Radiomic Analysis of Abdominal CT Plain Scans for Identifying Aortic Syndrome in Emergency Patients

Radiomic Analysis of Abdominal CT Plain Scans for Identifying Aortic Syndrome in Emergency Patients | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Radiomic Analysis of Abdominal CT Plain Scans for Identifying Aortic Syndrome in Emergency Patients Wenxiao Lin, Yifan Guo, Haonan Zhu, MengYuan Shen, Jiehui Su This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7095829/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background: This study aimed to assess the capability of radiomics based on abdominal CT scans in identifying aortic syndrome (AS) among patients in the emergency department. Methods: Emergency patients who underwent both plain and contrast-enhanced abdominal CT scans from August 2012 to October 2020 were retrospectively enrolled. These patients were classified based on the presence of abdominal AS. The dataset was randomly split into training, test, and external validation sets in a 3:1:1 ratio. Radiomic features were extracted from manually segmented regions of the abdominal aorta in the CT images. These features were used to create a radiomic model. The radiomic model was integrated with relevant clinical factors using multivariate logistic regression to develop a clinical-radiomics model. The diagnostic performance of the models was assessed through receiver operating characteristic (ROC) analysis. Results: A total of 188 patients were included in the study. Ten of 1794 radiomic features were selected for constructing the radiomic model. The SHapley Additive exPlanations (SHAP) analysis identified the wavelet-LHL_GLDM_SDLGLE feature as a significant contributor to the predictive accuracy of the radiomic model. The area under the curves (AUCs) for the clinical, radiomic, and clinical-radiomics models in the external validation set were 0.763, 0.891, and 0.860, respectively. The AUCs of the radiomic and clinical-radiomics models were significantly higher than that of the clinical model ( P 0.05). Conclusions: The study demonstrated that an abdominal CT plain scan-based radiomic model can effectively be utilized for preliminary diagnosis of abdominal AS in emergency patients. Biological sciences/Computational biology and bioinformatics Health sciences/Diseases Health sciences/Gastroenterology Health sciences/Medical research Abdomen Aortic Syndrome Emergencies Radiomics Computed Tomography Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Acute aortic syndrome (AS) encompasses a group of critical and life-threatening aortic diseases, including acute aortic dissection (AD), intramural hematoma (IMH), and penetrating aortic ulcer (PAU). The aforementioned conditions exhibit overlapping pathophysiological mechanisms, clinical manifestations, and diagnostic and therapeutic challenges 1 . Due to the non-specific nature of symptoms and signs associated with acute AS, its identification often requires a high degree of clinical suspicion. Misdiagnosis and delayed treatment can significantly exacerbate patient outcomes 1,2 . Acute AD is typically characterized by the sudden onset of severe chest or back pain, a symptom that is also commonly observed in IMH and PAU 3,4 . However, the occurrence of abdominal pain is less frequent 3,5,6 . Patients with abdominal manifestations of acute AS are more prone to being overlooked in emergency scenarios compared to those presenting with chest pain. The complexity and diversity of etiologies in patients presenting with acute abdominal pain in emergency settings pose significant diagnostic challenges 7,8 . Among these, acute AS is a critical condition that cannot be overlooked. Diagnosing acute AS in the context of acute abdominal pain is particularly challenging due to its less frequent occurrence and the non-specific nature of its symptoms compared to thoracic presentations 8,9 . The abdominal CT plain scan is a commonly used imaging modality for patients presenting with acute abdominal pain. If the utilization of abdominal computerized tomography (CT) plain scans can achieve accurate diagnosis of AS, it would facilitate prompt therapeutic interventions for patients, consequently improving their prognostic outcomes. The integration of advanced artificial intelligence technology with abdominal CT plain scan holds promise for achieving immediate diagnosis of acute AS during the scanning process, potentially mitigating severe adverse outcomes caused by missed or incorrect diagnoses. Recent studies have demonstrated that the application of radiomics or deep learning based on non-contrast chest CT scans can accurately predict AD 10,11 . Radiomics is a technique that facilitates the high-throughput extraction of quantitative features from medical images, enabling the conversion of these features into analyzable data 12,13 . This approach utilizes computational methods to identify patterns and characteristics within images that may not be discernible to the human eye. The heterogeneity of blood in the thoracic aorta caused by AD on CT images is different from that in normal arteries, which is thought to be one of the main reasons for the successful identification of AD in the thoracic aorta by radiomics 10 . The blood flow pattern of the abdominal aorta is simpler compared to the complex flow pattern in the thoracic aorta caused by the aortic arch 14,15 . The specific blood heterogeneity caused by AS in the abdominal aorta might be more easily captured by radiomic features. If a radiomic model based on non-contrast abdominal CT can accurately predict abdominal AS, it could alleviate the burden on radiologists dealing with patients presenting with abdominal pain in emergency departments. This is particularly important for patients presenting at night, reducing the risk of missed acute AS diagnoses due to radiologist fatigue 16 . The aim of this study is to apply radiomics to abdominal CT plain scans, enabling the immediate diagnosis of abdominal AS in emergency patients following their abdominal CT. This ensures that patients with acute AS receive timely and effective diagnosis and treatment. Methods Patients The study was approved by the Ethics Committee of The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University This study retrospectively collected emergency patients who underwent abdominal CT plain scans and contrast-enhanced CT scans at organization A (anonymous) and organization B (anonymous) between August 2012 and October 2020. Exclusion criteria were as follows: 1) The patient underwent separate sessions for the completion of the abdominal CT plain scan and contrast-enhanced CT scans, with a time interval exceeding 24 hours. 2) Metallic artifacts or motion artifacts present in the CT images hindered accurate diagnosis of the aorta. Baseline data collected from patients included gender, age, and abdominal pain. Figure 1 illustrates the flowchart detailing the process of patient enrollment. Image Acquisitions The abdominal CT scans were performed using two different CT machines. For the training and test sets, a Philips Brilliance 16 CT scanner was utilized with the following parameters: tube voltage of 120 kVp; tube current of 250 mAs; a field of view (FOV) of 350 mm; matrix size of 512 x 512; filtered back projection reconstruction filter; and a slice thickness of 5 mm. The external validation set was conducted on a Toshiba Aquilion ONE scanner, employing automatic tube current modulation for the tube current and a FOV of 400 mm. Other imaging parameters, including tube voltage, matrix size, reconstruction filter, and slice thickness, remained consistent with those used in the Brilliance 16 CT. Interpretation of abdominal CT scan images Two radiologists, each with over 10 years of experience in abdominal imaging diagnosis, initially interpreted the abdominal CT plain scans of patients, focusing on signs such as displaced intimal flap, displaced calcification, and high attenuation area. Subsequently, these radiologists reviewed the abdominal CT contrast-enhanced images and categorized all patients into either the AS group or non-AS group based on the presence of AS. In cases of diagnostic disagreement between the two radiologists, a third radiologist with more than 20 years of experience in abdominal imaging diagnosis was responsible for making the final decisions. Patients were stratified and randomly sampled according to their grouping, and then divided in a 3:1:1 ratio into training, test, and external validation sets. Image segmentation The volume of interest (VOI) for the abdominal aorta was segmented from abdominal CT images using ITK-SNAP software (version 5.0.0). The CT images in DICOM format were imported into the software, and manual segmentation was performed layer by layer to meticulously delineate the boundary of the abdominal aorta, encompassing calcified plaques. An experienced radiologist with five years of experience in abdominal imaging diagnosis initially delineated all VOIs of the abdominal aorta. To ensure reliability and reproducibility, this delineation process was repeated one month later by the same radiologist and another radiologist with a decade of experience in abdominal imaging diagnosis. Intraclass correlation coefficients (ICCs) were calculated to assess the reliability of radiomic feature extraction, evaluating intra-observer and inter-observer consistency. Radiomic features extraction A comprehensive suite of 1794 radiomic features was extracted utilizing the MATLAB (version 2022b). These features were systematically categorized into several key groups: first-order features, shape features, Gray Level Co-occurrence Matrix (GLCM), Gray Level Size Zone Matrix (GLSZM), Gray Level Run Length Matrix (GLRLM), Neighbouring Gray Tone Difference Matrix (NGTDM), and Gray Level Dependence Matrix (GLDM). The extraction process integrated a variety of filter parameters including wavelet transforms, laplacian of gaussian (LoG), square, square root, logarithmic, exponential, gradient, 2D local binary pattern (LBP2D), and 3D local binary pattern (LBP3D). The chosen bin width for the analysis was set at 25, with the voxel dimensions uniformly resampled to a 3 ´ 3 ´ 3 grid. In the application of the LoG filter, kernel sizes varied, ranging from 1 to 5. Radiomic model construction Radiomic features demonstrating intra-observer and inter-observer