Multi-parametric MRI Radiomics Predicts Different HER2 Expression in Breast Cancer

Multi-parametric MRI Radiomics Predicts Different HER2 Expression in Breast Cancer | 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 Research Article Multi-parametric MRI Radiomics Predicts Different HER2 Expression in Breast Cancer Siqi Zhao, Fan Wei, Yuanfei Li, Yueqi Wu, Moyun Zhang, Shuo Wang, and 7 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6646001/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 24 Dec, 2025 Read the published version in Cancer Imaging → Version 1 posted 8 You are reading this latest preprint version Abstract Purpose To develop and validate radiomic models using multi-parametric dynamic contrast-enhanced MRI (DCE-MRI) and intravoxel incoherent movement (IVIM)-based features for the preoperative differentiation of HER2 expressions levels in breast cancer. Materials and Methods This retrospectively study analyzed 227 female breast cancer patients who underwent breast 3.0TMRI examination at our institution from December 2019 to December 2023. The least absolute shrinkage and selection operator (LASSO) ten-fold cross-validation method was used to develop the radiomic features to identify HER2 positive and HER2 negative cancer(task 1), and further identify HER2 low and HER2 zero cancer(task 2). Then the radiomic features were selected and combined with clinical characteristics to construct predicting models using the logistic regression analysis. The area under the receiver operating characteristic curve(AUC), sensitivity, and specificity were used to evaluate the performance of radiomic models. Results For task 1, the AUCs of clinical model (histological grade and peritumoral edema), DCE combined IVIM(D + D*+f) radiomic model and clinic combined radiomic model were 0.785 (95%CI:0.713,0.846), 0.866 (95%CI:0.803,0.915) and 0.903 (95%CI:0.846,0.944) respectively. In the validation cohort, The AUCs were 0.751 (95%CI:0.633,0.848), 0.751 (95%CI:0.633,0.848) and 0.830 (95%CI:0.720,0.910) respectively. For task 2, the AUCs of DCE combined IVIM radiomic model in training and validation cohort were 0.951 (95%CI:0.888,0.984) and 0.853 (95%CI:0.712,0.942) respectively, and the radiomics score was independent predictors of HER2 low cancer. Conclusion The radiomic signature derived from multi-parametric MRI, together with peritumoral edema and histological grade, demonstrated strong performance in predicting HER2 expression preoperatively in breast cancer, which may support individualized treatment strategies. Breast cancer Human epidermal growth factor receptor 2 Magnetic resonance imaging Radiomics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction Breast cancer is one of the most commonly diagnosed malignancies among women worldwide, with its incidence continuing to rise steadily( 1 ). The high degree of heterogeneity in breast cancer presents significant challenges for effective treatment ( 2 ). Therefore, in order to alleviate the substantial disease burden, it is essential to adopt targeted policy interventions and individualized therapeutic strategies. Among the key biomarkers, human epidermal growth factor receptor 2 (HER2) plays a crucial role, as its expression status is associated with different clinical behaviors and treatment responsiveness ( 3 – 5 ). Traditional anti-HER2 targeted therapy have significantly improved both quality of life and overall survival in patients with HER2-positive breast cancer ( 6 – 8 ). Recent studies have shown that trastuzumab deruxtecan(T-DXd), a HER2 directed antibody–drug conjugate, improves the survival in patients with advanced or metastatic HER2 low breast cancer. This has opened a new treatment scheme and drawn increasing attention to the HER2-low subgroup( 9 – 11 ). Currently, HER2 expression is primarily assessed using immunochemistry (IHC) and Fluorescence in situ hybridization (FISH), which are widely accepted for their reliability, repeatability and diagnostic accuracy( 12 ). However, these methods are invasive, and face limitations in accurately distinguishing HER2 low from HER2 zero tumors( 13 , 14 ). To address this challenge, researchers have been exploring noninvasive, quantitative and reproducible techniques to assess HER2 expression. Radiomics has emerged as a promising tool in precision medicine of breast cancer. It enables the extraction of high-throughput quantitative features from medical images, and through machine learning, facilitates automated analysis and predictive modeling for lesion characterization ( 15 ). Although a few studies ( 16 – 19 ) have studied "HER2 low expression" in combination with breast cancer imaging, and attempted to predict HER2 expression levels using imaging modalities such as T2-weighted imaging (T2WI), dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI), or apparent diffusion coefficient (ADC). However, their diagnostic performance remains suboptimal. This may be due to difference in imaging modeling methods or the exclusion of key clinical predictors from the models. Moreover, while DCE-MRI and intravoxel incoherent motion (IVIM) imaging have shown value in distinguishing benign from malignant lesions and in predicting breast cancer prognosis, evidence regarding their utility in differentiating the three HER2 expression states remains limited. This study aims to apply radiomics based on multiparametric MRI, including DCE-MRI and IVIM, to preoperatively predict HER2 expression status in breast cancer. Furthermore, we aim to construct and validate predictive models by integrating radiomic features with clinical and radiological characteristics, thereby identifying the most effective radiomic signature for HER2 stratification. Materials and methods Patients The retrospective study was approved by the institutional ethics committee and the need for informed consent was waived off (IRB number: PJ- KS-KY-2024–69). Consecutive patients with newly diagnosed breast cancer who underwent 3.0T breast MRI examination at our institution between December 2019 and December 2023 were included in the study. The exclusion criteria were as follows: ( 1 ) lesion size smaller than 10mm ( 2 ) Neoadjuvant chemotherapy, vacuum-assisted resection, or puncture biopsy performed prior to MRI ( 3 ) poor imaging quality of MRI or lesions difficulty to delineate ( 4 ) incomplete pathological data. For patients with multiple lesions, the largest pathologically confirmed lesion was selected for analysis. In total, 227 patients with 227 lesions were finally selected for the study. All 227 lesions were analyzed in Task 1 (HER2 positive vs HER2 negative) and 143 lesions were included in task 2 (HER2 low vs HER2 zero)( Fig. 1 ). Clinical and radiological data All histopathological specimens were performed by two experienced breast specialist pathologists who were blind to the clinical and MRI imaging data. Evaluated parameters included histological grade, axillary lymph node status, and immunohistochemical indicators including HER2, estrogen receptor (ER), progesterone receptor (PR), and Ki-67 index. According to the American Society of Clinical Oncology/College of American Pathologists guidelines( 20 ), HER2 positive cancers were defined as IHC 3 + or IHC 2+/ with FISH amplification; HER2 low as IHC 1 + or 2 + without FISH amplification; and HER2 zero as IHC 0. ER, PR status were defined according to international guidelines( 21 ). Hormone receptor (HR) status was defined as positive if ER or PR positive, and negative if both were negative. The f Ki 67 index cutoff was set at 20%( 22 ). Two radiologist (Radiologist A with 5 years and Radiologist B with 20 years of breast imaging experience) independently evaluated MRI images according to BI-RADS 5th edition( 23 ), blinded to clinical or pathological information. The following imaging features were assessed: maximum lesion diameter, intratumoral necrocystosis, fibroglandular tissue content (FGT), background parenchymal enhancement (BPE), enhancement pattern, and time-intensity curve (TIC) type. Peritumoral edema was defined as signal intensity surrounding or posterior to the tumor mass equal to that of water( 24 ). MRI acquisition parameters are presented in Supplementary Table S1 . Image pre-processing and segmentation To minimize variability from different scanners and scanning parameters, image were standardized using Z-score normalization across DCE-MRI (early stage at 90s), and IVIM-derived D, D* and f images. All images were resampled to 1 mm× 1 mm× 1 mm resolution using 3D-Slicer software (version 4.10.1). Tumor segmentation was manually performed using the pre-processed DCE-MRI early and IVIM images (b values: 0, 20, 30, 50, 80,100, 120, 150, 180, 200, 300, 500, 800, 1000, 1200, 1500, 1800, 2000 and 2500 s/mm 2 ). Radiologist A, blinded to pathological results, delineated the regions of interest (ROI) slice by slice, avoiding the surrounding normal tissues, necrosis, bleeding, and liquid components. For IVIM (b value = 1000 s/mm 2 ), a three-dimensional volume of interest (VOI) was delineated and transferred to D, D*, and f maps with manual corrections as needed. To assess reproducibility, sixty randomly selected lesions were independently segmented by Radiologist B, and Radiologist repeated the segmentation one month later following the same procedure to evaluate intra- and inter-observer consistency. Feature extractions and dimension reduction All radiomic features were normalized using Z-score. The dataset was randomly split into training and validation cohort in a 7: 3 ratio (70% training and 30% validation), and features selection was conducted exclusively on the training cohort. The intra-class correlation coefficient (ICC) was used to evaluate the VOI segmentation variability in the radiomic features selection of DCE-MRI early stage, D, D* and f images, and the features of inter-reader and intra-reader ICC > 0.75 were retained. Mann-Whitney U test and Student t test were used to eliminate irrelevant features, focusing on the features with statistically significant differences between Tasks 1 and 2 ( P < 0.05). Spearman correlation analysis was used to delete redundant features. Feature selection based on mutual information further screened out features highly related to the target variable (k = 10). Least Absolute Shrinkage and Selection Operator (LASSO) with ten-fold cross-validation was applied to further refine the feature set by eliminating features with zero coefficient. Model development Logistic regression was used to construct radiomic models based on the selected radiomic features. Single sequence models were developed for DCE, D, D*, and f maps; Multiparametric models included a combined IVIM model (D + D*+f) and a DCE + IVIM model. For each patient, a radiomics score (radscore) was calculated according to the screened optimal radiomic features and their corresponding intercept and regression coefficient in logistic regression model. The best-performing radiomic model was selected for integration with clinical variables. All variables associated with HER2 expression at the P < 0.05 level in univariate analysis were included in multivariate logistic regression analysis to screen independent predictors and construct a clinical model. And multivariate logistic regression analysis continued to be used to combine the optimal radiomic model with the clinical model. The flow chart of radiomics is shown in Fig. 2 . Statistical analysis Mann-Whitney U test and Student t test were used to compare continuous variables. Categorical variables were analyzed using the χ2 test or Fisher exact test. Univariate and multivariate logistic regression analysis were used to screen out independent predictors related to HER2 expression. ICC was used to evaluate readers' consistency segmentation, and ICC ≥ 0.75 meaned excellent consistency. The area under the receiver operating characteristic curve (AUC) and 95% confidence interval were used to evaluate the performance of the models used to identify HER2 expressions, and the specificity and sensitivity of each model were calculated. Delong test was used to compare the performance of different models. Calibration curve and Hosmer-Lemeshow(H-L) test were used to evaluate the fitting degree of each model. Decision curve analysis (DCA) was used to determine the clinical utility of the model. Bilateral P < 0.05 was considered as an indicator of statistically significant difference. All statistical analysis was carried out using SPSS (version 22, IBM) software package, Medcalc (version 15.2.2) software, Python (version 3.7) and R software (version 4.4.2). Results Comparison of clinical and MRI features between HER2 positive and HER2 negative breast cancer In Task 1, the histological grade was significantly different between the training and validation cohorts ( P 0.05)(Table S2 ). In the training cohort, compared with HER2 negative breast cancer, HER2 positive breast cancer showed significantly higher