Development and Validation of a CEUS-Based Intratumoral–Peritumoral Radiomics Nomogram for Preoperative Prediction of HER2-Low Breast Cancer

Development and Validation of a CEUS-Based Intratumoral–Peritumoral Radiomics Nomogram for Preoperative Prediction of HER2-Low 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 Development and Validation of a CEUS-Based Intratumoral–Peritumoral Radiomics Nomogram for Preoperative Prediction of HER2-Low Breast Cancer Yachen Wang, meiqin Yu, Chun Li This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8786769/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objective This study aims to develop a preoperative diagnostic model for HER2-low breast cancer by integrating radiomics, conventional B-mode ultrasound (US), contrast-enhanced ultrasound (CEUS), and clinical features. Method This retrospective study included 175 patients with pathologically confirmed breast cancer (91 HER2-low; 84 HER2-zero). All patients underwent preoperative conventional B-mode ultrasound and CEUS examinations. Intratumoral and peritumoral (2 mm and 3 mm) regions of interest were manually segmented, and radiomic features were extracted using PyRadiomics. Feature selection was performed through t-tests, Pearson correlation analysis, mRMR（Max-Relevance and Min-Redundancy）and LASSO regression. Multiple radiomics models (US, CEUS, multi-region) and a clinical model were constructed and evaluated using ROC analysis. A combined clinical–radiomics nomogram was generated using multivariate logistic regression and was assessed with calibration curves and decision curve analysis. Result CEUS-based peritumoral radiomics demonstrated superior discriminative performance compared to conventional ultrasound. The combined CEUS + tumor and peritumoral 3-mm model achieved the highest predictive accuracy (AUC 0.892 in the training cohort; 0.802 in the test cohort). Microcalcification was identified as an independent clinical predictor of HER2-low disease, although the clinical model alone showed moderate performance (AUC 0.721 and 0.717). The integrated nomogram combining radiomics signatures and clinical factors yielded the best overall performance (AUC 0.921 and 0.850), with favorable calibration and superior net clinical benefit. Conclusion A CEUS-based multi-region radiomics nomogram that incorporates intratumoral and peritumoral features enables accurate and noninvasive preoperative prediction of HER2-low breast cancer. Breast cancer ultrasound radiomics HER 2-low Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction Breast cancer remains one of the most commonly diagnosed malignancy in women worldwide and remains a leading cause of cancer-related mortality[1, 2]. Molecular subtyping based on immunohistochemistry (IHC) plays a pivotal role in guiding treatment strategies and traditionally categorizes breast cancer into Luminal A, Luminal B, HER2-positive, and triple-negative subtypes according to estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), and Ki-67 expression profiles[3–5]. Historically, HER2 expression has been dichotomized as either positive or negative. However, emerging biological and clinical evidence indicates that this binary classification oversimplifies HER2 biology. Tumors with an IHC score of 1 + or 2 + with negative fluorescence in situ hybridization (FISH) results are now recognized as a distinct subgroup—HER2-low breast cancer[6, 7]. Compared with HER2-zero tumors, HER2-low lesions display unique molecular characteristics, including enrichment of luminal-related gene expression signatures and reduced expression of ERBB2-driven pathways. Compared with HER2-low breast cancer, HER2-zero tumors tend to exhibit relatively clearer and more regular margins and shapes[8]. Yin et al. reported a positive correlation between HER2-low expression and pixel-based boundary length, suggesting greater morphological irregularity, reflecting increased intratumoral heterogeneity and aiding in the differentiation between HER2-low and HER2-zero subtypes[9]. Clinically, a clinical trial demonstrated that patients with HER2-low metastatic breast cancer derive substantial survival benefits from trastuzumab deruxtecan [10, 11], highlighting HER2-low as a therapeutically actionable subtype. Consequently, accurate preoperative identification of HER2-low status has become increasingly important for treatment planning in the era of antibody–drug conjugates. Currently, HER2 assessment relies on immunohistochemical evaluation of core biopsy samples, an invasive procedure subject to sampling error, tumor heterogeneity, and processing delays[12, 13]. These limitations underscore the need for rapid, reproducible, and noninvasive approaches capable of capturing both the anatomical and functional characteristics of breast tumors[14]. Conventional B-mode Ultrasound (US) is widely used in breast cancer detection and offers advantages such as accessibility, low cost, and real-time imaging. Contrast-enhanced ultrasound (CEUS) further enhances diagnostic capability by visualizing microvascular perfusion and tumor neoangiogenesis, providing functional information complementary to US imaging[15]. While conventional US provides morphological details, its ability to reflect underlying molecular heterogeneity remains limited due to reliance on subjective descriptors. In contrast, CEUS captures dynamic perfusion patterns that may correlate with HER2-driven angiogenic signaling—a key biological distinction between HER2-zero and HER2-low phenotypes[16]. Radiomics enables high-throughput extraction of quantitative imaging features and has shown promise in characterizing tumor heterogeneity, predicting molecular subtypes, evaluating tumor aggressiveness, and estimating prognosis[17, 18]. Although US- and CEUS-based radiomics models have demonstrated potential across various oncologic applications, research specifically addressing radiomic signatures associated with HER2-low breast cancer remains scarce[19]. Integrating radiomic signatures with readily available clinical variables—such as menopausal status, tumor size, and BI-RADS category—may improve generalizability and facilitate translation into routine clinical workflows, where purely imaging-based models may lack context[20–22]. Recent spatial transcriptomic and multiplex immunohistochemistry studies have revealed that HER2-low tumors exhibit distinct peritumoral microenvironments, characterized by altered immune cell infiltration, stromal fibrosis, and heterogeneous angiogenesis—features potentially detectable through CEUS due to its sensitivity to microvascular changes[23, 24]. Furthermore, the potential added value of integrating intratumoral and peritumoral radiomics features—including microvascular and stromal alterations adjacent to tumors—has not been fully explored in the context of HER2-low prediction. Therefore, this study aims to compare the predictive performance of US– and CEUS–based radiomics models for differentiating HER2-zero from HER2-low breast cancer; evaluate the added value of incorporating peritumoral regions (2 mm and 3 mm) into radiomic feature extraction; and develop a clinical–radiomics model by integrating the optimal CEUS-derived radiomics signature with independent clinical predictors to enable noninvasive preoperative stratification of HER2 status. 2. Method and materials This retrospective study initially included 1,332 female patients with US- and CEUS-detected breast masses at the First Affiliated Hospital of Nanchang University between 2019 and 2025. After excluding 934 cases lacking histopathological confirmation or diagnosed with non-breast cancer lesions, 398 patients with surgically confirmed breast cancer were retained. Further exclusions were applied as follows: 42 non-mass lesions, 61 lesions ≥5 cm (incompatible with single-plane imaging), 57 patients who received pre-ultrasound neoadjuvant therapy or biopsy, and 63 cases with incomplete HER2 assessment (missing FISH validation for IHC 2+ results, heterogeneous FISH findings, or IHC 2+ without FISH testing). Ultimately, 175 patients met all inclusion criteria and were enrolled in the final analysis. 1.1 Inclusion criteria All participants should have histologically confirmed breast cancer with complete clinical, imaging, and pathological data. HER2 status was categorized into two groups: HER2-low (n = 91; IHC 1+ or IHC 2+/FISH-negative) and HER2-zero (n = 84; IHC 0). Patients were randomly assigned to a training cohort (n = 123) and test cohort (n = 52) at a 7:3 ratio, stratified by HER2 status to ensure balanced distribution. 1.2 Exclusion criteria Preoperative US and CEUS examinations were performed within one week of surgery for all enrolled patients. Inclusion criteria for imaging assessments included age between 18–80 years and no contraindications to SonoVue administration (e.g., allergy to contrast agents, acute coronary syndrome, severe arrhythmia, pulmonary hypertension, or respiratory distress syndrome). Exclusion criteria ensured homogeneity of the cohort by eliminating confounding factors such as prior interventions or ambiguous HER2 classifications. 2.1 Clinical data collection This retrospective study was approved by the Institutional Review Board, and the requirement for informed consent was waived. Clinical and pathological variables included age, menopausal status, clinical tumor stage, laterality and quadrant of the breast lesion, family history of cancer, and axillary lymph node (ALN) status. All patients had pathologically confirmed malignant breast tumors. 2.2 Ultrasound image acquisition Preoperative ultrasound examinations were performed using either a Siemens Acuson system (Siemens Healthineers, Germany), equipped with 18L6 or 10L4 linear array transducers. All scans were conducted by experienced sonographers with at least 5 years of clinical expertise. Patients were placed in the supine position with both breasts fully exposed during scanning. For each lesion, the largest cross-sectional image was acquired, and standard ultrasound features were documented according to the Breast Imaging Reporting and Data System (BI-RADS). US: Patients were placed in the supine position. Transverse and longitudinal scans of the target tumor were performed. Images of the largest tumor cross-section and characteristic 2D features were recorded. The imaging plane with the maximum tumor diameter and richest vascularity was selected. CEUS: Prior to contrast injection, grayscale and color Doppler US were used to identify the imaging plane showing the largest lesion diameter and most prominent vascularity. A bolus of 4.8 mL SonoVue (sulfur hexafluoride microbubbles) was administered intravenously, followed by a 5 mL saline flush. The largest cross-sectional view of the tumor was clearly displayed, and the system was switched to real-time dual-display contrast mode. The probe remained stable during imaging. A timer and cine storage function were activated at contrast onset, and dynamic perfusion within the lesion was recorded continuously for 90 seconds. The peak-enhancement frame was selected for analysis. 2.3 IHC: HER2-zero was defined as no membrane staining (IHC score 0). HER2-low was defined as weak and incomplete membrane staining without HER2 gene amplification by FISH (IHC score 1+ or 2+/FISH-negative). 2.4 Radiomics analysis Region of interest (ROI) delineation: Using ITK-SNAP software (version 3.8.0), two radiologists independently contoured the tumor boundaries on both 2D US and CEUS images, blinded to pathological results. In cases of disagreement, a senior radiologist with over 15 years of experience in breast ultrasound adjudicated to reach consensus. Shown in figure 1, ROIs encompassed the entire lesion. Radiomics Feature Extraction, Selection, and Model Construction: Ultrasound image features were extracted using PyRadiomics. These features included first-order statistics and higher-order features, specifically from the Gray-Level Co-occurrence Matrix (GLCM), Gray-Level Run-Length Matrix (GLRLM), Gray-Level Size Zone Matrix (GLSZM), Gray-Level Dependence Matrix (GLDM), Neighborhood Gray-Tone Difference Matrix (NGTDM), as well as features derived from filtered images. Only features with an intraclass correlation coefficient greater than 0.75 were retained for further analysis. The extracted features were then standardized using Z-score normalization. Patients were randomly assigned to training and test cohort in a 7:3 ratio. Feature selection was performed sequentially: first, independent-samples t-tests were used to identify features significantly associated with HER2-low expression; second, Pearson correlation analysis was applied to assess inter-feature correlations, and for any feature pair with a correlation coefficient ≥ 0.90, only one was retained to reduce redundancy; third, mRMR was employed to select the most discriminative feature subset from the remaining candidates; finally, Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation was used for hyperparameter tuning. The final optimal feature subset was used to construct three predictive models: a US-only model, a CEUS-only model, and a combined US+CEUS model. Clinical model construction: Univariate analysis was first performed on clinical variables (age, menopausal status, tumor stage, lesion location, family history, ALN status). Variables with P < 0.05 were entered into multivariate logistic regression to identify independent predictors, which were then used to build a clinical prediction model. Nomogram development: A final nomogram was constructed by integrating significant clinical predictors from multivariate logistic regression with the highest-performing radiomics model to predict HER2-low status in breast cancer. 