with EHPVO can restore the portal flow by endovascular t reatment, even with obliterated Rex vein. Thanks to its low invasivenes s, PVR may be regarded as the primary intervention, before considering surger y.

Limitations

Retrospective study with limited sample Funding for this study: None Ethics committee - additional information: EC of Bergamo - Portal-01 Author Disclosures: Sandro Sironi: Nothing to disclose Paolo Marra: Nothing to disclose Francesco Saverio Carbone: Nothing to disclose Riccardo Muglia: Nothing to disclose Lorenzo D'Antiga: Nothing to disclose Ludovico Dulcetta: Nothing to disclose Comparison of different techniques for transjugular intrahepatic portosystemic shunt creation in a retrospective ser ies of 51 paediatric patients *A. Princi*, P. Marra, F. S. Carbone, L. Dulcetta, R. Muglia, S. Sironi; Bergamo/IT ([email protected]) Purpose or Learning Objective: TIPS creation with the standard technique may be challenging in children with low body weight , unusual anatomy, and liver grafts. This study analyses different approac hes for TIPS creation in a retrospective cohort of paediatric patients.

Methods

or Background: The retrospective single-centre cohort included 47 patients who received TIPS either with the standard (sTIPS; n:30 median age 10 years IQR8-16) or with a transhepatic/transpleni c ‘’hybrid’’ (hTIPS; hTIPS n:17 13y IQR7-16) approaches from 2005 to 2023. Inc lusion criteria were age ≤ 18 years or liver graft transplanted in paediatric age. All the variables between the sTIPS and the hTIPS group were compared .

Results

or Findings: Technical, hemodynamic, and clinical success were 100%, 100%, and 83% in sTIPS (8 permanent, 22 as br idge) while 100%, 97%, and 81% in hTIPS (10 permanent, three bridge). Patients with liver grafts were 4 (13%) in the sTIPS and 5 in the hTIPS (30%) groups, respectively. Portal cavernoma, acute portal vein thrombosis, and Budd-Chiari syndrome were significantly more prevalent in the hTIPS grou p (76% vs 33%, p<0.05). Indications were comparable. Covered stents were re spectively employed in 90% sTIPS and 88% hTIPS, with primary-assisted pate ncy of 100% and 93% after a median follow-up of 34 and 14 months (IQR 2 1-52 and 12-20). Complications included one hemoperitoneum in each g roup conservatively managed, one liver failure (hTIPS group) requiring urgent transplantation and one septic shock (sTIPS group). Eight shunt dysfunc tions in sTIPS and 3 in hTIPS were all successfully treated.

Conclusion

In a retrospective cohort of children, hybrid techn iques for TIPS creation achieved high success rates comparable to the standard technique despite unfavourable baseline characteristics. Furt her studies may investigate the role of these approaches in expanding TIPS indi cation for complex scenarios in paediatric patients.

Limitations

Retrospective nature. Funding for this study: None. Ethics committee - additional information: Not applicable. Author Disclosures: Sandro Sironi: Nothing to disclose Paolo Marra: Nothing to disclose Francesco Saverio Carbone: Nothing to disclose Angelo Princi: Nothing to disclose Riccardo Muglia: Nothing to disclose Ludovico Dulcetta: Nothing to disclose Transjugular Intrahepatic Portosystemic Shunt (TIPS ): early laboratory changes and correlations with short-term mortality *F. Schön*, T. Helmberger, M. Berning, S. F. U. Blu m, C. Radosa, S. Löck, R-T. Hoffmann, J-P. Kühn; Dresden/DE Purpose or Learning Objective: To investigate early laboratory changes following Transjugular Intrahepatic Portosystemic S hunt (TIPS) and their associations with short-term mortality.

