Table 4 Registration performance on datasets with small and large transformations.
From: A two-step deep learning method for 3DCT-2DUS kidney registration during breathing
Methods | Metric (mm) | Group A | Group B | ||
|---|---|---|---|---|---|
CT–CT | CT–US | CT–CT | CT–US | ||
Ours | HD | 3.58 ± 1.37 | 3.85 ± 0.32 | 4.36 ± 1.46 | 3.75 ± 0.83 |
MCD | 1.02 ± 0.59 | 0.82 ± 0.04 | 1.28 ± 0.53 | 1.10 ± 0.20 | |
Voxel-Morph | HD | 7.97 ± 0.31 | 4.20 ± 0.34 | 8.34 ± 6.75 | 4.12 ± 0.94 |
MCD | 2.63 ± 0.02 | 0.95 ± 0.04 | 2.79 ± 2.49 | 1.15 ± 0.23 | |
C2FViT | HD | 8.01 ± 0.09 | 5.82 ± 0.88 | 11.85 ± 1.79 | 5.47 ± 0.09 |
MCD | 3.38 ± 0.65 | 1.37 ± 0.23 | 3.57 ± 0.58 | 1.53 ± 0.10 | |
VTN-Affine | HD | 5.84 ± 0.41 | 4.41 ± 0.47 | 9.48 ± 5.21 | 3.86 ± 1.09 |
MCD | 1.83 ± 0.13 | 0.95 ± 0.02 | 3.08 ± 1.94 | 1.09 ± 0.20 | |
ConvNet-Affine | HD | 4.32 ± 1.06 | 3.72 ± 0.01 | 4.80 ± 1.66 | 4.06 ± 1.04 |
MCD | 1.28 ± 0.43 | 0.83 ± 0.04 | 1.64 ± 0.72 | 1.16 ± 0.28 | |