3. Statistical analysis Statistical analyses were performed using R software and SPSS 26.0, with histopathology as the reference standard. Continuous variables were tested for normality and homogeneity of variance. Normally distributed data were compared using independent-sample t-tests; non-normally distributed data were analyzed with Mann–Whitney U tests. Categorical variables were compared using chi-square tests. Model performance was evaluated via receiver operating characteristic (ROC) curves, reporting area under the curve (AUC), accuracy, sensitivity, specificity, precision, positive predictive value (PPV), negative predictive value (NPV), recall, threshold and F1-score. Differences between models were assessed using DeLong’s test. Calibration curves evaluated nomogram reliability, and decision curve analysis (DCA) assessed clinical utility. A P-value < 0.05 was considered statistically significant. 4. Results Baseline characteristic A total of 175 patients with pathologically confirmed breast cancer during hospitalization were ultimately included. Pathological results showed that 91 patients (52.0%) had HER2-low expression and 84 patients (48.0%) had HER2-zero expression. This near-even distribution reflects the growing clinical relevance of distinguishing between these two subtypes, especially in light of emerging HER2-targeted therapies for HER2-low disease. Figure 2 shows the patient recruitment workflow. All cases were randomly assigned to a training cohort and a test cohort at a 7:3 ratio. Based on IHC or FISH results, the training cohort included 122 patients, while the test cohort comprised 53 patients. Differences in baseline characteristics between the two cohorts are presented in Table 1 . The mean age was 50.74 ± 10.77 years in the training group and 49.62 ± 10.75 years in the test group. The number of postmenopausal patients was 25 (47.2%) in the training group and 48 (39.3%) in the test group. This slight imbalance in menopausal status did not reach statistical significance, consistent with the overall comparability of the two groups. No statistically significant differences were observed between the two groups in terms of age, menopausal status, BI-RADS category, tumor maximum diameter, margin, orientation, posterior acoustic features, clinical stage, or other ultrasound characteristics (all P > 0.05). Table 1 Comparison of clinical and ultrasound morphological features between the training and testing cohorts. Features name Train Test P value age 49.62 ± 10.75 50.74 ± 10.77 0.446 length 1.97 ± 1.03 1.94 ± 1.00 0.902 widen 1.60 ± 0.89 1.48 ± 0.72 0.519 high 1.22 ± 0.55 1.20 ± 0.59 0.691 BI RADS 3.64 ± 0.93 3.68 ± 0.94 0.802 postmenopause 0.425 Premenopausal 74(60.66) 28(52.83) Postmenopausal 48(39.34) 25(47.17) Clinical Staging 0.943 I 69(56.56) 31(58.49) II 53(43.44) 22(41.51) Position 0.936 outerupperquadrant 58(47.54) 25(47.17) outerlowerquadrant 23(18.85) 12(22.64) upperinnerquadrant 7(5.74) 3(5.66) lowerinnerquadrant 34(27.87) 13(24.53) ALN 0.205 Yes 57(46.72) 31(58.49) No 65(53.28) 22(41.51) Direction 0.936 parallel 62(50.82) 28(52.83) Non-parallel 60(49.18) 25(47.17) Margin 0.088 clear 27(22.13) 19(35.85) fuzzy 95(77.87) 34(64.15) Posterior Acoustic Features 0.730 No Posterior Acoustic Features 49(40.16) 24(45.28) Posterior Acoustic Shadowing 60(49.18) 25(47.17) Posterior Acoustic Enhancement 13(10.66) 4(7.55) Microcalcification 0.760 No 76(62.30) 31(58.49) Yes 46(37.70) 22(41.51) The range was enlarged after enhancement 0.638 Stable 31(25.41) 16(30.19) Enlargement 91(74.59) 37(69.81) Perfusion defects 0.52 No 35(28.69) 12(22.64) Yes 87(71.31) 41(77.36) Direction of enhancement 1.0 No 17(13.93) 7(13.21) Yes 105(86.07) 46(86.79) Distribution of contrast media 0.55 No 37(30.33) 13(24.53) Yes 85(69.67) 40(75.47) Magnitude of enhancement 0.234 No 6(4.92) Yes 116(95.08) 53(100.00) crab foot sign 1.0 No 49(40.16) 22(41.51) Yes 73(59.84) 31(58.49) Clinical Model Clinical characteristics of 122 patients in the training cohort were included in a multivariate logistic regression analysis (Table 2 ). Univariate analysis found ALN status (OR = 2.286, 95% CI: 1.350–3.873; P = 0.01) and microcalcification (OR = 2.500, 95% CI: 1.537–4.067; P = 0.002) significantly associated with HER2 - low expression. Both were entered into the model. Final analysis showed microcalcification was independently associated with HER2 - low expression (adjusted OR = 2.053, 95% CI: 1.169–3.607; P = 0.036). This finding suggests that the presence of microcalcifications on conventional ultrasound may serve as a non-invasive imaging biomarker for HER2-low status, potentially aiding in preoperative risk stratification. Given the increasing therapeutic implications of HER2-low classification, such readily assessable sonographic features could streamline patient selection for targeted therapies without requiring repeated biopsies. Table 2 shows the results of univariate and multivariate analyses of the potential predictors based on the training cohort. Table 2 Results of the univariate and multivariate analyses based on the training cohort. Feature name Uni OR P value Multi OR P value The range was enlarged after enhancement 1.068 (0.757–1.508) 0.753 — — Distribution of contrast media 0.889 (0.622–1.271) 0.588 — — Direction of enhancement 0.909 (0.659–1.254) 0.626 — — Perfusion defects 0.851 (0.598–1.213) 0.453 — — Magnitude of enhancement 0.966 (0.712–1.311) 0.853 — — Crab foot sign 0.973 (0.662–1.430) 0.907 — — age 0.999 (0.994–1.005) 0.861 — — postmenopause 1.000 (0.622–1.608) 1 — — Clinical Staging 0.925 (0.760–1.125) 0.512 — — BI RADS 0.995 (0.919–1.078) 0.923 — — length 0.979 (0.856–1.120) 0.797 — — Margin 1.021 (0.729–1.432) 0.918 — — Posterior Acoustic Features 0.931 (0.682–1.271) 0.706 — — Position 1.035 (0.919–1.166) 0.637 — — Direction 1.308 (0.852–2.008) 0.303 — — ALN 2.286 (1.350–3.873) 0.01 1.510 (0.815–2.795) 0.271 Microcalcification 2.500 (1.537–4.067) 0.002 2.053 (1.169–3.607) 0.036 The range was enlarged after enhancement 1.068 (0.757–1.508) 0.753 Distribution of contrast media 0.889 (0.622–1.271) 0.588 Direction of enhancement 0.909 (0.659–1.254) 0.626 Perfusion defects 0.851 (0.598–1.213) 0.453 Magnitude of enhancement 0.966 (0.712–1.311) 0.853 Crab foot sign 0.973 (0.662–1.430) 0.907 Radiomics feature selection Features significantly associated with HER2-low expression were first identified using independent-samples t-tests. Pearson correlation analysis was then applied to remove highly redundant features (feature pairs with a correlation coefficient ≥ 0.90). Subsequently, the mRMR algorithm was used to select the most discriminative features. Finally, the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm was employed to identify the subset of features with the highest predictive value for model development, ultimately constructing a predictive model for HER2-low breast cancer (Fig. 3). Finally, ten retained non-zero coefficients were feature after dimensionality reduction (sFigure 1). Model Performance Evaluation The supplementary materials present a comprehensive comparison of radiomics models based on CEUS and conventional US, incorporating peritumoral margins of 2 mm and 3 mm in addition to the intratumoral region. Shown in Fig. 4, across all configurations, the Random Forest algorithm was employed to ensure consistent model architecture for fair comparison. In the training cohort, both CEUS- and US-based radiomics models demonstrated strong discriminative performance, with AUCs exceeding 0.87 across all peritumoral margin configurations. Specifically, CEUS models achieved AUCs of 0.892 (95% CI: 0.836–0.947) for the 3-mm margin and 0.870 (95% CI: 0.808–0.931) for the 2-mm margin. Similarly, US models yielded AUCs of 0.887 (95% CI: 0.827–0.947) (3-mm) and 0.895 (95% CI: 0.837–0.953) (2-mm), indicating excellent in-sample fitting regardless of imaging modality or margin width. However, a pronounced divergence emerged in the test cohort, highlighting stark differences in generalizability. CEUS models maintained robust performance: the 3-mm configuration achieved an AUC of 0.802 (95% CI: 0.677–0.926) with 81.1% accuracy, and the 2-mm version retained an AUC of 0.756 (95% CI: 0.623–0.888) with 71.7% accuracy. In contrast, US models exhibited substantial performance degradation—AUCs dropped to 0.670 (95% CI: 0.513–0.827; accuracy: 69.8%) for the 3-mm model and 0.674 (95% CI: 0.523–0.825; accuracy: 66.0%) for the 2-mm model. Exact result shown in sfigure2, sTable1 and sTable 2. These findings indicate that, despite comparable training performance—suggesting adequate model capacity—CEUS-derived radiomic signatures demonstrate markedly superior generalizability than those from conventional B-mode ultrasound. The significant decline in US model performance on external validation strongly suggests overfitting, likely due to the inherent instability of US features, which are highly sensitive to operator technique, gain settings, tissue attenuation, and other technical variables. In contrast, CEUS leverages microbubble-enhanced perfusion signals to capture biologically meaningful microvascular characteristics, yielding more reproducible and less redundant features that enhance model robustness. Consequently, the CEUS + tumor and peritumoral 3-mm model emerged as the optimal predictor, outperforming the best-performing US model (US + peritumoral 2-mm) in both AUC (0.802 vs. 0.674) and overall reliability. Table 3 Performance of the three prediction models in training and test cohort. Model name Accuracy AUC 95% CI Sensitivity Specificity PPV NPV Precision Recall F1 Threshold Clinic-train 0.721 0.720 0.6394–0.8002 0.690 0.750 0.714 0.727 0.714 0.690 0.702 0.720 IntraPeri3mm -train 0.803 0.892 0.8363–0.9471 0.793 0.812 0.793 0.812 0.793 0.793 0.793 0.489 Combined-train 0.869 0.921 0.8731–0.9690 0.862 0.875 0.862 0.875 0.862 0.862 0.862 0.501 Clinic-test 0.717 0.733 0.6119–0.8548 0.667 0.800 0.846 0.593 0.846 0.667 0.746 0.720 IntraPeri3mm -test 0.811 0.802 0.6770–0.9260 0.939 0.600 0.795 0.857 0.795 0.939 0.861 0.378 Combined-test 0.830 0.850 0.7384–0.9616 0.879 0.750 0.853 0.789 0.853 0.879 0.866 0.337 5. Discussion Traditionally, HER2 expression in breast cancer has been dichotomized as either negative or positive[25]. However, with evolving clinical and biological insights, HER2 status has been further refined in practice. HER2-low breast cancer—once regarded merely as a pathological category—has now emerged as a critical therapeutic target[26]. It is recognized as a distinct molecular, pathological, and clinical subtype, defined as tumors with an immunohistochemistry (IHC) score of 1 + or 2 + and negative FISH results. Notably, there was a trial demonstrated that trastuzumab deruxtecan confers significantly superior outcomes compared to standard chemotherapy in patients with HER2-low breast cancer, challenging the long-held notion that HER2-low tumors are unresponsive to targeted therapy[10, 27]. Nevertheless, the current gold standard for assessing HER2 status—IHC—is an invasive procedure with inherent clinical limitations. To address this, we developed and validated a non-invasive ultrasound-based radiomics nomogram model that integrates intratumoral and peritumoral features to preoperatively predict HER2-low status. Our CEUS–based intratumoral and peritumoral radiomics model demonstrated markedly superior predictive performance compared to the conventional ultrasound model, thereby offering a novel, non-invasive approach for the precise diagnosis and management of HER2-low breast cancer. An increasing body of evidence suggests that subtle structural alterations in tissue adjacent to tumors may provide crucial insights into tumor biology[28]. Medical imaging enables non-invasive capture of tumor heterogeneity, and radiomics facilitates high-throughput extraction of quantitative image features for in-depth tumor characterization[29, 30]. For instance, Ding et al. developed a model incorporating peritumoral features to predict sentinel lymph node (SLN) status in breast cancer and investigated the impact of peritumoral region size[31]. Similarly, Mao et al. designed a multimodal fusion model integrating both intratumoral and peritumoral radiomics features to predict response to neoadjuvant chemotherapy[32]. Collectively, these studies—and ours—support the efficacy of peritumoral features, either alone or in combination with intratumoral features, for predictive modeling. In our study, peritumoral regions of 2 mm and 3 mm demonstrated predictive value for HER2-low status. The CEUS + tumor and peritumoral 3-mm model demonstrated diagnostic performance in the training cohort that was broadly comparable to that of the US + tumor and peritumoral 2-mm model; however, it achieved significantly superior performance in the test cohort. This overfitting in the conventional ultrasound (US) model is likely attributable to the inherent variability of B-mode imaging, which is susceptible to operator-dependent factors such as probe pressure, gain settings, and tissue attenuation. These technical inconsistencies lead to unstable grayscale intensities and poor reproducibility of extracted texture and intensity features. Consequently, subsequent analyses focused on the 3-mm peritumoral region. This zone harbors biologically relevant information, including angiogenesis, lymphangiogenesis activity, peritumoral lymphovascular invasion, immune cell infiltration, edema, and cytokine release[33, 34]. While the optimal peritumoral margin for predicting HER2 status remains undefined, our findings underscore the utility of the 3-mm peritumoral region in this context. Our integrated radiomics model—combining intratumoral and 3-mm peritumoral features—achieved high diagnostic accuracy, with AUCs of 0.892 in the training cohort and 0.802 in the test cohort, establishing it as the optimal combined model. This provides valuable insights into peritumoral modeling in both US and CEUS imaging, with CEUS-based peritumoral features showing particularly robust performance. A key innovation of this study lies in the incorporation of CEUS-derived peritumoral features into