Methods

or Background: TIPS procedures from 2017 to 2023 were enrolled retrospectively. Laboratory parameters (INR, ALAT, ASAT, GGT and bilirubin) were assessed once pre-procedurally, on post-proced ural days 1 or 2, and once again between days 5 and 7. Percentage changes from baseline were calculated for each parameter. Temporal changes of the parameters were assessed using Kruskal-Wallis tests, and comparison s regarding 30-day mortality were evaluated using Mann-Whitney U tests .

Results

or Findings: A total of 245 TIPS procedures (161 men, mean age 59.8 +/- 10.9 years) were enrolled, with a technica l success rate of 95.5% (234/245). All laboratory parameters significantly increased post-procedurally (p < 0.001). ALAT and ASAT revealed the highest inc rease within the first two post-procedural days (+374 +/- 1118%, and +450 +/- 1079%, respectively), followed by decreases on days 5-7 (+279 +/- 568%, a nd +125 +/- 233%, respectively). Patients who died within 30 days (n = 17) had significantly higher baseline INR (p = 0.009), ASAT (p = 0.014) and bili rubin (p = 0.011), while GGT was lower (p = 0.012). 30-day mortality was ass ociated with a higher increase of ASAT and GGT on days 1/2 (+1361 +/- 220 2% vs. +415 +/- 954%, p = 0.029; and +56 +/- 101% vs. +21 +/- 51%, p = 0. 034, respectively).

Conclusion

TIPS significantly impacts liver function, with mar ked early increases of ALAT and ASAT levels. Pronounced incre ase of laboratory parameters within the first two post-procedural day s might help to identify high- risk patients in terms of short-term mortality.

Limitations

Retrospective study design with a relatively small number of patients. Funding for this study: Not applicable. Ethics committee - additional information: The present study was approved by the local ethics committee (BO-EK-501122023). Author Disclosures: Ralf-Thorsten Hoffmann: Nothing to disclose Marco Berning: Nothing to disclose Jens-Peter Kühn: Nothing to disclose Felix Schön: Nothing to disclose Christoph Radosa: Nothing to disclose Sophia Freya Ulrike Blum: Nothing to disclose Thomas Helmberger: Nothing to disclose Steffen Löck: Nothing to disclose Successful approach to giant hydatid cysts of liver *U. Koç*, C. Aydın, M. Özdemir; Ankara/TR Purpose or Learning Objective: We aimed to evaluate cases of giant liver hydatid cysts.

Methods

or Background: Between December 2020 and January 2023, out of 100 liver hydatid cysts treated with the percutaneo us approach in our interventional radiology department, 35 cases were more than 10 cm in one of the diameters. These giant hydatid cysts were treat ed with the catheterization approach which includes trochar style puncture, asp iration of the cavity, and installation with hypertonic saline solution. Then, ethanol installation of the cyst is the next step if the cavity is not connected wit h the biliary tree.

Results

or Findings: Out of 35 patients, 27 (78%) patients were treated successfully with catheter approach solely without the need of ERCP related interventions. 8 patients (22%) needed further inte rventions with ERCP because of cystobiliary fistula; 2 out of 8 patient s had biliary passage confirmed after contrast installation at catheteriz ation. Other 6 patients did not have opaque drainage into the biliary system, rathe r they were suspected of having a fistula because of yellow content or bioch emistry showing high bilirubin. These 8 patients had extended days of ex ternal catheter duration, about 20 days on average; while other patients had the external catheter removed approximately 2 days later. 1 patient had t o go through cystectomy Sunday Abstract-based Programme 292 operation because of insufficient drainage after re peated ERCP interventions, and this patient had the longest hospitalization du ration of 100 days; whilst other ERCP patients had 20 days of hospitalization on average.

Conclusion

Giant hydatid cysts are manageable with a percutane ous approach. Cystobiliary fistula must be kept in mind especially for the giant hydatid cysts, since size is an important predictor for this communication.

