Image-to-text Retrieval on MIMIC-CXR
44.7R@1MEDBind
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MEDBindvariant=BD2024.03 | 44.7 | — | 91 | |
| MEDBindvariant=NM2024.03 | 43.8 | — | 88.5 | |
| CXR-CLIP2024.03 | 21.6 | — | 60.2 | |
| MedCLIP2024.03 | 2.8 | — | 18 | |
| CLIP2024.03 | 1 | — | 10.5 | |
| CXR-CLIPBackbone=Swin-Tiny, Pre-train Dataset=MIMIC-CXR2023.10 | 0.216 | 0.489 | 0.602 | |
| CXR-CLIPBackbone=ResNet-50, Pre-train Dataset=MIMIC-CXR2023.10 | 0.214 | 0.46 | 0.592 | |
| CXR-CLIPBackbone=Swin-Tiny, Pre-train Dataset=MIMIC-CXR, CheXpert, Chest X-ray142023.10 | 0.209 | 0.462 | 0.588 | |
| CXR-CLIPBackbone=ResNet-50, Pre-train Dataset=MIMIC-CXR, CheXpert2023.10 | 0.202 | 0.459 | 0.582 | |
| CXR-CLIPBackbone=ResNet-50, Pre-train Dataset=MIMIC-CXR, CheXpert, Chest X-ray142023.10 | 0.197 | 0.444 | 0.564 | |
| CXR-CLIPBackbone=Swin-Tiny, Pre-train Dataset=MIMIC-CXR, CheXpert2023.10 | 0.196 | 0.442 | 0.571 | |
| GloRIABackbone=ResNet-50, Pre-train Dataset=CheXpert with reports2023.10 | 0.072 | 0.206 | 0.303 | |
| MedCLIPBackbone=Swin-Tiny, Pre-train Dataset=MIMIC-CXR, CheXpert2023.10 | 0.011 | 0.014 | 0.055 |