You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Make sure motion parameters are reported in the same convention for tortoise and eddy (#1084)
* Export DWI head-motion parameters in RAS
Motion confounds were reported in backend-specific frames: eddy/SHORELine
via get_fsl_motion_params (FSL) and DIFFPREP via raw TORTOISE Okan params
(LPS). Relative to the RAS world frame this flips trans_x/rot_x (eddy) and
trans_x/trans_y/rot_x/rot_y (DIFFPREP), so the two backends disagreed and
neither matched RAS-defined ground truth.
Add get_ras_motion_params (per-volume ITK/LPS transform -> RAS, decomposed
about the image center) and use it in CombineMotions (eddy/SHORELine);
negate the x,y components of the DIFFPREP Okan params (LPS -> RAS). The
confounds trans_*/rot_* columns are now RAS for every HMC backend.
Add test_get_ras_motion_params_no_axis_flip, which verifies no axis flip on
a radiological grid.
* also make the ecc parameters appear
* update diffprep param test
---------
Co-authored-by: Taylor Salo <salot@pennmedicine.upenn.edu>
0 commit comments