pRCC Subtyping on pRCC Subtyping Dataset (test)
93.01Accuracyi-ViT
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| i-ViTHidden dimensions=128, Heads=12, Layers=12, Sequence length=5002021.06 | 93.01 | 90 | 97.5 | 93.6 | 96.97 | 88.07 | 92.31 | |
| i-ViT-HInput=Ten cellular features of each nucleus, Hidden dimensions=32, Heads=2, Layers=1, Sequence length=5002021.06 | 89.18 | 91.3 | 87.5 | 89.36 | 86.84 | 90.83 | 88.79 | |
| GBDT-GHClassifier=GBDT, Features=150-dimensional hand-craft + 3 grade-distribution2021.06 | 88.21 | 94.29 | 82.5 | 88 | 83.06 | 94.5 | 88.41 | |
| DT-GClassifier=Decision Tree, Library=scikit-learn2021.06 | 80.35 | 85.05 | 75.83 | 80.18 | 76.23 | 85.32 | 80.52 | |
| CNN-ORIBackbone=ResNet-34, Pre-training=ImageNet2021.06 | 75.11 | 77.39 | 74.17 | 75.74 | 72.81 | 76.15 | 74.44 |