Multicontrast MRI-based radiomics for the prediction of pathological complete response to neoadjuvant chemotherapy in patients with early triple negative breast cancer

To assess pre-therapeutic MRI-based radiomic analysis to predict the pathological complete response neoadjuvant chemotherapy (NAC) in women with early triple negative breast cancer (TN). This monocentric retrospective study included 75 TN female patients MRI (T1-weighted, T2-weighted, diffusion-weighted and dynamic contrast enhancement images) performed before NAC. For each patient, tumor(s) parenchyma were independently segmented analyzed extract shape, size, texture features. Several sets of features realized based on 4 different sequence images. Performances classifiers (random forest, multilayer perceptron, support vector machine (SVM) linear or quadratic kernel) compared (defined excised tissues), 100 draws 75% as training set 25% test. The combination extracted from MR images improved classifier performance (more precisely, T1W, T2W DWI). SVM kernel showed best a mean AUC 0.83, sensitivity 0.85 specificity 0.75 test set. radiomics may be relevant NAC cancer. Our results promote use multi-contrast sources for radiomics, providing enrich source information enhance model generalization.