ICCs both exceeding 0.90 were retained for analysis. The preserved feature data underwent preprocessing, including outlier and missing value replacement with median values, followed by normalization to mitigate dimensionality effects. Within the training set, the top 10 features exhibiting high relevance to the target variable and minimal redundancy among themselves were selected using the minimum redundancy and maximum relevance (mRMR) method. Subsequently, a radiomic model was constructed using the least absolute shrinkage and selection operator (LASSO) logistic regression. The radiomics score (Rad-score) for each patient was calculated using the formula derived from the radiomic model. The Hosmer-Lemeshow test was employed to evaluate the potential overfitting of the model. The contribution of features in constructing the radiomic model to predictive outcomes was interpreted using SHapley Additive exPlanations (SHAP) analysis. SHAP provides an approach to elucidate the predictions made by machine learning models, assisting in comprehending the impact of each feature on the model output. Development of a clinical-radiomics model To evaluate differences in clinical factors between the AS and non-AS groups, univariate and multivariable analyses were conducted. Subsequently, independent clinical factors identified were utilized to establish a clinical model predicting abdominal AS via logistic regression. A combined clinical-radiomics model was constructed, incorporating both the radiomic features from the radiomic model and the clinical factors from the clinical model. This integrated model was initially developed using the training set and subsequently assessed in both the test and external validation sets. The workflow is shown in Figue 2. Statistical analysis Statistical analysis for this study was performed using R software (version 3.6.1). For comparing continuous variables, the independent sample t-test or Mann–Whitney U test was applied. Categorical variables were analyzed using the chi-square test or Fisher’s exact test. Receiver operating characteristic (ROC) analysis was conducted for all models, with the area under the curve (AUC) measuring their predictive performance. The Delong test was utilized to determine significant differences in AUC between models. Statistical significance was set at a two-sided p-value of less than 0.05. Results Experiment cohort The diagnosis of AS was confirmed in 66 out of the total cohort of 188 emergency patients. Within the AS group, AD was identified in 44 cases, IMH in 9 cases, and PAU in 5 cases. Notably, one patient presented with both AD and PAU, another with both AD and IMH, and four patients had concurrent IMH and PAU. From organization A (anonymous), 139 cases were segregated into a training group and a test group in a 3:1 ratio. Additionally, 39 patients from organization B (anonymous) served as an external validation group. There were significant differences between the AS group and the non-AS group in terms of gender, abdominal pain, displaced intimal flap, displaced calcification, and high attenuation area ( p 0.05). Table 1 presents the clinical factors of the participants across the training, test, and external validation sets. Radiomic model analysis In this study, a total of 1569 features showed excellent intra- and inter-observer consistency, with ICCs exceeding 0.9. Subsequent feature selection was conducted using the mRMR and LASSO algorithms, ultimately retaining 10 features for the construction of radiomic labels (Fig 2a). These features were then used to calculate the Rad-score for each patient. The AUC values of the radiomic model were 0.827, 0.865, and 0.891 for the training, test, and external validation sets, respectively (Table 2, Fig 3). The Hosmer-Lemeshow test indicated no evidence of overfitting in the radiomic model across the training, test, and external validation sets ( P > 0.05). SHAP (SHapley Additive exPlanations) methodology offered a quantitative interpretation of the radiomic model utilized in our analysis. The summary plots of SHAP, depicted in a concise visual format, highlighted the range and distribution of the importance each feature held in influencing the model's output. This approach also correlated the actual value of a feature with its relative impact on the model. The features were organized based on their overall significance, with the SHAP value of each feature from individual patients plotted horizontally. These points were vertically aligned to indicate the concentration of similar SHAP values. Color coding was used to represent the feature values, ranging from low (blue) to high (red). As illustrated in Fig. 2b, the radiomic feature wavelet-LHL_GLDM_SmallDependenceLowGrayLevelEmphasis (SDLGLE) emerged as the most crucial in distinguishing abdominal AS. This radiomic feature's plot revealed diverse SHAP values within the patient cohort. Notably, the color gradient demonstrated that lower values of this feature were associated with a higher model output, emphasizing its pivotal role in the diagnostic process. Clinical and clinical-radiomics model models In a univariate logistic analysis, abdominal pain, intimal flap and calcified plaque migration significantly influenced the prediction of abdominal AS. Multivariate logistic analysis further identified abdominal pain and calcified plaque migration as independent clinical risk factors for abdominal AS, leading to the development of a clinical model. The AUC of this clinical model was 0.776, 0.793, and 0.763 in the training, test, and external validation sets, respectively (Table 2, Fig 3). Additionally, a multifactorial logistic regression model incorporating Ras-score, intimal flap, and intramural hematoma was established. The clinical-radiomic model demonstrated excellent diagnostic capability, with AUCs of 0.866, 0.920, and 0.891 in the training, test, and external validation sets, respectively (Table 2, Fig 3). According to the Hosmer-Lemeshow test, the clinical-radiomics model showed no overfitting in the training, test, and external validation sets ( P > 0.05). Model comparison The results of the external validation group were used to assess the diagnostic capability of each model for abdominal AS. According to the Delong test, the AUC of both the radiomic model and the clinical-radiomics model was significantly higher than that of the clinical model alone ( P both < 0.001). The accuracy, sensitivity, specificity, negative predictive value, and positive prediction value of both the radiomic model and the clinical-radiomics model demonstrated consistency. This indicates that incorporating clinical features into the radiomic model did not alter the threshold but only had a slight impact on the overall performance of the model. This is consistent with the finding that there was no significant difference in AUC between the radiomic model and the clinical-radiomics model ( P > 0.05). Discussion In this study, the application of radiomics to abdominal CT scans has been demonstrated to effectively diagnose abdominal AS, exhibiting notably superior diagnostic efficacy compared to the clinical model and comparable diagnostic capabilities to the clinical-radiomics model. The radiomic model based on abdominal plain CT scans shows high sensitivity (0.929), although its specificity (0.760) is relatively lower. The proposed model is in good alignment with the screening requirements for AS in emergency patients. The high sensitivity of the radiomic model is crucial in preventing misdiagnoses of abdominal AS, which could lead to severe consequences or even mortality. Overall, the radiomic model based on abdominal plain CT scans meets the needs of emergency departments for screening abdominal AS. The urgency in treating AD is underscored by its steep mortality rate of approximately 1% per hour before receiving appropriate treatment 17 . The prognosis for other forms of AS, such as IMH of the descending aorta and symptomatic PAU, is equally concerning. The short-term mortality rate for patients diagnosed with IMH reaches approximately 20% 18-20 . Furthermore, the rate of rupture at the time of hospital admission in patients with PAU exceeds 30% 21,22 . Given these risks, the ability to rapidly identify abdominal AS through emergency CT scans becomes crucial. Timely diagnosis of abdominal AS is of significant importance for improving patient outcomes. The causes of abdominal pain in patients are numerous, and based solely on medical history, physical examination, and laboratory test results, accurate diagnosis can only be made for a small portion of patients 23 . Due to its rapid image acquisition speed and high spatial resolution, CT plain scan is widely utilized in emergency patients. Abdominal CT scan can provide rapid and accurate diagnosis for specific diseases, such as appendicitis and diverticulitis 24,25 . However, the diagnostic capabilities of an abdominal CT plain scan in identifying abdominal AS are considerably limited. The clinical model we constructed, based on clinical baseline characteristics and CT plain imaging manifestations of AS, did not achieve the expected diagnostic performance, with an AUC of only 0.763. The highly sensitive radiomic model developed in this study can assist emergency physicians and radiologists in the rapid diagnosis of abdominal AS, thereby reducing patient mortality through the facilitation of early intervention. The CT density of blood is influenced by the types and corresponding concentrations of its components 26 . When there are alterations in the distribution of blood components within the vasculature due to changes in blood flow, it leads to a modification in the heterogeneity of blood either within the affected area or throughout the entire vessel. Visual identification of this heterogeneity generated under different hemodynamic conditions in CT images is typically challenging; however, the blood heterogeneity generated by AD in the chest has been found to be able to be simulated by radiomics features and used for prediction 27 . This study not only developed a radiomic model for identifying abdominal AS but also conducted a SHAP analysis on the model. The SHAP analysis revealed that the feature wavelet-LHL_GLDM_SDLGLE significantly contributes to the predictive accuracy of the radiomic model. This feature combines wavelet