 |
 |
Hi, > Thanks for the response. We have tried registration of the FA map to the FMRIB58_FA template using FNIRT and the results do look better. However we are not sure how the Hardvard-Oxford's cortical ROIs line up with this space. They do not appear to align properly. Why don't you think that they align well? They should be in alignment, and it seems fine when I look at it in FSLView. > We have also tried FNIRT for co-registration of the DTI brain to the T1 brain without contrast brain with okay results (see code below). However we are still having the consistent problem of a larger output brain with all of our FNIRT attempts. > > fnirt --ref=T1_brain.nii.gz --in=nodif_brain.nii.gz --inmask=nodif_brain_mask.nii.gz --refmask=T1_brain_mask.nii.gz --iout=dti2t1_fnirt.nii.gz --aff=dti2t1_flirt.mat --cout=dti2t1_fnirt_cout > > …where we used both the affine transformation matrix (output of flirt) of the dti brian to the T1 brain and a BBR transformation matrix (output of FEAT) for the --aff input. Both yielding similar results. I would not expect this to work well, as fnirt is not designed to work with different modalities, and has difficulty if the edge of the brain is masked out from the calculation at all iterations. My advice is to stick with the registration of the FA to the FMRIB58. All the best, Mark > If you have any other information that could help, please let us know. > > Thank you, > Nicole >