Image Registration on OASIS (test)
87Dice CoefficientEfficientMorph-23
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| EfficientMorph-23stride=2x2x2, C=96, Param=2.8M, Seg Loss=with2024.03 | 87 | — | 0.017 | — | — | — | — | — | |
| EfficientMorph-11stride=2x2x2, C=96, Param=1.8M, Seg Loss=with2024.03 | 86.23 | — | 0.01 | — | — | — | — | — | |
| TransMorph-Lstride=4x4x4, C=128, Param=108.11M, Seg Loss=with2024.03 | 86.2 | — | 0.128 | — | — | — | — | — | |
| Fourier-NetParam=4.19M, Seg Loss=with2024.03 | 84.7 | — | — | — | — | — | — | — | |
| EfficientMorph-23stride=4x4x4, C=96, Param=2.8M, Seg Loss=with2024.03 | 84.6 | — | 0.125 | — | — | — | — | — | |
| TransMorphstride=4x4x4, C=96, Param=46.5M, Seg Loss=with2024.03 | 84.58 | — | 0.119 | — | — | — | — | — | |
| S+LCPrediction Source=Oracle (best-sample)2026.03 | 81.2 | — | — | — | — | — | — | 16.3 | |
| EfficientMorph-23stride=2x2x2, C=96, Param=2.8M, Seg Loss=w/o2024.03 | 81 | — | 0.01 | — | — | — | — | — | |
| S+LDPrediction Source=Oracle (best-sample)2026.03 | 80.9 | — | — | — | — | — | — | 9.6 | |
| DEQRegMem. (MB)=6925, Training Time (s)=1.09, Inference Time (s)=0.982025.07 | 80.7 | 2.43 | 0.053 | 0.542 | — | — | — | — | |
| V+CPrediction Source=Oracle (best-sample)2026.03 | 80.7 | — | — | — | — | — | — | 5.7 | |
| RIIRMem. (MB)=12542, Training Time (s)=0.60, Inference Time (s)=0.302025.07 | 80.6 | 2.43 | 0.058 | 0.556 | — | — | — | — | |
| V+LCPrediction Source=Oracle (best-sample)2026.03 | 80.6 | — | — | — | — | — | — | 5.6 | |
| V+DPrediction Source=Oracle (best-sample)2026.03 | 80.5 | — | — | — | — | — | — | 5.9 | |
| V+LDPrediction Source=Oracle (best-sample)2026.03 | 80.5 | — | — | — | — | — | — | 5.9 | |
| S+DPrediction Source=Oracle (best-sample)2026.03 | 80.5 | — | — | — | — | — | — | 6.2 | |
| S+CPrediction Source=Oracle (best-sample)2026.03 | 80.5 | — | — | — | — | — | — | 6.5 | |
| TransMorph-Lstride=4x4x4, C=128, Param=108.11M, Seg Loss=w/o2024.03 | 80.4 | — | 0.009 | — | — | — | — | — | |
| GraDIRNMem. (MB)=16619, Training Time (s)=0.59, Inference Time (s)=0.242025.07 | 80.3 | 2.47 | 0.071 | 0.568 | — | — | — | — | |
| EfficientMorph-11stride=2x2x2, C=96, Param=1.8M, Seg Loss=w/o2024.03 | 80.3 | — | 0.011 | — | — | — | — | — | |
| V+CPrediction Source=Mean prediction (mu_Z)2026.03 | 80.2 | — | — | — | — | — | — | 4.5 | |
| V+CPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 80.2 | — | — | — | — | — | — | 5.7 | |
| TransMorph-Tinystride=4x4x4, C=6, Param=0.24M, Seg Loss=with2024.03 | 80.1 | — | 0.081 | — | — | — | — | — | |
| TransMorphstride=4x4x4, C=96, Param=46.5M, Seg Loss=w/o2024.03 | 80.1 | — | 0.03 | — | — | — | — | — | |
| S+DPrediction Source=Mean prediction (mu_Z)2026.03 | 80.1 | — | — | — | — | — | — | 4.8 | |
| S+DPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 80.1 | — | — | — | — | — | — | 6.2 | |
| S+CPrediction Source=Mean prediction (mu_Z)2026.03 | 80.1 | — | — | — | — | — | — | 5.1 | |
| S+CPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 80.1 | — | — | — | — | — | — | 6.5 | |