Limitations

As a limitation, this study did not compare the met hod with others, but a future study will address this. Funding for this study: None Ethics committee - additional information: None Author Disclosures: Mustafa Özdemir: Nothing to disclose Ural Koç: Nothing to disclose Ceren Aydın: Nothing to disclose Incidence of Bleeding Between Percutaneous vs. Endo scopic Biliary Diversion in Patients with Biliary Tract Obstructio n: A Systematic Review and Meta-Analysis *E. D. L. A. Salazar Perez*, E. E. Lozada Hernandez , B. E. E. Retamoza Rojas; Leon/MX ([email protected]) Purpose or Learning Objective: This meta-analysis examines the incidence of bleeding in patients undergoing biliary diversio n via two alternatives to ERCP: percutaneous and endoscopic methods. The anal ysis aims to establish a foundation for identifying the most suitable trea tment by evaluating effectiveness and safety. The primary goal of manag ing biliary tract obstruction is to achieve safe and effective drainage, with pro per patient selection and diagnosis being essential for determining the best procedure.

Methods

or Background: Eleven studies were analyzed regarding the incidence of bleeding, pancreatitis, and reinterven tions associated with endoscopic or percutaneous biliary drainage procedu res from January 2010 to 2023. The evaluated aspects included sensitivity, r isk of bias ratio, odds ratio, and their 95% confidence interval using a random-ef fects model, with effects considered statistically significant if the confide nce interval was at 95%. The I² statistic was calculated to assess heterogeneity. B ias analysis was reported using funnel plot tables. Data processing was perfo rmed using the R programming language within the RStudio environment (version 4.1.0 CRAN). Any p-value less than 0.05 was considered statistic ally significant.

Results

or Findings: The study evaluated complications and reinterventio ns for two techniques of biliary diversion: the percut aneous method, involving 2,058 patients from 11 studies, and the endoscopic method, with 7,959 patients. The overall odds ratio for bleeding was 1 .81 (95% CI 0.43-7.60), with heterogeneity (I² = 74%). For acute pancreatitis, t he overall odds ratio was 0.15 (95% CI 0.05-0.47) with p = 0.03. Reinterventions s howed an odds ratio of 0.25, with a wide confidence interval [0.06; 1.51], reflecting high heterogeneity and variability among the studies.

Conclusion

The percutaneous technique, as a first-line option compared to many other emerging techniques, remains the ideal c hoice in many referral centers for diseases presenting with biliary tract obstruction, showing minimal complication rates.

Limitations

Heterogeneity. Funding for this study: There is no funding or conflict of interest. Ethics committee - additional information: The registration for the approval of the hospital was the CEI-004-2022 Author Disclosures: Beatriz Elena Elena Retamoza Rojas: Nothing to disc lose Estrella De Los Angeles Salazar Perez: Nothing to d isclose Edgard Efren Lozada Hernandez: Nothing to disclose Clinical Outcomes of Separate versus Single Tract T echniques in Percutaneous Radiologic Gastrostomy with Single Gas tropexy: A Multi-Center Retrospective Analysis *H. N. Lee*¹, S-J. Park², Y. Cho³, S. Lee¹; ¹Cheona n/KR, ²Ansan/KR, ³Gangneung/KR ([email protected]) Purpose or Learning Objective: To compare the clinical outcomes of separate versus single tract techniques and to inve stigate predictors of complications during percutaneous radiologic gastro stomy with single gastropexy.

Methods

or Background: Between January 2018 and January 2024, 241 consecutive patients (mean age: 68.8 ± 13.5 years; male: 73.4%) who underwent percutaneous radiologic gastrostomy with single gastropexy were enrolled. The patients were divided into two groups based on the anchoring

Method

the separate tract group (n = 136) and the single tract group (n = 105).

Results

or Findings: The technical success rate was 99.3% in the separat e tract group and 98.1% in the single tract group (p = 0.582). Four patients (3.81%) in the single tract group experienced intra -procedural anchor dislodgment. In 3 of these cases, technical success was achieved by deploying a second anchor. The 30-day complication rate was 1 2.5% in the separate tract group and 15.2% in the single tract group (p = 0.324). There was no procedure-related mortality. BMI (odds ratio 0.885, p = 0.021) was a negative predictor, while unfavorable anatomy on CT (odds ra tio 2.4, p = 0.033) was a positive predictor for complication.