rates of axillary lymph node metastasis and larger maximum diameter of lesion ( P 0.05). Across both cohorts, HER2 positive breast cancer had a lower HR positivity rate, higher Ki-67 expression, higher histological grade, and a greater incidence of peritumoral edema compared to HER2 negative breast cancer ( P 0.05) (Table 1 ). Table 1 Comparison of Clinical Pathological and Conventional Imaging Features of HER2 Positive Group and HER2 Negative Group Variables Training cohort Validation cohort HER2 positive (n = 61) HER2 negative (n = 97) P HER2 positive (n = 23) HER2 negative (n = 46) P Age 51 ± 9 48 ± 9 0.07 51 ± 10 50 ± 9 0.40 HR status 0.000 * 0.002 * Positive 22(36.1) 85(87.6) 14(60.9) 43(93.5) Negative 39(63.9) 12(12.4) 9(39.1) 3(6.5) Ki-67 expression 0.000 * 0.006 * 20% 59(96.7) 72(74.2) 22(95.7) 30(65.2) ALN 0.001 * 0.50 Absent 20(32.8) 59(60.8) 10(43.5) 24(52.2) Present 41(67.2) 38(39.2) 13(56.5) 22(47.8) Histological grade 0.000 * 0.000 * I-II 14(23.0) 58(58.8) 8(34.8) 36(78.3) III 47(77.0) 39(41.2) 15(65.2) 10(21.7) Maximum diameter of lesion 0.000 * 0.09 ≤ 2cm 9(14.8) 47(48.5) 7(30.4) 24(52.2) > 2cm 52(85.2) 50(51.5) 16(69.6) 22(47.8) Intratumoral necrocystosis 0.87 0.17 Absent 32(52.5) 53(53.6) 11(47.8) 30(65.2) Present 29(47.5) 44(45.4) 12(52.2) 16(34.8) Enhancement mode 0.39 0.47 Mass 49(80.3) 83(85.6) 19(82.6) 41(89.1) Non mass/Mass and Non mass 12(19.7) 14(14.4) 4(17.4) 5(10.9) FGT 0.65 0.71 Fatty/Scattered 15(24.6) 27(27.8) 6(26.1) 14(30.4) Heterogeneous/Extreme 46(75.4) 70(72.2) 17(73.9) 32(69.6) BPE 0.19 0.53 Minimal/Mild 59(96.7) 87(89.7) 23(100.0) 43(93.5) Moderate/Marked 2(3.3) 10(10.3) 0(0.0) 3(6.5) TIC 0.84 0.45 I/II 38(62.3) 62(63.9) 15(65.2) 34(73.9) III 23(37.7) 35(36.1) 8(34.8) 12(26.1) Peritumoral edema 0.000 0.002 * Absent 30(49.2) 86(88.7) 13(56.5) 42(91.3) Present 31(50.8) 11(11.3) 10(43.5) 4(8.7) Note: HR,hormone receptor; ALN,Axillary lymph node metastasis; FGT, amount of fibro glandular tissue; BPE, background parenchymal enhancement; TIC,time-signal intensity curve. Radiomic model for distinguishing HER2 positive from HER2 negative breast cancer The Lasso regression diagram for radiomic feature selection from DCE, D, D* and f images is shown in Fig. 3 . In the training cohort, the AUCs based on DCE, D, D* and f single-sequence radiomic models were 0.769 (95%CI: 0.696,0.833), 0.750(95%CI:0.675,0.815), 0.777 (95%CI:0.704,0.839) and 0.772(95%CI:0.699,0.835) respectively. In the validation cohort, AUCs were 0.736 (95%CI: 0.616,0.835), 0.786 (95%CI:0.671,0.876), 0.754 (95%CI:0.636,0.850) and 0.770 (95%CI:0.653,0.863) respectively. The multiparametric IVIM radiomic model achieved AUCs of 0.819 (95%CI:0.750,0.876) and 0.760 (95%CI:0.642,0.855) in the training and validation cohorts, respectively. The combined DCE + IVIM model showed the highest performance in the training cohort with an AUC of 0.866 (95%CI:0.803,0.915) and an AUC of 0.751 (95%CI:0.633,0.848) in the validation cohort (Table S3, Fig. 4 ). According to the Delong test, the DCE combined IVIM model significantly outperformed all single-sequence and IVIM-only models in the training cohort ( P 0.05). A radiomic feature correlation heatmap is provided in Fig. S2 . Clinical model for distinguishing HER2 positive from HER2 negative breast cancer Multivariate logistic regression analysis showed that histological grade (OR = 0.33, 95%CI:0.12,0.92), peritumoral edema (OR = 0.32, 95%CI:0.11,0.93), and radscore (OR = 2.96, 95%CI:1.88,4.68) were independent predictors of HER2 positive breast cancer(Table 2 ). Histological grade and peritumoral edema were used to construct the clinical model. The clinical model yielded AUCs of 0.785 (95%CI:0.713,0.846) and 0.751 (95%CI:0.633,0.848), sensitivity of 0.885 and 0.696, and specificity of 0.546 and 0.739 in the training and validation cohorts, respectively (Table 3 and Fig. 5 ). Table 2 Univariable and Multivariable Logistic Regression Analysis of Training Cohort to Evaluate the Variables Related to HER2 Positive Status Variables Univariable Multivariable OR(95%CI) P OR(95%CI) P HR status 0.25(0.11,0.56) 0.001 1.29(0.41,4.04) 0.67 Ki-67 expression 0.10(0.02,0.43) 0.002 0.08(0.01,1.06) 0.06 ALN 0.31(0.16,0.62) 0.001 0.79(0.29,2.14) 0.64 Histological grade 0.21(0.10,0.43) 0.000 0.33(0.12,0.92) 0.033 * Maximum diameter of lesion 0.18(0.08,0.42) 0.000 0.72(0.23,2.24) 0.57 Peritumoral edema 0.12(0.06,0.28) 0.000 0.32(0.11,0.93) 0.036 * Radscore 3.34(2.23,5.02) 0.000 2.96(1.88,4.68) 0.000 * Note: HR,hormone receptor; ALN,Axillary lymph node; OR,odds ratio; 95CI%,confidence interval. Table 3 Efficiency of Clinical Model, Radiomic Model and Clinic ombined Radiomic Model for Differentiating HER2 Positive and Negative Breast Cancer Models Cohort AUC(95%CI) Sensitivity Specificity Clinical model Training 0.785(0.713,0.846) 0.885 0.546 Validation 0.751(0.633,0.848) 0.696 0.739 Radiomic model Training 0.866(0.803,0.915) 0.672 0.928 Validation 0.751(0.633,0.848) 0.739 0.717 Clinic combined radiomic model Training 0.903(0.846,0.944) 0.918 0.753 Validation 0.830(0.720,0.910) 0.696 0.891 Clinic combined radiomic model for distinguishing HER2 positive from HER2 negative breast cancer The combined radiomic model for predicting HER2 positive breast cancer achieved the highest performance, with an AUC 0.903 (95%CI:0.846,0.944), sensitivity of 0.918, and specificity of 0.753 in the training cohort and an AUC of 0.830 (95%CI:0.720,0.910), sensitivity of 0.696, and specificity of 0.891 in the validation cohort (Table 3 , Fig. 5 ). In the training cohort, Delong test showed the clinical combined radiomic model significantly outperformed both the clinical model and radiomic models ( P 0.05)(Table S5). DCA demonstrated that compared with clinical model and radiomic models, clinic combined radiomic model provided greater net benefit across threshold probabilities, indicating better clinical utility(Fig. S3). Comparison of clinical and MRI characteristics of HER2 low and HER2 zero breast cancer In Task2, there were no significant differences in clinical pathological and conventional imaging features between the training and validation cohorts (Table S6). In the training cohort, HER2 low cancer had higher HR positivity, lower Ki-67 expression, and lower histological grade compared to HER2 zero cancer ( P < 0.05). In the validation cohort, only FGT showed a statistically significant difference between the two group ( P < 0.05)(Table S7). Radiomics model for distinguishing HER2 low from HER2 zero breast cancer The Lasso regression diagram of DCE, D, D* and f images is shown in Fig. 6 . In the training cohort, the AUCs of distinguishing HER2 low cancer from HER2 zero cancer based on DCE, D, D*, f single sequence radiomic model were 0.894 (95%CI:0.817,0.947), 0.838 (95%CI:0.751,0.904), 0.846 (95%CI:0.760,0.911) and 0.829 (95%CI:0.741,0.897), respectively. And in the validation cohort, the same radiomic signatures achieved AUCs of 0.789 (95%CI:0.638,0.898), 0.722 (95%CI:0.565,0.848), 0.827 (95%CI:0.681,0.925) and 0.800(95%CI:0.650,0.906), respectively. The AUCs of the multiparametric IVIM radiomic model in the training and validation cohorts were 0.929 (95%CI:0.859,0.970) and 0.816(95%CI: 0.668,0.917), respectively. The AUCs of DCE combined IVIM radiomic model predicting HER2 low expression were 0.951 (95%CI:0.888,0.984) in the training cohort and 0.853 (95%CI:0.712,0.942) in the validation cohort (Table S8 and Fig. 7 ). Table S9 shows the final screening of 7 DCE-MRI early, 5 D, 5 D* and 5 f radiomic features. In the training cohort, Delong test results showed DCE combined IVIM radiomic model performed well in differentiating HER2 low cancer and HER2 zero cancer, which was higher than other single sequence radiomic models ( P 0.05). In the validation cohort, there was no significant difference between DCE combined with IVIM radiomic model and other single sequence and multiparametric IVIM radiomic models ( P > 0.05)( Table S10). The calibration curve and H-L test confirmed good model fit of DCE combined with IVIM radiomic model in both cohorts ( P > 0.05) (Fig. S4). DCA showed that compared with clinical model and radiomic model, clinic combined radiomic model provides better net income, which indicated that clinic combined radiomic model has better clinical benefits(Fig. S5). Logistic regression analysis was used to evaluate the variables related to the low expression of HER2 in breast cancer. Multivariable logistic analysis showed that in both the training and validation cohorts, only DCE combined with IVIM radscore was an independent predictive factors of HER2 low status (training cohort: OR = 4.50, 95%CI:2.18,9.29; validation cohort: OR = 2.05, 95%CI: 1.30,3.22) (Table 4 ). Table 4 Univariable and Multivariable Logistic Regression Analysis of Training Cohort to Evaluate the Variables Related to HER2 Low Expression Variables Univariable Multivariable OR(95%CI) P OR(95%CI) P Training Cohort HR status 0.12(0.03,0.58) 0.008 0.03(0.001,1.12) 0.06 Ki-67 expression 0.28(0.10,0.77) 0.014 0.31(0.02,4.60) 0.40 Histological grade 3.88(1.67,9.10) 0.002 4.31(0.77,24.10) 0.10 Radscore 3.34(2.23,5.02) 0.000 4.50(2.18,9.29) 0.000 * Validation Cohort FGT 4.62(1.06,20.01) 0.04 3.23(0.55,19.64) 0.19 Radscore 2.10(1.35,3.28) 0.001 2.05(1.30,3.22) 0.002 * Note: HR,hormone receptor; FGT, amount of fibro glandular tissue; BPE, background parenchymal enhancement; TIC,time-signal intensity curve. Discussion This study demonstrated that a clinical model incorporating peritumoral edema and histological grade, when combined with a radiomics model based on multiparametric DCE and IVIM imaging, achieved favorable performance in differentiating the positive and negative expression of HER2 in breast cancer. The combined model yielded AUCs of 0.903 and 0.830 in the training and validation cohorts, respectively. Additionally, we developed a multiparametric MRI radiomic model based on DCE and IVIM sequences to identify HER2 low cancer and HER2 zero cancer. This model achieved excellent diagnostic performance, with AUCs of 0.951 and 0.853 in the training and validation cohorts, respectively, outperforming single-sequence radiomic models. Moreover, the radscore was identified as an independent predictor of HER2 low status. At present, researchers are gradually paying attention to the category of breast cancer with low expression of HER2, and trying to predict three different expression States of HER2 based on multi-parametric MRI radiomics( 16 – 19 ). The multi-parametric MRI radiomic model (including DWI, ADC and DCE sequences) developed by Zheng et al.( 16 ) had a low performance in the differential diagnosis of HER2 positive and HER2 negative breast cancer, and the AUC in the validation cohort was 0.725. Furthermore, the AUCs of the radiomic models for differentiating HER2 low breast cancer from HER2 positive/zero cancer and differentiating HER2 zero from HER2 positive/low breast cancer were 0.782 and 0.813 respectively. Peng et al.( 17 ) constructed a radiomic signature built from T2WI, DWI, ADC and T1 weighted delayed contrast enhanced images to distinguish HER2 zero cancer from HER2 positive/low cancer, and its AUC was 0.780 in the validation cohort, and Nomogram of multi-parametric MRI radiomics combined with pathological features (histochemistry, Ki-67 index, and PR status) was developed for differentiating HER2 positive from HER2 low cancer, with an AUC of 0.800 in the validation cohort. Ramtohul et al.