radiomics analysis. Our results indicate that CEUS peritumoral characteristics possess independent predictive value for differentiating HER2-low from HER2-zero status. Compared to conventional US, CEUS not only depicts morphological features but also dynamically visualizes microvascular perfusion, thereby providing complementary functional information[19, 35, 36]. Prior studies have linked CEUS parameters to prognostic factors in breast cancer, suggesting that tumor microcirculation may reflect underlying molecular subtypes[37]. Thus, the intratumoral and peritumoral CEUS features extracted in our study likely captured subtle differences in microvascular perfusion associated with HER2-low expression—information largely inaccessible via US alone. We also constructed a clinical prediction model incorporating microcalcification, achieving AUCs of 0.720 (95% CI: 0.6394–0.8002) in the training cohort and 0.733 (95% CI: 0.6119–0.8548) in the test cohort. Univariate and multivariate logistic regression analyses confirmed microcalcification as an independent risk factor for HER2-low breast cancer. Although the molecular mechanism remains incompletely understood, it may involve the SPCA2–Orai1 pathway, wherein HER2 amplification activates Orai channels, triggering intracellular calcium release[38]. Our finding aligns with Zhang et al., who similarly identified microcalcification as a risk factor for HER2-low disease[8]. However, most conventional ultrasound features—both US and CEUS—are subject to substantial interobserver variability due to reliance on subjective interpretation by sonographers, which may partly explain the modest performance of our clinical model. To enhance predictive accuracy, we integrated radiomics features with clinical data and developed a comprehensive nomogram by combining the optimal radiomics model with the clinical model. Our results demonstrate that this integrated approach significantly improves preoperative prediction of HER2-low status compared to the clinical model alone. The combined model achieved higher AUCs in both training and test cohorts than any individual model, underscoring its superior performance. Furthermore, DCA and calibration curves confirmed its strong clinical utility and reliability for preoperative HER2-low prediction. This study has several limitations that should be acknowledged. First, the relatively small sample size may introduce selection bias and limit the generalizability of our findings. Second, the number of clinical variables included was limited, which is partly attributable to the nascent stage of HER2-low breast cancer research—key clinical predictors may not yet be fully established or captured in our cohort. Third, although regions of interest were manually delineated independently by two radiologists with consensus adjudication, this process remains inherently subjective; future studies could benefit from automated or AI-assisted segmentation to enhance reproducibility and reduce interobserver variability. Fourth, as a single-center retrospective analysis, our study population is relatively homogeneous, potentially limiting external validity despite robust internal validation performance. Finally, while multimodal ultrasound techniques are known to provide complementary diagnostic information and improve accuracy, only US and contrast-enhanced ultrasound were utilized in this work; incorporating (additional ultrasound modalities in future research may further refine predictive performance. Abbreviations US B-mode ultrasound CEUS Contrast-enhanced ultrasound mRMR Max-Relevance and Min-Redundancy IHC immunohistochemistry FISH fluorescence in situ hybridization ER estrogen receptor PR progesterone receptor HER2 human epidermal growth factor receptor 2 ROI Region of interest GLCM Gray-Level Co-occurrence Matrix GLRLM Gray-Level Run-Length Matrix GLSZM Gray-Level Size Zone Matrix GLDM Gray-Level Dependence Matrix NGTDM Neighborhood Gray-Tone Difference Matrix () ALN axillary lymph node LASSO Least Absolute Shrinkage and Selection Operator AUC area under the curve DCA Decision curve analysis Declarations Funding None Author information Authors and Affiliations Department of ultrasound, the 1st affiliated hospital, Jiangxi Medical College, Nanchang University Yachen Wang, Meiqin Yu and Chun Li. Contributions YW collected data, performed the statistical analysis and wrote the first draft of the manuscript, LC revised the manuscript. MY contributed constructive suggestions. All authors contributed to the article and approved the submitted version. Corresponding author Correspondence to Chun Li Ethics declarations Conflict of interest The authors declare that they have no competing interests. Ethical approval The study design and protocol were approved by the Ethics Committee of the 1st affiliated hospital, Jiangxi Medical College, Nanchang University, (IIT20250747) and the requirement for written informed consent was waived. Consent for publication Not applicable. Author Contribution YW collected data, performed the statistical analysis and wrote the first draft of the manuscript, LC revised the manuscript. MY contributed constructive suggestions. All authors contributed to the article and approved the submitted version. Acknowledgements None. Data Availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. References Loibl S, Poortmans P, Morrow M, Denkert C, Curigliano G. Breast cancer. Lancet. 2021;397(10286):1750-69.https://doi.org/10.1016/s0140-6736(20)32381-3 Xiong X, Zheng LW, Ding Y, Chen YF, Cai YW, Wang LP, Huang L, Liu CC, Shao ZM, Yu KD. Breast cancer: pathogenesis and treatments. Signal Transduct Target Ther. 2025;10(1):49.https://doi.org/10.1038/s41392-024-02108-4 Lebeau A, Deimling D, Kaltz C, Sendelhofert A, Iff A, Luthardt B, Untch M, Löhrs U. Her-2/neu analysis in archival tissue samples of human breast cancer: comparison of immunohistochemistry and fluorescence in situ hybridization. J Clin Oncol. 2001;19(2):354-63.https://doi.org/10.1200/jco.2001.19.2.354 Prat A, Pineda E, Adamo B, Galván P, Fernández A, Gaba L, Díez M, Viladot M, Arance A, Muñoz M. Clinical implications of the intrinsic molecular subtypes of breast cancer. Breast. 2015;24 Suppl 2:S26-35.https://doi.org/10.1016/j.breast.2015.07.008 Barzaman K, Karami J, Zarei Z, Hosseinzadeh A, Kazemi MH, Moradi-Kalbolandi S, Safari E, Farahmand L. Breast cancer: Biology, biomarkers, and treatments. Int Immunopharmacol. 2020;84:106535.https://doi.org/10.1016/j.intimp.2020.106535 Marchiò C, Annaratone L, Marques A, Casorzo L, Berrino E, Sapino A. Evolving concepts in HER2 evaluation in breast cancer: Heterogeneity, HER2-low carcinomas and beyond. Semin Cancer Biol. 2021;72:123-35.https://doi.org/10.1016/j.semcancer.2020.02.016 Hamilton E, Shastry M, Shiller SM, Ren R. Targeting HER2 heterogeneity in breast cancer. Cancer Treat Rev. 2021;100:102286.https://doi.org/10.1016/j.ctrv.2021.102286 Zhang X, Wu S, Zu X, Li X, Zhang Q, Ren Y, Qian X, Tong S, Li H. Ultrasound-based radiomics nomogram for predicting HER2-low expression breast cancer. Front Oncol. 2024;14:1438923.https://doi.org/10.3389/fonc.2024.1438923 Yin L, Zhang Y, Wei X, Shaibu Z, Xiang L, Wu T, Zhang Q, Qin R, Shan X. Preliminary study on DCE-MRI radiomics analysis for differentiation of HER2-low and HER2-zero breast cancer. Front Oncol. 2024;14:1385352.https://doi.org/10.3389/fonc.2024.1385352 Modi S, Jacot W, Yamashita T, Sohn J, Vidal M, Tokunaga E, Tsurutani J, Ueno NT, Prat A, Chae YS, Lee KS, Niikura N, Park YH, Xu B, Wang X, Gil-Gil M, Li W, Pierga JY, Im SA, Moore HCF, Rugo HS, Yerushalmi R, Zagouri F, Gombos A, Kim SB, Liu Q, Luo T, Saura C, Schmid P, Sun T, Gambhire D, Yung L, Wang Y, Singh J, Vitazka P, Meinhardt G, Harbeck N, Cameron DA. Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer. N Engl J Med. 2022;387(1):9-20.https://doi.org/10.1056/NEJMoa2203690 Modi S, Park H, Murthy RK, Iwata H, Tamura K, Tsurutani J, Moreno-Aspitia A, Doi T, Sagara Y, Redfern C, Krop IE, Lee C, Fujisaki Y, Sugihara M, Zhang L, Shahidi J, Takahashi S. Antitumor Activity and Safety of Trastuzumab Deruxtecan in Patients With HER2-Low-Expressing Advanced Breast Cancer: Results From a Phase Ib Study. J Clin Oncol. 2020;38(17):1887-96.https://doi.org/10.1200/jco.19.02318 Idossa D, Borrero M, Blaes A. ERBB2-Low (Also Known as HER2-Low) Breast Cancer. JAMA Oncol. 2023;9(4):576.https://doi.org/10.1001/jamaoncol.2022.6889 Ahuja S, Khan AA, Zaheer S. Understanding the spectrum of HER2 status in breast cancer: From HER2-positive to ultra-low HER2. Pathol Res Pract. 2024;262:155550.https://doi.org/10.1016/j.prp.2024.155550 Robbins CJ, Bates KM, Rimm DL. HER2 testing: evolution and update for a companion diagnostic assay. Nat Rev Clin Oncol. 2025;22(6):408-23.https://doi.org/10.1038/s41571-025-01016-y Xu M, Liu Y, Zeng S, Li F. Development of an Ultrasound-Based Radiomics Nomogram for Preoperative Prediction of HER-2 Status in Invasive Breast Cancer. Acad Radiol. 2025;32(6):3160-9.https://doi.org/10.1016/j.acra.2024.12.059 Li H, Zhang CT, Shao HG, Pan L, Li Z, Wang M, Xu SH. Prediction models of breast cancer molecular subtypes based on multimodal ultrasound and clinical features. BMC Cancer. 2025;25(1):886.https://doi.org/10.1186/s12885-025-14233-6 Conti A, Duggento A, Indovina I, Guerrisi M, Toschi N. Radiomics in breast cancer classification and prediction. Semin Cancer Biol. 2021;72:238-50.https://doi.org/10.1016/j.semcancer.2020.04.002 Liu H, Xia H, Yin X, Qin A, Zhang W, Feng S, Jin J. Study on the Differentiation of Infiltrating Breast Cancer Molecular Subtypes Based on Ultrasound Radiomics. Clin Breast Cancer. 2025;25(4):e450-e60.https://doi.org/10.1016/j.clbc.2025.01.005 Gong X, Li Q, Gu L, Chen C, Liu X, Zhang X, Wang B, Sun C, Yang D, Li L, Wang Y. Conventional ultrasound and contrast-enhanced ultrasound radiomics in breast cancer and molecular subtype diagnosis. Front Oncol. 2023;13:1158736.https://doi.org/10.3389/fonc.2023.1158736 Quan MY, Huang YX, Wang CY, Zhang Q, Chang C, Zhou SC. Deep learning radiomics model based on breast ultrasound video to predict HER2 expression status. Front Endocrinol (Lausanne). 2023;14:1144812.https://doi.org/10.3389/fendo.2023.1144812 Hu M, Zhang L, Wang X, Xiao X. Enhanced HER-2 prediction in breast cancer through synergistic integration of deep learning, ultrasound radiomics, and clinical data. Sci Rep. 2025;15(1):26992.https://doi.org/10.1038/s41598-025-12825-7 Wu J, Guo Y, Wu C, Wang Z, Sun Y, Xu D. Integration of Longitudinal and Transverse Radiomics from Ultrasound Images with Clinical Factors for HER-2 Status Prediction in Invasive Breast Cancer Patients. J Invest Surg. 2024;37(1):2436050.https://doi.org/10.1080/08941939.2024.2436050 Zhuo X, Lv J, Chen B, Liu J, Luo Y, Liu J, Xie X, Lu J, Zhao N. Combining conventional ultrasound and ultrasound elastography to predict HER2 status in patients with breast cancer. Front Physiol. 2023;14:1188502.https://doi.org/10.3389/fphys.2023.1188502 Zhou J, Jin AQ, Zhou SC, Li JW, Zhi WX, Huang YX, Zhu Q, Qian L, Wu J, Chang C. Application of preoperative ultrasound features combined with clinical factors in predicting HER2-positive subtype (non-luminal) breast cancer. BMC Med Imaging. 2021;21(1):184.https://doi.org/10.1186/s12880-021-00714-0 Wang J, Liu Y, Zhang Q, Li W, Feng J, Wang X, Fang J, Han Y, Xu B. Disitamab vedotin, a HER2-directed antibody-drug conjugate, in patients with HER2-overexpression and HER2-low advanced breast cancer: a phase I/Ib study. Cancer Commun (Lond). 2024;44(7):833-51.https://doi.org/10.1002/cac2.12577 Ivanova M, Porta FM, D'Ercole M, Pescia C, Sajjadi E, Cursano G, De Camilli E, Pala O, Mazzarol G, Venetis K, Guerini-Rocco E, Curigliano G, Viale G, Fusco N. Standardized pathology report for HER2 testing in compliance with 2023 ASCO/CAP updates and 2023 ESMO consensus statements on HER2-low breast cancer. Virchows Arch. 2024;484(1):3-14.https://doi.org/10.1007/s00428-023-03656-w Kang S, Kim SB. HER2-Low Breast Cancer: Now and in the Future. Cancer Res Treat. 2024;56(3):700-20.https://doi.org/10.4143/crt.2023.1138 Domínguez-Cejudo MA, Gil-Torralvo A, Cejuela M, Molina-Pinelo S, Salvador Bofill J. Targeting the Tumor Microenvironment in Breast Cancer: Prognostic and Predictive Significance and Therapeutic Opportunities. Int J Mol Sci. 2023;24(23).https://doi.org/10.3390/ijms242316771 Lambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P, Zegers CM, Gillies R, Boellard R, Dekker A, Aerts HJ. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012;48(4):441-6.https://doi.org/10.1016/j.ejca.2011.11.036 Liu Z, Wang S, Dong D, Wei J, Fang C, Zhou X, Sun K, Li L, Li B, Wang M, Tian J. The Applications of Radiomics in Precision Diagnosis and Treatment of Oncology: Opportunities and Challenges. Theranostics. 2019;9(5):1303-22.https://doi.org/10.7150/thno.30309 Ding J, Chen S, Serrano Sosa M, Cattell R, Lei L, Sun J, Prasanna P, Liu C, Huang C. 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.https://doi.org/10.1016/j.acra.2020.10.015 Mao N, Shi Y, Lian C, Wang Z, Zhang K, Xie H, Zhang H, Chen Q, Cheng G, Xu C, Dai Y. Intratumoral and peritumoral radiomics for preoperative prediction of neoadjuvant chemotherapy effect in breast cancer based on contrast-enhanced spectral mammography. Eur Radiol. 2022;32(5):3207-19.https://doi.org/10.1007/s00330-021-08414-7 Li C, Song L, Yin J. Intratumoral and Peritumoral Radiomics Based on Functional Parametric Maps from Breast DCE-MRI for Prediction of HER-2 and Ki-67 Status. J Magn Reson Imaging. 2021;54(3):703-14.https://doi.org/10.1002/jmri.27651 Uematsu T. Focal breast edema associated with malignancy on T2-weighted images of breast MRI: peritumoral edema, prepectoral edema, and subcutaneous edema. Breast Cancer. 2015;22(1):66-70.https://doi.org/10.1007/s12282-014-0572-9 Liang X, Li Z, Zhang L, Wang D, Tian J. Application of Contrast-Enhanced Ultrasound in the Differential Diagnosis of Different Molecular Subtypes of Breast Cancer. Ultrason Imaging. 