Machine Learning Predicts Pathologic Complete Response to Neoadjuvant Chemotherapy for ER+HER2- Breast Cancer: Integrating Tumoral and Peritumoral MRI Radiomic Features

Background: This study aimed to predict pathologic complete response (pCR) in neoadjuvant chemotherapy for ER+HER2- locally advanced breast cancer (LABC), a subtype with limited treatment response. Methods: We included 265 LABC patients (2010–2020) pre-treatment MRI, chemotherapy, and confirmed pathology. Using data from January 2016, we divided them into training validation cohorts. Volumes of interest (VOI) the tumoral peritumoral regions were segmented on preoperative MRI three sequences: T1-weighted early delayed contrast-enhanced sequences T2-weighted fat-suppressed sequence (T2FS). constructed seven machine learning models using tumoral, peritumoral, combined texture features within across sequences, evaluated their pCR prediction performance AUC values. Results: The best single model was SVM 1 mm tumor-to-peritumor VOI phase (AUC = 0.9447). Among combinations, top-performing K-Nearest Neighbor, 3 T2FS 0.9631). Conclusions: suggest that integrates radiomic different can provide more accurate pretreatment LABC.