( 18 ) identified HER2 low cancer and HER2 positive cancer based on the radiomic signatures of T2WI and DCE-MRI, and achieved an AUC of 0.800 in the validation cohort. In addition, the study also showed that HER2 low and HER2 zero cancer in luminal and triple negative breast cancer subtypes have similar radiomic features, so it is impossible to distinguish them effectively. Compared with the existing radiomic models, a new multi-parametric IVIM radiomic model has been developed in this study. The activation of HER2 factor is related to the up-regulation of vascular endothelial growth factor, which can further promote the formation of tumor neovascularization( 25 ). IVIM can both reflect the diffusion movement of water molecules and microcirculation perfusion in tissues( 26 ). By combining it with DCE-MRI sequence, the functional information can be complemented and fused. The combination of these multi-parametric MRI radiomic models improves the differential diagnosis ability of the models in the expression level of HER2, and at the same time provides more new imaging biomarkers for clinic, which is helpful to identify the breast cancer patients who are resistant to HER2 treatment more accurately. In this study, we also found that the radiomic features used to distinguish the different expressions of HER2 in breast cancer were mainly wavelet features, and there was cross-sequence feature repetition in multi-parametric MRI sequences, which is consistent with the previous radiomic research result( 27 ). Wavelet features are the intensity and texture features of the original image obtained by wavelet decomposition calculation, which can help us quantify the multi-frequency information of tumors in different wavelet transform frequency ranges, thus reflecting tumor heterogeneity( 28 ). This is consistent with the demonstration of the key role of wavelet features in imageology in the existing literature, which further confirms its importance( 28 , 29 ). The cross-model stability and cross-sequence repeatability of wavelet first-order features suggest that it may be a potential image biomarker and provide a new way for noninvasive quantitative evaluation of HER2 expression level. Secondly, although there are overlapping features, the differences in imaging principles of different MRI sequences may make these features contain specific biological information. Therefore, this study proposes a modeling strategy based on DCE-MRI and IVIM multi-parametric MRI sequences, which can enhance the model's ability to capture the heterogeneity of tumor microenvironment by mining their complementary associations, and finally improve the diagnostic efficiency of HER2 expression level. The current research( 16 – 18 ) mostly focused on the evaluation of HER2 expression level by intratumoral imaging. Peritumoral region also represents a unique microenvironment, which is related to biological information such as tumor growth and invasion, and may be used as a potential biomarker( 30 ). The latest progress in breast cancer research has identified various key findings in the peritumor tissue, such as angiogenesis factors, lymphatic and vascular invasion around the tumor and peritumor edema( 31 ). Bian et al.( 19 ) extracted the intra-tumor and peritumoral features of ADC and T1WI contrast-enhanced imaging to distinguish between HER2 positive and HER2 negative cancer, and between HER2 low and HER2 zero cancer. The research found that the AUC in the external validation cohort was 0.760 and 0.710, respectively. Although they additionally included the analysis of peritumoral information, their diagnostic efficiency was still lower than the results of this study, especially in identifying the more challenging HER2 low and HER2 zero cancer. We speculated that it may be related to the different construction methods of the radiomic model and the different scope of feature extraction around the tumor. In our previous research on peritumoral region and prognosis of breast cancer, we recommend 2–20 mm as a study-worthy peritumoral region range for MR researchers( 32 ). Therefore, in the future, we will continue to explore the best size of peritumoral area in order to better evaluate the peritumoral microenvironment and predict the expression level of HER2 in breast cancer more accurately. The limitations of this study were as follows: firstly, this study was a retrospective single-center design, and the sample size was still small. In the future, we should further expand the sample size and conduct multi-center research. Secondly, although we included two MRI scanners with different equipment, they were all carried out with 3.0T magnetic field intensity. Therefore, it is necessary to measure under different magnetic field strengths to verify the universality of our research. Thirdly, IVIM selected more b values. Although it pays attention to perfusion information, it leads to a long scanning time. In the future, we will further standardize and standardize the key b values. In a word, the results of this study showed that multi-parametric MRI radiomics based on IVIM and DCE-MRI can effectively predict the expression level of HER2 in breast cancer from a non-invasive point of view, thus better assisting pathologists to detect the expression of HER2 and make personalized treatment plans for patients. Abbreviations HER2 human epidermal growth factor receptor 2 DCE-MRI dynamic contrast enhanced-magnetic resonance imaging IVIM intravoxel incoherent motion HR Hormone receptor FGT fibroglandular tissue content BPE background parenchymal enhancement TIC time-intensity curve ICC intra-class correlation coefficient LASSO Least Absolute Shrinkage and Selection Operator AUC area under the receiver operating characteristic curve DCA Decision curve analysis Declarations Ethics approval and consent to participate: The study was approved by the Ethics Committee of First Affiliated Hospital of Dalian Medical University [IRB number: PJ- KS-KY-2024-69]. Given the retrospective nature of the study, written informed consent was not required. Consent for publication: Not applicable. Availability of data and materials: The datasets generated and/or analysed during the current study are not publicly available due [REASON WHY DATA ARE NOT PUBLIC] but are available from the corresponding author on reasonable request. Competing interests: The authors declare that they have no competing interests. Funding: This work were supported by [Research on Digitalization Empowering the Reform of Pathology Teaching] and [A project of the "14th Five-Year Plan" for Educational Science in Liaoning Province]. Authors' contributions: Lina Zhang is the study project leader; she conceived and participated in the study design and coordination. Siqi Zhao conceived the study and drafted the manuscript. Lina Zhang and Siqi Zhao analyzed radiological and clinical data. Fan Wei and Hui Liu provided technical support.Yueqi Wu, Yuanfei Li and Moyun Zhang collected the clinical materials and radiological data. Shuo Wang, Xinyue Yin and Zhitian Guo participated in the sequence alignment and scanning. Xue Gao reviewed the pathological data. Jie Yang and Haonan Guan revised the manuscript and gave statistical support. All authors read and approved the final manuscript. Acknowledgements: Not applicable. References Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229-63. Jokar N, Velikyan I, Ahmadzadehfar H, et al. Theranostic Approach in Breast Cancer: A Treasured Tailor for Future Oncology. Clin Nucl Med. 2021;46(8):e410-e20. Pernas S, Tolaney SM. Targeting HER2 heterogeneity in early-stage breast cancer. Curr Opin Oncol. 2020;32(6):545-54. Li Z, Metzger Filho O, Viale G, et al. HER2 heterogeneity and treatment response-associated profiles in HER2-positive breast cancer in the NCT02326974 clinical trial. J Clin Invest. 2024;134(7). Trastuzumab for early-stage, HER2-positive breast cancer: a meta-analysis of 13 864 women in seven randomised trials. Lancet Oncol. 2021;22(8):1139-50. Godoy-Ortiz A, Sanchez-Muñoz A, Chica Parrado MR, et al. Deciphering HER2 Breast Cancer Disease: Biological and Clinical Implications. Front Oncol. 2019;9:1124. Yoon J, Oh DY. HER2-targeted therapies beyond breast cancer - an update. Nat Rev Clin Oncol. 2024;21(9):675-700. Wuerstlein R, Harbeck N. Neoadjuvant Therapy for HER2-positive Breast Cancer. Rev Recent Clin Trials. 2017;12(2):81-92. Modi S, Jacot W, Yamashita T, et al. Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer. N Engl J Med. 2022;387(1):9-20. Corti C, Giugliano F, Nicolò E, et al. HER2-Low Breast Cancer: a New Subtype? Curr Treat Options Oncol. 2023;24(5):468-78. Tarantino P, Viale G, Press MF, et al. ESMO expert consensus statements (ECS) on the definition, diagnosis, and management of HER2-low breast cancer. Ann Oncol. 2023;34(8):645-59. Press OA, Guzman R, Cervantes M, et al. Characterization of HER2 status by fluorescence in situ hybridization (FISH) and immunohistochemistry (IHC). Methods Mol Biol. 2014;1180:181-207. Robbins CJ, Fernandez AI, Han G, et al. Multi-institutional Assessment of Pathologist Scoring HER2 Immunohistochemistry. Mod Pathol. 2023;36(1):100032. Schnitt SJ, Tarantino P, Collins LC. The American Society of Clinical Oncology-College of American Pathologists Guideline Update for Human Epidermal Growth Factor Receptor 2 Testing in Breast Cancer. Arch Pathol Lab Med. 2023;147(9):991-2. Lee SH, Park H, Ko ES. Radiomics in Breast Imaging from Techniques to Clinical Applications: A Review. Korean J Radiol. 2020;21(7):779-92. Zheng S, Yang Z, Du G, et al. Discrimination between HER2-overexpressing, -low-expressing, and -zero-expressing statuses in breast cancer using multiparametric MRI-based radiomics. Eur Radiol. 2024;34(9):6132-44. Peng Y, Zhang X, Qiu Y, et al. Development and Validation of MRI Radiomics Models to Differentiate HER2-Zero, -Low, and -Positive Breast Cancer. AJR Am J Roentgenol. 2024;222(4):e2330603. Ramtohul T, Djerroudi L, Lissavalid E, et al. Multiparametric MRI and Radiomics for the Prediction of HER2-Zero, -Low, and -Positive Breast Cancers. Radiology. 2023;308(2):e222646. Bian X, Du S, Yue Z, et al. Potential Antihuman Epidermal Growth Factor Receptor 2 Target Therapy Beneficiaries: The Role of MRI-Based Radiomics in Distinguishing Human Epidermal Growth Factor Receptor 2-Low Status of Breast Cancer. J Magn Reson Imaging. 2023;58(5):1603-14. Wolff AC, Somerfield MR, Dowsett M, et al. Human Epidermal Growth Factor Receptor 2 Testing in Breast Cancer: ASCO-College of American Pathologists Guideline Update. J Clin Oncol. 2023;41(22):3867-72. Allison KH, Hammond MEH, Dowsett M, et al. Estrogen and Progesterone Receptor Testing in Breast Cancer: ASCO/CAP Guideline Update. J Clin Oncol. 2020;38(12):1346-66. Goldhirsch A, Winer EP, Coates AS, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol. 2013;24(9):2206-23. Spak DA, Plaxco JS, Santiago L, et al. BI-RADS(®) fifth edition: A summary of changes. Diagn Interv Imaging. 2017;98(3):179-90. Panzironi G, Moffa G, Galati F, et al. Peritumoral edema as a biomarker of the aggressiveness of breast cancer: results of a retrospective study on a 3 T scanner. Breast Cancer Res Treat. 2020;181(1):53-60. Kourea HP, Dimitrakopoulos FI, Koliou GA, et al. Clinical Significance of Major Angiogenesis-Related Effectors in Patients with Metastatic Breast Cancer Treated with Trastuzumab-Based Regimens. Cancer Res Treat. 2022;54(4):1053-1064. Iima M, Honda M, Sigmund EE, et al. Diffusion MRI of the breast: Current status and future directions. J Magn Reson Imaging. 2020;52(1):70-90. Yu Y, Tan Y, Xie C, et al. Development and Validation of a Preoperative Magnetic Resonance Imaging Radiomics-Based Signature to Predict Axillary Lymph Node Metastasis and Disease-Free Survival in Patients With Early-Stage Breast Cancer. JAMA Netw Open. 2020;3(12):e2028086. Wu Q, Wang S, Chen X, et al. Radiomics analysis of magnetic resonance imaging improves diagnostic performance of lymph node metastasis in patients with cervical cancer. Radiother Oncol. 2019;138:141-8. Zhou X, Yi Y, Liu Z, et al. Radiomics-Based Pretherapeutic Prediction of Non-response to Neoadjuvant Therapy in Locally Advanced Rectal Cancer. Ann Surg Oncol. 2019;26(6):1676-84. Jiang W, Meng R, Cheng Y,et al. Intra- and Peritumoral Based Radiomics for Assessment of Lymphovascular Invasion in Invasive Breast Cancer. J Magn Reson Imaging. 2024;59(2):613-625. Ding J, Chen S, Serrano Sosa M, et al. Optimizing the Peritumoral Region Size in Radiomics Analysis for Sentinel Lymph Node Status Prediction in Breast Cancer. Acad Radiol. 2022;29 Suppl 1(Suppl 1):S223-s8. Zhao S, Li Y, Ning N, et al. Association of peritumoral region features assessed on breast MRI and prognosis of breast cancer: a systematic review and meta-analysis. Eur Radiol. 