Medical Image Analysis on 26 Medical Downstream Tasks
78.4Average PerformanceDirector-Experts (DEX)
Evaluation Results
| Method | Links | |
|---|---|---|
| Director-Experts (DEX)Modality=MM, Backbone=ViT-B, Data size=4.97M2026.05 | 78.4 | |
| PanDermModality=MM, Backbone=ViT-B, Data size=2.15M2026.05 | 75.4 | |
| LVMMedModality=MM, Backbone=ViT-B, Data size=1.3M2026.05 | 75.4 | |
| DINOv3Modality=NI, Backbone=ViT-B, Data size=1.7B2026.05 | 75.2 | |
| CONCHModality=Path, Backbone=ViT-B, Data size=1.17M, Supervision Type=paired text or labels2026.05 | 74.2 | |
| RETFoundModality=OCT, Backbone=ViT-L, Data size=736K2026.05 | 73.6 | |
| MedSAMModality=MM, Backbone=ViT-B, Data size=1.57M, Supervision Type=paired text or labels2026.05 | 73.5 | |
| MAEModality=NI, Backbone=ViT-B, Data size=1.28M2026.05 | 73.3 | |
| MAE-MedVerseModality=MM, Backbone=ViT-B, Data size=4.97M2026.05 | 73.1 | |
| DINOv2Modality=NI, Backbone=ViT-B, Data size=120M2026.05 | 72.7 | |
| USFMModality=US, Backbone=ViT-B, Data size=2.19M2026.05 | 70.1 | |
| PLIPModality=Path, Backbone=ViT-B, Data size=208K, Supervision Type=paired text or labels2026.05 | 66.5 | |
| scratchModality=scratch, Backbone=ViT-B, Data size=-2026.05 | 59 |