Fig. 4

a Comparison of ROC in the training group of three models (LR, SVM, LightGBM) based on T1WI; b Comparison of ROC in the test group of three models (LR, SVM, LightGBM) based on T1WI; c Comparison of ROC in the training group of three models (LR, SVM, LightGBM) based on T2WI; d Comparison of ROC in the test group based on three T2WI models (LR, SVM, LightGBM); e Comparison of ROC in the training group based on three bimodal models (LR, SVM, LightGBM); f Comparison of ROC in the test group based on three bimodal models (LR, SVM, LightGBM); g Comparison of ROC in the training group of three models based on SVM (T1WI, T2WI, bimodal); h Comparison of ROC in the test group of three models based on SVM (T1WI, T2WI, bimodal)