Brain Tissue Segmentation on iSeg 2019 (test)
96.65Dice (CSF)DuMeta++
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DuMeta++Backbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Meta-trained on IBIS and OASIS32026.02 | 96.65 | 93.64 | 91.96 | 0.1043 | 0.2887 | 0.3291 | |
| DuMetaBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Meta-learning (DuMeta)2026.02 | 96.11 | 93.13 | 91.45 | 0.1082 | 0.2916 | 0.3318 | |
| nnU-Net2026.01 | 95.6 | 91.1 | 92.8 | — | — | — | |
| HDAN2026.01 | 95.3 | 93.7 | 94.7 | — | — | — | |
| LSRLBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Longitudinally consistent representation learning2026.02 | 95.08 | 91.02 | 88.23 | 0.1147 | 0.386 | 0.5294 | |
| LSORBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Longitudinally consistent representation learning2026.02 | 94.84 | 90.75 | 88.22 | 0.1717 | 0.3901 | 0.5339 | |
| ReptileBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Meta-learning (Reptile)2026.02 | 94.5 | 90.47 | 87.77 | 0.1173 | 0.3896 | 0.5324 | |
| PatchNCEBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Contrastive learning-based PatchNCE2026.02 | 94.44 | 90.43 | 87.82 | 0.136 | 0.382 | 0.452 | |
| MAMLBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Meta-learning (MAML)2026.02 | 94.33 | 90.25 | 87.68 | 0.1744 | 0.3931 | 0.5374 | |
| CC-3D-FCN2026.01 | 93.4 | 90.3 | 91.9 | — | — | — | |
| 3D-SkipDensSeg2026.01 | 93.4 | 90.1 | 92 | — | — | — | |
| LNEBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Longitudinally consistent representation learning2026.02 | 93.15 | 89.37 | 86.54 | 0.1753 | 0.4482 | 0.5274 | |
| HyperDenseNet2026.01 | 92.6 | 88.9 | 88.9 | — | — | — | |
| Swin-UNETR2026.01 | 92.2 | 88.2 | 85.1 | — | — | — | |
| DeepMedic2026.01 | 91.7 | 85.9 | 84.4 | — | — | — | |
| VoxResNet2026.01 | 91.7 | 89.5 | 88.9 | — | — | — | |
| GLCLBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Contrastive learning-based GLCL2026.02 | 91.41 | 87.18 | 83.37 | 0.2337 | 0.5017 | 0.6596 | |
| PCLBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Contrastive learning-based PCL2026.02 | 91.39 | 86.88 | 83.13 | 0.2326 | 0.5161 | 0.6692 | |
| DenseVoxNet2026.01 | 90.9 | 87.9 | 85.4 | — | — | — | |
| Context RestoreBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Pretext-task-based Context Restore2026.02 | 90.7 | 86.15 | 82.45 | 0.2641 | 0.542 | 0.6298 | |
| RandInitUnet.3DBackbone=3D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Random Initialization2026.02 | 90.29 | 86.16 | 82 | 0.2652 | 0.5497 | 0.6487 | |
| 3D-UNet2026.01 | 89.7 | 86.2 | 83.6 | — | — | — | |
| LINKS2026.01 | 88.5 | 85.7 | 83.9 | — | — | — | |
| RandInitUnet.2DBackbone=2D U-Net, Evaluation Protocol=One-shot, Pre-training Strategy=Random Initialization2026.02 | 87.67 | 84.02 | 79.65 | 0.3317 | 0.5578 | 0.8479 | |
| RF2026.01 | 85.9 | 83.9 | 82.1 | — | — | — | |
| CC-2D-FCN2026.01 | 85.4 | 81.4 | 75.6 | — | — | — | |
| MV2026.01 | 71.3 | 71 | 56.2 | — | — | — | |
| FAST2026.01 | 55.7 | 39.3 | 45.3 | — | — | — | |
| 3D-SkipDensSeg2026.01 | — | — | — | 0.256 | 0.174 | 0.209 | |
| 3D-UNet2026.01 | — | — | — | 0.483 | 0.463 | 0.602 | |
| CC-2D-FCN2026.01 | — | — | — | 0.567 | 0.672 | 0.789 | |
| CC-3D-FCN2026.01 | — | — | — | 0.285 | 0.179 | 0.259 | |
| DeepMedic2026.01 | — | — | — | 0.475 | 0.499 | 0.519 | |
| DenseVoxNet2026.01 | — | — | — | 0.497 | 0.456 | 0.582 | |
| FAST2026.01 | — | — | — | 1.838 | 1.012 | 1.714 | |
| HDAN2026.01 | — | — | — | 0.131 | 0.102 | 0.129 | |
| HyperDenseNet2026.01 | — | — | — | 0.314 | 0.289 | 0.335 | |
| LINKS2026.01 | — | — | — | 0.439 | 0.489 | 0.569 | |
| MV2026.01 | — | — | — | 1.963 | 1.196 | 1.771 | |
| nnU-Net2026.01 | — | — | — | 0.194 | 0.148 | 0.178 | |
| RF2026.01 | — | — | — | 0.475 | 0.594 | 0.695 | |
| Swin-UNETR2026.01 | — | — | — | 0.562 | 0.494 | 0.578 | |
| VoxResNet2026.01 | — | — | — | 0.428 | 0.483 | 0.546 |