Conclusion

Although anchor dislodgment was a complication uniq ue to the single tract technique, the two groups showed no si gnificant difference in overall clinical outcomes. This study highlights th at BMI and CT findings are more crucial predictors of complications than the c hoice of technique.

Limitations

The retrospective nature of the study leads to seve ral biases. Funding for this study: This study was not supported by any funding. Ethics committee - additional information: The Institutional Review Board of tertiary care hospitals approved this retrospective study and waived written informed consent for using clinical and imaging dat a. Author Disclosures: Sung-Joon Park: Nothing to disclose Youngjong Cho: Nothing to disclose Hyoung Nam Lee: Nothing to disclose Sangjoon Lee: Nothing to disclose 11:30-12:30 Research Stage 4 Research Presentation Session: Musculoskeletal RPS 2410 Imaging of musculoskeletal tumours Moderator I.-M. Noebauer-Huhmann; Vienna/AT ([email protected]) Dual-energy computed tomography parameters for the differentiation of vertebral small osteolytic metastases (SOMs) and SO M-mimics *J. Li*¹, J. Liu²; ¹Fujian/CN, ²Xiamen/CN ([email protected]) Purpose or Learning Objective: To evaluate the value of dual energy computed tomography (DECT) quantitative parameters for the differentiation of small osteolytic metastases (SOMs) and SOM-mimics s uch as osteopenia, osteoporosis, and Schmorl's nodule.

Methods

or Background: Fat(HAP), fat(calcium), hydroxyapatite(fat), and calcium(fat) densities [Dfat(HAP), Dfat(calcium), D HAP(fat), and Dcalcium(fat)], as well as CT value were collected. Comparisons wer e made using the independent sample T test. Diagnostic performance w as assessed in terms of area under the receiver operating characteristic cu rve (AUC). The sensitivity, specificity, positive predictive value (PPV), negat ive predictive value (NPV), and accuracy of each parameter was assessed as well .

Results

or Findings: A total of 106 patients were included, of whom 24 h ad SOMs (lesion, n = 48), while 82 had SOM-mimics (les ion, n = 202). SOMs associated with significantly higher CT value, Dfat (calcium), and Dfat(HAP) compared to SOM-mimics (P < .001). The AUCs were 0. 674, 0.879, and 0.887, respectively. The sensitivity, specificity, PPV, NP V, and accuracy of fat(calcium) were 77.1%, 85.1%, 55.2%, 94.0%, 83.6% , respectively; while those for Dfat(HAP) were 83.3%, 80.7%, 50.6%, 95.3% , 81.2%, respectively. The optimal diagnostic cutoffs for Dfat(calcium) an d Dfat(HAP) were ≥ 1000.0 mg/cm3 and ≥ 966.9 mg/cm3, respectively, which achieved consist ent diagnostic results among 89.6% lesions (n = 224). T he combined use of Dfat(HAP) and Dfat(calcium) achieved significantly better diagnostic performance, with AUC, sensitivity, specificity, PP V, NPV, and accuracy of 0.910, 82.2%, 87.2%, 61.7%, 95.1%, and 86.2%, respe ctively.

