Radiomics-Based Multiparametric MRI for Differentiating Luminal A and Luminal B Breast Cancer: A Multi-Center Study

Motivation: This study aims to improve the non-invasive classification of Luminal A and Luminal B breast cancer subtypes using a multiparametric radiomics model integrating T2WI, DWI, and DCE-MRI. Goal(s): The goal is to improve non-invasive breast cancer classification using a multiparametric radiomics model. Approach: This study analyzed T2WI, DWI, and DCE-MRI data from 130 patients, developing a model evaluated with AUC, DCA, and DeLong test. Results: The multiparametric model achieved AUCs of 0.9625, 0.8276, and 0.8750 in the training and test sets, outperforming single-sequence models in accuracy and clinical benefit. DeLong test confirmed significant differences, supporting the superior performance of the combined approach. Impact: This study enhances breast cancer diagnosis by providing a noninvasive method to differentiate Luminal A and Luminal B subtypes. This could potentially improve clinical decision-making and enable personalized treatment plans for patients.