Table 3 Performance of the machine learning-based classifications
AUC | Accuracy | Sensitivity | Specificity | |
|---|---|---|---|---|
T2WI | ||||
Logistic regression | 0.746 (0.630, 0.863) | 69.1 (68.5, 69.7) | 83.3 (70.0, 96.7) | 57.9 (42.2, 73.6) |
Random forest | 0.754 (0.635, 0.872) | 73.5 (73.0, 74.1) | 63.3 (46.1, 80.6) | 81.6 (69.3, 93.9) |
SVM | 0.688 (0.556, 0.820) | 70.6 (70.0, 71.2) | 73.3 (57.5, 89.2) | 68.4 (53.6, 83.2) |
ceT1WI | ||||
Logistic regression | 0.820 (0.718, 0.922) | 77.9 (77.4,78.4) | 56.7 (38.9, 74.4) | 94.7 (87.6, 100) |
Random forest | 0.771 (0.659, 0.884) | 72.1 (71.5,72.6) | 80.0 (65.7, 94.3) | 65.8 (50.7, 80.9) |
SVM | 0.782 (0.661, 0.902) | 77.9 (77.4, 78.4) | 83.3 (70.0, 96.7) | 73.7 (59.7, 87.7) |
T2WI + ceT1WI | ||||
Logistic regression | 0.846 (0.750, 0.943) | 80.9 (80.4, 81.3) | 80.0 (65.7, 94.3) | 81.6 (69.3, 93.9) |
Random forest | 0.813 (0.703, 0.924) | 79.4 (78.9, 79.9) | 73.3 (57.5, 89.2) | 84.2 (72.6, 95.8) |
SVM | 0.822 (0.721, 0.923) | 77.9 (77.4, 78.4) | 76.7 (61.5, 91.8) | 78.9 (66.0, 91.9) |