Conclusion

Dfat(calcium) and Dfat(HAP) on DECT carry the poten tial as parameters for the discrimination of SOMs from SOM- mimics

Limitations

This was a retrospective study with a relatively sm all sample size. In addition, the focus on thoracolumbar lesions lim its the generalizability of our results. Funding for this study: Natural Science Foundation of Fujian Province, Chin a (grant numbers: 2023J01181) Ethics committee - additional information: Fujian Cancer Hospital Ethics Committee (K2023-198-01) Author Disclosures: Jianfang Liu: Nothing to disclose Jie Li: Nothing to disclose Sunday Abstract-based Programme 293 Multimodal machine learning method for the identifi cation of prognostic and predictive biomarkers in Adolescent and Young A dults (AYA) sarcoma: a pilot study *S. Lusi*, R. Romanelli, A. Marzullo, A. Laffi, A. Santoro, L. Balzarini, M. Francone, A. F. Bertuzzi; Milan/IT Purpose or Learning Objective: This study relies on the use of Artificial Intelligence (AI) to develop and validate a multimo dal machine learning method that could provide a prognostic model in AYA patien ts affected by sarcomas, exploring the clinical, radiomic and pathological f eatures that may be predictive of disease outcome.

Methods

or Background: The study is a monocentric retrospective cohort study involving 245 patients with sarcomas. Our pre liminary and full results were performed on a smaller cohort of 22 patients w ith soft tissue sarcoma of the extremities (13 non-AYA and 9 AYA) for whom cli nical data (using Excel form), radiomic features (from a pre-treatment MRI) and histopathological features (extracted using a foundation model) were collected. All this data was then used to match similar patient profiles in the two groups using logistic regression propensity scores. Disease-free survival of matched patients was described using a Kaplan-Meier curve.

Results

or Findings: Statistical analysis didn't identify any correlatio n between clinical and radiological features that cou ld explain the differences in prognosis between the two groups, probably due to t he small cohort size. However, AI analysis using a Kaplan-Meier curve sho wed that AYA patients had a worse prognosis than non-AYA patients (p < 0. 05), confirming for the first time, to our knowledge, by deep machine learn ing what is observed in clinical practice.

Conclusion

Despite the limitations of these preliminary result s based on a small cohort of patients, our findings provide valu able insights into the differential prognosis that characterises these two groups. AI holds promise for uncovering hidden characteristics, with future rese arch potentially incorporating biological markers to further explore therapeutic t argets.

Limitations

The small cohort size of the study limits its stati stical power. The retrospective design may introduce selection bias. Future studies in larger populations are needed. Funding for this study: No funding was received for this study. Ethics committee - additional information: The study is retrospective. Author Disclosures: Alexia Francesca Bertuzzi: Nothing to disclose Armando Santoro: Nothing to disclose Aldo Marzullo: Nothing to disclose Alice Laffi: Nothing to disclose Marco Francone: Nothing to disclose Roberta Romanelli: Nothing to disclose Stefano Lusi: Nothing to disclose Luca Balzarini: Nothing to disclose Radiomics in MRI to improve the characterization of cartilaginous bone tumours *Q. Bui*, M. Lacroix, L. S. Fournier, F. Larousseri e, A. Feydy; Paris/FR Purpose or Learning Objective: In long bones, distinguishing between enchondromas and chondrosarcomas before surgery is often challenging and may require invasive biopsy for accurate diagnosis. The purpose of this work was to assess the performance of MRI radiomics-base d machine learning in classifying enchondromas and chondrosarcomas in lon g bones.

Methods

or Background: Ninety-eight patients with pathology-proven cartilaginous tumours of long bones were retrospect ively included from a tertiary bone tumour centre. The training set consi sted of 81 MRI (n = 33 enchondromas; n = 48 chondrosarcomas). The internal test set consisted of 17 MRI (n = 7 enchondromas; n = 10 chondrosarcomas). 3 D segmentation was performed on T1-weighted and fat-suppressed T2-weig hted MRI images and radiomics features were extracted. Dimensionality r eduction was performed based on reproducibility, redundancy and feature im portance. Different models were tested, including multiparametric, single sequ ence and sequential. A Random Forest classifier was tuned on the training set using five-fold cross- validation and tested on the internal test set.

Results

or Findings: The Random Forest classifier with the T2 then T1 sequential model, which was the best-performing mod el, achieved an AUC of 0.943 [0.832 – 1.000] on the internal test set. Its accuracy in correctly classifying enchondromas and chondrosarcomas was 71 % (5/7) and 100% (10/10), respectively.

