Figure 2.

Comprehensive workflow diagram of the prediction model. Tumor segmentation in MR images is the first step. After that, MRI feature extraction was conducted separately via radiomics and neural convolutional networks. Student's t-test, Mann-Whitney U test and least absolute shrinkage and selection operator (LASSO) were used to feature selection, sequentially. The receiver operating characteristic (ROC) curve and the area under the curve (AUC) were calculated to evaluate the prediction efficiency of the radiomic features. Finally, a nomogram was developed and evaluated.