Table 1.
Three-fold cross-validation results on a slide level.
| Model | Point | Average: cross validation (CV) | IT1 | IT2 | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SE | SP | ACC | AUC | SE | SP | ACC | AUC | SE | SP | ACC | AUC | ||
| U-Net | P1 | 1 | 0.75 | 0.82 | 0.98 ± 0.05 | 1 | 0.41 | 0.81 | 0.92 ± 0.03 | 1 | 0.05 | 0.66 | 0.83 ± 0.06 |
| P2 | 0.5 | 1 | 0.87 | 0.64 | 1 | 0.75 | 0.33 | 1 | 0.57 | ||||
| P3 | 0.85 | 0.97 | 0.94 | 0.92 | 0.65 | 0.83 | 0.88 | 0.53 | 0.75 | ||||
| DenseNetFCN | P1 | 1 | 0.41 | 0.56 | 0.97 ± 0.08 | – | – | – | – | – | – | – | – |
| P2 | 0.31 | 1 | 0.82 | – | – | – | – | – | – | – | – | ||
| P3 | 0.91 | 0.93 | 0.92 | – | – | – | – | – | – | – | – | ||
| EfficientNet | P1 | 1 | 0.28 | 0.47 | 0.97 ± 0.09 | – | – | – | – | – | – | – | – |
| P2 | 0.36 | 1 | 0.84 | – | – | – | – | – | – | – | – | ||
| P3 | 0.89 | 0.97 | 0.95 | – | – | – | – | – | – | – | – | ||
Results are presented for three points of the ROC curve: P1—specificity equal 1, P2—sensitivity equal 1, P3—the best accuracy, where: SE—sensitivity, SP—specificity, ACC—accuracy.