Conclusion

This work shows that MRI radiomics can accurately d ifferentiate between benign and malignant cartilaginous tumours. Although comparisons among various models did not achieve statistical si gnificance, the data suggest that a sequential approach using single sequence mo dels might outperform a multiparametric model. Further studies with larger sample sizes are needed to integrate these findings into clinical practice and improve preoperative diagnosis of cartilaginous tumours within the conte xt of personalised medicine.

Limitations

Limitations of the study were the small sample size and the lack of an external test set. Funding for this study: The author received a research grant from Societé Française de Radiologie and Assistance Publique - H ôpitaux de Paris. Ethics committee - additional information: This study involved a retrospective analysis of anonymised data collected as part of routine care. Author Disclosures: Laure S. Fournier: Nothing to disclose Antoine Feydy: Nothing to disclose Quentin Bui: Research/Grant Support: Société França ise de Radiologie Research/Grant Support: Assistance Publique - Hôpit aux de Paris Frederique Larousserie: Nothing to disclose Maxime Lacroix: Nothing to disclose Optimizing Cryoablation Outcomes in Desmoid Tumors: A Machine Learning-Driven Radiomic Analysis *M. E. Chevasco Hanze*, L. Ponsa Cobas, J. A. Narvá ez García, D. A. Sandoval Díaz, J. Hernández Gañan, J. C. Sard iñas Barrero; Barcelona/ES ([email protected]) Purpose or Learning Objective: Desmoid tumors (DT) are locally aggressive, infiltrative neoplasms with a high risk of local re currence. Recently, percutaneous cryoablation has emerged as an alterna tive therapy, though its role as a salvage treatment remains unclear. This s tudy aimed to evaluate disease progression after cryoablation at a 1-year follow-up and develop a predictive model using clinical and radiomic variab les.

Methods

or Background: We performed a retrospective analysis of patients treated with cryoablation for extra-abdominal DT fr om January 2018 to September 2023. Pre- and post-cryoablation contrast -enhanced MRIs were reviewed, and disease progression was defined as le ss than 90% necrosis or non-enhancement at follow-up. Radiomics features we re extracted from T2- weighted pre-cryoablation MRIs, and data were filte red based on correlation matrices and statistical tests (T-Student, Mann-Whi tney U, Chi-square). Prediction models, including LASSO, Random Forest, XGBoost, SVM, and KNN, were evaluated using ROC analysis and 5-fold c ross-validation to determine the optimal approach.

Results

or Findings: Twenty-eight patients were included (median age 43; 67% women), with a no disease progression rate of 6 0.71%, significantly associated with partial response on mRECIST criteri a (p = 0.022). The Random Forest model showed the best performance (AU C = 0.77). Key predictive features included tumor diameter, spheri city, major axis length, minimum intensity, kurtosis, interquartile range, a nd tumor location. Tumors >61 mm, with ellipsoid shape (major axis length >79 mm) and regular form (sphericity <0.7), predicted no disease progression . Similarly, tumors <61 mm with a regular shape (sphericity <0.5) and fibrous matrix (minimum <175) predicted favorable outcomes.

Conclusion

Cryoablation therapy has demonstrated a good therap y for DT treatment. Radiomics shape and first order features have shown their relevance in cryoablation therapy planning as it se rve as a patient selection tool.

Limitations

Small sample size. No split train-test approach. Funding for this study: No funding was received Ethics committee - additional information: No intervention, just an observational study Author Disclosures: Daniel Alejandro Sandoval Díaz: Nothing to disclose Laia Ponsa Cobas: Nothing to disclose Miguel Emilio Chevasco Hanze: Nothing to disclose Javier Hernández Gañan: Nothing to disclose José Antonio Narváez García: Nothing to disclose Juan Carlos Sardiñas Barrero: Nothing to disclose “Pseudo-CT” MRI sequences and detection of lytic le sions in multiple myeloma *C. Chabot*; Brussels/BE ([email protected]) Purpose or Learning Objective: To assess the diagnostic accuracy, repeatability, and reproducibility of pseudo-CT MRI sequences (ZTE, BB) in detecting osteolytic lesions in MM using WB-CT as t he reference standard.