transforms and the analysis of texture through the SDLGLE metric, a component derived from the GLDM 28 . Specifically, it employs wavelet transforms in the LHL (Low-High-Low frequency) sub-band to identify subtle structural variations in images. Following this, the SDLGLE metric, which originates from GLDM, is applied to these sub-bands. This step involves assessing the joint distribution of areas in the image where there is a small dependence between pixel values and where these values are of lower gray-levels. Essentially, SDLGLE from GLDM is used to analyze the textures captured by the LHL sub-band wavelet transforms. A lower value of the wavelet-LHL_GLDM_SDLGLE feature is indicative of a more heterogeneous distribution of blood components in CT images, which may signal a higher risk of abdominal AS. The current study is subject to several limitations. First and foremost, the study was conducted with a relatively small cohort, necessitating larger-scale studies to further validate the diagnostic performance of the radiomic model. In addition, the selection criteria of this study, which required abdominal plain and contrast-enhanced CT scans for emergency patients, may have inadvertently led to a higher incidence rate of abdominal AS than is typically observed in the broader population, thereby introducing a potential selection bias. Finally, the process of manually outlining the VOI in the abdominal aorta proved to be labor-intensive. The development and application of an intelligent segmentation algorithm for the abdominal aorta in CT images could potentially address this issue. Conclusions In conclusion, this study integrated radiomics with abdominal CT plain scans for application in emergency settings and demonstrated that the radiomic model exhibits substantial diagnostic capability for abdominal AS even without incorporating clinical characteristics. With its high sensitivity, the radiomic model is suitable for emergency clinical contexts and may aid clinicians in promptly diagnosing abdominal AS to enhance patient prognosis. Abbreviations AD aortic dissection, AS aortic syndrome, AUC area under the curves, CT computerized tomography, FOV field of view, GLCM Gray Level Co-occurrence Matrix, GLDM Gray Level Dependence Matrix, GLSZM Gray Level Size Zone Matrix, GLRLM Gray Level Run Length Matrix, ICC intraclass correlation coefficient, IMH intramural hematoma, LBP2D 2D local binary pattern, LBP3D 3D local binary pattern, LoG laplacian of gaussian, NGTDM Neighbouring Gray Tone Difference Matrix, PAU penetrating aortic ulcer, ROC receiver operating characteristic, SDLGLE SmallDependenceLowGrayLevelEmphasis, SHAP SHapley Additive exPlanations, VOI volume of interest. Declarations Competing interests The authors declare no competing interests. Ethics aproval The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University(No.2024-K-046-01) ,and informed consent was taken from all individual participants. Funding This work was supported by Wenzhou Science and Technology Bureau in China (Y20220450). Author Contribution W.X.L. and J.H.S. designed the study and drafted the manuscript. J.H.S. and Y.F.G. provided study materials or patients.Y.F.G.,H.N.Z.,M.Y.S. and J.H.S. collected and analyzed the data. W.X.L. and J.H.S. provided critical revisions for intellectual content. All authors reviewed and approved the final manuscript. Data Availability The datasets generated during and/or analysed during the current study are not publicly available due to the protection of hospital and patient data but are available from the corresponding author on reasonable request. References Bossone, E., LaBounty, T. M. & Eagle, K. A. Acute aortic syndromes: diagnosis and management, an update. Eur Heart J 39 , 739–749d, doi:10.1093/eurheartj/ehx319 (2018). Salmasi, M. Y. et al. The risk of misdiagnosis in acute thoracic aortic dissection: a review of current guidelines. Heart 106 , 885–891, doi:10.1136/heartjnl-2019-316322 (2020). Erbel, R. et al. 2014 ESC Guidelines on the diagnosis and treatment of aortic diseases: Document covering acute and chronic aortic diseases of the thoracic and abdominal aorta of the adult. The Task Force for the Diagnosis and Treatment of Aortic Diseases of the European Society of Cardiology (ESC). Eur Heart J 35 , 2873–2926, doi:10.1093/eurheartj/ehu281 (2014). Clough, R. E. & Nienaber, C. A. Management of acute aortic syndrome. Nat Rev Cardiol 12 , 103–114, doi:10.1038/nrcardio.2014.203 (2015). Tolenaar, J. L. et al. The differences and similarities between intramural hematoma of the descending aorta and acute type B dissection. J Vasc Surg 58 , 1498–1504, doi:10.1016/j.jvs.2013.05.099 (2013). Nathan, D. P. et al. Presentation, complications, and natural history of penetrating atherosclerotic ulcer disease. J Vasc Surg 55 , 10–15, doi:10.1016/j.jvs.2011.08.005 (2012). Cartwright, S. L. & Knudson, M. P. Evaluation of acute abdominal pain in adults. Am Fam Physician 77 , 971–978 (2008). Takayama, Y. et al. Etiologies and outcomes of emergency surgery for acute abdominal pain: an audit of 1456 cases in a single center. Eur J Trauma Emerg Surg 46 , 363–369, doi:10.1007/s00068-018-1051-3 (2020). Pape, L. A. et al. Presentation, Diagnosis, and Outcomes of Acute Aortic Dissection: 17-Year Trends From the International Registry of Acute Aortic Dissection. J Am Coll Cardiol 66 , 350–358, doi:10.1016/j.jacc.2015.05.029 (2015). Guo, Y. et al. Non-contrast CT-based radiomic signature for screening thoracic aortic dissections: a multicenter study. Eur Radiol 31 , 7067–7076, doi:10.1007/s00330-021-07768-2 (2021). Hata, A. et al. Deep learning algorithm for detection of aortic dissection on non-contrast-enhanced CT. Eur Radiol 31 , 1151–1159, doi:10.1007/s00330-020-07213-w (2021). Lambin, P. et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer 48 , 441–446, doi:10.1016/j.ejca.2011.11.036 (2012). Gillies, R. J., Kinahan, P. E. & Hricak, H. Radiomics: Images Are More than Pictures, They Are Data. Radiology 278 , 563–577, doi:10.1148/radiol.2015151169 (2016). Klipstein, R. H., Firmin, D. N., Underwood, S. R., Rees, R. S. & Longmore, D. B. Blood flow patterns in the human aorta studied by magnetic resonance. Br Heart J 58 , 316–323, doi:10.1136/hrt.58.4.316 (1987). Kilner, P. J., Yang, G. Z., Mohiaddin, R. H., Firmin, D. N. & Longmore, D. B. Helical and retrograde secondary flow patterns in the aortic arch studied by three-directional magnetic resonance velocity mapping. Circulation 88 , 2235–2247, doi:10.1161/01.cir.88.5.2235 (1993). Hanna, T. N. et al. The Effects of Fatigue From Overnight Shifts on Radiology Search Patterns and Diagnostic Performance. J Am Coll Radiol 15 , 1709–1716, doi:10.1016/j.jacr.2017.12.019 (2018). Mészáros, I. et al. Epidemiology and clinicopathology of aortic dissection. Chest 117 , 1271–1278, doi:10.1378/chest.117.5.1271 (2000). Maraj, R., Rerkpattanapipat, P., Jacobs, L. E., Makornwattana, P. & Kotler, M. N. Meta-analysis of 143 reported cases of aortic intramural hematoma. Am J Cardiol 86 , 664–668, doi:10.1016/s0002-9149(00)01049-3 (2000). Evangelista, A. et al. Prognostic value of clinical and morphologic findings in short-term evolution of aortic intramural haematoma. Therapeutic implications. Eur Heart J 25 , 81–87, doi:10.1016/j.ehj.2003.10.011 (2004). von Kodolitsch, Y. et al. Intramural hematoma of the aorta: predictors of progression to dissection and rupture. Circulation 107 , 1158–1163, doi:10.1161/01.cir.0000052628.77047.ea (2003). Tittle, S. L. et al. Midterm follow-up of penetrating ulcer and intramural hematoma of the aorta. J Thorac Cardiovasc Surg 123 , 1051–1059, doi:10.1067/mtc.2002.121681 (2002). Chou, A. S. et al. Long-term behavior of aortic intramural hematomas and penetrating ulcers. J Thorac Cardiovasc Surg 151 , 361–372, 373.e361, doi:10.1016/j.jtcvs.2015.09.012 (2016). Stoker, J., van Randen, A., Laméris, W. & Boermeester, M. A. Imaging patients with acute abdominal pain. Radiology 253 , 31–46, doi:10.1148/radiol.2531090302 (2009). Laméris, W. et al. Graded compression ultrasonography and computed tomography in acute colonic diverticulitis: meta-analysis of test accuracy. Eur Radiol 18 , 2498–2511, doi:10.1007/s00330-008-1018-6 (2008). van Randen, A. et al. Acute appendicitis: meta-analysis of diagnostic performance of CT and graded compression US related to prevalence of disease. Radiology 249 , 97–106, doi:10.1148/radiol.2483071652 (2008). New, P. F. & Aronow, S. Attenuation measurements of whole blood and blood fractions in computed tomography. Radiology 121 , 635–640, doi:10.1148/121.3.635 (1976). Touvron, H. et al. Llama: Open and efficient foundation language models. Preprint at https://arxiv.org/abs/2302.13971 (2023). Zwanenburg, A. et al. The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping. Radiology 295 , 328–338, doi:10.1148/radiol.2020191145 (2020). Tables Table 1 Clinical factors of patients. Clinical factors Entire set (n = 188) Training set (n = 112) Test set (n = 37) Validation set (n = 39) p -value Age, years 60.6 (17.9) 63.0 (19.0) 56.6 (16.2) 57.5 (15.4) 0.866 Gender 0.012 Male 123 73 (65.2) 27 (73.0) 23 (59.0) Female 65 39 (34.8) 10 (27.0) 16 (41.0) Abdominal pain 0.005 Yes 89 51 (45.5) 18 (48.6) 20 (51.3) No 99 61 (54.5) 19 (51.4) 19 (48.7) Displaced intimal flap 0.003 Yes 11 5 (4.5) 5 (13.5) 1 (2.6) No 177 107 (95.5) 32 (86.5) 38 (97.4) Displaced calcification < 0.001 Yes 46 24 (21.4) 10 (27.0) 12 (30.8) No 142 88 (78.6) 27 (73.0) 27 (69.2) High attenuation area 0.037 Yes 6 1 (0.9) 3 (8.1) 2 (5.1) No 182 111 (99.1) 34 (91.9) 37 (94.9) The data represent the patient count, with percentages provided in parentheses where applicable. SD standard deviation. Table 2 The diagnostic performance of radiomic model in the training, test, and external validation sets. Sets AUC ACC SEN SPE PPV NPV Training set Clinical model 0.776 0.777 0.487 0.932 0.792 0.773 Radiomic model 0.827 0.750 0.872 0.685 0.596 0.909 Clinical-radiomics model 0.866 0.804 0.821 0.795 0.681 0.892 Test set Clinical model 0.793 0.757 0.538 0.875 0.700 0.778 Radiomic model 0.865 0.730 0.692 0.750 0.600 0.818 Clinical-radiomics model 0.920 0.865 0.750 0.952 0.923 0.833 External validation set Clinical model 0.763 0.744 0.571 0.840 0.667 0.778 Radiomic model 0.891 0.821 0.929 0.760 0.950 0.684 Clinical-radiomics model 0.860 0.821 0.929 0.760 0.950 0.684 ACC Accuracy, AUC area under the curve, CI confidence interval, NPV negative predictive value, PPV positive prediction value, SEN sensitivity, SPE specificity. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 24 Nov, 2025 Reviews received at journal 19 Nov, 2025 Reviewers agreed at journal 29 Oct, 2025 Reviewers agreed at journal 08 Sep, 2025 Reviews received at journal 23 Aug, 2025 Reviewers agreed at journal 03 Aug, 2025 Reviewers invited by journal 03 Aug, 2025 Editor assigned by journal 24 Jul, 2025 Submission checks completed at journal 17 Jul, 2025 First submitted to journal 17 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7095829","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":495615684,"identity":"d708a39e-11e6-45dd-97c9-3743a6050ae2","order_by":0,"name":"Wenxiao Lin","email":"","orcid":"","institution":"The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Wenxiao","middleName":"","lastName":"Lin","suffix":""},{"id":495615685,"identity":"9b3b4c59-acee-42a2-aa1e-bff6c1eb2324","order_by":1,"name":"Yifan Guo","email":"","orcid":"","institution":"The First School of Clinical Medicine, Zhejiang Chinese Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yifan","middleName":"","lastName":"Guo","suffix":""},{"id":495615686,"identity":"7e9914c2-0a5f-4075-bac1-954c7712869b","order_by":2,"name":"Haonan Zhu","email":"","orcid":"","institution":"The First School of Clinical Medicine, Zhejiang Chinese Medical University","correspondingAuthor":false,"prefix":"","firstName":"Haonan","middleName":"","lastName":"Zhu","suffix":""},{"id":495615687,"identity":"f224d67f-4ef6-450e-a4c3-bf937582b96a","order_by":3,"name":"MengYuan Shen","email":"","orcid":"","institution":"The First School of Clinical Medicine, Zhejiang Chinese Medical University","correspondingAuthor":false,"prefix":"","firstName":"MengYuan","middleName":"","lastName":"Shen","suffix":""},{"id":495615688,"identity":"8cb6b0ce-11ad-428f-a42e-246230b3802c","order_by":4,"name":"Jiehui Su","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1klEQVRIiWNgGAWjYHCChAOMDQw8/BAOMwlaJBtI0MLAANTCYHCAWC3yMxIeHvi447CM8Y3sNAmGCuvEBvazB/BqMbiRkHBw5pnDPGY3crdJMJxJT2zgyUvAr0UiIeEwbxtQy22gFsa2w4kNEjwGhByWcPgvUIvxbJCWf0RoYQA67DDQcB4DaZCWBiK0GJx5kHCwty2dR+L+280WCcfSjdt4cgg4rD0n+cPPNmt7/p6zG298qLGW7Wc/Q8BhDDwJCDaIyUZAPRCwHyCsZhSMglEwCkY2AACa1kl/Nt+yLgAAAABJRU5ErkJggg==","orcid":"","institution":"The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University","correspondingAuthor":true,"prefix":"","firstName":"Jiehui","middleName":"","lastName":"Su","suffix":""}],"badges":[],"createdAt":"2025-07-10 19:38:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7095829/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7095829/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":88489227,"identity":"7fe443d8-84d5-49a1-95dd-5bc5f0302a49","added_by":"auto","created_at":"2025-08-07 04:03:42","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":258140,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the patient enrollment process.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-7095829/v1/a5f809673a21724051bb45b0.png"},{"id":88490431,"identity":"2be43010-d779-4a07-8531-274cffe6077f","added_by":"auto","created_at":"2025-08-07 04:11:42","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":246910,"visible":true,"origin":"","legend":"\u003cp\u003eThe workflow of the developmental process for radiomics model.\u003cem\u003e AUC\u003c/em\u003e area under the curve,\u003cem\u003e LASSO\u003c/em\u003e least absolute shrinkage and selection operator,\u003cem\u003eLBP2D \u003c/em\u003e2D local binary pattern,\u003cem\u003e LBP3D \u003c/em\u003e3D local binary pattern,\u003cem\u003e LoG \u003c/em\u003elaplacian of gaussian, \u003cem\u003emRMR\u003c/em\u003e minimum redundancy and maximum relevance,\u003cem\u003e ROC\u003c/em\u003e receiver operating characteristic, \u003cem\u003eSHAP\u003c/em\u003e SHapley Additive exPlanations.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-7095829/v1/bb774e46680d415c3cabc634.png"},{"id":88491227,"identity":"8d723b1a-eed0-418b-b1f5-fc38ec000b9e","added_by":"auto","created_at":"2025-08-07 04:19:42","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":607788,"visible":true,"origin":"","legend":"\u003cp\u003eThe ROC curves for all models are presented, with the AUC detailed for each. \u003cem\u003eAUC\u003c/em\u003earea under the curve,\u003cem\u003e ROC\u003c/em\u003e receiver operating characteristic.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-7095829/v1/dcd6d95552173fabd58e48f0.png"},{"id":88490433,"identity":"f3111158-5a35-4882-b7fa-62c371cbb038","added_by":"auto","created_at":"2025-08-07 04:11:42","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":663160,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP summary plots were created for the radiomic model, illustrating both the importance of individual features and their collective contributions to diagnostic performance. \u003cem\u003eDN\u003c/em\u003e Dependence Non-Uniformity, \u003cem\u003eGLCM\u003c/em\u003e Gray Level Co-occurrence Matrix, \u003cem\u003eGLDM\u003c/em\u003e Gray Level Dependence Matrix, \u003cem\u003eGLSZM\u003c/em\u003e Gray Level Size Zone Matrix, \u003cem\u003eIMC\u003c/em\u003e Informational Measure of Correlation, \u003cem\u003eLoG\u003c/em\u003eLaplacian of Gaussian, \u003cem\u003eLAHGLE\u003c/em\u003e Large Area High Gray Level Emphasis, \u003cem\u003eMCC\u003c/em\u003eMaximal Correlation Coefficient, \u003cem\u003eSDLGLE\u003c/em\u003e Small Dependence Low Gray Level Emphasis, \u003cem\u003eSHAP\u003c/em\u003e SHapley Additive exPlanations, \u003cem\u003eSZNN\u003c/em\u003e Size-Zone Non-Uniformity Normalized.\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-7095829/v1/83b016156c2efae3c366a868.png"},{"id":88491924,"identity":"0fd9182b-835d-4bbc-ae37-492ae836a592","added_by":"auto","created_at":"2025-08-07 04:27:48","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1462073,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7095829/v1/84a79e01-28db-43c0-a039-3a966d7946b8.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Radiomic Analysis of Abdominal CT Plain Scans for Identifying Aortic Syndrome in Emergency Patients","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAcute aortic syndrome (AS) encompasses a group of critical and life-threatening aortic diseases, including acute aortic dissection (AD), intramural hematoma (IMH), and penetrating aortic ulcer (PAU).\u0026nbsp;The aforementioned conditions exhibit overlapping pathophysiological mechanisms, clinical manifestations, and diagnostic and therapeutic challenges \u003csup\u003e1\u003c/sup\u003e. Due to the non-specific nature of symptoms and signs associated with acute AS, its identification often requires a high degree of clinical suspicion. Misdiagnosis and delayed treatment can significantly exacerbate patient outcomes \u003csup\u003e1,2\u003c/sup\u003e. Acute AD is typically characterized by the sudden onset of severe chest or back pain, a symptom that is also commonly observed in IMH and PAU \u003csup\u003e3,4\u003c/sup\u003e. However, the occurrence of abdominal pain is less frequent \u003csup\u003e3,5,6\u003c/sup\u003e. Patients with abdominal manifestations of acute AS are more prone to being overlooked in emergency scenarios compared to those presenting with chest pain.\u003c/p\u003e\n\u003cp\u003eThe complexity and diversity of etiologies in patients presenting with acute abdominal pain in emergency settings pose significant diagnostic challenges \u003csup\u003e7,8\u003c/sup\u003e. Among these, acute AS is a critical condition that cannot be overlooked. Diagnosing acute AS in the context of acute abdominal pain is particularly challenging due to its less frequent occurrence and the non-specific nature of its symptoms compared to thoracic presentations \u003csup\u003e8,9\u003c/sup\u003e. The abdominal CT plain scan is a commonly used imaging modality for patients presenting with acute abdominal pain. If the utilization of abdominal computerized tomography (CT) plain scans can achieve accurate diagnosis of AS, it would facilitate prompt therapeutic interventions for patients, consequently improving their prognostic outcomes. The integration of advanced artificial intelligence technology with abdominal CT plain scan holds promise for achieving immediate diagnosis of acute AS during the scanning process, potentially mitigating severe adverse outcomes caused by missed or incorrect diagnoses.\u003c/p\u003e\n\u003cp\u003eRecent studies have demonstrated that the application of radiomics or deep learning based on non-contrast chest CT scans can accurately predict AD \u003csup\u003e10,11\u003c/sup\u003e. Radiomics is a technique that facilitates the high-throughput extraction of quantitative features from medical images, enabling the conversion of these features into analyzable data \u003csup\u003e12,13\u003c/sup\u003e. This approach utilizes computational methods to identify patterns and characteristics within images that may not be discernible to the human eye. The heterogeneity of blood in the thoracic aorta caused by AD on CT images is different from that in normal arteries, which is thought to be one of the main reasons for the successful identification of AD in the thoracic aorta by radiomics \u003csup\u003e10\u003c/sup\u003e. The blood flow pattern of the abdominal aorta is simpler compared to the complex flow pattern in the thoracic aorta caused by the aortic arch \u003csup\u003e14,15\u003c/sup\u003e. The specific blood heterogeneity caused by AS in the abdominal aorta might be more easily captured by radiomic features. If a radiomic model based on non-contrast abdominal CT can accurately predict abdominal AS, it could alleviate the burden on radiologists dealing with patients presenting with abdominal pain in emergency departments. This is particularly important for patients presenting at night, reducing the risk of missed acute AS diagnoses due to radiologist fatigue \u003csup\u003e16\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe aim of this study is to apply radiomics to abdominal CT plain scans, enabling the immediate diagnosis of abdominal AS in emergency patients following their abdominal CT. This ensures that patients with acute AS receive timely and effective diagnosis and treatment.