Table 4.
Diagnostic performance of the difference models in predicting MVI in HCC within Milan criteria.
| Different models | Training group |
P value | Test group |
||||||
|---|---|---|---|---|---|---|---|---|---|
| Accuracy | Sensitivity | Specificity | AUC (95%CI) | Accuracy | Sensitivity | Specificity | AUC (95%CI) | ||
| MRI features | 75.65% | 89.13% | 66.67% | 0.85 (0.78–0.92) | P1= 0.30 | 76.00% | 80.00% | 73.33% | 0.85 (0.74–0.95) |
| Rad-Score | 76.52% | 80.43% | 73.91% | 0.82 (0.73–0.90) | P2= 0.13 | 82.00% | 60.00% | 96.67% | 0.80 (0.67–0.93) |
| Nomogram | 80.87 | 80.43% | 81.16% | 0.87 (0.81–0.94) | P3= 0.02 | 82.00% | 70.00% | 90.00% | 0.89 (0.81–0.98) |
P1, AUC comparison between MRI features and Rad-score; P2, MRI features vs. Nomogram; P3, MRI Rad-score vs. Nomogram.