2020;42(6):261-70.https://doi.org/10.1177/0161734620959780 Wan CF, Du J, Fang H, Li FH, Zhu JS, Liu Q. Enhancement patterns and parameters of breast cancers at contrast-enhanced US: correlation with prognostic factors. Radiology. 2012;262(2):450-9.https://doi.org/10.1148/radiol.11110789 Zhao YX, Liu S, Hu YB, Ge YY, Lv DM. Diagnostic and prognostic values of contrast-enhanced ultrasound in breast cancer: a retrospective study. Onco Targets Ther. 2017;10:1123-9.https://doi.org/10.2147/ott.S124134 Feng M, Grice DM, Faddy HM, Nguyen N, Leitch S, Wang Y, Muend S, Kenny PA, Sukumar S, Roberts-Thomson SJ, Monteith GR, Rao R. Store-independent activation of Orai1 by SPCA2 in mammary tumors. Cell. 2010;143(1):84-98.https://doi.org/10.1016/j.cell.2010.08.040 Additional Declarations No competing interests reported. Supplementary Files supplymentarymaterial.docx Cite Share Download PDF Status: Posted Version 1 posted 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-8786769","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":586246208,"identity":"f54b7ddb-4cd2-4231-a48b-29cc6e9898c1","order_by":0,"name":"Yachen Wang","email":"","orcid":"","institution":"First Affiliated Hospital of Nanchang University","correspondingAuthor":false,"prefix":"","firstName":"Yachen","middleName":"","lastName":"Wang","suffix":""},{"id":586246209,"identity":"5998eceb-6c2b-4348-a328-198483590ba8","order_by":1,"name":"meiqin Yu","email":"","orcid":"","institution":"First Affiliated Hospital of Nanchang University","correspondingAuthor":false,"prefix":"","firstName":"meiqin","middleName":"","lastName":"Yu","suffix":""},{"id":586246210,"identity":"0309f307-4b8a-450c-b726-7671968d5ac1","order_by":2,"name":"Chun Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA80lEQVRIiWNgGAWjYFCCBIYPDAY2zPaHmQ9ABA4Q1sI4g6EijZ3heFsCKVrOHOZnOHPGgDgt5uw5hs28bczSjDNyvj342cYgx3cjgfFzAR4tlj1vQFrYjJklcrcb9rYxGEveSGCWnoFHi8GNHPPHvG08yWwSudukGdsYEjfcSGBj5sGvBWSLRH2PRM4zkJZ64rTwnDFgluA5wwbSkmBAUMuZZ4WNcyoSmA3Y28wke85JGM4887BZGq+W48kbG94Y/Gc2YGZ+JvGjzEae73jywc/4tDAwcBgwISmQAGLGBrwaGBjYHzD+IKBkFIyCUTAKRjgAAAxRTPS+pjX5AAAAAElFTkSuQmCC","orcid":"","institution":"First Affiliated Hospital of Nanchang University","correspondingAuthor":true,"prefix":"","firstName":"Chun","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2026-02-04 12:55:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8786769/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8786769/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102296802,"identity":"902f3f26-3ddb-45c1-b692-77c9a99dae6d","added_by":"auto","created_at":"2026-02-10 10:21:48","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1500035,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Fig1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/d1d1a3d25a0e0b37642920a0.jpg"},{"id":102297138,"identity":"8bfebff6-e5d2-44c1-95d8-dec8b8b49802","added_by":"auto","created_at":"2026-02-10 10:26:02","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":482325,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Fig2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/0a8bdb64957aaf97f8592615.jpg"},{"id":102215003,"identity":"20ee49aa-2628-46be-960a-378060eb3638","added_by":"auto","created_at":"2026-02-09 12:48:01","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":843629,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Fig3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/b618ed70514cdddd4f3b7a8d.jpg"},{"id":102215008,"identity":"2984defa-1865-49e7-b7bc-d27bbdc18f87","added_by":"auto","created_at":"2026-02-09 12:48:01","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1683144,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Fig4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/b8012c8715c59ac072ed75cd.jpg"},{"id":102297648,"identity":"01a1f57a-06dc-4012-8ce2-a67f411e22b7","added_by":"auto","created_at":"2026-02-10 10:28:40","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":579359,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Fig5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/b2104141d63703acd125b8cd.jpg"},{"id":102215009,"identity":"6a3b7db2-32e4-4cb4-b125-3290a2363daf","added_by":"auto","created_at":"2026-02-09 12:48:01","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":642939,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Fig6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/a401e2248693e79b27651c71.jpg"},{"id":102215007,"identity":"6b728d8a-ae8a-43d5-9384-db0f5101d584","added_by":"auto","created_at":"2026-02-09 12:48:01","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":472205,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Fig7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/0d54fe3423bdd7f064f72926.jpg"},{"id":105838323,"identity":"c5518aeb-1e1d-4780-8f60-ac72be13be80","added_by":"auto","created_at":"2026-03-31 15:57:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6948576,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/f7188e33-4847-46f4-ade7-fd543897db1e.pdf"},{"id":102215005,"identity":"35aa9917-b7b9-458f-897d-3fec378cf148","added_by":"auto","created_at":"2026-02-09 12:48:01","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":517852,"visible":true,"origin":"","legend":"","description":"","filename":"supplymentarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-8786769/v1/75bc7623f4381a42787ec81f.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Development and Validation of a CEUS-Based Intratumoral–Peritumoral Radiomics Nomogram for Preoperative Prediction of HER2-Low Breast Cancer","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eBreast cancer remains one of the most commonly diagnosed malignancy in women worldwide and remains a leading cause of cancer-related mortality[1, 2]. Molecular subtyping based on immunohistochemistry (IHC) plays a pivotal role in guiding treatment strategies and traditionally categorizes breast cancer into Luminal A, Luminal B, HER2-positive, and triple-negative subtypes according to estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), and Ki-67 expression profiles[3–5]. Historically, HER2 expression has been dichotomized as either positive or negative. However, emerging biological and clinical evidence indicates that this binary classification oversimplifies HER2 biology. Tumors with an IHC score of 1 + or 2 + with negative fluorescence in situ hybridization (FISH) results are now recognized as a distinct subgroup—HER2-low breast cancer[6, 7]. Compared with HER2-zero tumors, HER2-low lesions display unique molecular characteristics, including enrichment of luminal-related gene expression signatures and reduced expression of ERBB2-driven pathways. Compared with HER2-low breast cancer, HER2-zero tumors tend to exhibit relatively clearer and more regular margins and shapes[8]. Yin et al. reported a positive correlation between HER2-low expression and pixel-based boundary length, suggesting greater morphological irregularity, reflecting increased intratumoral heterogeneity and aiding in the differentiation between HER2-low and HER2-zero subtypes[9]. Clinically, a clinical trial demonstrated that patients with HER2-low metastatic breast cancer derive substantial survival benefits from trastuzumab deruxtecan [10, 11], highlighting HER2-low as a therapeutically actionable subtype. Consequently, accurate preoperative identification of HER2-low status has become increasingly important for treatment planning in the era of antibody–drug conjugates.\u003c/p\u003e \u003cp\u003eCurrently, HER2 assessment relies on immunohistochemical evaluation of core biopsy samples, an invasive procedure subject to sampling error, tumor heterogeneity, and processing delays[12, 13]. These limitations underscore the need for rapid, reproducible, and noninvasive approaches capable of capturing both the anatomical and functional characteristics of breast tumors[14]. Conventional B-mode Ultrasound (US) is widely used in breast cancer detection and offers advantages such as accessibility, low cost, and real-time imaging. Contrast-enhanced ultrasound (CEUS) further enhances diagnostic capability by visualizing microvascular perfusion and tumor neoangiogenesis, providing functional information complementary to US imaging[15]. While conventional US provides morphological details, its ability to reflect underlying molecular heterogeneity remains limited due to reliance on subjective descriptors. In contrast, CEUS captures dynamic perfusion patterns that may correlate with HER2-driven angiogenic signaling—a key biological distinction between HER2-zero and HER2-low phenotypes[16].\u003c/p\u003e \u003cp\u003eRadiomics enables high-throughput extraction of quantitative imaging features and has shown promise in characterizing tumor heterogeneity, predicting molecular subtypes, evaluating tumor aggressiveness, and estimating prognosis[17, 18]. Although US- and CEUS-based radiomics models have demonstrated potential across various oncologic applications, research specifically addressing radiomic signatures associated with HER2-low breast cancer remains scarce[19]. Integrating radiomic signatures with readily available clinical variables—such as menopausal status, tumor size, and BI-RADS category—may improve generalizability and facilitate translation into routine clinical workflows, where purely imaging-based models may lack context[20–22]. Recent spatial transcriptomic and multiplex immunohistochemistry studies have revealed that HER2-low tumors exhibit distinct peritumoral microenvironments, characterized by altered immune cell infiltration, stromal fibrosis, and heterogeneous angiogenesis—features potentially detectable through CEUS due to its sensitivity to microvascular changes[23, 24]. Furthermore, the potential added value of integrating intratumoral and peritumoral radiomics features—including microvascular and stromal alterations adjacent to tumors—has not been fully explored in the context of HER2-low prediction.\u003c/p\u003e \u003cp\u003eTherefore, this study aims to compare the predictive performance of US– and CEUS–based radiomics models for differentiating HER2-zero from HER2-low breast cancer; evaluate the added value of incorporating peritumoral regions (2 mm and 3 mm) into radiomic feature extraction; and develop a clinical–radiomics model by integrating the optimal CEUS-derived radiomics signature with independent clinical predictors to enable noninvasive preoperative stratification of HER2 status.\u003c/p\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"2. Method and materials","content":"\u003cp\u003eThis retrospective study initially included 1,332 female patients with US- and CEUS-detected breast masses at the First Affiliated Hospital of Nanchang University between 2019 and 2025. After excluding 934 cases lacking histopathological confirmation or diagnosed with non-breast cancer lesions, 398 patients with surgically confirmed breast cancer were retained. Further exclusions were applied as follows: 42 non-mass lesions, 61 lesions \u0026ge;5 cm (incompatible with single-plane imaging), 57 patients who received pre-ultrasound neoadjuvant therapy or biopsy, and 63 cases with incomplete HER2 assessment (missing FISH validation for IHC 2+ results, heterogeneous FISH findings, or IHC 2+ without FISH testing). Ultimately, 175 patients met all inclusion criteria and were enrolled in the final analysis.\u003c/p\u003e\n\u003cp\u003e1.1 Inclusion criteria\u003c/p\u003e\n\u003cp\u003eAll participants should have histologically confirmed breast cancer with complete clinical, imaging, and pathological data. HER2 status was categorized into two groups: HER2-low (n = 91; IHC 1+ or IHC 2+/FISH-negative) and HER2-zero (n = 84; IHC 0). Patients were randomly assigned to a training cohort (n = 123) and test cohort (n = 52) at a 7:3 ratio, stratified by HER2 status to ensure balanced distribution.\u003c/p\u003e\n\u003cp\u003e1.2 Exclusion criteria\u003c/p\u003e\n\u003cp\u003ePreoperative US and CEUS examinations were performed within one week of surgery for all enrolled patients. Inclusion criteria for imaging assessments included age between 18\u0026ndash;80 years and no contraindications to SonoVue administration (e.g., allergy to contrast agents, acute coronary syndrome, severe arrhythmia, pulmonary hypertension, or respiratory distress syndrome). Exclusion criteria ensured homogeneity of the cohort by eliminating confounding factors such as prior interventions or ambiguous HER2 classifications.\u003c/p\u003e\n\u003cp\u003e2.1 Clinical data collection\u003c/p\u003e\n\u003cp\u003eThis retrospective study was approved by the Institutional Review Board, and the requirement for informed consent was waived. Clinical and pathological variables included age, menopausal status, clinical tumor stage, laterality and quadrant of the breast lesion, family history of cancer, and axillary lymph node (ALN) status. All patients had pathologically confirmed malignant breast tumors.\u003c/p\u003e\n\u003cp\u003e2.2 Ultrasound image acquisition\u003c/p\u003e\n\u003cp\u003ePreoperative ultrasound examinations were performed using either a Siemens Acuson system (Siemens Healthineers, Germany), equipped with 18L6 or 10L4 linear array transducers. All scans were conducted by experienced sonographers with at least 5 years of clinical expertise. Patients were placed in the supine position with both breasts fully exposed during scanning. For each lesion, the largest cross-sectional image was acquired, and standard ultrasound features were documented according to the Breast Imaging Reporting and Data System (BI-RADS).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eUS: Patients were placed in the supine position. Transverse and longitudinal scans of the target tumor were performed. Images of the largest tumor cross-section and characteristic 2D features were recorded. The imaging plane with the maximum tumor diameter and richest vascularity was selected.