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003ePatients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was approved by\u0026nbsp;the Ethics Committee of The Second Affiliated Hospital and Yuying Children\u0026apos;s Hospital of Wenzhou Medical \u0026nbsp;University This study retrospectively collected emergency patients who underwent abdominal CT plain scans and contrast-enhanced CT scans at organization A (anonymous) and organization B\u0026nbsp;(anonymous)\u0026nbsp;between August 2012 and October 2020. Exclusion criteria were as follows: 1) The patient underwent separate sessions for the completion of the abdominal CT plain scan and contrast-enhanced CT scans, with a time interval exceeding 24 hours. 2) Metallic artifacts or motion artifacts present in the CT images hindered accurate diagnosis of the aorta. Baseline data collected from patients included gender, age, and abdominal pain. Figure 1 illustrates the flowchart detailing the process of patient enrollment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImage Acquisitions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe abdominal CT scans were performed using two different CT machines. For the training and test sets, a Philips Brilliance 16 CT scanner was utilized with the following parameters: tube voltage of 120 kVp; tube current of 250 mAs; a field of view (FOV) of 350 mm; matrix size of 512 x 512; filtered back projection reconstruction filter; and a slice thickness of 5 mm. The external validation set was conducted on a Toshiba Aquilion ONE scanner, employing automatic tube current modulation for the tube current and a FOV of 400 mm. Other imaging parameters, including tube voltage, matrix size, reconstruction filter, and slice thickness, remained consistent with those used in the Brilliance 16 CT.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInterpretation of abdominal CT scan images\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTwo radiologists, each with over 10 years of experience in abdominal imaging\u0026nbsp;diagnosis, initially interpreted the abdominal CT plain scans of patients, focusing on signs such as\u0026nbsp;displaced intimal flap, displaced calcification, and high attenuation area. Subsequently, these radiologists reviewed the abdominal CT contrast-enhanced images and categorized all patients into either the AS group or non-AS group based on the presence of AS. In cases of diagnostic disagreement between the two radiologists, a third radiologist with more than 20 years of experience in abdominal imaging diagnosis was responsible for making the final decisions. Patients were stratified and randomly sampled according to their grouping, and then divided in a 3:1:1 ratio into training, test, and external validation sets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImage segmentation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe volume of interest (VOI) for the abdominal aorta was segmented from abdominal CT images using ITK-SNAP software (version 5.0.0). The CT images in DICOM format were imported into the software, and manual segmentation was performed layer by layer to meticulously delineate the boundary of the abdominal aorta, encompassing calcified plaques. An experienced radiologist with five years of experience in abdominal imaging diagnosis initially delineated all VOIs of the abdominal aorta. To ensure reliability and reproducibility, this delineation process was repeated one month later by the same radiologist and another radiologist with a decade of experience in abdominal imaging diagnosis. Intraclass correlation coefficients (ICCs) were calculated to assess the reliability of radiomic feature extraction, evaluating intra-observer and inter-observer consistency.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRadiomic features extraction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA comprehensive suite of 1794 radiomic features was extracted utilizing the MATLAB (version 2022b). These features were systematically categorized into several key groups: first-order features, shape features, Gray Level Co-occurrence Matrix (GLCM), Gray Level Size Zone Matrix (GLSZM), Gray Level Run Length Matrix (GLRLM), Neighbouring Gray Tone Difference Matrix (NGTDM), and Gray Level Dependence Matrix (GLDM). The extraction process integrated a variety of filter parameters including wavelet transforms, laplacian of gaussian (LoG), square, square root, logarithmic, exponential, gradient, 2D local binary pattern (LBP2D), and 3D local binary pattern (LBP3D). The chosen bin width for the analysis was set at 25, with the voxel dimensions uniformly resampled to a 3\u0026nbsp;\u0026acute;\u0026nbsp;3\u0026nbsp;\u0026acute;\u0026nbsp;3 grid. In the application of the LoG filter, kernel sizes varied, ranging from 1 to 5.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRadiomic model construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRadiomic features demonstrating intra-observer and inter-observer ICCs both exceeding 0.90 were retained for analysis. The preserved feature data underwent preprocessing, including outlier and missing value replacement with median values, followed by normalization to mitigate dimensionality effects. Within the training set, the top 10 features exhibiting high relevance to the target variable and minimal redundancy among themselves were selected using the minimum redundancy and maximum relevance (mRMR) method. Subsequently, a radiomic model was constructed using the least absolute shrinkage and selection operator (LASSO) logistic regression. The radiomics score (Rad-score) for each patient was calculated using the formula derived from the radiomic model. The Hosmer-Lemeshow test was employed to evaluate the potential overfitting of the model. The contribution of features in constructing the radiomic model to predictive outcomes was interpreted using SHapley Additive exPlanations (SHAP)\u0026nbsp;analysis. SHAP provides an approach to elucidate the predictions made by machine learning models, assisting in comprehending the impact of each feature on the model output.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDevelopment of a clinical-radiomics model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo evaluate differences in clinical factors between the AS and non-AS groups, univariate and multivariable analyses were conducted. Subsequently, independent clinical factors identified were utilized to establish a clinical model predicting abdominal AS via logistic regression. A combined clinical-radiomics model was constructed, incorporating both the radiomic features from the radiomic model and the clinical factors from the clinical model. This integrated model was initially developed using the training set and subsequently assessed in both the test and external validation sets. The workflow is shown in Figue 2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStatistical analysis for this study was performed using R software (version 3.6.1). For comparing continuous variables, the independent sample t-test or Mann\u0026ndash;Whitney U test was applied. Categorical variables were analyzed using the chi-square test or Fisher\u0026rsquo;s exact test. Receiver operating characteristic (ROC) analysis was conducted for all models, with the area under the curve (AUC) measuring their predictive performance. The Delong test was utilized to determine significant differences in AUC between models. Statistical significance was set at a two-sided p-value of less than 0.05.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eExperiment cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe diagnosis of AS was confirmed in 66 out of the total cohort of 188 emergency patients. Within the AS group, AD was identified in 44 cases, IMH in 9 cases, and PAU in 5 cases. Notably, one patient presented with both AD and PAU, another with both AD and IMH, and four patients had concurrent IMH and PAU. From organization A (anonymous), 139 cases were segregated into a training group and a test group in a 3:1 ratio. Additionally, 39 patients from organization B (anonymous) served as an external validation group. There were significant differences between the AS group and the non-AS group in terms of gender, abdominal pain, displaced intimal flap, displaced calcification, and high attenuation area (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05). The age difference between the AS group and the non-AS group was not statistically significant (\u003cem\u003ep\u003c/em\u003e \u0026gt; 0.05).\u0026nbsp;Table 1 presents the clinical factors of the participants across the training, test, and external validation sets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRadiomic model analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, a total of 1569 features showed excellent intra- and inter-observer consistency, with ICCs exceeding 0.9. Subsequent feature selection was conducted using the mRMR and LASSO algorithms, ultimately retaining 10 features for the construction of radiomic labels (Fig 2a). These features were then used to calculate the Rad-score for each patient. The AUC values of the radiomic model were 0.827, 0.865, and 0.891 for the training, test, and external validation sets, respectively (Table 2, Fig 3). The Hosmer-Lemeshow test indicated no evidence of overfitting in the radiomic model across the training, test, and external validation sets (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSHAP (SHapley Additive exPlanations) methodology offered a quantitative interpretation of the radiomic model utilized in our analysis. The summary plots of SHAP, depicted in a concise visual format, highlighted the range and distribution of the importance each feature held in influencing the model\u0026apos;s output. This approach also correlated the actual value of a feature with its relative impact on the model. The features were organized based on their overall significance, with the SHAP value of each feature from individual patients plotted horizontally. These points were vertically aligned to indicate the concentration of similar SHAP values. Color coding was used to represent the feature values, ranging from low (blue) to high (red). As illustrated in Fig. 2b, the radiomic feature wavelet-LHL_GLDM_SmallDependenceLowGrayLevelEmphasis (SDLGLE) emerged as the most crucial in distinguishing abdominal AS. This radiomic feature\u0026apos;s plot revealed diverse SHAP values within the patient cohort. Notably, the color gradient demonstrated that lower values of this feature were associated with a higher model output, emphasizing its pivotal role in the diagnostic process.