VoxelMorph: A Learning Framework for Deformable Medical Image Registration

14 Sep 2018Guha BalakrishnanAmy ZhaoMert R. SabuncuJohn GuttagAdrian V. Dalca

We present VoxelMorph, a fast learning-based framework for deformable, pairwise medical image registration. Traditional registration methods optimize an objective function for each pair of images, which can be time-consuming for large datasets or rich deformation models... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK USES EXTRA
TRAINING DATA
COMPARE
Diffeomorphic Medical Image Registration Automatic Cardiac Diagnosis Challenge (ACDC) VoxelMorph Dice 0.790 # 3
Diffeomorphic Medical Image Registration Automatic Cardiac Diagnosis Challenge (ACDC) VoxelMorph RMSE 0.24 # 1
Diffeomorphic Medical Image Registration Automatic Cardiac Diagnosis Challenge (ACDC) VoxelMorph Hausdorff Distance (mm) 8.4 # 1
Diffeomorphic Medical Image Registration Automatic Cardiac Diagnosis Challenge (ACDC) VoxelMorph Grad Det-Jac 9.2 # 1
Diffeomorphic Medical Image Registration CUMC12 VoxelMorph Mean target overlap ratio 0.517 # 2
Diffeomorphic Medical Image Registration OASIS+ADIBE+ADHD200+MCIC+PPMI+HABS+HarvardGSP VoxelMorph (CC) Dice (Average) 0.753 # 3
Diffeomorphic Medical Image Registration OASIS+ADIBE+ADHD200+MCIC+PPMI+HABS+HarvardGSP VoxelMorph (CC) GPU sec 0.45 # 2
Diffeomorphic Medical Image Registration OASIS+ADIBE+ADHD200+MCIC+PPMI+HABS+HarvardGSP VoxelMorph (CC) CPU (sec) 57 # 2
Diffeomorphic Medical Image Registration OASIS+ADIBE+ADHD200+MCIC+PPMI+HABS+HarvardGSP VoxelMorph (CC) Neg Jacob Det 19715 # 4
Diffeomorphic Medical Image Registration OASIS+ADIBE+ADHD200+MCIC+PPMI+HABS+HarvardGSP VoxelMorph (CC) Dice (SE) 0.145 # 2