back to Papers


Evaluation of volume-based and surface-based brain image registration methods.
Arno Klein, Satrajit S. Ghosh, Brian Avants, B.T.T. Yeo, Bruce Fischl, Babak Ardekani, James C. Gee, J.J. Mann, Ramin V. Parsey.
2010. NeuroImage. 51: 214-220.

[PDF]    [doi:10.1016/j.neuroimage.2010.01.091]

Supplementary Material


Code (convert data, call registration packages, evaluate and analyze results):

Code (Matlab and Python) run on the LPBA40 individuals: LPBA40.tar.gz

Code (Matlab and Python) run on the FS40 individuals: FS40.tar.gz

Template building scripts:
FreeSurfer (Python): make_freesurfer_template.txt
ANTS/SyN (the buildtemplateparallel.sh script supercedes this implementation): make_syn_templates.tar.gz
Spherical Demons software (complete version used in this study): SDv1.1-svn593-klein.tar.gz

Data:

LPBA label list and (unlabeled) LPBA40 custom templates (95 MB)


Updates:

2/2010: We have since found much better affine initialization for SyN registrations
by including the following argument at the end of the "ANTS" command in Appendix A:
--number-of-affine-iterations 10000x10000x10000x10000x10000

7/2010: NEW standalone python algorithm for "paired indifference-zone" ranking: rank_paired_indifferencezone.py
(We used this new program to corroborate some of the permutation test results in the study.)


Registration evaluation studies conducted since this article was published:

EMPIRE: Evaluation of Methods for Pulmonary Image Registration 2010 (MICCAI 2010 challenge workshop)