Closed gkiar closed 8 years ago
Reporting here for clarity. Current progress has been identifying source of poor derivatives is the registration between the B0 DTI volume at the MPRAGE volume for several subjects. First pass at fixing was to introduce skull stripping on the MPRAGE volume prior to alignment. This corrected the problem in some scans but introduced the problem in others. Next steps are to investigate datatypes and do a minor parameter sweep of FLIRT, of course seeking the most general solution possible and sticking to linear registration.
Fixed with these lines. I will refactor code in next days/week to be better looking and more modular, but not changing the function until I detect the next bug 😄