2024;34(9):6108-20. Additional Declarations No competing interests reported. Supplementary Files TableS1S10.docx FigS1S5.docx Cite Share Download PDF Status: Published Journal Publication published 24 Dec, 2025 Read the published version in Cancer Imaging → Version 1 posted Editorial decision: Accepted 17 Dec, 2025 Reviews received at journal 12 Sep, 2025 Reviewers agreed at journal 28 Aug, 2025 Reviewers agreed at journal 21 Aug, 2025 Reviewers invited by journal 14 May, 2025 Editor assigned by journal 14 May, 2025 Submission checks completed at journal 13 May, 2025 First submitted to journal 12 May, 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-6646001","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":456618870,"identity":"dc9cd35f-f110-4577-82e1-6a2dddf9d0ff","order_by":0,"name":"Siqi Zhao","email":"","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Siqi","middleName":"","lastName":"Zhao","suffix":""},{"id":456618871,"identity":"05440236-ed67-4ff2-907e-457d664f7ced","order_by":1,"name":"Fan Wei","email":"","orcid":"","institution":"School of Control Science and Engineering, Dalian University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Fan","middleName":"","lastName":"Wei","suffix":""},{"id":456618872,"identity":"594bba59-de45-4078-a96c-ad93097161c1","order_by":2,"name":"Yuanfei Li","email":"","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yuanfei","middleName":"","lastName":"Li","suffix":""},{"id":456618873,"identity":"49d5f251-1f20-4894-a0d0-82ab3a65f9a4","order_by":3,"name":"Yueqi Wu","email":"","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yueqi","middleName":"","lastName":"Wu","suffix":""},{"id":456618874,"identity":"148f58cf-5b3a-42ea-b22c-29c07913fe8b","order_by":4,"name":"Moyun Zhang","email":"","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Moyun","middleName":"","lastName":"Zhang","suffix":""},{"id":456618875,"identity":"3f21bd2b-cfc5-4659-b298-2a4f85a7e51a","order_by":5,"name":"Shuo Wang","email":"","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shuo","middleName":"","lastName":"Wang","suffix":""},{"id":456618876,"identity":"4a9df571-b744-4b9f-97b0-b0b1f3683e2d","order_by":6,"name":"Xinyue Yin","email":"","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xinyue","middleName":"","lastName":"Yin","suffix":""},{"id":456618877,"identity":"953c2bbd-02fe-4eb7-b1a8-b7b9061a00c9","order_by":7,"name":"Zhitian Guo","email":"","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhitian","middleName":"","lastName":"Guo","suffix":""},{"id":456618878,"identity":"edc6dc85-5d1c-4213-8718-eae1d79d2620","order_by":8,"name":"Jie Yang","email":"","orcid":"","institution":"School of Public Health, Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jie","middleName":"","lastName":"Yang","suffix":""},{"id":456618879,"identity":"218d9497-da6f-4e6f-a94e-77591549738e","order_by":9,"name":"Xue Gao","email":"","orcid":"","institution":"Department of Pathology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xue","middleName":"","lastName":"Gao","suffix":""},{"id":456618880,"identity":"7b18e31e-2042-4159-884a-4958dbc3230d","order_by":10,"name":"Haonan Guan","email":"","orcid":"","institution":"GE Healthcare, MR Research China, Beijing","correspondingAuthor":false,"prefix":"","firstName":"Haonan","middleName":"","lastName":"Guan","suffix":""},{"id":456618881,"identity":"64aab427-6fa6-4749-8eda-fa1966959c44","order_by":11,"name":"Hui Liu","email":"","orcid":"","institution":"School of Control Science and Engineering, Dalian University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Hui","middleName":"","lastName":"Liu","suffix":""},{"id":456618882,"identity":"6a7baf15-7bd7-47ed-b3cf-3f763110f14c","order_by":12,"name":"Lina Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0UlEQVRIie3QvQrCMBDA8SsHdokfY0pBn0A4CSiC1FdpKNTV0c2WQrv4AILv4Vzp4CsoLu0bZHRwMLMgiZtD/tyYH9wFwOX6x3zw2pg4rqs875QVQUBS25VPrCkEtyS94KjSIfFNOWI2YlL065BREy6CrgQO0XiaGYhXDGKhiVieZNluIRHz2rQXMko0SeAhK+JQy7OJ9DRp9Oyz+6XkzIYwZLP8SCnSzbMkHJkARSsMDlJ/MlncMqkOi2f84jj0r12ndtHYSD6i3567XC6X60tvovg8KREQx9UAAAAASUVORK5CYII=","orcid":"","institution":"Department of Radiology, First Affiliated Hospital of Dalian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Lina","middleName":"","lastName":"Zhang","suffix":""}],"badges":[],"createdAt":"2025-05-12 11:08:44","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6646001/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6646001/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s40644-025-00981-y","type":"published","date":"2025-12-24T15:58:35+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":83032396,"identity":"e41018a9-f1c7-4856-83f1-9693ff615d5c","added_by":"auto","created_at":"2025-05-19 09:21:04","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":110862,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/6f8875416d509536533a09f1.png"},{"id":83032398,"identity":"65ad8445-62b2-4355-b567-b73653c47283","added_by":"auto","created_at":"2025-05-19 09:21:04","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":539861,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/7747a63565518a320b75e2d9.png"},{"id":83034017,"identity":"35c5a780-7296-4385-98eb-03d5cc265cca","added_by":"auto","created_at":"2025-05-19 09:37:04","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":341692,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/ff7f849af012d58ad78f9865.jpeg"},{"id":83032401,"identity":"b3548fa5-4671-40b0-aada-72283a371f92","added_by":"auto","created_at":"2025-05-19 09:21:04","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":361481,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/cc2b12d0ac2453fdfcf14f6f.jpeg"},{"id":83032408,"identity":"95ef574f-a6f8-4953-8fad-c2176be5a309","added_by":"auto","created_at":"2025-05-19 09:21:04","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":338403,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/0199e0e25f08b9f4b604f19a.jpeg"},{"id":83032404,"identity":"9f914b9b-7fb6-4136-835b-24e1a6c093ce","added_by":"auto","created_at":"2025-05-19 09:21:04","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":338493,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/c21c0d03aaa2cc0e2d9cfe5b.jpeg"},{"id":83034018,"identity":"73793f28-0b5f-4309-a9c4-ee32e8b1bb55","added_by":"auto","created_at":"2025-05-19 09:37:04","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":395183,"visible":true,"origin":"","legend":"\u003cp\u003eLegend not included with this version\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/1e81ee9c2cac4a27992ce584.jpeg"},{"id":99172529,"identity":"3e417d5f-14a0-4cd1-a2a6-177a860453d7","added_by":"auto","created_at":"2025-12-29 16:10:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3746093,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/18dd0d22-8c39-4961-ab49-df7b774d7acc.pdf"},{"id":83032400,"identity":"76801961-bd4d-4fdb-9ce9-548adcd54342","added_by":"auto","created_at":"2025-05-19 09:21:04","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":47483,"visible":true,"origin":"","legend":"","description":"","filename":"TableS1S10.docx","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/41194aff4b369c84e7e2e516.docx"},{"id":83032416,"identity":"8ffbf16d-6484-4fec-b1f9-04dfb9c63b79","added_by":"auto","created_at":"2025-05-19 09:21:04","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":463763,"visible":true,"origin":"","legend":"","description":"","filename":"FigS1S5.docx","url":"https://assets-eu.researchsquare.com/files/rs-6646001/v1/8c8a764d12e79f34ade2d18a.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Multi-parametric MRI Radiomics Predicts Different HER2 Expression in Breast Cancer","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBreast cancer is one of the most commonly diagnosed malignancies among women worldwide, with its incidence continuing to rise steadily(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). The high degree of heterogeneity in breast cancer presents significant challenges for effective treatment (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Therefore, in order to alleviate the substantial disease burden, it is essential to adopt targeted policy interventions and individualized therapeutic strategies. Among the key biomarkers, human epidermal growth factor receptor 2 (HER2) plays a crucial role, as its expression status is associated with different clinical behaviors and treatment responsiveness (\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Traditional anti-HER2 targeted therapy have significantly improved both quality of life and overall survival in patients with HER2-positive breast cancer (\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). Recent studies have shown that trastuzumab deruxtecan(T-DXd), a HER2 directed antibody\u0026ndash;drug conjugate, improves the survival in patients with advanced or metastatic HER2 low breast cancer. This has opened a new treatment scheme and drawn increasing attention to the HER2-low subgroup(\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Currently, HER2 expression is primarily assessed using immunochemistry (IHC) and Fluorescence in situ hybridization (FISH), which are widely accepted for their reliability, repeatability and diagnostic accuracy(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). However, these methods are invasive, and face limitations in accurately distinguishing HER2 low from HER2 zero tumors(\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo address this challenge, researchers have been exploring noninvasive, quantitative and reproducible techniques to assess HER2 expression. Radiomics has emerged as a promising tool in precision medicine of breast cancer. It enables the extraction of high-throughput quantitative features from medical images, and through machine learning, facilitates automated analysis and predictive modeling for lesion characterization (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAlthough a few studies (\u003cspan additionalcitationids=\"CR17 CR18\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e) have studied \"HER2 low expression\" in combination with breast cancer imaging, and attempted to predict HER2 expression levels using imaging modalities such as T2-weighted imaging (T2WI), dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI), or apparent diffusion coefficient (ADC). However, their diagnostic performance remains suboptimal. This may be due to difference in imaging modeling methods or the exclusion of key clinical predictors from the models. Moreover, while DCE-MRI and intravoxel incoherent motion (IVIM) imaging have shown value in distinguishing benign from malignant lesions and in predicting breast cancer prognosis, evidence regarding their utility in differentiating the three HER2 expression states remains limited.\u003c/p\u003e \u003cp\u003eThis study aims to apply radiomics based on multiparametric MRI, including DCE-MRI and IVIM, to preoperatively predict HER2 expression status in breast cancer. Furthermore, we aim to construct and validate predictive models by integrating radiomic features with clinical and radiological characteristics, thereby identifying the most effective radiomic signature for HER2 stratification.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePatients\u003c/h2\u003e \u003cp\u003e The retrospective study was approved by the institutional ethics committee and the need for informed consent was waived off (IRB number: PJ- KS-KY-2024\u0026ndash;69). Consecutive patients with newly diagnosed breast cancer who underwent 3.0T breast MRI examination at our institution between December 2019 and December 2023 were included in the study.\u003c/p\u003e \u003cp\u003eThe exclusion criteria were as follows: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) lesion size smaller than 10mm (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) Neoadjuvant chemotherapy, vacuum-assisted resection, or puncture biopsy performed prior to MRI (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) poor imaging quality of MRI or lesions difficulty to delineate (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) incomplete pathological data. For patients with multiple lesions, the largest pathologically confirmed lesion was selected for analysis. In total, 227 patients with 227 lesions were finally selected for the study. All 227 lesions were analyzed in Task 1 (HER2 positive vs HER2 negative) and 143 lesions were included in task 2 (HER2 low vs HER2 zero)( Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eClinical and radiological data\u003c/h3\u003e\n\u003cp\u003eAll histopathological specimens were performed by two experienced breast specialist pathologists who were blind to the clinical and MRI imaging data. Evaluated parameters included histological grade, axillary lymph node status, and immunohistochemical indicators including HER2, estrogen receptor (ER), progesterone receptor (PR), and Ki-67 index. According to the American Society of Clinical Oncology/College of American Pathologists guidelines(\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e), HER2 positive cancers were defined as IHC 3\u0026thinsp;+\u0026thinsp;or IHC 2+/ with FISH amplification; HER2 low as IHC 1\u0026thinsp;+\u0026thinsp;or 2\u0026thinsp;+\u0026thinsp;without FISH amplification; and HER2 zero as IHC 0. ER, PR status were defined according to international guidelines(\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). Hormone receptor (HR) status was defined as positive if ER or PR positive, and negative if both were negative. The f Ki 67 index cutoff was set at 20%(\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTwo radiologist (Radiologist A with 5 years and Radiologist B with 20 years of breast imaging experience) independently evaluated MRI images according to BI-RADS 5th edition(\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e), blinded to clinical or pathological information. The following imaging features were assessed: maximum lesion diameter, intratumoral necrocystosis, fibroglandular tissue content (FGT), background parenchymal enhancement (BPE), enhancement pattern, and time-intensity curve (TIC) type. Peritumoral edema was defined as signal intensity surrounding or posterior to the tumor mass equal to that of water(\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). MRI acquisition parameters are presented in Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e\n\u003ch3\u003eImage pre-processing and segmentation\u003c/h3\u003e\n\u003cp\u003eTo minimize variability from different scanners and scanning parameters, image were standardized using Z-score normalization across DCE-MRI (early stage at 90s), and IVIM-derived D, D* and f images. All images were resampled to 1 mm\u0026times; 1 mm\u0026times; 1 mm resolution using 3D-Slicer software (version 4.10.1). Tumor segmentation was manually performed using the pre-processed DCE-MRI early and IVIM images (b values: 0, 20, 30, 50, 80,100, 120, 150, 180, 200, 300, 500, 800, 1000, 1200, 1500, 1800, 2000 and 2500 s/mm\u003csup\u003e2\u003c/sup\u003e). Radiologist A, blinded to pathological results, delineated the regions of interest (ROI) slice by slice, avoiding the surrounding normal tissues, necrosis, bleeding, and liquid components. For IVIM (b value\u0026thinsp;=\u0026thinsp;1000 s/mm\u003csup\u003e2\u003c/sup\u003e), a three-dimensional volume of interest (VOI) was delineated and transferred to D, D*, and f maps with manual corrections as needed. To assess reproducibility, sixty randomly selected lesions were independently segmented by Radiologist B, and Radiologist repeated the segmentation one month later following the same procedure to evaluate intra- and inter-observer consistency.