Hi,
Processing functional and anatomical data by means of some package
resulted
in the following aligned data (all in standard axial orientation):
- anatomical: anat.nii.gz (176x240x256, 1x1x1 mm^3 voxel size)
- functional: funct.nii.gz cropped to 35x49x27, 3.5x3.5x5.2 voxel size)
I extracted the first volume from the functional data, called
funct0.nii.gz
Using
flirt -in funct0 -r anat -applyxfm -usesqform -omat funct0_2_anat -
out funct0_2_anat
resampled the functional to the anatomical and made sure that the mm
coordinates
aligned. I checked this and it looked perfect.
Then I wanted to register the functional data to a standard Talairach
template, called tal.nii.gz.
For that purpose I first registered the anatomical data to this
template using flirt and fnirt:
flirt -in anat -ref tal -omat anat2tal -out anat2tal_flirt
fnirt --in=anat --ref=tal --aff=anat2tal
applywarp -i anat -r tal -w anat_warpcoef -o anat2tal_fnirt
The resulting anat2tal_fnirt looked very similar to tal. So far so good.
But now comes the hard part.
First I combined the two matrices into a single matrix by
convert_xfm -omat funct0_2_tal -concat anat2tal funct0_2_anat
and then I used
applywarp -i funct0 -r tal --premat=funct0_2_tal -w anat_warpcoef -o
funct2tal_fnirt.
The resulting funct2tal_fnirt looked bad.
Using only the linear registration:
applywarp -i funct0 -r tal --premat=funct0_2_tal -o funct2tal_flirt
produced a result that was roughly similar to the tal template. So I
guess that
the combined matrix is OK.
What is wrong with this approach?
Cheers,
Ed
|