Fig. 1.

Lesion classification pipeline based on diagnostic images. Radiomic features were extracted from DCE, T2w, and DWI sequences. The mpMRI information was incorporated in two different ways: feature fusion, i.e., merging radiomic features extracted from all sequences to train an SVM classifier and classifier fusion, i.e., aggregating the PM output from all single-parametric classifiers via soft voting. Parentheses contain the numbers of features extracted from each sequence. The dashed lines for DWI indicate that the DWI sequence was only included in the classification process when it was available, while the DCE and T2w sequences were available for all lesions and thus were always included. ADC, apparent diffusion coefficient and ROC, receiver operating characteristic.