\u003c/p\u003e\n\u003ch3\u003eFeature extractions and dimension reduction\u003c/h3\u003e\n\u003cp\u003eAll radiomic features were normalized using Z-score. The dataset was randomly split into training and validation cohort in a 7: 3 ratio (70% training and 30% validation), and features selection was conducted exclusively on the training cohort. The intra-class correlation coefficient (ICC) was used to evaluate the VOI segmentation variability in the radiomic features selection of DCE-MRI early stage, D, D* and f images, and the features of inter-reader and intra-reader ICC\u0026thinsp;\u0026gt;\u0026thinsp;0.75 were retained. Mann-Whitney U test and Student \u003cem\u003et\u003c/em\u003e test were used to eliminate irrelevant features, focusing on the features with statistically significant differences between Tasks 1 and 2 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Spearman correlation analysis was used to delete redundant features. Feature selection based on mutual information further screened out features highly related to the target variable (k\u0026thinsp;=\u0026thinsp;10). Least Absolute Shrinkage and Selection Operator (LASSO) with ten-fold cross-validation was applied to further refine the feature set by eliminating features with zero coefficient.\u003c/p\u003e\n\u003ch3\u003eModel development\u003c/h3\u003e\n\u003cp\u003eLogistic regression was used to construct radiomic models based on the selected radiomic features. Single sequence models were developed for DCE, D, D*, and f maps; Multiparametric models included a combined IVIM model (D\u0026thinsp;+\u0026thinsp;D*+f) and a DCE\u0026thinsp;+\u0026thinsp;IVIM model. For each patient, a radiomics score (radscore) was calculated according to the screened optimal radiomic features and their corresponding intercept and regression coefficient in logistic regression model. The best-performing radiomic model was selected for integration with clinical variables. All variables associated with HER2 expression at the \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 level in univariate analysis were included in multivariate logistic regression analysis to screen independent predictors and construct a clinical model. And multivariate logistic regression analysis continued to be used to combine the optimal radiomic model with the clinical model. The flow chart of radiomics is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eMann-Whitney U test and Student \u003cem\u003et\u003c/em\u003e test were used to compare continuous variables. Categorical variables were analyzed using the χ2 test or Fisher exact test. Univariate and multivariate logistic regression analysis were used to screen out independent predictors related to HER2 expression. ICC was used to evaluate readers' consistency segmentation, and ICC\u0026thinsp;\u0026ge;\u0026thinsp;0.75 meaned excellent consistency. The area under the receiver operating characteristic curve (AUC) and 95% confidence interval were used to evaluate the performance of the models used to identify HER2 expressions, and the specificity and sensitivity of each model were calculated. Delong test was used to compare the performance of different models. Calibration curve and Hosmer-Lemeshow(H-L) test were used to evaluate the fitting degree of each model. Decision curve analysis (DCA) was used to determine the clinical utility of the model. Bilateral \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered as an indicator of statistically significant difference. All statistical analysis was carried out using SPSS (version 22, IBM) software package, Medcalc (version 15.2.2) software, Python (version 3.7) and R software (version 4.4.2).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eComparison of clinical and MRI features between HER2 positive and HER2 negative breast cancer\u003c/h2\u003e \u003cp\u003eIn Task 1, the histological grade was significantly different between the training and validation cohorts (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). However, there was no significant difference in age, HR status, Ki-67 expression, axillary lymph node metastasis and MRI findings between the two cohorts (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05)(Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e). In the training cohort, compared with HER2 negative breast cancer, HER2 positive breast cancer showed significantly higher rates of axillary lymph node metastasis and larger maximum diameter of lesion (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). These differences were not statistically significant in the validation cohort (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05). Across both cohorts, HER2 positive breast cancer had a lower HR positivity rate, higher Ki-67 expression, higher histological grade, and a greater incidence of peritumoral edema compared to HER2 negative breast cancer (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). No significant differences were found in age, intratumoral necrocystosis, enhancement mode, FGT, BPE, and TIC types between the two groups (all \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05) (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparison of Clinical Pathological and Conventional Imaging Features of HER2 Positive Group and HER2 Negative Group\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eVariables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eTraining cohort\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c7\" namest=\"c5\"\u003e \u003cp\u003eValidation cohort\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHER2\u003c/p\u003e \u003cp\u003epositive\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;61)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHER2\u003c/p\u003e \u003cp\u003enegative\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;97)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eHER2\u003c/p\u003e \u003cp\u003epositive (n\u0026thinsp;=\u0026thinsp;23)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eHER2\u003c/p\u003e \u003cp\u003enegative\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;46)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e51\u0026thinsp;\u0026plusmn;\u0026thinsp;9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e48\u0026thinsp;\u0026plusmn;\u0026thinsp;9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e51\u0026thinsp;\u0026plusmn;\u0026thinsp;10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e50\u0026thinsp;\u0026plusmn;\u0026thinsp;9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHR status\u003c/b\u003e 0.000\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.002\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePositive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e22(36.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e85(87.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e14(60.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e43(93.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNegative\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e39(63.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12(12.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e9(39.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e3(6.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eKi-67 expression\u003c/b\u003e 0.000\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.006\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;20%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2(3.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25(25.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1(4.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e16(34.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;20%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e59(96.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e72(74.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e22(95.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e30(65.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eALN 0.001\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20(32.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e59(60.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10(43.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e24(52.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e41(67.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e38(39.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e13(56.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e22(47.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHistological grade\u003c/b\u003e 0.000\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.000\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eI-II\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e14(23.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e58(58.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8(34.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e36(78.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIII\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e47(77.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e39(41.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e15(65.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e10(21.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMaximum diameter of lesion\u003c/b\u003e 0.000\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.09\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026le;\u0026thinsp;2cm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9(14.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e47(48.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e7(30.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e24(52.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;2cm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e52(85.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50(51.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e16(69.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e22(47.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eIntratumoral necrocystosis\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.17\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e32(52.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e53(53.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11(47.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e30(65.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e29(47.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e44(45.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e12(52.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e16(34.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEnhancement mode\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.47\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMass\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49(80.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e83(85.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e19(82.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e41(89.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNon mass/Mass and Non mass\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12(19.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14(14.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4(17.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5(10.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFGT\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.71\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFatty/Scattered\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15(24.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e27(27.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e6(26.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e14(30.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHeterogeneous/Extreme\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e46(75.