Table 1 Presents result on LPBA 40 dataset, including average Dice scores and percentage of locations with non-positive Jacobian. Standard deviations are presented in parentheses.
From: Dual attention network for unsupervised medical image registration based on VoxelMorph
Method | Dice | % of \({\mathcal{N}}\) |
|---|---|---|
NiftyReg | 0.694 (0.122) | 0.549 |
VoxelMorph | 0.703 (0.125) | 0.355 |
ANTs SyN | 0.707 (0.123) | 0.137 |
CycleMorph | 0.705 (0.133) | 0.157 |
DAVoxelMorph (Bending Penalty only) | 0.709 (0.122) | 0.255 |
DAVoxelMorph (Dual Attention CNN Architecture only) | 0.712 (0.126) | 0.365 |
DAVoxelMorph (Dual Attention CNN Architecture and Bending Penalty) | 0.714 (0.127) | 0.345 |
