When registering images to target FMRIB58_FA each image in all_FA as well as
mean_FA lack great brain areas.
Many of the original images have a lack of top slices, registration does not
account for that? Still the problem also occurs in inferior slices.
Furthermore brain-sizes are quite variable (6y-20y old subjects).
Using your "old style" (formerly recommended) n x n registration in FSl 3.3
registration was fine, voxels were not deleted, i.e. extend of each brain in
all_FA differed a bit due to their lacking slices (in contrast to the
FMRIB58_FA registration: extend of each partly-registered img in all_FA is
exactly the same).
Any solutions for a whole brain registration to FMRIB58_FA?
Cheers,
Stefan
|