Medical Image Classification on NoduleMNIST3D
94.3AUCCdTransformer + CcCL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CdTransformer + CcCL2026.02 | 94.3 | 90.3 | |
| auto-sklearn2026.02 | 91.4 | 87.4 | |
| CoralBayU96H (FT)# pre-trained data=11k, Evaluation Protocol=Fine-tuned2026.06 | 91 | — | |
| AC-Tiny# of trainable params=9.5M2026.02 | 90.4 | 88.1 | |
| AC-Base# of trainable params=427M2026.02 | 90.2 | 89 | |
| BSDA# of trainable params=33M2026.02 | 89.2 | 86.1 | |
| LMTTM-VMI# of trainable params=21M2026.02 | 89.1 | 88.3 | |
| ResNet-50+ACS2026.02 | 88.6 | 84.1 | |
| ViViT-M + R-LLM# of trainable params=295M2026.02 | 88.5 | 87.4 | |
| Medformer2026.02 | 87.7 | 84.2 | |
| ResNet-50+3D2026.02 | 87.5 | 84.7 | |
| ResNet-18+ACS2026.02 | 87.3 | 84.7 | |
| ResNet-18+3D2026.02 | 86.3 | 84.4 | |
| Official# of trainable params=33M2026.02 | 86.3 | 84.4 | |
| AutoKeras2026.02 | 84.4 | 83.4 | |
| ResNet-18+2.5D2026.02 | 83.8 | 83.5 | |
| G-LoG bi-filtration (MLP)sigma=12026.02 | 83.8 | 85.2 | |
| ResNet-50+2.5D2026.02 | 83.5 | 84.8 | |
| SwinUNETR# pre-trained data=—, Evaluation Protocol=Fine-tuned2026.06 | 83 | — | |
| AC-Small# of trainable params=216M2026.02 | 82.1 | 87.4 | |
| Topo-Med (MLP)2026.02 | 80.8 | 73.6 | |
| G-LoG bi-filtration (MLP)sigma=1.52026.02 | 80.3 | 84.8 | |
| CoralBayU96H# pre-trained data=11k, Evaluation Protocol=Linear Probing2026.06 | 80 | — | |
| CoralBayU96B# pre-trained data=11k, Evaluation Protocol=Linear Probing2026.06 | 79 | — | |
| G-LoG bi-filtration (MLP)sigma=0.52026.02 | 78.9 | 85.2 | |
| Universal Model# pre-trained data=2.1k, Evaluation Protocol=Linear Probing2026.06 | 78 | — | |
| VoCo-H# pre-trained data=160k, Evaluation Protocol=Linear Probing2026.06 | 76 | — | |
| VoCo-B# pre-trained data=160k, Evaluation Protocol=Linear Probing2026.06 | 74 | — | |
| SuPreM# pre-trained data=2.1k, Evaluation Protocol=Linear Probing2026.06 | 71 | — | |
| G-LoG bi-filtration (MLP)sigma=02026.02 | 56.4 | 77.4 |