Figure 3.

Radiomics workflow and model selection process of the current study. (I) Segmentation and preprocessing of images using filter and normalization techniques. (II) Feature extraction (shape, first order and texture features). (III) Model training and evaluation of different model configurations. T2w, T2-weighted; ADC, apparent diffusion coefficient; GLCM, gray-level co-occurence matrix; GLDM, gray-level dependence matrix; GLRLM, gray-level run-length matrix; NGTDM, neighboring gray-tone difference matrix; GLSZM, gray-level size-zone matrix; mADC, mean apparent diffusion coefficient; PSAD, prostate specific antigen density; PI-RADS, Prostate Imaging Reporting and Data System; CV, cross validation; MRMR, minimal redundancy maximal relevance; cw, class weights; SMOTE, synthetic minority oversampling technique.