From imaging to management: Enhancing the Breast Imaging Reporting and Data System with contrast-enhanced ultrasound for stratification of breast lesions

Selective vacuum-assisted biopsy (VAB) improves diagnostic accuracy for intermediate breast lesions—including B3, imaging-pathologic discordant, and high-risk lesions—while enabling complete resection. We evaluated the integrating of the Breast Imaging Reporting and Data System (BI-RADS) with contrast-enhanced ultrasound (CEUS) to optimize pre-biopsy VAB selection. This retrospective study analyzed 647 patients (2016–2023) who underwent conventional ultrasound, CEUS, core needle biopsy, and surgical excision. A triple-classification BI-RADS model was developed based on the original BI-RADS. Multivariate logistic regression analysis was conducted to develop a logistic regression model. CEUS was incorporated into BI-RADS to establish two models: the CEUS-adjusted BI-RADS model and the CEUS-BI-RADS scoring model. Overall accuracy, sensitivity, specificity, underdiagnosis rate of malignancies, overdiagnosis rate of benign lesions, and simulated secondary excision rate for intermediate lesions were compared across models. Multivariate analysis showed that irregular shape, earlier enhancement, and iso-/hyperenhancement predicted intermediate lesions, whereas perfusion defects, irregular shape, earlier enhancement, and post-enhancement size enlargement predicted of malignancies. The CEUS-adjusted BI-RADS model achieved an overall accuracy of 73.8 %, demonstrating sensitivities of 64.5 % (benign), 52.2 % (intermediate), and 93.0 % (malignant), with corresponding specificities of 83.9 %, 43.2 %, and 84.2 %. Compared to the original BI-RADS, it improved sensitivity for malignancies (87.8 %→93.0 %), reduced the overdiagnosis rate of benign lesions (25.1 %→13.3 %) and the simulated secondary resection of intermediate lesions (19.4 %→6.5 %), while maintaining the same underdiagnosis for malignant lesions (1.2 %→1.2 %). CEUS-BI-RADS integration demonstrates potential for tripartite classification of breast lesions before biopsy, offering valuable guidance for optimal biopsy strategy and enhanced clinical management efficiency.