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical and clinical-radiomics model models\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn a univariate logistic analysis, abdominal pain, intimal flap and calcified plaque migration significantly influenced the prediction of abdominal AS. Multivariate logistic analysis further identified abdominal pain and calcified plaque migration as independent clinical risk factors for abdominal AS, leading to the development of a clinical model. The AUC of this clinical model was 0.776, 0.793, and 0.763 in the training, test, and external validation sets, respectively (Table 2, Fig 3). Additionally, a multifactorial logistic regression model incorporating Ras-score, intimal flap, and intramural hematoma was established. The clinical-radiomic model demonstrated excellent diagnostic capability, with AUCs of 0.866, 0.920, and 0.891 in the training, test, and external validation sets, respectively (Table 2, Fig 3). According to the Hosmer-Lemeshow test, the clinical-radiomics model showed no overfitting in the training, test, and external validation sets (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel comparison\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe results of the external validation group were used to assess the diagnostic capability of each model for abdominal AS. According to the Delong test, the AUC of both the radiomic model and the clinical-radiomics model was significantly higher than that of the clinical model alone (\u003cem\u003eP\u003c/em\u003e both \u0026lt; 0.001). The accuracy, sensitivity, specificity, negative predictive value, and positive prediction value of both the radiomic model and the clinical-radiomics model demonstrated consistency. This indicates that incorporating clinical features into the radiomic model did not alter the threshold but only had a slight impact on the overall performance of the model. This is consistent with the finding that there was no significant difference in AUC between the radiomic model and the clinical-radiomics model (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, the application of radiomics to abdominal CT scans has been demonstrated to effectively diagnose abdominal AS, exhibiting notably superior diagnostic efficacy compared to the clinical model and comparable diagnostic capabilities to the clinical-radiomics model. The radiomic model based on abdominal plain CT scans shows high sensitivity (0.929), although its specificity (0.760) is relatively lower. The proposed model is in good alignment with the screening requirements for AS in emergency patients. The high sensitivity of the radiomic model is crucial in preventing misdiagnoses of abdominal AS, which could lead to severe consequences or even mortality. Overall, the radiomic model based on abdominal plain CT scans meets the needs of emergency departments for screening abdominal AS.\u003c/p\u003e\n\u003cp\u003eThe urgency in treating AD is underscored by its steep mortality rate of approximately 1% per hour before receiving appropriate treatment \u003csup\u003e17\u003c/sup\u003e. The prognosis for other forms of AS, such as IMH of the descending aorta and symptomatic PAU, is equally concerning. The short-term mortality rate for patients diagnosed with IMH reaches approximately 20% \u003csup\u003e18-20\u003c/sup\u003e. Furthermore, the rate of rupture at the time of hospital admission in patients with PAU exceeds 30% \u003csup\u003e21,22\u003c/sup\u003e. Given these risks, the ability to rapidly identify abdominal AS through emergency CT scans becomes crucial.\u0026nbsp;Timely diagnosis of abdominal AS is of significant importance for improving patient outcomes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe causes of abdominal pain in patients are numerous, and based solely on medical history, physical examination, and laboratory test results, accurate diagnosis can only be made for a small portion of patients \u003csup\u003e23\u003c/sup\u003e. Due to its rapid image acquisition speed and high spatial resolution, CT plain scan is widely utilized in emergency patients. Abdominal CT scan can provide rapid and accurate diagnosis for specific diseases, such as appendicitis and diverticulitis \u003csup\u003e24,25\u003c/sup\u003e.\u0026nbsp;However, the diagnostic capabilities of an abdominal CT plain scan in identifying abdominal AS are considerably limited. The clinical model we constructed, based on clinical baseline characteristics and CT plain imaging manifestations of AS, did not achieve the expected diagnostic performance, with an AUC of only 0.763. The highly sensitive radiomic model developed in this study can assist emergency physicians and radiologists in the rapid diagnosis of abdominal AS, thereby reducing patient mortality through the facilitation of early intervention.\u003c/p\u003e\n\u003cp\u003eThe CT density of blood is influenced by the types and corresponding concentrations of its components \u003csup\u003e26\u003c/sup\u003e. When there are alterations in the distribution of blood components within the vasculature due to changes in blood flow, it leads to a modification in the heterogeneity of blood either within the affected area or throughout the entire vessel. Visual identification of this heterogeneity generated under different hemodynamic conditions in CT images is typically challenging; however, the blood heterogeneity generated by AD in the chest has been found to be able to be simulated by radiomics features and used for prediction \u003csup\u003e27\u003c/sup\u003e. This study not only developed a radiomic model for identifying abdominal AS but also conducted a SHAP analysis on the model. The SHAP analysis revealed that the feature wavelet-LHL_GLDM_SDLGLE significantly contributes to the predictive accuracy of the radiomic model. This feature combines wavelet transforms and the analysis of texture through the SDLGLE metric, a component derived from the GLDM \u003csup\u003e28\u003c/sup\u003e. Specifically, it employs wavelet transforms in the LHL (Low-High-Low frequency) sub-band to identify subtle structural variations in images. Following this, the SDLGLE metric, which originates from GLDM, is applied to these sub-bands. This step involves assessing the joint distribution of areas in the image where there is a small dependence between pixel values and where these values are of lower gray-levels. Essentially, SDLGLE from GLDM is used to analyze the textures captured by the LHL sub-band wavelet transforms. A lower value of the wavelet-LHL_GLDM_SDLGLE feature is indicative of a more heterogeneous distribution of blood components in CT images, which may signal a higher risk of abdominal AS.\u003c/p\u003e\n\u003cp\u003eThe current study is subject to several limitations. First and foremost, the study was conducted with a relatively small cohort, necessitating larger-scale studies to further validate the diagnostic performance of the radiomic model. In addition, the selection criteria of this study, which required abdominal plain and contrast-enhanced CT scans for emergency patients, may have inadvertently led to a higher incidence rate of abdominal AS than is typically observed in the broader population, thereby introducing a potential selection bias. Finally, the process of manually outlining the VOI in the abdominal aorta proved to be labor-intensive. The development and application of an intelligent segmentation algorithm for the abdominal aorta in CT images could potentially address this issue.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn conclusion, this study integrated radiomics with abdominal CT plain scans for application in emergency settings and demonstrated that the radiomic model exhibits substantial diagnostic capability for abdominal AS even without incorporating clinical characteristics. With its high sensitivity, the radiomic model is suitable for emergency clinical contexts and may aid clinicians in promptly diagnosing abdominal AS to enhance patient prognosis.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eAD aortic dissection, AS aortic syndrome, AUC area under the curves, CT computerized tomography, FOV field of view, GLCM Gray Level Co-occurrence Matrix, GLDM Gray Level Dependence Matrix, GLSZM Gray Level Size Zone Matrix, GLRLM Gray Level Run Length Matrix, ICC intraclass correlation coefficient, IMH intramural hematoma, LBP2D 2D local binary pattern, LBP3D 3D local binary pattern, LoG laplacian of gaussian, NGTDM Neighbouring Gray Tone Difference Matrix, PAU penetrating aortic ulcer, ROC receiver operating characteristic, SDLGLE SmallDependenceLowGrayLevelEmphasis, SHAP SHapley Additive exPlanations, VOI volume of interest.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003ch2\u003eEthics aproval\u003c/h2\u003e\n\u003cp\u003eThe authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The Second Affiliated Hospital and Yuying Children\u0026apos;s Hospital of Wenzhou Medical University(No.2024-K-046-01) ,and informed consent was taken from all individual participants.\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis work was supported by Wenzhou Science and Technology Bureau in China (Y20220450).\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\n\u003cp\u003eW.X.L. and J.H.S. designed the study and drafted the manuscript. J.H.S. and Y.F.G. provided study materials or patients.Y.F.G.,H.N.Z.,M.Y.S. and J.H.S. collected and analyzed the data. W.X.L. and J.H.S. provided critical revisions for intellectual content. All authors reviewed and approved the final manuscript.\u003c/p\u003e\n\u003ch2\u003eData Availability\u003c/h2\u003e\n\u003cp\u003eThe datasets generated during and/or analysed during the current study are not publicly available due to the protection of hospital and patient data but are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBossone, E., LaBounty, T. M. \u0026amp; Eagle, K. A. Acute aortic syndromes: diagnosis and management, an update. \u003cem\u003eEur Heart J\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 739\u0026ndash;749d, doi:10.1093/eurheartj/ehx319 (2018).\u003c/li\u003e\n\u003cli\u003eSalmasi, M. Y.