Subtyping on CPTAC NSCLC
99.5AUPRCCOBRA-HO (multi-FM)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| COBRA-HO (multi-FM)Training Dataset=TCGA, Magnification=20x, Backbone=H-Optimus-0, Inference Mode=Eq. (8)2024.11 | 99.5 | — | |
| COBRA-HOTraining Dataset=TCGA, Magnification=20x, Backbone=H-Optimus-0, Inference Mode=Eq. (6)2024.11 | 99.4 | — | |
| PRISMTraining Dataset=TCGA, Magnification=20x2024.11 | 99.3 | — | |
| COBRA-UNITraining Dataset=TCGA, Magnification=20x, Backbone=UNI, Inference Mode=Eq. (6)2024.11 | 99.1 | — | |
| COBRA-UNI (multi-FM)Training Dataset=TCGA, Magnification=20x, Backbone=UNI, Inference Mode=Eq. (8)2024.11 | 98.9 | — | |
| COBRA-GPTraining Dataset=TCGA, Magnification=20x, Backbone=GigaPath, Inference Mode=Eq. (6)2024.11 | 98.9 | — | |
| COBRA-V2 (multi-FM)Training Dataset=TCGA, Magnification=20x, Backbone=Virchow-2, Inference Mode=Eq. (8)2024.11 | 98.4 | — | |
| COBRA-V2Training Dataset=TCGA, Magnification=20x, Backbone=Virchow-2, Inference Mode=Eq. (6)2024.11 | 98.1 | — | |
| ConcatenatedTraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 97.4 | — | |
| H-OptimusTraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 97.3 | — | |
| Ensemble PredictionTraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 97.3 | — | |
| CONCHTraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 96.7 | — | |
| GigaPathTraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 96.6 | — | |
| COBRA-CTP (multi-FM)Training Dataset=TCGA, Magnification=20x, Backbone=CTransPath, Inference Mode=Eq. (8)2024.11 | 96.6 | — | |
| COBRA-CTPTraining Dataset=TCGA, Magnification=20x, Backbone=CTransPath, Inference Mode=Eq. (6)2024.11 | 96.4 | — | |
| UNITraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 96.1 | — | |
| Virchow-2Training Dataset=TCGA, Magnification=20x2024.11 | 95.9 | — | |
| CHIEFTraining Dataset=TCGA, Magnification=20x2024.11 | 94.7 | — | |
| MADELEINETraining Dataset=TCGA, Magnification=20x2024.11 | 94.6 | — | |
| COBRA-ENCTraining Dataset=TCGA, Magnification=20x2024.11 | 92.7 | — | |
| GigaPath-SETraining Dataset=TCGA, Magnification=20x2024.11 | 91.3 | — | |
| VirchowTraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 90.3 | — | |
| CTransPathTraining Dataset=TCGA, Magnification=20x, Inference Mode=Mean over patch embeddings2024.11 | 87.9 | — | |
| CHIEFBackbone=CHIEF [42]2024.11 | — | 82.3 | |
| COBRA-CTPBackbone=CTransPath [41], Inference Mode=Equation (6)2024.11 | — | 85.5 | |
| COBRA-ENCSlide Encoder=True, Inference Mode=Equation (6)2024.11 | — | 84.9 | |
| COBRA-H0Backbone=H-Optimus-0 [32], Inference Mode=Equation (6)2024.11 | — | 94.6 | |
| COBRA-UNIBackbone=UNI [4], Inference Mode=Equation (6)2024.11 | — | 90 | |
| COBRA-V2Backbone=Virchow 2 [47], Inference Mode=Equation (6)2024.11 | — | 89.9 | |
| COBRA†-CTPBackbone=CTransPath [41], Inference Mode=Equation (8)2024.11 | — | 85.4 | |
| COBRA†-GPBackbone=GigaPath [44], Inference Mode=Equation (8)2024.11 | — | 92.2 | |
| COBRA†-H0Backbone=H-Optimus-0 [32], Inference Mode=Equation (8)2024.11 | — | 94.2 | |
| COBRA†-UNIBackbone=UNI [4], Inference Mode=Equation (8)2024.11 | — | 90.5 | |
| COBRA†-V2Backbone=Virchow 2 [47], Inference Mode=Equation (8)2024.11 | — | 89.7 | |
| Concatenated2024.11 | — | 78 | |
| CONCHBackbone=CONCH [24]2024.11 | — | 89.4 | |
| CTransPathBackbone=CTransPath [41]2024.11 | — | 75.9 | |
| Ensemble Prediction2024.11 | — | 81.2 | |
| GigaPathBackbone=GigaPath [44]2024.11 | — | 84.3 | |
| GigaPath-SEBackbone=GigaPath [44], Slide Encoder=True2024.11 | — | 70.6 | |
| H-OptimusBackbone=H-Optimus-0 [32]2024.11 | — | 85.4 | |
| MADELEINEBackbone=MADELEINE [18]2024.11 | — | 84.9 | |
| PRISMBackbone=PRISM [34]2024.11 | — | 96.2 | |
| UNIBackbone=UNI [4]2024.11 | — | 74.6 | |
| VirchowBackbone=Virchow [39]2024.11 | — | 78.2 | |
| Virchow2Backbone=Virchow 2 [47]2024.11 | — | 80.8 |