Hello FSL Experts,
I have run bedpostx and FS recon -all for all subjects with usable DTI and T1 data. I am now running probtrackx and I’m interested in using seed/target ROIs along the white matter surface to generate N x N connectivity matrices. I am wondering if there are any definitive benefits of seeding from vertices on the FreeSurfer white matter surface, compared to seeding from voxels in a WM surface volume.
Previously, I converted FreeSurfer lh.white and rh.white surfaces to volumes (.nii.gz), re-sliced to subject T1 space, and then masked my dilated GM parcellation by this WM boundary to define regional seed volumes along the WM surface in native space. These WM boundary ROIs were then co-registered to diffusion space using a rigid-body transform, and defined as seeds in probtrackx calls for each subject.
What would I gain (if anything) from instead using surface-based seeding? (defining seeds as cortical surface files in GIFTI format, --meshspace=freesurfer, etc.) Would you expect the volume-based approach described above to perform worse, and if so, why?
Thanks very much for your time!
Graham
|