\u003cem\u003e et al.\u003c/em\u003e The risk of misdiagnosis in acute thoracic aortic dissection: a review of current guidelines. \u003cem\u003eHeart\u003c/em\u003e \u003cstrong\u003e106\u003c/strong\u003e, 885\u0026ndash;891, doi:10.1136/heartjnl-2019-316322 (2020).\u003c/li\u003e\n\u003cli\u003eErbel, R.\u003cem\u003e et al.\u003c/em\u003e 2014 ESC Guidelines on the diagnosis and treatment of aortic diseases: Document covering acute and chronic aortic diseases of the thoracic and abdominal aorta of the adult. The Task Force for the Diagnosis and Treatment of Aortic Diseases of the European Society of Cardiology (ESC). \u003cem\u003eEur Heart J\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 2873\u0026ndash;2926, doi:10.1093/eurheartj/ehu281 (2014).\u003c/li\u003e\n\u003cli\u003eClough, R. E. \u0026amp; Nienaber, C. A. Management of acute aortic syndrome. \u003cem\u003eNat Rev Cardiol\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 103\u0026ndash;114, doi:10.1038/nrcardio.2014.203 (2015).\u003c/li\u003e\n\u003cli\u003eTolenaar, J. L.\u003cem\u003e et al.\u003c/em\u003e The differences and similarities between intramural hematoma of the descending aorta and acute type B dissection. \u003cem\u003eJ Vasc Surg\u003c/em\u003e \u003cstrong\u003e58\u003c/strong\u003e, 1498\u0026ndash;1504, doi:10.1016/j.jvs.2013.05.099 (2013).\u003c/li\u003e\n\u003cli\u003eNathan, D. P.\u003cem\u003e et al.\u003c/em\u003e Presentation, complications, and natural history of penetrating atherosclerotic ulcer disease. \u003cem\u003eJ Vasc Surg\u003c/em\u003e \u003cstrong\u003e55\u003c/strong\u003e, 10\u0026ndash;15, doi:10.1016/j.jvs.2011.08.005 (2012).\u003c/li\u003e\n\u003cli\u003eCartwright, S. L. \u0026amp; Knudson, M. P. Evaluation of acute abdominal pain in adults. \u003cem\u003eAm Fam Physician\u003c/em\u003e \u003cstrong\u003e77\u003c/strong\u003e, 971\u0026ndash;978 (2008).\u003c/li\u003e\n\u003cli\u003eTakayama, Y.\u003cem\u003e et al.\u003c/em\u003e Etiologies and outcomes of emergency surgery for acute abdominal pain: an audit of 1456 cases in a single center. \u003cem\u003eEur J Trauma Emerg Surg\u003c/em\u003e \u003cstrong\u003e46\u003c/strong\u003e, 363\u0026ndash;369, doi:10.1007/s00068-018-1051-3 (2020).\u003c/li\u003e\n\u003cli\u003ePape, L. A.\u003cem\u003e et al.\u003c/em\u003e Presentation, Diagnosis, and Outcomes of Acute Aortic Dissection: 17-Year Trends From the International Registry of Acute Aortic Dissection. \u003cem\u003eJ Am Coll Cardiol\u003c/em\u003e \u003cstrong\u003e66\u003c/strong\u003e, 350\u0026ndash;358, doi:10.1016/j.jacc.2015.05.029 (2015).\u003c/li\u003e\n\u003cli\u003eGuo, Y.\u003cem\u003e et al.\u003c/em\u003e Non-contrast CT-based radiomic signature for screening thoracic aortic dissections: a multicenter study. \u003cem\u003eEur Radiol\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 7067\u0026ndash;7076, doi:10.1007/s00330-021-07768-2 (2021).\u003c/li\u003e\n\u003cli\u003eHata, A.\u003cem\u003e et al.\u003c/em\u003e Deep learning algorithm for detection of aortic dissection on non-contrast-enhanced CT. \u003cem\u003eEur Radiol\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 1151\u0026ndash;1159, doi:10.1007/s00330-020-07213-w (2021).\u003c/li\u003e\n\u003cli\u003eLambin, P.\u003cem\u003e et al.\u003c/em\u003e Radiomics: extracting more information from medical images using advanced feature analysis. \u003cem\u003eEur J Cancer\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, 441\u0026ndash;446, doi:10.1016/j.ejca.2011.11.036 (2012).\u003c/li\u003e\n\u003cli\u003eGillies, R. J., Kinahan, P. E. \u0026amp; Hricak, H. Radiomics: Images Are More than Pictures, They Are Data. \u003cem\u003eRadiology\u003c/em\u003e \u003cstrong\u003e278\u003c/strong\u003e, 563\u0026ndash;577, doi:10.1148/radiol.2015151169 (2016).\u003c/li\u003e\n\u003cli\u003eKlipstein, R. H., Firmin, D. N., Underwood, S. R., Rees, R. S. \u0026amp; Longmore, D. B. Blood flow patterns in the human aorta studied by magnetic resonance. \u003cem\u003eBr Heart J\u003c/em\u003e \u003cstrong\u003e58\u003c/strong\u003e, 316\u0026ndash;323, doi:10.1136/hrt.58.4.316 (1987).\u003c/li\u003e\n\u003cli\u003eKilner, P. J., Yang, G. Z., Mohiaddin, R. H., Firmin, D. N. \u0026amp; Longmore, D. B. Helical and retrograde secondary flow patterns in the aortic arch studied by three-directional magnetic resonance velocity mapping. \u003cem\u003eCirculation\u003c/em\u003e \u003cstrong\u003e88\u003c/strong\u003e, 2235\u0026ndash;2247, doi:10.1161/01.cir.88.5.2235 (1993).\u003c/li\u003e\n\u003cli\u003eHanna, T. N.\u003cem\u003e et al.\u003c/em\u003e The Effects of Fatigue From Overnight Shifts on Radiology Search Patterns and Diagnostic Performance. \u003cem\u003eJ Am Coll Radiol\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 1709\u0026ndash;1716, doi:10.1016/j.jacr.2017.12.019 (2018).\u003c/li\u003e\n\u003cli\u003eM\u0026eacute;sz\u0026aacute;ros, I.\u003cem\u003e et al.\u003c/em\u003e Epidemiology and clinicopathology of aortic dissection. \u003cem\u003eChest\u003c/em\u003e \u003cstrong\u003e117\u003c/strong\u003e, 1271\u0026ndash;1278, doi:10.1378/chest.117.5.1271 (2000).\u003c/li\u003e\n\u003cli\u003eMaraj, R., Rerkpattanapipat, P., Jacobs, L. E., Makornwattana, P. \u0026amp; Kotler, M. N. Meta-analysis of 143 reported cases of aortic intramural hematoma. \u003cem\u003eAm J Cardiol\u003c/em\u003e \u003cstrong\u003e86\u003c/strong\u003e, 664\u0026ndash;668, doi:10.1016/s0002-9149(00)01049-3 (2000).\u003c/li\u003e\n\u003cli\u003eEvangelista, A.\u003cem\u003e et al.\u003c/em\u003e Prognostic value of clinical and morphologic findings in short-term evolution of aortic intramural haematoma. Therapeutic implications. \u003cem\u003eEur Heart J\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 81\u0026ndash;87, doi:10.1016/j.ehj.2003.10.011 (2004).\u003c/li\u003e\n\u003cli\u003evon Kodolitsch, Y.\u003cem\u003e et al.\u003c/em\u003e Intramural hematoma of the aorta: predictors of progression to dissection and rupture. \u003cem\u003eCirculation\u003c/em\u003e \u003cstrong\u003e107\u003c/strong\u003e, 1158\u0026ndash;1163, doi:10.1161/01.cir.0000052628.77047.ea (2003).\u003c/li\u003e\n\u003cli\u003eTittle, S. L.\u003cem\u003e et al.\u003c/em\u003e Midterm follow-up of penetrating ulcer and intramural hematoma of the aorta. \u003cem\u003eJ Thorac Cardiovasc Surg\u003c/em\u003e \u003cstrong\u003e123\u003c/strong\u003e, 1051\u0026ndash;1059, doi:10.1067/mtc.2002.121681 (2002).\u003c/li\u003e\n\u003cli\u003eChou, A. S.\u003cem\u003e et al.\u003c/em\u003e Long-term behavior of aortic intramural hematomas and penetrating ulcers. \u003cem\u003eJ Thorac Cardiovasc Surg\u003c/em\u003e \u003cstrong\u003e151\u003c/strong\u003e, 361\u0026ndash;372, 373.e361, doi:10.1016/j.jtcvs.2015.09.012 (2016).\u003c/li\u003e\n\u003cli\u003eStoker, J., van Randen, A., Lam\u0026eacute;ris, W. \u0026amp; Boermeester, M. A. Imaging patients with acute abdominal pain. \u003cem\u003eRadiology\u003c/em\u003e \u003cstrong\u003e253\u003c/strong\u003e, 31\u0026ndash;46, doi:10.1148/radiol.2531090302 (2009).\u003c/li\u003e\n\u003cli\u003eLam\u0026eacute;ris, W.\u003cem\u003e et al.\u003c/em\u003e Graded compression ultrasonography and computed tomography in acute colonic diverticulitis: meta-analysis of test accuracy. \u003cem\u003eEur Radiol\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 2498\u0026ndash;2511, doi:10.1007/s00330-008-1018-6 (2008).\u003c/li\u003e\n\u003cli\u003evan Randen, A.\u003cem\u003e et al.\u003c/em\u003e Acute appendicitis: meta-analysis of diagnostic performance of CT and graded compression US related to prevalence of disease. \u003cem\u003eRadiology\u003c/em\u003e \u003cstrong\u003e249\u003c/strong\u003e, 97\u0026ndash;106, doi:10.1148/radiol.2483071652 (2008).\u003c/li\u003e\n\u003cli\u003eNew, P. F. \u0026amp; Aronow, S. Attenuation measurements of whole blood and blood fractions in computed tomography. \u003cem\u003eRadiology\u003c/em\u003e \u003cstrong\u003e121\u003c/strong\u003e, 635\u0026ndash;640, doi:10.1148/121.3.635 (1976).\u003c/li\u003e\n\u003cli\u003eTouvron, H.\u003cem\u003e et al.\u003c/em\u003e Llama: Open and efficient foundation language models. Preprint at https://arxiv.org/abs/2302.13971 (2023).\u003c/li\u003e\n\u003cli\u003eZwanenburg, A.\u003cem\u003e et al.\u003c/em\u003e The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping. \u003cem\u003eRadiology\u003c/em\u003e \u003cstrong\u003e295\u003c/strong\u003e, 328\u0026ndash;338, doi:10.1148/radiol.2020191145 (2020).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1\u0026nbsp;\u003c/strong\u003eClinical factors of patients.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"655\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClinical factors\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEntire set\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(n = 188)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTraining set\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(n = 112)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTest set\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(n = 37)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eValidation set\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(n = 39)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003ep\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eAge, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e60.6 (17.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e63.0 (19.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e56.6 (16.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e57.5 (15.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e0.866\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e0.012\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e123\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e73 (65.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e27 (73.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e23 (59.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e39 (34.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e10 (27.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e16 (41.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eAbdominal pain\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e51 (45.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e18 (48.