| S+LCPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 80.1 | — | — | — | — | — | — | 11.5 | |
| V+DPrediction Source=Mean prediction (mu_Z)2026.03 | 80 | — | — | — | — | — | — | 4.6 | |
| V+DPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 80 | — | — | — | — | — | — | 5.9 | |
| V+LCPrediction Source=Mean prediction (mu_Z)2026.03 | 80 | — | — | — | — | — | — | 4.4 | |
| V+LCPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 80 | — | — | — | — | — | — | 5.6 | |
| S+LDPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 80 | — | — | — | — | — | — | 8.5 | |
| V+LDPrediction Source=Mean prediction (mu_Z)2026.03 | 79.9 | — | — | — | — | — | — | 4.6 | |
| V+LDPrediction Source=Mean of resampled fields (Z_bar)2026.03 | 79.8 | — | — | — | — | — | — | 5.9 | |
| S+LDPrediction Source=Mean prediction (mu_Z)2026.03 | 79.8 | — | — | — | — | — | — | 3.2 | |
| EfficientMorph-23stride=4x4x4, C=96, Param=2.8M, Seg Loss=w/o2024.03 | 79.6 | — | 0.091 | — | — | — | — | — | |
| S+LCPrediction Source=Mean prediction (mu_Z)2026.03 | 79.4 | — | — | — | — | — | — | 1.1 | |
| TraditionalConditioning=Concatenation-based2021.06 | 78 | — | — | 0.97 | 4.62 | 28.4 | 0.212 | — | |
| VoxelMorphParam=0.063M, Seg Loss=with2024.03 | 78 | — | 0.1304 | — | — | — | — | — | |
| VoxelMorphMem. (MB)=2645, Training Time (s)=0.20, Inference Time (s)=0.142025.07 | 77.8 | 2.74 | 0.063 | 0.601 | — | — | — | — | |
| BaselineDescription=Non-conditional registration2021.06 | 77 | — | — | 1.157 | — | 200.3 | 0.204 | — | |
| CIR-DMMapping network=Distributed mapping network2021.06 | 77 | — | — | 0.963 | -3.78 | 28.5 | 0.216 | — | |
| Fourier-NetParam=4.19M, Seg Loss=w/o2024.03 | 77 | — | 0.031 | — | — | — | — | — | |
| CIR-CMMapping network=8-layer MLP centralized, Latent space=2562021.06 | 76.7 | — | — | 0.9 | -5.23 | 28.8 | 0.227 | — | |
| deedsBCVCPU time [s]=33.18 ± 0.00, Objective Function=Self-similarity context (SSC), Grid spacing=6×5×4×3×2, Search-radius=6×5×4×3×2, Quantization-step pyramids=5×4×3×2×12025.11 | 76.38 | — | 23 | — | — | — | — | — | |
| SyNCPU time [s]=48.66 ± 0.00, Objective Function=Normalized cross-correlation (CC), Multi-resolution schedule=(60, 40, 20)2025.11 | 75.53 | — | 0 | — | — | — | — | — | |
| elastixInference Time (s)=35.452025.07 | 75.2 | 2.89 | 0.068 | 0.551 | — | — | — | — | |
| VoxelMorphParam=0.063M, Seg Loss=w/o2024.03 | 67.83 | — | 2.981 | — | — | — | — | — | |
| Affine2025.07 | 61.2 | 8.56 | — | — | — | — | — | — | |
| InitialType=Spatial normalization2021.06 | 55.2 | — | — | — | — | — | — | — | |
| ViT-V-Netstride=8x8x8, C=252, Param=9.8M, Seg Loss=with2024.03 | 46.59 | — | 0.1272 | — | — | — | — | — | |
| TransMatchstride=4x4x4, C=96, Param=26.39M, Seg Loss=with2024.03 | 46.12 | — | 0.0546 | — | — | — | — | — | |
| TransMorph-Tinystride=4x4x4, C=6, Param=0.24M, Seg Loss=w/o2024.03 | 44.1 | — | 0.013 | — | — | — | — | — | |
| TransMatchstride=4x4x4, C=96, Param=26.39M, Seg Loss=w/o2024.03 | 40.37 | — | 0.1167 | — | — | — | — | — | |
| ViT-V-Netstride=8x8x8, C=252, Param=9.8M, Seg Loss=w/o2024.03 | 36.32 | — | 0.0149 | — | — | — | — | — |