Hi,
I am working with a number of images that have been converted from DICOM
(so voxel data is radiological orientation). To segment them (MFAST) and
coregister them (FLIRT) to a neurological-space template, I wanted to
have the originals (rad) as well as the x-swapped (neu) images.
Converted via Analyze, the original images had no NifTI header information.
What I thought was handiest was
ln -s originalFile.nii originalRadiologicalFile.nii
avworient \
-forceradiological originalRadiologicalFile.nii
avwswapdim \
originalRadiologicalFile.nii -x y z originalNeurologicalFile.nii
But then, if the voxel sizes are sx sy sz, and the centres of gravity cx
cy cz, the affine transformation in the sform of
originalRadiologicalFile.nii has the form
sx 0 0 cx
0 sy 0 cy
0 0 sz cz
and the affine transformation in the sform of
originalNeurologicalFile.nii has the form
-sx 0 0 cx
0 sy 0 cy
0 0 sz cz
According to the NifTI specifications (radiological -> negative x
scaling, neurological -> positive x scaling) that should be the other
way round, shouldn't it?
best wishes
Alle Meije
|