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e20 (51.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e61 (54.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e19 (51.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e19 (48.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eDisplaced intimal flap\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e5 (4.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e5 (13.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e1 (2.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e177\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e107 (95.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e32 (86.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e38 (97.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eDisplaced calcification\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026lt; 0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e24 (21.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e10 (27.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e12 (30.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e88 (78.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e27 (73.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e27 (69.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eHigh attenuation area\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e0.037\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e1 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e3 (8.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e2 (5.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e111 (99.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e34 (91.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e37 (94.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eThe data represent the patient count, with percentages provided in parentheses where applicable. \u003cem\u003eSD\u003c/em\u003e standard deviation.\u003cstrong\u003e\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2\u0026nbsp;\u003c/strong\u003eThe diagnostic performance of radiomic model in the training, test, and external validation sets.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSets\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eACC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSEN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSPE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePPV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNPV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eTraining set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eClinical model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.776\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.777\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.487\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.932\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.792\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.773\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eRadiomic model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.827\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.872\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.685\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.596\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.909\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eClinical-radiomics model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.866\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.795\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.681\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.892\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eTest set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eClinical model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.793\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.757\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.538\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.875\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.700\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.778\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eRadiomic model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.730\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.692\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.600\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.818\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eClinical-radiomics model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.920\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.750\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.952\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.923\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.833\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eExternal validation set\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eClinical model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.763\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.744\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.840\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.667\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.778\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eRadiomic model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.891\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.760\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.950\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.684\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 42px;\"\u003e\n \u003cp\u003eClinical-radiomics model\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.860\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.760\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.950\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 9px;\"\u003e\n \u003cp\u003e0.684\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eACC\u0026nbsp;\u003c/em\u003eAccuracy,\u003cem\u003e\u0026nbsp;AUC\u003c/em\u003e area under the curve, \u003cem\u003eCI\u003c/em\u003e confidence interval, \u003cem\u003eNPV\u003c/em\u003e negative predictive value, \u003cem\u003ePPV\u003c/em\u003e positive prediction value, \u003cem\u003eSEN\u003c/em\u003e sensitivity, \u003cem\u003eSPE\u003c/em\u003e specificity.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Abdomen, Aortic Syndrome, Emergencies, Radiomics, Computed Tomography","lastPublishedDoi":"10.21203/rs.3.rs-7095829/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7095829/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e This study aimed to assess the capability of radiomics based on abdominal CT scans in identifying aortic syndrome (AS) among patients in the emergency department.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e Emergency patients who underwent both plain and contrast-enhanced abdominal CT scans from August 2012 to October 2020 were retrospectively enrolled. These patients were classified based on the presence of abdominal AS. The dataset was randomly split into training, test, and external validation sets in a 3:1:1 ratio. Radiomic features were extracted from manually segmented regions of the abdominal aorta in the CT images. These features were used to create a radiomic model. The radiomic model was integrated with relevant clinical factors using multivariate logistic regression to develop a clinical-radiomics model. The diagnostic performance of the models was assessed through receiver operating characteristic (ROC) analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eA total of 188 patients were included in the study. Ten of 1794 radiomic features were selected for constructing the radiomic model. The SHapley Additive exPlanations (SHAP) analysis identified the wavelet-LHL_GLDM_SDLGLE feature as a significant contributor to the predictive accuracy of the radiomic model. The area under the curves (AUCs) for the clinical, radiomic, and clinical-radiomics models in the external validation set were 0.763, 0.891, and 0.860, respectively. The AUCs of the radiomic and clinical-radiomics models were significantly higher than that of the clinical model (\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05). However, no statistically significant difference was observed between the AUCs of the radiomic model and the clinical-radiomics model (\u003cem\u003eP\u003c/em\u003e \u0026gt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions: \u003c/strong\u003eThe study demonstrated that an abdominal CT plain scan-based radiomic model can effectively be utilized for preliminary diagnosis of abdominal AS in emergency patients.\u003c/p\u003e","manuscriptTitle":"Radiomic Analysis of Abdominal CT Plain Scans for Identifying Aortic Syndrome in Emergency Patients","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-07 04:03:37","doi":"10.21203/rs.3.rs-7095829/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-11-24T06:43:01+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-19T15:47:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"41096984621906209699071516101377046856","date":"2025-10-29T15:01:34+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"171509562111125091016370344323661294450","date":"2025-09-08T07:40:22+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-23T22:34:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"76438622401461106194594598630774707026","date":"2025-08-03T18:00:43+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-03T17:58:16+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-24T05:58:43+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-17T14:45:31+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-07-17T09:28:03+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0d86fee4-7d09-4348-96c6-e2f21a5e3b39","owner":[],"postedDate":"August 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":52636090,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":52636091,"name":"Health sciences/Diseases"},{"id":52636092,"name":"Health sciences/Gastroenterology"},{"id":52636093,"name":"Health sciences/Medical research"}],"tags":[],"updatedAt":"2026-04-29T05:09:46+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-07 04:03:37","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7095829","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7095829","identity":"rs-7095829","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}