Methods

or Background: In this prospective study, consecutive patients wer e enrolled in our academic hospital. Inclusion criter ia were newly diagnosed MM, monoclonal gammopathy of undetermined significance at high risk for MM, or suspicion of progressive MM. Participants underwent ZTE and BB sequences covering the lumbar spine, pelvis, and proximal fem urs as part of 3T WB-MRI examinations, as well as clinically indicated 18F-F DG PET/CT examination that included optimized WB-CT. Ten bone regions and two scores (categorical score/semiquantitative score) were assessed by thre e radiologists on the ZTE, BB, and WB-CT images. The accuracy, repeatability, and reproducibility of categorical scores (Gwet agreement coefficients AC1 and AC2) and Sunday Abstract-based Programme 294 differences in semiquantitative scores were assesse d at per-sequence, per- region, and per-patient levels.

Results

or Findings: 47 participants were included. In experienced reade rs, BB and ZTE showed 98% accuracy per-patient, while B B accuracy ranged from 83%–100% and ZTE from 74%–94% per-region. Incr eased false-negative findings in the spine ranging from 17%-23%, accordi ng to the lumbar vertebra, was observed using ZTE(P<.013). Regardless of the r egion (except coxal bones), differences in the BB score minus the ZTE s core were positively skewed(P<.021). Repeatability was very good(AC1 ≥0.87), while reproducibility was at least good(AC2≥0.63).

Conclusion

Both MRI-based ZTE and BB pseudo-CT sequences of th e lumbar spine, pelvis, and femurs demonstrated high diagnostic accuracy in detecting osteolytic lesions in MM. Compared with B B, the ZTE sequence yielded more FN findings in the spine.

Limitations

Pseudo-CT sequences were limited to the lumbar spin e, pelvis, and femurs; the reference CT required optimization from 18F-FDG PET/CT; focus was on detecting osteolytic lesions, includin g nonactive ones that may persist post-treatment. Funding for this study: None Ethics committee - additional information: This prospective study was approved by the institutional ethics committee (202 0/27JUL/380) and is registered on ClinicalTrials.gov (no. NCT05381077). Written informed consent was obtained from all participants. Author Disclosures: Caroline Chabot: Nothing to disclose Implementing tin prefiltration in routine clinical CT scans of the lower extremity: Impact on radiation dose *T. Marth*, C. Stern, R. Sutter; Zürich/CH ([email protected]) Purpose or Learning Objective: Several studies have demonstrated the potential of tin prefiltration to reduce the radiat ion dose while maintaining diagnostic quality for musculoskeletal imaging. Sti ll, no study has reported data on the impact of radiation dose reduction on clinic al routine scanning.

Methods

or Background: Retrospective inclusion of 300 patients who underwent clinically indicated CT scans of the pelv is, knee, or ankle on a single CT scanner (SOMATOM Definition AS, Siemens Healthin eers) before January 2020 (without tin filter) and after December 2020 ( with tin filter). For each joint, 50 patients with tin prefiltration and 50 patients without tin prefiltration were selected.

Results

or Findings: The CTDIvol, DLP, and effective dose were reduced significantly in all tin-prefiltered examinations c ompared to the conventional, non-tin-prefiltered examinations (p<.001). Tin-pref iltered CT scans had a significantly lower CTDIvol and mean effective dose (all p<.001): CTDIvol was 65% lower in the pelvis, 73% in the knee, and 54% i n the ankle. This resulted in a reduction of the effective dose of 61%, 71%, a nd 60%, respectively. In absolute numbers, the reduction of the median effec tive dose delivered in a single scan of the pelvis was 2.29 mSv, 0.15 mSv fo r the knee, and 0.03 mSv in the ankle.

