Broad Diagnosis (4-Class Structure) on Test Set (test)
98.5SensitivityMS-AQNet
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MS-AQNetClass=Severe ROP2026.02 | 98.5 | 87.2 | 31 | 47.1 | 95.1 | — | |
| CAA EnsembleClass=Mild / Treated2026.02 | 97.1 | 96.8 | 86.2 | 91.3 | 99.6 | — | |
| CAA EnsembleClass=Normal2026.02 | 95.4 | 96.3 | 96.9 | 96.2 | 99.4 | — | |
| CAA EnsembleClass=Other2026.02 | 93.7 | 98.9 | 96.2 | 95 | 99.7 | — | |
| Baseline CNNClass=Severe ROP, Backbone=EfficientNet-B0, Fusion=Late fusion by concatenation2026.02 | 93.5 | 86.6 | 50.9 | 66 | 94.7 | — | |
| CAA EnsembleClass=All Classes (Macro Average)2026.02 | 91.6 | 97.9 | 93.1 | 92.2 | 99.6 | 94.2 | |
| MS-AQNetClass=Other2026.02 | 90.8 | 82 | 59.6 | 71.9 | 89.7 | — | |
| Baseline CNNClass=Other, Backbone=EfficientNet-B0, Fusion=Late fusion by concatenation2026.02 | 80.9 | 88.4 | 61.9 | 70.2 | 93.3 | — | |
| CAA EnsembleClass=Severe ROP2026.02 | 80.3 | 99.7 | 93 | 86.2 | 99.5 | — | |
| Baseline CNNClass=All Classes (Macro Average), Backbone=EfficientNet-B0, Fusion=Late fusion by concatenation2026.02 | 67.7 | 89.8 | 59.4 | 61.2 | 89.6 | 55.3 | |
| MS-AQNetClass=All Classes (Macro Average)2026.02 | 67 | 87.6 | 52.2 | 53.3 | 86.6 | 44.3 | |
| Baseline CNNClass=Normal, Backbone=EfficientNet-B0, Fusion=Late fusion by concatenation2026.02 | 64.8 | 91.4 | 86.1 | 74 | 89.4 | — | |
| MS-AQNetClass=Normal2026.02 | 57.3 | 93.6 | 91.5 | 70.5 | 83.3 | — | |
| Baseline CNNClass=Mild / Treated, Backbone=EfficientNet-B0, Fusion=Late fusion by concatenation2026.02 | 31.6 | 92.7 | 38.9 | 34.9 | 81.1 | — | |
| MS-AQNetClass=Mild / Treated2026.02 | 21.4 | 87.8 | 26.7 | 23.7 | 78.4 | — |