Dear John and SPMers,
A slightly strange question. Not sure if this is a bug in SPM or simply me getting confused (most likely the latter).
When I am trying to normalise images to MNI space using DARTEL if my initial (and then segmented) image is 160x240x256 1/1/1 voxelsize, using the default bounding box NaN settings and keeping voxelsize at 1/1/1, the final image ends up as 181x217x181. Ie. Different dimensions to the standard 1mm cubed MNI brain at 182x218x182 - this means that they can't be viewed neatly together in fslview. I would be keen to end up with 182x218x182 dimensions as the SPM output.
However there is no issue if I use the same starting image and go for a 2/2/2 voxelsize, which produces a final image of 91x109x91, nicely aligning to the 2mm cubed MNI brain.
I think I previously 'fixed' this problem in the 1/1/1mm situation by using a custom bounding box but am not sure how legitimate that is as a solution.
%matlabbatch{1}.spm.tools.dartel.mni_norm.bb = [-91 -126 -72
% 90 91 109];
Have you or others come across this previously and could you comment on the best way to manage it?
Many thanks indeed
Neil
Clinical Research Fellow
C3NL
Imperial College
|