Chest radiograph classification on MIMIC-CXR (test)
0.827AUROCConvNeXt-B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ConvNeXt-BInitialization strategy=DINOv3, Resolution=512x512, Evaluation protocol=full fine-tuning2025.10 | 0.827 | — | |
| ViT-BInitialization strategy=DINOv3, Resolution=512x512, Evaluation protocol=full fine-tuning2025.10 | 0.825 | — | |
| ConvNeXt-BInitialization strategy=ImageNet, Resolution=512x512, Evaluation protocol=full fine-tuning2025.10 | 0.822 | 0.001 | |
| ViT-BInitialization strategy=ImageNet, Resolution=512x512, Evaluation protocol=full fine-tuning2025.10 | 0.814 | 0.006 | |
| ConvNeXt-BInitialization strategy=DINOv3, Resolution=224x224, Evaluation protocol=full fine-tuning2025.10 | 0.812 | — | |
| ViT-BInitialization strategy=DINOv2, Resolution=512x512, Evaluation protocol=full fine-tuning2025.10 | 0.812 | 0.044 | |
| ViT-BInitialization strategy=DINOv2, Resolution=224x224, Evaluation protocol=full fine-tuning2025.10 | 0.809 | 0.006 | |
| ViT-BInitialization strategy=DINOv3, Resolution=224x224, Evaluation protocol=full fine-tuning2025.10 | 0.808 | — | |
| ConvNeXt-BInitialization strategy=ImageNet, Resolution=224x224, Evaluation protocol=full fine-tuning2025.10 | 0.804 | 0.001 | |
| ViT-BInitialization strategy=ImageNet, Resolution=224x224, Evaluation protocol=full fine-tuning2025.10 | 0.799 | 0.006 |