Table 3.
Model | AUC | Sensitivity | Specificity | Accuracy | P value | |
---|---|---|---|---|---|---|
Training cohort | ||||||
Clinical model | 0.723 | 0.795 | 0.594 | 0.669 | ||
Radiomics model | T2WI | 0.839 | 0.833 | 0.800 | 0.814 | |
ADC | 0.938 | 0.897 | 0.848 | 0.864 | ||
ADC+T2WI | 0.942 | 0.841 | 0.905 | 0.881 | <0.001 d | |
Combined model | 0.955 | 0.886 | 0.905 | 0.898 | 0.214 e | |
Validation cohort | ||||||
Clinical model | 0.745 | 0.632 | 0.844 | 0.765 | ||
Radiomics model | T2WI | 0.809 | 0.867 | 0.722 | 0.765 | |
ADC | 0.869 | 0.833 | 0.815 | 0.824 | ||
ADC+T2WI | 0.910 | 0.947 | 0.844 | 0.882 | 0.026 d | |
Combined model | 0.931 | 0.895 | 0.875 | 0.882 | 0.277 e |
Comparison of ROC curve performance between clinical model and radiomics model using DeLong test.
Comparison of ROC curve performance between radiomics model and radiomics-Clinical combined model using DeLong test.