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e70(72.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e17(73.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e32(69.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBPE\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.53\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMinimal/Mild\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e59(96.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e87(89.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e23(100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e43(93.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModerate/Marked\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2(3.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10(10.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0(0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e3(6.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eTIC\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.45\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eI/II\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e38(62.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e62(63.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e15(65.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e34(73.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIII\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e23(37.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e35(36.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8(34.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e12(26.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePeritumoral edema\u003c/b\u003e 0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.002\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30(49.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e86(88.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e13(56.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e42(91.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePresent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31(50.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11(11.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10(43.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e4(8.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"7\"\u003eNote: HR,hormone receptor; ALN,Axillary lymph node metastasis; FGT, amount of fibro glandular tissue; BPE, background parenchymal enhancement; TIC,time-signal intensity curve.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eRadiomic model for distinguishing HER2 positive from HER2 negative breast cancer\u003c/h2\u003e \u003cp\u003eThe Lasso regression diagram for radiomic feature selection from DCE, D, D* and f images is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. In the training cohort, the AUCs based on DCE, D, D* and f single-sequence radiomic models were 0.769 (95%CI: 0.696,0.833), 0.750(95%CI:0.675,0.815), 0.777 (95%CI:0.704,0.839) and 0.772(95%CI:0.699,0.835) respectively. In the validation cohort, AUCs were 0.736 (95%CI: 0.616,0.835), 0.786 (95%CI:0.671,0.876), 0.754 (95%CI:0.636,0.850) and 0.770 (95%CI:0.653,0.863) respectively. The multiparametric IVIM radiomic model achieved AUCs of 0.819 (95%CI:0.750,0.876) and 0.760 (95%CI:0.642,0.855) in the training and validation cohorts, respectively. The combined DCE\u0026thinsp;+\u0026thinsp;IVIM model showed the highest performance in the training cohort with an AUC of 0.866 (95%CI:0.803,0.915) and an AUC of 0.751 (95%CI:0.633,0.848) in the validation cohort (Table S3, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAccording to the Delong test, the DCE combined IVIM model significantly outperformed all single-sequence and IVIM-only models in the training cohort (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Table S4). The calibration curve (Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and H-L test showed that the fitting degree of DCE combined with IVIM radiomic model in both cohorts is good (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05). A radiomic feature correlation heatmap is provided in Fig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eClinical model for distinguishing HER2 positive from HER2 negative breast cancer\u003c/h2\u003e \u003cp\u003eMultivariate logistic regression analysis showed that histological grade (OR\u0026thinsp;=\u0026thinsp;0.33, 95%CI:0.12,0.92), peritumoral edema (OR\u0026thinsp;=\u0026thinsp;0.32, 95%CI:0.11,0.93), and radscore (OR\u0026thinsp;=\u0026thinsp;2.96, 95%CI:1.88,4.68) were independent predictors of HER2 positive breast cancer(Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Histological grade and peritumoral edema were used to construct the clinical model. The clinical model yielded AUCs of 0.785 (95%CI:0.713,0.846) and 0.751 (95%CI:0.633,0.848), sensitivity of 0.885 and 0.696, and specificity of 0.546 and 0.739 in the training and validation cohorts, respectively (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eUnivariable and Multivariable Logistic Regression Analysis of Training Cohort to Evaluate the Variables Related to HER2 Positive Status\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eVariables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eUnivariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eMultivariable\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOR(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOR(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHR status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.25(0.11,0.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.29(0.41,4.04)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.67\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKi-67 expression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.10(0.02,0.43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.08(0.01,1.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.06\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eALN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.31(0.16,0.62)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.79(0.29,2.14)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.64\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHistological grade\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.21(0.10,0.43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.33(0.12,0.92)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.033\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMaximum diameter of lesion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.18(0.08,0.42)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.72(0.23,2.24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePeritumoral edema\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.12(0.06,0.28)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.32(0.11,0.93)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.036\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRadscore\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.34(2.23,5.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.96(1.88,4.68)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.000\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNote: HR,hormone receptor; ALN,Axillary lymph node; OR,odds ratio; 95CI%,confidence interval.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEfficiency of Clinical Model, Radiomic Model and Clinic ombined Radiomic Model for Differentiating HER2 Positive and Negative Breast Cancer\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModels\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCohort\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAUC(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eClinical model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTraining\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.785(0.713,0.846)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.885\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.546\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eValidation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.751(0.633,0.848)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.696\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.739\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eRadiomic model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTraining\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.866(0.803,0.915)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.672\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.928\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eValidation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.751(0.633,0.848)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.739\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.717\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eClinic combined radiomic model\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTraining\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.903(0.846,0.944)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.918\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.753\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eValidation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.830(0.720,0.910)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.696\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.891\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eClinic combined radiomic model for distinguishing HER2 positive from HER2 negative breast cancer\u003c/h2\u003e \u003cp\u003eThe combined radiomic model for predicting HER2 positive breast cancer achieved the highest performance, with an AUC 0.903 (95%CI:0.846,0.944), sensitivity of 0.918, and specificity of 0.753 in the training cohort and an AUC of 0.830 (95%CI:0.720,0.910), sensitivity of 0.696, and specificity of 0.891 in the validation cohort (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). In the training cohort, Delong test showed the clinical combined radiomic model significantly outperformed both the clinical model and radiomic models (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) However, in the validation cohort, the differences were not statistically significant (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05)(Table S5). DCA demonstrated that compared with clinical model and radiomic models, clinic combined radiomic model provided greater net benefit across threshold probabilities, indicating better clinical utility(Fig. S3).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eComparison of clinical and MRI characteristics of HER2 low and HER2 zero breast cancer\u003c/h2\u003e \u003cp\u003eIn Task2, there were no significant differences in clinical pathological and conventional imaging features between the training and validation cohorts (Table S6). In the training cohort, HER2 low cancer had higher HR positivity, lower Ki-67 expression, and lower histological grade compared to HER2 zero cancer (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). In the validation cohort, only FGT showed a statistically significant difference between the two group (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05)(Table S7).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eRadiomics model for distinguishing HER2 low from HER2 zero breast cancer\u003c/h2\u003e \u003cp\u003eThe Lasso regression diagram of DCE, D, D* and f images is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e. In the training cohort, the AUCs of distinguishing HER2 low cancer from HER2 zero cancer based on DCE, D, D*, f single sequence radiomic model were 0.894 (95%CI:0.817,0.947), 0.838 (95%CI:0.751,0.904), 0.846 (95%CI:0.760,0.911) and 0.829 (95%CI:0.741,0.897), respectively. And in the validation cohort, the same radiomic signatures achieved AUCs of 0.789 (95%CI:0.638,0.898), 0.722 (95%CI:0.565,0.848), 0.827 (95%CI:0.681,0.925) and 0.800(95%CI:0.650,0.906), respectively. The AUCs of the multiparametric IVIM radiomic model in the training and validation cohorts were 0.929 (95%CI:0.859,0.970) and 0.816(95%CI: 0.668,0.917), respectively. The AUCs of DCE combined IVIM radiomic model predicting HER2 low expression were 0.951 (95%CI:0.888,0.984) in the training cohort and 0.853 (95%CI:0.712,0.942) in the validation cohort (Table S8 and Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e). Table S9 shows the final screening of 7 DCE-MRI early, 5 D, 5 D* and 5 f radiomic features.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn the training cohort, Delong test results showed DCE combined IVIM radiomic model performed well in differentiating HER2 low cancer and HER2 zero cancer, which was higher than other single sequence radiomic models (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05), but there was no statistically significant difference compared with multiparametric IVIM radiomic model (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05). In the validation cohort, there was no significant difference between DCE combined with IVIM radiomic model and other single sequence and multiparametric IVIM radiomic models (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05)( Table S10).