Classification on CT-RATE
0.87AUCVoxelFM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VoxelFM2026.04 | 0.87 | — | |
| ASAPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.841 | — | |
| ASAPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.835 | — | |
| HLIPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.83 | — | |
| SimCroPBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.829 | — | |
| SimCroPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.829 | — | |
| fVLMBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.827 | — | |
| HLIPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.825 | — | |
| SimCroPBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.824 | — | |
| ViSDBBackbone=Other (not 3D ViT-B), Linear probing=true, Supervision Level=100%2026.05 | 0.824 | — | |
| SimCroPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.824 | — | |
| fVLMBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.822 | — | |
| ASAPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.822 | — | |
| MRMBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.821 | — | |
| MRMBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.821 | — | |
| fVLMBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.821 | — | |
| ViSDBBackbone=Other (not 3D ViT-B), Linear probing=true, Supervision Level=10%2026.05 | 0.82 | — | |
| PCRLv2Backbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.819 | — | |
| fVLMBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.818 | — | |
| MRMBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.817 | — | |
| MRMBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.817 | — | |
| PCRLv2Backbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.813 | — | |
| M3AEBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.81 | — | |
| SimCroPBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.81 | — | |
| M3AEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.81 | — | |
| SimCroPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.81 | — | |
| M3DBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.807 | — | |
| M3DBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.807 | — | |
| HLIPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.807 | — | |
| HEMAEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.806 | — | |
| ViSDBBackbone=Other (not 3D ViT-B), Linear probing=true, Supervision Level=1%2026.05 | 0.805 | — | |
| CT-CLIPBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.804 | — | |
| CT-CLIPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.804 | — | |
| fVLMBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.802 | — | |
| MAEBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.799 | — | |
| MAEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.799 | — | |
| Merlin2026.04 | 0.798 | — | |
| MerlinBackbone=Other (not 3D ViT-B), Linear probing=true, Supervision Level=100%2026.05 | 0.796 | — | |
| M3AEBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.795 | — | |
| M3AEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.795 | — | |
| fVLMBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.794 | — | |
| M3DBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.791 | — | |
| M3DBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.791 | — | |
| HEMAEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.789 | — | |
| CT-CLIPBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.786 | — | |
| CT-CLIPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.786 | — | |
| MerlinBackbone=Other (not 3D ViT-B), Linear probing=true, Supervision Level=10%2026.05 | 0.783 | — | |
| MAEBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.781 | — | |
| MAEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.781 | — | |
| MRMBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.777 | — | |
| MRMBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.777 | — | |
| M3AEBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.772 | — | |
| M3AEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.772 | — | |
| PCRLv2Backbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.771 | — | |
| HEMAEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.767 | — | |
| RadFM2026.04 | 0.759 | — | |
| MerlinBackbone=Other (not 3D ViT-B), Linear probing=true, Supervision Level=1%2026.05 | 0.758 | — | |
| MAEBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.756 | — | |
| MAEBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.756 | — | |
| M3DBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.743 | — | |
| CT-CLIPBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.7411 | — | |
| CT-CLIPBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.741 | — | |
| M3D2026.04 | 0.74 | — | |
| M3DBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.739 | — | |
| Random initBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=100%2025.09 | 0.638 | — | |
| Random initBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=100%2026.05 | 0.638 | — | |
| Random initBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=10%2025.09 | 0.609 | — | |
| Random initBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=10%2026.05 | 0.609 | — | |
| Random initBackbone=3D ViT-B, Evaluation Protocol=Linear probing, Label Percentage=1%2025.09 | 0.582 | — | |
| Random initBackbone=3D ViT-B [82], Linear probing=true, Supervision Level=1%2026.05 | 0.582 | — | |
| CT-CLIP2026.04 | 0.574 | — | |
| COLIPRIZero-shot=true2026.06 | — | 0.785 | |
| COLIPRIEvaluation protocol=Downstream task evaluation (task-specific heads trained)2026.06 | — | 0.876 | |
| CT-CLIPZero-shot=true2026.06 | — | 0.731 | |
| CT-CLIPEvaluation protocol=Downstream task evaluation (task-specific heads trained)2026.06 | — | 0.767 | |
| fVLMZero-shot=true2026.06 | — | 0.778 | |
| fVLMEvaluation protocol=Downstream task evaluation (task-specific heads trained)2026.06 | — | 0.807 | |
| GLINT-CTZero-shot=true2026.06 | — | 0.85 | |
| GLINT-CTEvaluation protocol=Downstream task evaluation (task-specific heads trained)2026.06 | — | 0.877 | |
| MerlinZero-shot=true2026.06 | — | 0.728 | |
| RadZero3DZero-shot=true2026.06 | — | 0.762 | |
| V-JEPA 2.1Evaluation protocol=Downstream task evaluation (task-specific heads trained)2026.06 | — | 0.859 | |
| ViSD-BoostZero-shot=true2026.06 | — | 0.79 | |
| ViSD-BoostEvaluation protocol=Downstream task evaluation (task-specific heads trained)2026.06 | — | 0.745 |