\u003c/p\u003e\n\u003cp\u003eCEUS: Prior to contrast injection, grayscale and color Doppler US were used to identify the imaging plane showing the largest lesion diameter and most prominent vascularity. A bolus of 4.8 mL SonoVue (sulfur hexafluoride microbubbles) was administered intravenously, followed by a 5 mL saline flush. The largest cross-sectional view of the tumor was clearly displayed, and the system was switched to real-time dual-display contrast mode. The probe remained stable during imaging. A timer and cine storage function were activated at contrast onset, and dynamic perfusion within the lesion was recorded continuously for 90 seconds. The peak-enhancement frame was selected for analysis.\u003c/p\u003e\n\u003cp\u003e2.3 IHC:\u003c/p\u003e\n\u003cp\u003eHER2-zero was defined as no membrane staining (IHC score 0). HER2-low was defined as weak and incomplete membrane staining without HER2 gene amplification by FISH (IHC score 1+ or 2+/FISH-negative).\u003c/p\u003e\n\u003cp\u003e2.4 Radiomics analysis\u003c/p\u003e\n\u003cp\u003eRegion of interest (ROI) delineation: Using ITK-SNAP software (version 3.8.0), two radiologists independently contoured the tumor boundaries on both 2D US and CEUS images, blinded to pathological results. In cases of disagreement, a senior radiologist with over 15 years of experience in breast ultrasound adjudicated to reach consensus. Shown in figure 1, ROIs encompassed the entire lesion.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRadiomics Feature Extraction, Selection, and Model Construction: Ultrasound image features were extracted using PyRadiomics. These features included first-order statistics and higher-order features, specifically from the Gray-Level Co-occurrence Matrix (GLCM), Gray-Level Run-Length Matrix (GLRLM), Gray-Level Size Zone Matrix (GLSZM), Gray-Level Dependence Matrix (GLDM), Neighborhood Gray-Tone Difference Matrix (NGTDM), as well as features derived from filtered images. Only features with an intraclass correlation coefficient greater than 0.75 were retained for further analysis. The extracted features were then standardized using Z-score normalization. Patients were randomly assigned to training and test cohort in a 7:3 ratio. Feature selection was performed sequentially: first, independent-samples t-tests were used to identify features significantly associated with HER2-low expression; second, Pearson correlation analysis was applied to assess inter-feature correlations, and for any feature pair with a correlation coefficient \u0026ge; 0.90, only one was retained to reduce redundancy; third, mRMR was employed to select the most discriminative feature subset from the remaining candidates; finally, Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation was used for hyperparameter tuning. The final optimal feature subset was used to construct three predictive models: a US-only model, a CEUS-only model, and a combined US+CEUS model.\u003c/p\u003e\n\u003cp\u003eClinical model construction: Univariate analysis was first performed on clinical variables (age, menopausal status, tumor stage, lesion location, family history, ALN status). Variables with P \u0026lt; 0.05 were entered into multivariate logistic regression to identify independent predictors, which were then used to build a clinical prediction model.\u003c/p\u003e\n\u003cp\u003eNomogram development: A final nomogram was constructed by integrating significant clinical predictors from multivariate logistic regression with the highest-performing radiomics model to predict HER2-low status in breast cancer.\u003c/p\u003e"},{"header":"3. Statistical analysis","content":"\u003cp\u003eStatistical analyses were performed using R software and SPSS 26.0, with histopathology as the reference standard. Continuous variables were tested for normality and homogeneity of variance. Normally distributed data were compared using independent-sample t-tests; non-normally distributed data were analyzed with Mann–Whitney U tests. Categorical variables were compared using chi-square tests. Model performance was evaluated via receiver operating characteristic (ROC) curves, reporting area under the curve (AUC), accuracy, sensitivity, specificity, precision, positive predictive value (PPV), negative predictive value (NPV), recall, threshold and F1-score. Differences between models were assessed using DeLong’s test. Calibration curves evaluated nomogram reliability, and decision curve analysis (DCA) assessed clinical utility. A P-value \u0026lt; 0.05 was considered statistically significant.\u003c/p\u003e"},{"header":"4. Results","content":"\u003cp\u003eBaseline characteristic\u003c/p\u003e\u003cp\u003eA total of 175 patients with pathologically confirmed breast cancer during hospitalization were ultimately included. Pathological results showed that 91 patients (52.0%) had HER2-low expression and 84 patients (48.0%) had HER2-zero expression. This near-even distribution reflects the growing clinical relevance of distinguishing between these two subtypes, especially in light of emerging HER2-targeted therapies for HER2-low disease. Figure\u0026nbsp;2 shows the patient recruitment workflow. All cases were randomly assigned to a training cohort and a test cohort at a 7:3 ratio. Based on IHC or FISH results, the training cohort included 122 patients, while the test cohort comprised 53 patients.\u003c/p\u003e\u003cp\u003eDifferences in baseline characteristics between the two cohorts are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The mean age was 50.74 ± 10.77 years in the training group and 49.62 ± 10.75 years in the test group. The number of postmenopausal patients was 25 (47.2%) in the training group and 48 (39.3%) in the test group. This slight imbalance in menopausal status did not reach statistical significance, consistent with the overall comparability of the two groups. No statistically significant differences were observed between the two groups in terms of age, menopausal status, BI-RADS category, tumor maximum diameter, margin, orientation, posterior acoustic features, clinical stage, or other ultrasound characteristics (all P \u0026gt; 0.05).\u003c/p\u003e\u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\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\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 and ultrasound morphological features between the training and testing cohorts.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFeatures name\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTrain\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTest\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eage\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49.62 ± 10.75\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50.74 ± 10.77\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.446\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003elength\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.97 ± 1.03\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.94 ± 1.00\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.902\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ewiden\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.60 ± 0.89\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.48 ± 0.72\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.519\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ehigh\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.22 ± 0.55\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.20 ± 0.59\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.691\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBI RADS\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.64 ± 0.93\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.68 ± 0.94\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.802\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003epostmenopause\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.425\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePremenopausal\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e74(60.66)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e28(52.83)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePostmenopausal\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e48(39.34)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25(47.17)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClinical Staging\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.943\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eI\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e69(56.56)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e31(58.49)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eII\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e53(43.44)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22(41.51)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosition\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.936\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eouterupperquadrant\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e58(47.54)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25(47.17)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eouterlowerquadrant\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e23(18.85)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12(22.64)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eupperinnerquadrant\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7(5.74)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3(5.66)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003elowerinnerquadrant\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e34(27.87)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13(24.53)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eALN\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.205\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e57(46.72)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e31(58.49)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e65(53.28)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22(41.51)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDirection\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.936\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eparallel\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e62(50.82)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e28(52.83)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNon-parallel\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e60(49.18)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25(47.17)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMargin\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.088\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eclear\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27(22.13)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19(35.85)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003efuzzy\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e95(77.87)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e34(64.15)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosterior Acoustic Features\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.730\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo Posterior Acoustic Features\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49(40.16)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24(45.28)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosterior Acoustic Shadowing\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e60(49.18)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25(47.17)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosterior Acoustic Enhancement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13(10.66)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4(7.55)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMicrocalcification\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.760\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e76(62.30)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e31(58.49)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e46(37.70)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22(41.51)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThe range was enlarged after\u003c/p\u003e \u003cp\u003eenhancement\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.638\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStable\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31(25.41)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e16(30.19)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEnlargement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e91(74.59)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e37(69.81)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerfusion defects\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.52\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e35(28.69)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12(22.64)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e87(71.31)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e41(77.36)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDirection of enhancement\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.0\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e17(13.93)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7(13.21)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e105(86.07)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e46(86.79)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDistribution of contrast media\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e37(30.33)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13(24.53)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e85(69.67)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e40(75.47)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMagnitude of enhancement\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.234\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6(4.92)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e116(95.08)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e53(100.00)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecrab foot sign\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.0\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49(40.16)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22(41.51)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e73(59.84)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e31(58.49)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/table\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eClinical Model\u003c/p\u003e\u003cp\u003eClinical characteristics of 122 patients in the training cohort were included in a multivariate logistic regression analysis (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Univariate analysis found ALN status (OR = 2.286, 95% CI: 1.350–3.873; P = 0.01) and microcalcification (OR = 2.500, 95% CI: 1.537–4.067; P = 0.002) significantly associated with HER2 - low expression. Both were entered into the model. Final analysis showed microcalcification was independently associated with HER2 - low expression (adjusted OR = 2.053, 95% CI: 1.169–3.607; P = 0.036). This finding suggests that the presence of microcalcifications on conventional ultrasound may serve as a non-invasive imaging biomarker for HER2-low status, potentially aiding in preoperative risk stratification. Given the increasing therapeutic implications of HER2-low classification, such readily assessable sonographic features could streamline patient selection for targeted therapies without requiring repeated biopsies. