Table 3 The results of clinical model, DTL model and DTLR nomogram.
From: An MRI-based deep transfer learning radiomics nomogram for predicting meningioma grade
Accuracy | AUC (95% CI) | Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | |
|---|---|---|---|---|---|---|
Training set | ||||||
Clinical model | 0.703 | 0.745 (0.6933–0.7969) | 0.682 (0.6546–0.7205) | 0.724 (0.6984–0.7532) | 0.712 (0.6865–0.7592) | 0.695 (0.6743–0.7423) |
Radiomics model | 0.844 | 0.909 (0.8772–0.9410) | 0.859 (0.8364–0.9034) | 0.829 (0.7985–0.8761) | 0.834 (0.8023–0.8986) | 0.855 (0.8201–0.8869) |
DTL model | 0.874 | 0.942 (0.9179–0.9660) | 0.824 (0.7895–0.8598) | 0.924 (0.8947–0.9478) | 0.915 (0.8859–0.9482) | 0.84 (0.8219–0.8739) |
DTLR nomogram | 0.765 | 0.848 (0.8075–0.8883) | 0.582 (0.5697–0.6987) | 0.947 (0.9234–0.9684) | 0.917 (0.8769–0.9347) | 0.694 (0.6749–0.7436) |
Test set | ||||||
Clinical model | 0.745 | 0.788 (0.6996–0.8756) | 0.941 (0.9218–0.9573) | 0.549 (0.5294–0.6539) | 0.676 (0.6395–0.7345) | 0.903 (0.8762–0.9247) |
Radiomics model | 0.735 | 0.805 (0.7211–0.8883) | 0.569 (0.5438–0.6258) | 0.902 (0.8729–0.9324) | 0.853 (0.8327–0.8819) | 0.676 (0.6285–0.7027) |
DTL model | 0.745 | 0.78 (0.6890–0.8712) | 0.686 (0.6649–0.7125) | 0.804 (0.7839–0.8246) | 0.778 (0.7537–0.8073) | 0.719 (0.6972–0.7238) |
DTLR nomogram | 0.804 | 0.866 (0.7984–0.9340) | 0.745 (0.7259–0.8114) | 0.863 (0.8249–0.9024) | 0.844 (0.8253–0.9116) | 0.772 (0.7028–0.8129) |