\u003c/p\u003e \u003cp\u003eThe calibration curve and H-L test confirmed good model fit of DCE combined with IVIM radiomic model in both cohorts (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05) (Fig. S4). DCA showed that compared with clinical model and radiomic model, clinic combined radiomic model provides better net income, which indicated that clinic combined radiomic model has better clinical benefits(Fig. S5).\u003c/p\u003e \u003cp\u003e \u003cb\u003eLogistic regression analysis was used to evaluate the variables related to the low expression of HER2 in breast cancer.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eMultivariable logistic analysis showed that in both the training and validation cohorts, only DCE combined with IVIM radscore was an independent predictive factors of HER2 low status (training cohort: OR\u0026thinsp;=\u0026thinsp;4.50, 95%CI:2.18,9.29; validation cohort: OR\u0026thinsp;=\u0026thinsp;2.05, 95%CI: 1.30,3.22) (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eUnivariable and Multivariable Logistic Regression Analysis of Training Cohort to Evaluate the Variables Related to HER2 Low Expression\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eUnivariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eMultivariable\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOR(95%CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOR(95%CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTraining Cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHR status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.12(0.03,0.58)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.008\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.03(0.001,1.12)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.06\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKi-67 expression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.28(0.10,0.77)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.31(0.02,4.60)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHistological grade\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.88(1.67,9.10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.31(0.77,24.10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRadscore\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.34(2.23,5.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.50(2.18,9.29)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.000\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eValidation Cohort\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFGT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.62(1.06,20.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.23(0.55,19.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.19\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRadscore\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.10(1.35,3.28)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.05(1.30,3.22)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.002\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNote: HR,hormone receptor; FGT, amount of fibro glandular tissue; BPE, background parenchymal enhancement; TIC,time-signal intensity curve.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study demonstrated that a clinical model incorporating peritumoral edema and histological grade, when combined with a radiomics model based on multiparametric DCE and IVIM imaging, achieved favorable performance in differentiating the positive and negative expression of HER2 in breast cancer. The combined model yielded AUCs of 0.903 and 0.830 in the training and validation cohorts, respectively. Additionally, we developed a multiparametric MRI radiomic model based on DCE and IVIM sequences to identify HER2 low cancer and HER2 zero cancer. This model achieved excellent diagnostic performance, with AUCs of 0.951 and 0.853 in the training and validation cohorts, respectively, outperforming single-sequence radiomic models. Moreover, the radscore was identified as an independent predictor of HER2 low status.\u003c/p\u003e \u003cp\u003eAt present, researchers are gradually paying attention to the category of breast cancer with low expression of HER2, and trying to predict three different expression States of HER2 based on multi-parametric MRI radiomics(\u003cspan additionalcitationids=\"CR17 CR18\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). The multi-parametric MRI radiomic model (including DWI, ADC and DCE sequences) developed by Zheng et al.(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e) had a low performance in the differential diagnosis of HER2 positive and HER2 negative breast cancer, and the AUC in the validation cohort was 0.725. Furthermore, the AUCs of the radiomic models for differentiating HER2 low breast cancer from HER2 positive/zero cancer and differentiating HER2 zero from HER2 positive/low breast cancer were 0.782 and 0.813 respectively. Peng et al.(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e) constructed a radiomic signature built from T2WI, DWI, ADC and T1 weighted delayed contrast enhanced images to distinguish HER2 zero cancer from HER2 positive/low cancer, and its AUC was 0.780 in the validation cohort, and Nomogram of multi-parametric MRI radiomics combined with pathological features (histochemistry, Ki-67 index, and PR status) was developed for differentiating HER2 positive from HER2 low cancer, with an AUC of 0.800 in the validation cohort. Ramtohul et al.(\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e) identified HER2 low cancer and HER2 positive cancer based on the radiomic signatures of T2WI and DCE-MRI, and achieved an AUC of 0.800 in the validation cohort. In addition, the study also showed that HER2 low and HER2 zero cancer in luminal and triple negative breast cancer subtypes have similar radiomic features, so it is impossible to distinguish them effectively.\u003c/p\u003e \u003cp\u003eCompared with the existing radiomic models, a new multi-parametric IVIM radiomic model has been developed in this study. The activation of HER2 factor is related to the up-regulation of vascular endothelial growth factor, which can further promote the formation of tumor neovascularization(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). IVIM can both reflect the diffusion movement of water molecules and microcirculation perfusion in tissues(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). By combining it with DCE-MRI sequence, the functional information can be complemented and fused. The combination of these multi-parametric MRI radiomic models improves the differential diagnosis ability of the models in the expression level of HER2, and at the same time provides more new imaging biomarkers for clinic, which is helpful to identify the breast cancer patients who are resistant to HER2 treatment more accurately.\u003c/p\u003e \u003cp\u003eIn this study, we also found that the radiomic features used to distinguish the different expressions of HER2 in breast cancer were mainly wavelet features, and there was cross-sequence feature repetition in multi-parametric MRI sequences, which is consistent with the previous radiomic research result(\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Wavelet features are the intensity and texture features of the original image obtained by wavelet decomposition calculation, which can help us quantify the multi-frequency information of tumors in different wavelet transform frequency ranges, thus reflecting tumor heterogeneity(\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). This is consistent with the demonstration of the key role of wavelet features in imageology in the existing literature, which further confirms its importance(\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). The cross-model stability and cross-sequence repeatability of wavelet first-order features suggest that it may be a potential image biomarker and provide a new way for noninvasive quantitative evaluation of HER2 expression level. Secondly, although there are overlapping features, the differences in imaging principles of different MRI sequences may make these features contain specific biological information. Therefore, this study proposes a modeling strategy based on DCE-MRI and IVIM multi-parametric MRI sequences, which can enhance the model's ability to capture the heterogeneity of tumor microenvironment by mining their complementary associations, and finally improve the diagnostic efficiency of HER2 expression level.\u003c/p\u003e \u003cp\u003eThe current research(\u003cspan additionalcitationids=\"CR17\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e) mostly focused on the evaluation of HER2 expression level by intratumoral imaging. Peritumoral region also represents a unique microenvironment, which is related to biological information such as tumor growth and invasion, and may be used as a potential biomarker(\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). The latest progress in breast cancer research has identified various key findings in the peritumor tissue, such as angiogenesis factors, lymphatic and vascular invasion around the tumor and peritumor edema(\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e). Bian et al.(\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e) extracted the intra-tumor and peritumoral features of ADC and T1WI contrast-enhanced imaging to distinguish between HER2 positive and HER2 negative cancer, and between HER2 low and HER2 zero cancer. The research found that the AUC in the external validation cohort was 0.760 and 0.710, respectively. Although they additionally included the analysis of peritumoral information, their diagnostic efficiency was still lower than the results of this study, especially in identifying the more challenging HER2 low and HER2 zero cancer. We speculated that it may be related to the different construction methods of the radiomic model and the different scope of feature extraction around the tumor. In our previous research on peritumoral region and prognosis of breast cancer, we recommend 2\u0026ndash;20 mm as a study-worthy peritumoral region range for MR researchers(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). Therefore, in the future, we will continue to explore the best size of peritumoral area in order to better evaluate the peritumoral microenvironment and predict the expression level of HER2 in breast cancer more accurately.\u003c/p\u003e \u003cp\u003eThe limitations of this study were as follows: firstly, this study was a retrospective single-center design, and the sample size was still small. In the future, we should further expand the sample size and conduct multi-center research. Secondly, although we included two MRI scanners with different equipment, they were all carried out with 3.0T magnetic field intensity. Therefore, it is necessary to measure under different magnetic field strengths to verify the universality of our research. Thirdly, IVIM selected more b values. Although it pays attention to perfusion information, it leads to a long scanning time. In the future, we will further standardize and standardize the key b values.\u003c/p\u003e \u003cp\u003eIn a word, the results of this study showed that multi-parametric MRI radiomics based on IVIM and DCE-MRI can effectively predict the expression level of HER2 in breast cancer from a non-invasive point of view, thus better assisting pathologists to detect the expression of HER2 and make personalized treatment plans for patients.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eHER2 \u0026nbsp;human epidermal growth factor receptor 2\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDCE-MRI \u0026nbsp; dynamic contrast enhanced-magnetic resonance imaging\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIVIM \u0026nbsp;intravoxel incoherent motion\u003c/p\u003e\n\u003cp\u003eHR Hormone receptor\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFGT \u0026nbsp;fibroglandular tissue content\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBPE \u0026nbsp;background parenchymal enhancement\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTIC \u0026nbsp;time-intensity curve\u003c/p\u003e\n\u003cp\u003eICC intra-class correlation coefficient\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLASSO \u0026nbsp;Least Absolute Shrinkage and Selection Operator\u003c/p\u003e\n\u003cp\u003eAUC \u0026nbsp;area under the receiver operating characteristic curve\u003c/p\u003e\n\u003cp\u003eDCA \u0026nbsp;Decision curve analysis\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate:\u003c/strong\u003e The study was approved by the Ethics Committee of First Affiliated Hospital of Dalian Medical University [IRB number: PJ- KS-KY-2024-69]. Given the retrospective nature of the study, written informed consent was not required.