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the results of univariate and multivariate analyses of the potential predictors based on the training cohort.\u003c/p\u003e\u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\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\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\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\u003eResults of the univariate and multivariate analyses based on the training cohort.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFeature name\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUni OR\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMulti OR\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003eP value\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThe range was enlarged after enhancement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.068 (0.757–1.508)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.753\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDistribution of contrast media\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.889 (0.622–1.271)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.588\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDirection of\u003c/p\u003e \u003cp\u003eenhancement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.909 (0.659–1.254)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.626\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerfusion defects\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.851 (0.598–1.213)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.453\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMagnitude of\u003c/p\u003e \u003cp\u003eenhancement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.966 (0.712–1.311)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.853\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCrab foot sign\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.973 (0.662–1.430)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.907\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eage\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.999 (0.994–1.005)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.861\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003epostmenopause\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.000 (0.622–1.608)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClinical Staging\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.925 (0.760–1.125)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.512\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBI RADS\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.995 (0.919–1.078)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.923\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003elength\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.979 (0.856–1.120)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.797\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMargin\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.021 (0.729–1.432)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.918\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosterior Acoustic\u003c/p\u003e \u003cp\u003eFeatures\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.931 (0.682–1.271)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.706\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePosition\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.035 (0.919–1.166)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.637\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDirection\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.308 (0.852–2.008)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.303\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e—\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e—\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=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.286 (1.350–3.873)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e1.510 (0.815–2.795)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.271\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMicrocalcification\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.500 (1.537–4.067)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e \u003cp\u003e2.053 (1.169–3.607)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.036\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThe range was enlarged after enhancement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.068 (0.757–1.508)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.753\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e\u0026nbsp;\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\u003eDistribution of contrast media\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.889 (0.622–1.271)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.588\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e\u0026nbsp;\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\u003eDirection of\u003c/p\u003e \u003cp\u003eenhancement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.909 (0.659–1.254)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.626\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e\u0026nbsp;\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\u003ePerfusion defects\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.851 (0.598–1.213)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.453\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e\u0026nbsp;\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\u003eMagnitude\u003c/p\u003e \u003cp\u003eof enhancement\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.966 (0.712–1.311)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.853\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e\u0026nbsp;\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\u003eCrab foot sign\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.973 (0.662–1.430)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.907\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colspan=\"3\" nameend=\"c6\" namest=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/table\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eRadiomics feature selection\u003c/p\u003e\u003cp\u003eFeatures significantly associated with HER2-low expression were first identified using independent-samples t-tests. Pearson correlation analysis was then applied to remove highly redundant features (feature pairs with a correlation coefficient ≥ 0.90). Subsequently, the mRMR algorithm was used to select the most discriminative features. Finally, the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm was employed to identify the subset of features with the highest predictive value for model development, ultimately constructing a predictive model for HER2-low breast cancer (Fig.\u0026nbsp;3). Finally, ten retained non-zero coefficients were feature after dimensionality reduction (sFigure 1).\u003c/p\u003e\u003cp\u003e \u003c/p\u003e\u003cp\u003eModel Performance Evaluation\u003c/p\u003e\u003cp\u003eThe supplementary materials present a comprehensive comparison of radiomics models based on CEUS and conventional US, incorporating peritumoral margins of 2 mm and 3 mm in addition to the intratumoral region. Shown in Fig.\u0026nbsp;4, across all configurations, the Random Forest algorithm was employed to ensure consistent model architecture for fair comparison.\u003c/p\u003e\u003cp\u003eIn the training cohort, both CEUS- and US-based radiomics models demonstrated strong discriminative performance, with AUCs exceeding 0.87 across all peritumoral margin configurations. Specifically, CEUS models achieved AUCs of 0.892 (95% CI: 0.836–0.947) for the 3-mm margin and 0.870 (95% CI: 0.808–0.931) for the 2-mm margin. Similarly, US models yielded AUCs of 0.887 (95% CI: 0.827–0.947) (3-mm) and 0.895 (95% CI: 0.837–0.953) (2-mm), indicating excellent in-sample fitting regardless of imaging modality or margin width.\u003c/p\u003e\u003cp\u003eHowever, a pronounced divergence emerged in the test cohort, highlighting stark differences in generalizability. CEUS models maintained robust performance: the 3-mm configuration achieved an AUC of 0.802 (95% CI: 0.677–0.926) with 81.1% accuracy, and the 2-mm version retained an AUC of 0.756 (95% CI: 0.623–0.888) with 71.7% accuracy. In contrast, US models exhibited substantial performance degradation—AUCs dropped to 0.670 (95% CI: 0.513–0.827; accuracy: 69.8%) for the 3-mm model and 0.674 (95% CI: 0.523–0.825; accuracy: 66.0%) for the 2-mm model. Exact result shown in sfigure2, sTable1 and sTable 2.\u003c/p\u003e\u003cp\u003eThese findings indicate that, despite comparable training performance—suggesting adequate model capacity—CEUS-derived radiomic signatures demonstrate markedly superior generalizability than those from conventional B-mode ultrasound. The significant decline in US model performance on external validation strongly suggests overfitting, likely due to the inherent instability of US features, which are highly sensitive to operator technique, gain settings, tissue attenuation, and other technical variables. In contrast, CEUS leverages microbubble-enhanced perfusion signals to capture biologically meaningful microvascular characteristics, yielding more reproducible and less redundant features that enhance model robustness. Consequently, the CEUS + tumor and peritumoral 3-mm model emerged as the optimal predictor, outperforming the best-performing US model (US + peritumoral 2-mm) in both AUC (0.802 vs. 0.674) and overall reliability.\u003c/p\u003e\u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\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\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\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\u003ePerformance of the three prediction models in training and test cohort.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"12\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel name\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePPV\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eNPV\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eRecall\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eF1\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c12\"\u003e \u003cp\u003eThreshold\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClinic-train\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.721\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.720\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.6394–0.8002\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.690\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.750\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.714\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.727\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.714\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e0.690\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.702\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e0.720\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntraPeri3mm -train\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.803\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.892\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.8363–0.9471\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.812\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.812\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e0.489\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCombined-train\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.869\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.921\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.8731–0.9690\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.862\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.875\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.862\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.875\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.862\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e0.862\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.862\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e0.501\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClinic-test\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.717\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.733\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.6119–0.8548\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.667\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.593\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e0.667\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.746\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e0.720\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntraPeri3mm -test\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.802\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.6770–0.9260\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.939\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.600\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.795\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.857\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.795\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e0.939\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.861\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e0.378\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCombined-test\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.850\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.7384–0.9616\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.879\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.750\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.853\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.789\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.853\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e0.879\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.866\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e0.337\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/table\u003e\u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eTraditionally, HER2 expression in breast cancer has been dichotomized as either negative or positive[25]. However, with evolving clinical and biological insights, HER2 status has been further refined in practice. HER2-low breast cancer\u0026mdash;once regarded merely as a pathological category\u0026mdash;has now emerged as a critical therapeutic target[26]. It is recognized as a distinct molecular, pathological, and clinical subtype, defined as tumors with an immunohistochemistry (IHC) score of 1\u0026thinsp;+\u0026thinsp;or 2\u0026thinsp;+\u0026thinsp;and negative FISH results. Notably, there was a trial demonstrated that trastuzumab deruxtecan confers significantly superior outcomes compared to standard chemotherapy in patients with HER2-low breast cancer, challenging the long-held notion that HER2-low tumors are unresponsive to targeted therapy[10, 27].\u003c/p\u003e \u003cp\u003eNevertheless, the current gold standard for assessing HER2 status\u0026mdash;IHC\u0026mdash;is an invasive procedure with inherent clinical limitations. To address this, we developed and validated a non-invasive ultrasound-based radiomics nomogram model that integrates intratumoral and peritumoral features to preoperatively predict HER2-low status. Our CEUS\u0026ndash;based intratumoral and peritumoral radiomics model demonstrated markedly superior predictive performance compared to the conventional ultrasound model, thereby offering a novel, non-invasive approach for the precise diagnosis and management of HER2-low breast cancer.\u003c/p\u003e \u003cp\u003eAn increasing body of evidence suggests that subtle structural alterations in tissue adjacent to tumors may provide crucial insights into tumor biology[28]. Medical imaging enables non-invasive capture of tumor heterogeneity, and radiomics facilitates high-throughput extraction of quantitative image features for in-depth tumor characterization[29, 30]. For instance, Ding et al. developed a model incorporating peritumoral features to predict sentinel lymph node (SLN) status in breast cancer and investigated the impact of peritumoral region size[31]. Similarly, Mao et al. designed a multimodal fusion model integrating both intratumoral and peritumoral radiomics features to predict response to neoadjuvant chemotherapy[32]. Collectively, these studies\u0026mdash;and ours\u0026mdash;support the efficacy of peritumoral features, either alone or in combination with intratumoral features, for predictive modeling.\u003c/p\u003e \u003cp\u003eIn our study, peritumoral regions of 2 mm and 3 mm demonstrated predictive value for HER2-low status. The CEUS\u0026thinsp;+\u0026thinsp;tumor and peritumoral 3-mm model demonstrated diagnostic performance in the training cohort that was broadly comparable to that of the US\u0026thinsp;+\u0026thinsp;tumor and peritumoral 2-mm model; however, it achieved significantly superior performance in the test cohort. This overfitting in the conventional ultrasound (US) model is likely attributable to the inherent variability of B-mode imaging, which is susceptible to operator-dependent factors such as probe pressure, gain settings, and tissue attenuation. These technical inconsistencies lead to unstable grayscale intensities and poor reproducibility of extracted texture and intensity features. Consequently, subsequent analyses focused on the 3-mm peritumoral region. This zone harbors biologically relevant information, including angiogenesis, lymphangiogenesis activity, peritumoral lymphovascular invasion, immune cell infiltration, edema, and cytokine release[33, 34]. While the optimal peritumoral margin for predicting HER2 status remains undefined, our findings underscore the utility of the 3-mm peritumoral region in this context.\u003c/p\u003e \u003cp\u003eOur integrated radiomics model\u0026mdash;combining intratumoral and 3-mm peritumoral features\u0026mdash;achieved high diagnostic accuracy, with AUCs of 0.892 in the training cohort and 0.802 in the test cohort, establishing it as the optimal combined model. This provides valuable insights into peritumoral modeling in both US and CEUS imaging, with CEUS-based peritumoral features showing particularly robust performance.\u003c/p\u003e \u003cp\u003eA key innovation of this study lies in the incorporation of CEUS-derived peritumoral features into radiomics analysis. Our results indicate that CEUS peritumoral characteristics possess independent predictive value for differentiating HER2-low from HER2-zero status. Compared to conventional US, CEUS not only depicts morphological features but also dynamically visualizes microvascular perfusion, thereby providing complementary functional information[19, 35, 36]. Prior studies have linked CEUS parameters to prognostic factors in breast cancer, suggesting that tumor microcirculation may reflect underlying molecular subtypes[37]. Thus, the intratumoral and peritumoral CEUS features extracted in our study likely captured subtle differences in microvascular perfusion associated with HER2-low expression\u0026mdash;information largely inaccessible via US alone.\u003c/p\u003e \u003cp\u003eWe also constructed a clinical prediction model incorporating microcalcification, achieving AUCs of 0.720 (95% CI: 0.6394\u0026ndash;0.8002) in the training cohort and 0.733 (95% CI: 0.6119\u0026ndash;0.8548) in the test cohort. Univariate and multivariate logistic regression analyses confirmed microcalcification as an independent risk factor for HER2-low breast cancer. Although the molecular mechanism remains incompletely understood, it may involve the SPCA2\u0026ndash;Orai1 pathway, wherein HER2 amplification activates Orai channels, triggering intracellular calcium release[38]. Our finding aligns with Zhang et al., who similarly identified microcalcification as a risk factor for HER2-low disease[8]. However, most conventional ultrasound features\u0026mdash;both US and CEUS\u0026mdash;are subject to substantial interobserver variability due to reliance on subjective interpretation by sonographers, which may partly explain the modest performance of our clinical model.\u003c/p\u003e \u003cp\u003eTo enhance predictive accuracy, we integrated radiomics features with clinical data and developed a comprehensive nomogram by combining the optimal radiomics model with the clinical model. Our results demonstrate that this integrated approach significantly improves preoperative prediction of HER2-low status compared to the clinical model alone. The combined model achieved higher AUCs in both training and test cohorts than any individual model, underscoring its superior performance. Furthermore, DCA and calibration curves confirmed its strong clinical utility and reliability for preoperative HER2-low prediction.\u003c/p\u003e \u003cp\u003eThis study has several limitations that should be acknowledged. First, the relatively small sample size may introduce selection bias and limit the generalizability of our findings. Second, the number of clinical variables included was limited, which is partly attributable to the nascent stage of HER2-low breast cancer research\u0026mdash;key clinical predictors may not yet be fully established or captured in our cohort. Third, although regions of interest were manually delineated independently by two radiologists with consensus adjudication, this process remains inherently subjective; future studies could benefit from automated or AI-assisted segmentation to enhance reproducibility and reduce interobserver variability. Fourth, as a single-center retrospective analysis, our study population is relatively homogeneous, potentially limiting external validity despite robust internal validation performance. Finally, while multimodal ultrasound techniques are known to provide complementary diagnostic information and improve accuracy, only US and contrast-enhanced ultrasound were utilized in this work; incorporating (additional ultrasound modalities in future research may further refine predictive performance.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eUS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eB-mode ultrasound\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eCEUS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eContrast-enhanced ultrasound\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003emRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eMax-Relevance and Min-Redundancy\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eIHC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eimmunohistochemistry\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eFISH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003efluorescence in situ hybridization\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eER\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eestrogen receptor\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003ePR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eprogesterone receptor\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eHER2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003ehuman epidermal growth factor receptor 2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eROI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eRegion of interest\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eGLCM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eGray-Level Co-occurrence Matrix\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eGLRLM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eGray-Level Run-Length Matrix\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eGLSZM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eGray-Level Size Zone Matrix\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eGLDM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eGray-Level Dependence Matrix\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eNGTDM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eNeighborhood Gray-Tone Difference Matrix ()\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eALN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eaxillary lymph node\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eLASSO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eLeast Absolute Shrinkage and Selection Operator\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003earea under the curve\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eDCA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 459px;\"\u003e\n \u003cp\u003eDecision curve analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eNone\u003c/p\u003e \u003cp\u003eAuthor information\u003c/p\u003e \u003cp\u003eAuthors and Affiliations\u003c/p\u003e \u003cp\u003eDepartment of ultrasound, the 1st affiliated hospital, Jiangxi Medical College, Nanchang University\u003c/p\u003e \u003cp\u003eYachen Wang, Meiqin Yu and Chun Li.\u003c/p\u003e \u003cp\u003eContributions\u003c/p\u003e \u003cp\u003eYW collected data, performed the statistical analysis and wrote the first draft of the manuscript, LC revised the manuscript. MY contributed constructive suggestions. All authors contributed to the article and approved the submitted version.\u003c/p\u003e \u003cp\u003eCorresponding author\u003c/p\u003e \u003cp\u003eCorrespondence to Chun Li\u003c/p\u003e \u003cp\u003eEthics declarations\u003c/p\u003e \u003cp\u003eConflict of interest\u003c/p\u003e \u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e \u003cp\u003eEthical approval\u003c/p\u003e \u003cp\u003e The study design and protocol were approved by the Ethics Committee of the 1st affiliated hospital, Jiangxi Medical College, Nanchang University, (IIT20250747) and the requirement for written informed consent was waived.\u003c/p\u003e \u003cp\u003e Consent for publication\u003c/p\u003e \u003cp\u003eNot applicable.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eYW collected data, performed the statistical analysis and wrote the first draft of the manuscript, LC revised the manuscript. MY contributed constructive suggestions. All authors contributed to the article and approved the submitted version.