Conclusion

The implementation of tin prefiltration in clinical routine scan protocols significantly reduced the effective radia tion dose for unenhanced CT scans of the pelvis (61% reduction, 2.29 mSv), the knee (71% reduction, 0.15 mSv), and the ankle (60% reduction, 0.03 mSv).

Limitations

It would be possible to reduce radiation dose even more by applying deep learning-based denoising algorithms, however this was not yet available in clinical routine during the data acqui sition. Funding for this study: No specific funding. Ethics committee - additional information: BASEC-ID 2024-01094 Kantonale Ethikkommission Zürich Author Disclosures: Thomas Marth: Other: Balgrist University Hospital a nd Balgrist Campus each have an academic research collaboration with Siemen s Healthineers. Balgrist University Hospital also has an academic research c ollaboration with Bayer. Christoph Stern: Other: Balgrist University Hospita l has an academic research collaboration with Siemens Healthineers. Balgrist U niversity Hospital also has an academic research collaboration with Bayer. Reto Sutter: Other: Balgrist University Hospital ha s an academic research collaboration with Siemens Healthineers. Balgrist U niversity Hospital also has an academic research collaboration with Bayer. Evaluation of the diagnostic potential of a Tomosyn thesis system for MSK Y. Beer¹, *N. Shabshin*², L. Copel¹, Y. Kimmel³, R. Ophir¹, Y. S. Schiffenbauer³, S. Tal¹; ¹Zrifin/IL, ²Afula/I L, ³Petach Tikva/IL ([email protected]) Purpose or Learning Objective: Digital tomosynthesis (DTS) is a well- established technology that has become the gold sta ndard for breast mammography. In recent years its benefits in muscul oskeletal (MSK) imaging have been acknowledged, leading to a rapid increase in its utilization. It improves detection, localization and characterizati on of subtle fractures. In some patients DTS can alleviate the need for CT wit h lower complexity. Recently a new technology based on Cold Cathode Xra y tube with a multi tube set-up has made this technology more affordable and accessible. The purpose of this study is to evaluate the diagnostic potenti al of the cold-cathode multi- tube DTS

Methods

or Background: The study included 19 patients with suspected fractures who underwent CT and radiographs (XR). Pa tients were scanned using the cold-cathode DTS . 15 patients had imagin g performed with Cast or metal. Images were evaluated by an MSK radiologist and orthopedic surgeon in consensus with CT as the gold standard. Studies were evaluated for presence, location, intraarticular involvement, dis placed fragments and incidental lesions. The surgeon was asked whether D TS provided valuable information and whether it increased the confidence of the final diagnosis.

Results

or Findings: In 17/19 studies DTS added value to the XR. In 7, D TS found fractures occult in XR. In 3, DTS was able to clear a suspected fracture. in 5, DTS was able to better localize the fracture. In 1, DTS was able to determine fracture age and in 1 study DTS found a s clerotic lesion obscured in XR. In addition, on XR the cast limited evaluation of fine bony details, however there was no such limitation with DTS

Conclusion

Cold cathode DTS provides high quality tomography o f anatomies enabling depiction of occult pathologies, localization, characterization and resolution of questionable fin dings

Limitations

Initial study results Funding for this study: Study was funded by Nanox-x Imaging Ltd. Ethics committee - additional information: Study was approved by the local Ethics committee in the institution and each patien t underwent informed consent Author Disclosures: Robenpour Ophir: Nothing to disclose Laurian Copel: Nothing to disclose Yiftah Beer: Nothing to disclose Yotam Kimmel: Nothing to disclose Yael S Schiffenbauer: Nothing to disclose Nogah Shabshin: Consultant: Nano-x imaging Sigal Tal: Nothing to disclose