Image Classification on NIH ChestX-ray
88.52AccuracyM&M
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| M&Mprotocol=zero-shot2024.07 | 88.52 | 77.92 | 27.55 | |
| LGDEAZero-shot=true, Group=Single Domain, Paired training data percentage=10%2026.02 | 87.06 | — | — | |
| MedKLIPprotocol=zero-shot2024.07 | 86.19 | 76.76 | 25.25 | |
| LGDEAZero-shot=true, Group=Single Domain, Paired training data percentage=5%2026.02 | 85.88 | — | — | |
| PRIORprotocol=zero-shot2024.07 | 84.41 | 74.51 | 23.29 | |
| MedAlignerZero-shot=true, Group=Paired Baseline2026.02 | 84.11 | — | — | |
| AFLocZero-shot=true, Group=Paired Baseline2026.02 | 83.87 | — | — | |
| CARZeroZero-shot=true, Group=Paired Baseline2026.02 | 83.16 | — | — | |
| MAVLZero-shot=true, Group=Paired Baseline2026.02 | 82.77 | — | — | |
| PRIORZero-shot=true, Group=Paired Baseline2026.02 | 80.84 | — | — | |
| MedKLIPZero-shot=true, Group=Paired Baseline2026.02 | 79.4 | — | — | |
| BioViLprotocol=zero-shot2024.07 | 79.16 | 69.12 | 19.31 | |
| CLEFTZero-shot=true, Group=Paired Baseline2026.02 | 78.95 | — | — | |
| LGDEAZero-shot=true, Group=Cross Domain, Paired training data percentage=10%2026.02 | 78.56 | — | — | |
| LGDEAZero-shot=true, Group=Cross Domain, Paired training data percentage=5%2026.02 | 77.06 | — | — | |
| GLORIAprotocol=zero-shot2024.07 | 77 | 66.1 | 17.32 | |
| ConVIRTprotocol=zero-shot2024.07 | 71.02 | 61.01 | 16.28 | |
| MGCAZero-shot=true, Group=Paired Baseline2026.02 | 68.4 | — | — | |
| BioViLZero-shot=true, Group=Paired Baseline2026.02 | 61.33 | — | — | |
| GLoRIAZero-shot=true, Group=Paired Baseline2026.02 | 59.47 | — | — | |
| MedCLIPZero-shot=true, Group=Paired Baseline2026.02 | 58.84 | — | — | |
| C2LFine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 79 | — | |
| From scratchFine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 75.4 | — | |
| MGFine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 77.3 | — | |
| PCRLv1Fine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 79.9 | — | |
| PCRLv2Fine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 81.5 | — | |
| SimSiamFine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 79.4 | — | |
| TransVWFine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 77.6 | — | |
| VoCoFine-tuning protocol=60% data fine-tuning, Backbone=3D-UNet2024.02 | — | 82.02 | — |