\u003c/p\u003e\u003ch2\u003eAcknowledgements\u003c/h2\u003e \u003cp\u003eNone.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLoibl S, Poortmans P, Morrow M, Denkert C, Curigliano G. Breast cancer. Lancet. 2021;397(10286):1750-69.https://doi.org/10.1016/s0140-6736(20)32381-3 \u003c/li\u003e\n\u003cli\u003eXiong X, Zheng LW, Ding Y, Chen YF, Cai YW, Wang LP, Huang L, Liu CC, Shao ZM, Yu KD. Breast cancer: pathogenesis and treatments. Signal Transduct Target Ther. 2025;10(1):49.https://doi.org/10.1038/s41392-024-02108-4 \u003c/li\u003e\n\u003cli\u003eLebeau A, Deimling D, Kaltz C, Sendelhofert A, Iff A, Luthardt B, Untch M, L\u0026ouml;hrs U. Her-2/neu analysis in archival tissue samples of human breast cancer: comparison of immunohistochemistry and fluorescence in situ hybridization. J Clin Oncol. 2001;19(2):354-63.https://doi.org/10.1200/jco.2001.19.2.354 \u003c/li\u003e\n\u003cli\u003ePrat A, Pineda E, Adamo B, Galv\u0026aacute;n P, Fern\u0026aacute;ndez A, Gaba L, D\u0026iacute;ez M, Viladot M, Arance A, Mu\u0026ntilde;oz M. Clinical implications of the intrinsic molecular subtypes of breast cancer. Breast. 2015;24 Suppl 2:S26-35.https://doi.org/10.1016/j.breast.2015.07.008 \u003c/li\u003e\n\u003cli\u003eBarzaman K, Karami J, Zarei Z, Hosseinzadeh A, Kazemi MH, Moradi-Kalbolandi S, Safari E, Farahmand L. Breast cancer: Biology, biomarkers, and treatments. Int Immunopharmacol. 2020;84:106535.https://doi.org/10.1016/j.intimp.2020.106535 \u003c/li\u003e\n\u003cli\u003eMarchi\u0026ograve; C, Annaratone L, Marques A, Casorzo L, Berrino E, Sapino A. Evolving concepts in HER2 evaluation in breast cancer: Heterogeneity, HER2-low carcinomas and beyond. Semin Cancer Biol. 2021;72:123-35.https://doi.org/10.1016/j.semcancer.2020.02.016 \u003c/li\u003e\n\u003cli\u003eHamilton E, Shastry M, Shiller SM, Ren R. Targeting HER2 heterogeneity in breast cancer. Cancer Treat Rev. 2021;100:102286.https://doi.org/10.1016/j.ctrv.2021.102286 \u003c/li\u003e\n\u003cli\u003eZhang X, Wu S, Zu X, Li X, Zhang Q, Ren Y, Qian X, Tong S, Li H. Ultrasound-based radiomics nomogram for predicting HER2-low expression breast cancer. Front Oncol. 2024;14:1438923.https://doi.org/10.3389/fonc.2024.1438923 \u003c/li\u003e\n\u003cli\u003eYin L, Zhang Y, Wei X, Shaibu Z, Xiang L, Wu T, Zhang Q, Qin R, Shan X. Preliminary study on DCE-MRI radiomics analysis for differentiation of HER2-low and HER2-zero breast cancer. Front Oncol. 2024;14:1385352.https://doi.org/10.3389/fonc.2024.1385352 \u003c/li\u003e\n\u003cli\u003eModi S, Jacot W, Yamashita T, Sohn J, Vidal M, Tokunaga E, Tsurutani J, Ueno NT, Prat A, Chae YS, Lee KS, Niikura N, Park YH, Xu B, Wang X, Gil-Gil M, Li W, Pierga JY, Im SA, Moore HCF, Rugo HS, Yerushalmi R, Zagouri F, Gombos A, Kim SB, Liu Q, Luo T, Saura C, Schmid P, Sun T, Gambhire D, Yung L, Wang Y, Singh J, Vitazka P, Meinhardt G, Harbeck N, Cameron DA. Trastuzumab Deruxtecan in Previously Treated HER2-Low Advanced Breast Cancer. N Engl J Med. 2022;387(1):9-20.https://doi.org/10.1056/NEJMoa2203690 \u003c/li\u003e\n\u003cli\u003eModi S, Park H, Murthy RK, Iwata H, Tamura K, Tsurutani J, Moreno-Aspitia A, Doi T, Sagara Y, Redfern C, Krop IE, Lee C, Fujisaki Y, Sugihara M, Zhang L, Shahidi J, Takahashi S. Antitumor Activity and Safety of Trastuzumab Deruxtecan in Patients With HER2-Low-Expressing Advanced Breast Cancer: Results From a Phase Ib Study. J Clin Oncol. 2020;38(17):1887-96.https://doi.org/10.1200/jco.19.02318 \u003c/li\u003e\n\u003cli\u003eIdossa D, Borrero M, Blaes A. ERBB2-Low (Also Known as HER2-Low) Breast Cancer. JAMA Oncol. 2023;9(4):576.https://doi.org/10.1001/jamaoncol.2022.6889 \u003c/li\u003e\n\u003cli\u003eAhuja S, Khan AA, Zaheer S. Understanding the spectrum of HER2 status in breast cancer: From HER2-positive to ultra-low HER2. Pathol Res Pract. 2024;262:155550.https://doi.org/10.1016/j.prp.2024.155550 \u003c/li\u003e\n\u003cli\u003eRobbins CJ, Bates KM, Rimm DL. HER2 testing: evolution and update for a companion diagnostic assay. Nat Rev Clin Oncol. 2025;22(6):408-23.https://doi.org/10.1038/s41571-025-01016-y \u003c/li\u003e\n\u003cli\u003eXu M, Liu Y, Zeng S, Li F. Development of an Ultrasound-Based Radiomics Nomogram for Preoperative Prediction of HER-2 Status in Invasive Breast Cancer. Acad Radiol. 2025;32(6):3160-9.https://doi.org/10.1016/j.acra.2024.12.059 \u003c/li\u003e\n\u003cli\u003eLi H, Zhang CT, Shao HG, Pan L, Li Z, Wang M, Xu SH. Prediction models of breast cancer molecular subtypes based on multimodal ultrasound and clinical features. BMC Cancer. 2025;25(1):886.https://doi.org/10.1186/s12885-025-14233-6 \u003c/li\u003e\n\u003cli\u003eConti A, Duggento A, Indovina I, Guerrisi M, Toschi N. Radiomics in breast cancer classification and prediction. Semin Cancer Biol. 2021;72:238-50.https://doi.org/10.1016/j.semcancer.2020.04.002 \u003c/li\u003e\n\u003cli\u003eLiu H, Xia H, Yin X, Qin A, Zhang W, Feng S, Jin J. Study on the Differentiation of Infiltrating Breast Cancer Molecular Subtypes Based on Ultrasound Radiomics. Clin Breast Cancer. 2025;25(4):e450-e60.https://doi.org/10.1016/j.clbc.2025.01.005 \u003c/li\u003e\n\u003cli\u003eGong X, Li Q, Gu L, Chen C, Liu X, Zhang X, Wang B, Sun C, Yang D, Li L, Wang Y. Conventional ultrasound and contrast-enhanced ultrasound radiomics in breast cancer and molecular subtype diagnosis. Front Oncol. 2023;13:1158736.https://doi.org/10.3389/fonc.2023.1158736 \u003c/li\u003e\n\u003cli\u003eQuan MY, Huang YX, Wang CY, Zhang Q, Chang C, Zhou SC. Deep learning radiomics model based on breast ultrasound video to predict HER2 expression status. Front Endocrinol (Lausanne). 2023;14:1144812.https://doi.org/10.3389/fendo.2023.1144812 \u003c/li\u003e\n\u003cli\u003eHu M, Zhang L, Wang X, Xiao X. Enhanced HER-2 prediction in breast cancer through synergistic integration of deep learning, ultrasound radiomics, and clinical data. Sci Rep. 2025;15(1):26992.https://doi.org/10.1038/s41598-025-12825-7 \u003c/li\u003e\n\u003cli\u003eWu J, Guo Y, Wu C, Wang Z, Sun Y, Xu D. Integration of Longitudinal and Transverse Radiomics from Ultrasound Images with Clinical Factors for HER-2 Status Prediction in Invasive Breast Cancer Patients. J Invest Surg. 2024;37(1):2436050.https://doi.org/10.1080/08941939.2024.2436050 \u003c/li\u003e\n\u003cli\u003eZhuo X, Lv J, Chen B, Liu J, Luo Y, Liu J, Xie X, Lu J, Zhao N. Combining conventional ultrasound and ultrasound elastography to predict HER2 status in patients with breast cancer. Front Physiol. 2023;14:1188502.https://doi.org/10.3389/fphys.2023.1188502 \u003c/li\u003e\n\u003cli\u003eZhou J, Jin AQ, Zhou SC, Li JW, Zhi WX, Huang YX, Zhu Q, Qian L, Wu J, Chang C. Application of preoperative ultrasound features combined with clinical factors in predicting HER2-positive subtype (non-luminal) breast cancer. BMC Med Imaging. 2021;21(1):184.https://doi.org/10.1186/s12880-021-00714-0 \u003c/li\u003e\n\u003cli\u003eWang J, Liu Y, Zhang Q, Li W, Feng J, Wang X, Fang J, Han Y, Xu B. Disitamab vedotin, a HER2-directed antibody-drug conjugate, in patients with HER2-overexpression and HER2-low advanced breast cancer: a phase I/Ib study. Cancer Commun (Lond). 2024;44(7):833-51.https://doi.org/10.1002/cac2.12577 \u003c/li\u003e\n\u003cli\u003eIvanova M, Porta FM, D\u0026apos;Ercole M, Pescia C, Sajjadi E, Cursano G, De Camilli E, Pala O, Mazzarol G, Venetis K, Guerini-Rocco E, Curigliano G, Viale G, Fusco N. Standardized pathology report for HER2 testing in compliance with 2023 ASCO/CAP updates and 2023 ESMO consensus statements on HER2-low breast cancer. Virchows Arch. 2024;484(1):3-14.https://doi.org/10.1007/s00428-023-03656-w \u003c/li\u003e\n\u003cli\u003eKang S, Kim SB. HER2-Low Breast Cancer: Now and in the Future. Cancer Res Treat. 2024;56(3):700-20.https://doi.org/10.4143/crt.2023.1138 \u003c/li\u003e\n\u003cli\u003eDom\u0026iacute;nguez-Cejudo MA, Gil-Torralvo A, Cejuela M, Molina-Pinelo S, Salvador Bofill J. Targeting the Tumor Microenvironment in Breast Cancer: Prognostic and Predictive Significance and Therapeutic Opportunities. Int J Mol Sci. 2023;24(23).https://doi.org/10.3390/ijms242316771 \u003c/li\u003e\n\u003cli\u003eLambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P, Zegers CM, Gillies R, Boellard R, Dekker A, Aerts HJ. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012;48(4):441-6.https://doi.org/10.1016/j.ejca.2011.11.036 \u003c/li\u003e\n\u003cli\u003eLiu Z, Wang S, Dong D, Wei J, Fang C, Zhou X, Sun K, Li L, Li B, Wang M, Tian J. The Applications of Radiomics in Precision Diagnosis and Treatment of Oncology: Opportunities and Challenges. Theranostics. 2019;9(5):1303-22.https://doi.org/10.7150/thno.30309 \u003c/li\u003e\n\u003cli\u003eDing J, Chen S, Serrano Sosa M, Cattell R, Lei L, Sun J, Prasanna P, Liu C, Huang C. 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.https://doi.org/10.1016/j.acra.2020.10.015 \u003c/li\u003e\n\u003cli\u003eMao N, Shi Y, Lian C, Wang Z, Zhang K, Xie H, Zhang H, Chen Q, Cheng G, Xu C, Dai Y. Intratumoral and peritumoral radiomics for preoperative prediction of neoadjuvant chemotherapy effect in breast cancer based on contrast-enhanced spectral mammography. Eur Radiol. 2022;32(5):3207-19.https://doi.org/10.1007/s00330-021-08414-7 \u003c/li\u003e\n\u003cli\u003eLi C, Song L, Yin J. Intratumoral and Peritumoral Radiomics Based on Functional Parametric Maps from Breast DCE-MRI for Prediction of HER-2 and Ki-67 Status. J Magn Reson Imaging. 2021;54(3):703-14.https://doi.org/10.1002/jmri.27651 \u003c/li\u003e\n\u003cli\u003eUematsu T. Focal breast edema associated with malignancy on T2-weighted images of breast MRI: peritumoral edema, prepectoral edema, and subcutaneous edema. Breast Cancer. 2015;22(1):66-70.https://doi.org/10.1007/s12282-014-0572-9 \u003c/li\u003e\n\u003cli\u003eLiang X, Li Z, Zhang L, Wang D, Tian J. Application of Contrast-Enhanced Ultrasound in the Differential Diagnosis of Different Molecular Subtypes of Breast Cancer. Ultrason Imaging. 2020;42(6):261-70.https://doi.org/10.1177/0161734620959780 \u003c/li\u003e\n\u003cli\u003eWan CF, Du J, Fang H, Li FH, Zhu JS, Liu Q. Enhancement patterns and parameters of breast cancers at contrast-enhanced US: correlation with prognostic factors. Radiology. 2012;262(2):450-9.https://doi.org/10.1148/radiol.11110789 \u003c/li\u003e\n\u003cli\u003eZhao YX, Liu S, Hu YB, Ge YY, Lv DM. Diagnostic and prognostic values of contrast-enhanced ultrasound in breast cancer: a retrospective study. Onco Targets Ther. 2017;10:1123-9.https://doi.org/10.2147/ott.S124134 \u003c/li\u003e\n\u003cli\u003eFeng M, Grice DM, Faddy HM, Nguyen N, Leitch S, Wang Y, Muend S, Kenny PA, Sukumar S, Roberts-Thomson SJ, Monteith GR, Rao R. Store-independent activation of Orai1 by SPCA2 in mammary tumors. Cell. 2010;143(1):84-98.https://doi.org/10.1016/j.cell.2010.08.040 \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Breast cancer, ultrasound, radiomics, HER 2-low","lastPublishedDoi":"10.21203/rs.3.rs-8786769/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8786769/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study aims to develop a preoperative diagnostic model for HER2-low breast cancer by integrating radiomics, conventional B-mode ultrasound (US), contrast-enhanced ultrasound (CEUS), and clinical features.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethod\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis retrospective study included 175 patients with pathologically confirmed breast cancer (91 HER2-low; 84 HER2-zero). All patients underwent preoperative conventional B-mode ultrasound and CEUS examinations. Intratumoral and peritumoral (2 mm and 3 mm) regions of interest were manually segmented, and radiomic features were extracted using PyRadiomics. Feature selection was performed through t-tests, Pearson correlation analysis, mRMR（Max-Relevance and Min-Redundancy）and LASSO regression. Multiple radiomics models (US, CEUS, multi-region) and a clinical model were constructed and evaluated using ROC analysis. A combined clinical–radiomics nomogram was generated using multivariate logistic regression and was assessed with calibration curves and decision curve analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResult\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCEUS-based peritumoral radiomics demonstrated superior discriminative performance compared to conventional ultrasound. The combined CEUS + tumor and peritumoral 3-mm model achieved the highest predictive accuracy (AUC 0.892 in the training cohort; 0.802 in the test cohort). Microcalcification was identified as an independent clinical predictor of HER2-low disease, although the clinical model alone showed moderate performance (AUC 0.721 and 0.717). The integrated nomogram combining radiomics signatures and clinical factors yielded the best overall performance (AUC 0.921 and 0.850), with favorable calibration and superior net clinical benefit.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA CEUS-based multi-region radiomics nomogram that incorporates intratumoral and peritumoral features enables accurate and noninvasive preoperative prediction of HER2-low breast cancer.\u003c/p\u003e","manuscriptTitle":"Development and Validation of a CEUS-Based Intratumoral–Peritumoral Radiomics Nomogram for Preoperative Prediction of HER2-Low Breast Cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-09 12:47:56","doi":"10.21203/rs.3.rs-8786769/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a7fa8e83-0ebd-483e-87d2-cd6ba0bf9314","owner":[],"postedDate":"February 9th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-31T15:56:59+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-09 12:47:56","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8786769","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8786769","identity":"rs-8786769","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}