Slide-Level Subtyping on TCGA-NSCLC (25% train)
0.923Macro AUCHIPT
Evaluation results
| Method | Links | |
|---|---|---|
| HIPTArchitecture=HIPT, Evaluation Protocol=Weakly-supervised Fine-tuning2022.06 | 0.923 | |
| MILArchitecture=MIL, Evaluation Protocol=Weakly-supervised Fine-tuning2022.06 | 0.857 | |
| CLAM-SBArchitecture=CLAM-SB, Evaluation Protocol=Weakly-supervised Fine-tuning2022.06 | 0.852 | |
| DS-MILArchitecture=DS-MIL, Evaluation Protocol=Weakly-supervised Fine-tuning2022.06 | 0.787 | |
| ViT_4096-256 (Mean)Architecture=ViT_4096-256, Evaluation Protocol=K-Nearest Neighbors (KNN), Aggregation Method=Mean Pooling2022.06 | 0.773 | |
| GCN-MILArchitecture=GCN-MIL, Evaluation Protocol=Weakly-supervised Fine-tuning2022.06 | 0.748 | |
| ResNet-50IN (Mean)Architecture=ResNet-50IN, Evaluation Protocol=K-Nearest Neighbors (KNN), Aggregation Method=Mean Pooling2022.06 | 0.696 | |
| DeepAttnMISLArchitecture=DeepAttnMISL, Evaluation Protocol=Weakly-supervised Fine-tuning2022.06 | 0.663 | |
| ViT_256-16 (Mean)Architecture=ViT_256-16, Evaluation Protocol=K-Nearest Neighbors (KNN), Aggregation Method=Mean Pooling2022.06 | 0.622 |