Table 2 The classification performance of visual inspection, GPnorm, and ResNet18 in separating ABE from non-ABE HB patients, as evaluated by sensitivity, specificity, precision, F1-score, Accuracy, AUC
Methods | Sensitivity | Specificity | Precision | F1-Score | Accuracy | AUC |
|---|---|---|---|---|---|---|
Visual inspection | 52.48 ± 13.58% | 55.21 ± 7.86% | 62.95 ± 3.58% | 56.79 ± 8.90% | 53.58 ± 5.71% | 53.87 ± 4.11% |
GPnorm | 68.10% | 78.10% | 82.05% | 74.42% | 72.15% | 76.90% |
ResNet18 | 78.95 ± 17.85% | 45.26 ± 19.19% | 59.58 ± 7.09% | 67.11 ± 8.28% | 62.11 ± 8.03% | 68.92 ± 11.06% |