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials:\u0026nbsp;\u003c/strong\u003eThe datasets generated and/or analysed during the current study are not publicly available due [REASON WHY DATA ARE NOT PUBLIC] but are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u003c/strong\u003e The authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This work were supported by [Research on Digitalization Empowering the Reform of Pathology Teaching]\u0026nbsp;and\u0026nbsp;[A project of the \u0026quot;14th Five-Year Plan\u0026quot; for Educational Science in Liaoning Province].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions:\u003c/strong\u003e Lina Zhang is the study project leader; she conceived and\u003c/p\u003e\n\u003cp\u003eparticipated in the study design and coordination. Siqi Zhao conceived the study and\u003c/p\u003e\n\u003cp\u003edrafted the manuscript. Lina Zhang and Siqi Zhao analyzed radiological and\u003c/p\u003e\n\u003cp\u003eclinical data. Fan Wei and Hui Liu provided technical support.Yueqi Wu, Yuanfei Li and Moyun Zhang collected the clinical\u0026nbsp;materials and radiological data. Shuo Wang, Xinyue Yin and Zhitian Guo\u0026nbsp;participated in the sequence alignment and scanning. Xue Gao reviewed the pathological data. Jie Yang and Haonan Guan revised the manuscript and gave\u0026nbsp;statistical support. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229-63.\u003c/li\u003e\n\u003cli\u003eJokar N, Velikyan I, Ahmadzadehfar H, et al. Theranostic Approach in Breast Cancer: A Treasured Tailor for Future Oncology. Clin Nucl Med. 2021;46(8):e410-e20.\u003c/li\u003e\n\u003cli\u003ePernas S, Tolaney SM. Targeting HER2 heterogeneity in early-stage breast cancer. Curr Opin Oncol. 2020;32(6):545-54.\u003c/li\u003e\n\u003cli\u003eLi Z, Metzger Filho O, Viale G, et al. HER2 heterogeneity and treatment response-associated profiles in HER2-positive breast cancer in the NCT02326974 clinical trial. J Clin Invest. 2024;134(7).\u003c/li\u003e\n\u003cli\u003eTrastuzumab for early-stage, HER2-positive breast cancer: a meta-analysis of 13 864 women in seven randomised trials. Lancet Oncol. 2021;22(8):1139-50.\u003c/li\u003e\n\u003cli\u003eGodoy-Ortiz A, Sanchez-Mu\u0026ntilde;oz A, Chica Parrado MR, et al. Deciphering HER2 Breast Cancer Disease: Biological and Clinical Implications. Front Oncol. 2019;9:1124.\u003c/li\u003e\n\u003cli\u003eYoon J, Oh DY. HER2-targeted therapies beyond breast cancer - an update. Nat Rev Clin Oncol. 2024;21(9):675-700. \u003c/li\u003e\n\u003cli\u003eWuerstlein R, Harbeck N. Neoadjuvant Therapy for HER2-positive Breast Cancer. Rev Recent Clin Trials. 2017;12(2):81-92.\u003c/li\u003e\n\u003cli\u003eModi S, Jacot W, Yamashita T, et al. Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer. N Engl J Med. 2022;387(1):9-20.\u003c/li\u003e\n\u003cli\u003eCorti C, Giugliano F, Nicol\u0026ograve; E, et al. HER2-Low Breast Cancer: a New Subtype? Curr Treat Options Oncol. 2023;24(5):468-78.\u003c/li\u003e\n\u003cli\u003eTarantino P, Viale G, Press MF, et al. ESMO expert consensus statements (ECS) on the definition, diagnosis, and management of HER2-low breast cancer. Ann Oncol. 2023;34(8):645-59.\u003c/li\u003e\n\u003cli\u003ePress OA, Guzman R, Cervantes M, et al. Characterization of HER2 status by fluorescence in situ hybridization (FISH) and immunohistochemistry (IHC). Methods Mol Biol. 2014;1180:181-207.\u003c/li\u003e\n\u003cli\u003eRobbins CJ, Fernandez AI, Han G, et al. Multi-institutional Assessment of Pathologist Scoring HER2 Immunohistochemistry. Mod Pathol. 2023;36(1):100032.\u003c/li\u003e\n\u003cli\u003eSchnitt SJ, Tarantino P, Collins LC. The American Society of Clinical Oncology-College of American Pathologists Guideline Update for Human Epidermal Growth Factor Receptor 2 Testing in Breast Cancer. Arch Pathol Lab Med. 2023;147(9):991-2.\u003c/li\u003e\n\u003cli\u003eLee SH, Park H, Ko ES. Radiomics in Breast Imaging from Techniques to Clinical Applications: A Review. Korean J Radiol. 2020;21(7):779-92.\u003c/li\u003e\n\u003cli\u003eZheng S, Yang Z, Du G, et al. Discrimination between HER2-overexpressing, -low-expressing, and -zero-expressing statuses in breast cancer using multiparametric MRI-based radiomics. Eur Radiol. 2024;34(9):6132-44.\u003c/li\u003e\n\u003cli\u003ePeng Y, Zhang X, Qiu Y, et al. Development and Validation of MRI Radiomics Models to Differentiate HER2-Zero, -Low, and -Positive Breast Cancer. AJR Am J Roentgenol. 2024;222(4):e2330603.\u003c/li\u003e\n\u003cli\u003eRamtohul T, Djerroudi L, Lissavalid E, et al. Multiparametric MRI and Radiomics for the Prediction of HER2-Zero, -Low, and -Positive Breast Cancers. Radiology. 2023;308(2):e222646.\u003c/li\u003e\n\u003cli\u003eBian X, Du S, Yue Z, et al. Potential Antihuman Epidermal Growth Factor Receptor 2 Target Therapy Beneficiaries: The Role of MRI-Based Radiomics in Distinguishing Human Epidermal Growth Factor Receptor 2-Low Status of Breast Cancer. J Magn Reson Imaging. 2023;58(5):1603-14.\u003c/li\u003e\n\u003cli\u003eWolff AC, Somerfield MR, Dowsett M, et al. Human Epidermal Growth Factor Receptor 2 Testing in Breast Cancer: ASCO-College of American Pathologists Guideline Update. J Clin Oncol. 2023;41(22):3867-72.\u003c/li\u003e\n\u003cli\u003eAllison KH, Hammond MEH, Dowsett M, et al. Estrogen and Progesterone Receptor Testing in Breast Cancer: ASCO/CAP Guideline Update. J Clin Oncol. 2020;38(12):1346-66.\u003c/li\u003e\n\u003cli\u003eGoldhirsch A, Winer EP, Coates AS, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol. 2013;24(9):2206-23.\u003c/li\u003e\n\u003cli\u003eSpak DA, Plaxco JS, Santiago L, et al. BI-RADS(\u0026reg;) fifth edition: A summary of changes. Diagn Interv Imaging. 2017;98(3):179-90.\u003c/li\u003e\n\u003cli\u003ePanzironi G, Moffa G, Galati F, et al. Peritumoral edema as a biomarker of the aggressiveness of breast cancer: results of a retrospective study on a 3 T scanner. Breast Cancer Res Treat. 2020;181(1):53-60.\u003c/li\u003e\n\u003cli\u003eKourea HP, Dimitrakopoulos FI, Koliou GA, et al. Clinical Significance of Major Angiogenesis-Related Effectors in Patients with Metastatic Breast Cancer Treated with Trastuzumab-Based Regimens. Cancer Res Treat. 2022;54(4):1053-1064. \u003c/li\u003e\n\u003cli\u003eIima M, Honda M, Sigmund EE, et al. Diffusion MRI of the breast: Current status and future directions. J Magn Reson Imaging. 2020;52(1):70-90.\u003c/li\u003e\n\u003cli\u003eYu Y, Tan Y, Xie C, et al. Development and Validation of a Preoperative Magnetic Resonance Imaging Radiomics-Based Signature to Predict Axillary Lymph Node Metastasis and Disease-Free Survival in Patients With Early-Stage Breast Cancer. JAMA Netw Open. 2020;3(12):e2028086.\u003c/li\u003e\n\u003cli\u003eWu Q, Wang S, Chen X, et al. Radiomics analysis of magnetic resonance imaging improves diagnostic performance of lymph node metastasis in patients with cervical cancer. Radiother Oncol. 2019;138:141-8.\u003c/li\u003e\n\u003cli\u003eZhou X, Yi Y, Liu Z, et al. Radiomics-Based Pretherapeutic Prediction of Non-response to Neoadjuvant Therapy in Locally Advanced Rectal Cancer. Ann Surg Oncol. 2019;26(6):1676-84.\u003c/li\u003e\n\u003cli\u003eJiang W, Meng R, Cheng Y,et al. Intra- and Peritumoral Based Radiomics for Assessment of Lymphovascular Invasion in Invasive Breast Cancer. J Magn Reson Imaging. 2024;59(2):613-625.\u003c/li\u003e\n\u003cli\u003eDing J, Chen S, Serrano Sosa M, et al. Optimizing the Peritumoral Region Size in Radiomics Analysis for Sentinel Lymph Node Status Prediction in Breast Cancer. Acad Radiol. 2022;29 Suppl 1(Suppl 1):S223-s8.\u003c/li\u003e\n\u003cli\u003eZhao S, Li Y, Ning N, et al. Association of peritumoral region features assessed on breast MRI and prognosis of breast cancer: a systematic review and meta-analysis. Eur Radiol. 2024;34(9):6108-20.\u003c/li\u003e\n\u003c/ol\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":"cancer-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"caig","sideBox":"Learn more about [Cancer Imaging](https://cancerimagingjournal.biomedcentral.com/)","snPcode":"40644","submissionUrl":"https://submission.nature.com/new-submission/40644/3","title":"Cancer Imaging","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Breast cancer, Human epidermal growth factor receptor 2, Magnetic resonance imaging, Radiomics","lastPublishedDoi":"10.21203/rs.3.rs-6646001/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6646001/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003ePurpose\u003c/b\u003e\u003c/p\u003e \u003cp\u003eTo develop and validate radiomic models using multi-parametric dynamic contrast-enhanced MRI (DCE-MRI) and intravoxel incoherent movement (IVIM)-based features for the preoperative differentiation of HER2 expressions levels in breast cancer.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMaterials and Methods\u003c/b\u003e\u003c/p\u003e \u003cp\u003eThis retrospectively study analyzed 227 female breast cancer patients who underwent breast 3.0TMRI examination at our institution from December 2019 to December 2023. The least absolute shrinkage and selection operator (LASSO) ten-fold cross-validation method was used to develop the radiomic features to identify HER2 positive and HER2 negative cancer(task 1), and further identify HER2 low and HER2 zero cancer(task 2). Then the radiomic features were selected and combined with clinical characteristics to construct predicting models using the logistic regression analysis. The area under the receiver operating characteristic curve(AUC), sensitivity, and specificity were used to evaluate the performance of radiomic models.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e \u003cp\u003eFor task 1, the AUCs of clinical model (histological grade and peritumoral edema), DCE combined IVIM(D\u0026thinsp;+\u0026thinsp;D*+f) radiomic model and clinic combined radiomic model were 0.785 (95%CI:0.713,0.846), 0.866 (95%CI:0.803,0.915) and 0.903 (95%CI:0.846,0.944) respectively. In the validation cohort, The AUCs were 0.751 (95%CI:0.633,0.848), 0.751 (95%CI:0.633,0.848) and 0.830 (95%CI:0.720,0.910) respectively. For task 2, the AUCs of DCE combined IVIM radiomic model in training and validation cohort were 0.951 (95%CI:0.888,0.984) and 0.853 (95%CI:0.712,0.942) respectively, and the radiomics score was independent predictors of HER2 low cancer.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusion\u003c/b\u003e\u003c/p\u003e \u003cp\u003eThe radiomic signature derived from multi-parametric MRI, together with peritumoral edema and histological grade, demonstrated strong performance in predicting HER2 expression preoperatively in breast cancer, which may support individualized treatment strategies.\u003c/p\u003e","manuscriptTitle":"Multi-parametric MRI Radiomics Predicts Different HER2 Expression in Breast Cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-19 09:20:59","doi":"10.21203/rs.3.rs-6646001/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Accepted","date":"2025-12-17T14:52:27+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-12T20:17:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"227589892590100425603120813835807456256","date":"2025-08-28T08:16:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"336538313892764604758527766674294171993","date":"2025-08-21T14:20:23+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-05-14T14:51:47+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-05-14T04:37:00+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-05-13T04:12:00+00:00","index":"","fulltext":""},{"type":"submitted","content":"Cancer Imaging","date":"2025-05-12T11:00:34+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"cancer-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"caig","sideBox":"Learn more about [Cancer Imaging](https://cancerimagingjournal.biomedcentral.com/)","snPcode":"40644","submissionUrl":"https://submission.nature.com/new-submission/40644/3","title":"Cancer Imaging","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a1316121-e9c6-4ff6-8fa9-a54c76e410ef","owner":[],"postedDate":"May 19th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-29T16:06:36+00:00","versionOfRecord":{"articleIdentity":"rs-6646001","link":"https://doi.org/10.1186/s40644-025-00981-y","journal":{"identity":"cancer-imaging","isVorOnly":false,"title":"Cancer Imaging"},"publishedOn":"2025-12-24 15:58:35","publishedOnDateReadable":"December 24th, 2025"},"versionCreatedAt":"2025-05-19 09:20:59","video":"","vorDoi":"10.1186/s40644-025-00981-y","vorDoiUrl":"https://doi.org/10.1186/s40644-025-00981-y","workflowStages":[]},"version":"v1","identity":"rs-6646001","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6646001","identity":"rs-6646001","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}