3D Medical Image Classification on VLM3D Challenge
98.5AUC (Bowel T1)AnyMC3D
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AnyMC3DBackbone=MII, Trainable Parameters (M)=1.32, Frozen Backbone=true2025.12 | 98.5 | 93.9 | 98.8 | 95.7 | 95.7 | 67.8 | 88.8 | 86.5 | 88.9 | 79.5 | 89.4 | 1.7 | |
| MSTTrainable Parameters (M)=23.05, Frozen Backbone=false2025.12 | 95.6 | 90.7 | 97.5 | 93.3 | 92 | 67.7 | 83.4 | 85.2 | 85.9 | 77.8 | 86.9 | 4 | |
| AnyMC3DBackbone=DINOv3, Trainable Parameters (M)=1.2, Frozen Backbone=true2025.12 | 95.4 | 92.2 | 98.4 | 95.3 | 96.2 | 72.9 | 90.3 | 85.6 | 89.2 | 79.3 | 89.4 | 2 | |
| AnyMC3DBackbone=MedGemma, Trainable Parameters (M)=2.01, Frozen Backbone=true2025.12 | 95.1 | 86.5 | 96.7 | 94.2 | 93.4 | 59.5 | 89.7 | 84 | 87.4 | 79.5 | 86.6 | 5.2 | |
| 3D DenseNetTrainable Parameters (M)=11.24, Frozen Backbone=false2025.12 | 92.6 | 84.2 | 95.1 | 92.5 | 91.6 | 65.4 | 90.3 | 72.4 | 84.9 | 63.7 | 83.3 | 6.4 | |
| MIITrainable Parameters (M)=0.03, Frozen Backbone=true2025.12 | 86.1 | 79.7 | 88.1 | 83.1 | 89.1 | 61.4 | 83.1 | 71.5 | 70.1 | 72.7 | 78.5 | 8.7 | |
| 3D ResNetTrainable Parameters (M)=33.14, Frozen Backbone=false2025.12 | 83.7 | 75.6 | 90.4 | 92.2 | 89.2 | 67.3 | 89.7 | 69.9 | 82.6 | 68.6 | 80.9 | 7.6 | |
| M3TTrainable Parameters (M)=29.23, Frozen Backbone=false2025.12 | 79 | 82.7 | 83.1 | 92.3 | 89.9 | 67.3 | 90.8 | 70.2 | 83.8 | 65.8 | 80.5 | 7.5 | |
| MedicalNetTrainable Parameters (M)=46.16, Frozen Backbone=false2025.12 | 78.1 | 74.4 | 87.3 | 89.9 | 86.1 | 62.4 | 88.7 | 67.9 | 80.1 | 67.6 | 78.3 | 8.1 | |
| MedGemmaTrainable Parameters (M)=0.03, Frozen Backbone=true2025.12 | 69.7 | 66.2 | 72.9 | 75.3 | 74 | 59.5 | 75.4 | 66.9 | 65.2 | 64.9 | 69 | 10.9 | |
| VoCoTrainable Parameters (M)=50.49, Frozen Backbone=false2025.12 | 66.4 | 68.8 | 93.9 | 91.9 | 92.6 | 59 | 89.9 | 74.6 | 85.2 | 70.2 | 79.3 | 7 | |
| RSNA-KaggleTrainable Parameters (M)=25.04, Frozen Backbone=false2025.12 | 65.9 | 83.6 | 98 | 95.7 | 95.7 | 60.1 | 86.4 | 78.2 | 88.2 | 58.4 | 81 | 6.8 | |
| 3D ConvNeXtTrainable Parameters (M)=31.32, Frozen Backbone=false2025.12 | 64.1 | 72.9 | 60.4 | 81.7 | 70.9 | 56.7 | 85.7 | 62.2 | 69.8 | 58.7 | 68.3 | 11.5 |