Abdominal Organ Segmentation on BTCV (val)
85.64Mean DiceTeacher
Evaluation results
| Method | Links | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TeacherBackbone=nnU-Net with a residual encoder, Capacity=full capacity2026.01 | 85.64 | 96.39 | 94.72 | 94.92 | 73.84 | 80.32 | 97.33 | 87.16 | 90.68 | 87.82 | 75.21 | 84.51 | 72.94 | 77.4 | 4.66 | |
| ReCo-KDBackbone=lightweight nnU-Net, channel scaling (t)=2, knowledge distillation=region- and context-aware distillation2026.01 | 85.01 | 95.94 | 93.85 | 94.72 | 75.57 | 78.11 | 96.97 | 89.83 | 91.58 | 87.19 | 72.5 | 80.72 | 71.26 | 76.91 | 8.89 | |
| MedNeXt2026.01 | 82.98 | 90.08 | 86.96 | 88.96 | 77.27 | 78.16 | 96.91 | 84.24 | 92.16 | 88.65 | 75.45 | 80.13 | 68.83 | 70.87 | 5.45 | |
| Physics-inspired Spatially-Decoupled Synthesis frameworkPre-train Data=Synthetic, Data Amount=5,0002026.03 | 81.51 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SwinUNETRv22026.01 | 81.26 | 90.78 | 86.29 | 85.62 | 79.2 | 75.9 | 95.21 | 78.9 | 90 | 86.25 | 72.61 | 74.5 | 71.44 | 69.66 | 12.86 | |
| UNETR++2026.01 | 80.92 | 94.69 | 85.99 | 86.9 | 78.83 | 73.79 | 96.22 | 83.27 | 91.2 | 87.02 | 72.4 | 71.82 | 67.64 | 62.19 | 9.59 | |
| SwinUNETRPre-train Data=Real (CT), Data Amount=5,0002026.03 | 80.56 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Student baselineBackbone=lightweight nnU-Net, channel scaling (t)=2, knowledge distillation=none2026.01 | 80.38 | 89.3 | 93.52 | 93.33 | 58.82 | 77.41 | 96.11 | 74.97 | 89.78 | 86.5 | 65.58 | 81.19 | 66.79 | 71.66 | 15.19 | |
| SwinUNETR2026.01 | 80.13 | 88.56 | 85.92 | 86.03 | 79.4 | 75.5 | 95.41 | 79.58 | 90.13 | 86.18 | 71.12 | 69.36 | 69.35 | 65.19 | 14.01 | |
| 3D UX-Net2026.01 | 79.74 | 92.47 | 84.39 | 86.54 | 78.72 | 74.16 | 95.44 | 82.47 | 90.93 | 85.03 | 70.56 | 64.6 | 66.49 | 64.85 | 12.43 | |
| TransBTS2026.01 | 78.87 | 89.92 | 84.31 | 85.59 | 73.86 | 72.09 | 96.2 | 81.49 | 89.81 | 85.23 | 65.67 | 71.2 | 63.74 | 66.15 | 14.47 | |
| nnFormer2026.01 | 78.28 | 92.4 | 83.31 | 85.39 | 72.02 | 73.16 | 94.85 | 83.61 | 89.56 | 81.64 | 67.63 | 70.76 | 61.58 | 61.71 | 10.57 | |
| nnU-Net2026.01 | 77.82 | 88 | 84.78 | 85.47 | 71.85 | 75.44 | 95.32 | 76.54 | 90.93 | 85.14 | 68.91 | 62.19 | 66.9 | 60.19 | 10.91 | |
| SwinMMPre-train Data=Real (CT), Data Amount=5,0002026.03 | 76.72 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UNETR2026.01 | 74.96 | 88.36 | 82.86 | 84.1 | 66.28 | 70.88 | 95.07 | 78.27 | 86.82 | 79.93 | 63.41 | 63.68 | 58.43 | 56.41 | 21.78 | |
| SegFormer3D2026.01 | 74.34 | 87.42 | 82.72 | 84.92 | 70.82 | 69.42 | 93.83 | 79.05 | 87.82 | 81.3 | 63.18 | 64.47 | 52.76 | 48.75 | 13.41 | |
| SlimUNETR2026.01 | 71.54 | 84.88 | 81.79 | 83.05 | 64.63 | 65.39 | 94.61 | 76.13 | 87.34 | 80.08 | 58.96 | 57.37 | 49.34 | 46.45 | 12.34 |