Dear Eddy,
> While waiting for future developments of FNIRT,
> I was wondering if a possible solution would be to find a way for
> applying
> the TBSS warping only to non-gray matter areas.
> This might be achieved by masking the warping image (computed by
> first step
> of TBSS) by imposing a null deformation on the gray matter area
> obtained by
> VBM?.
I am not sure I fully understand your suggestion. Are you suggesting
to mask out the cortical areas while doing the FA-based registration
of the white matter?
If so, I am afraid that wouldn't really work either. Masking only
works to exclude areas from the sum-of-squared-differences part of the
cost-function. The regularization (which is what will relax the warps
from the previous gray matter registration) will always be calculated
based on the full FOV.
I am afraid that I cannot really think of a way to currently achieve
what you are after. In general people tend to base their registration
on their response variable, so that for a VBM style analysis the
registration will be based on the T1 data and for a TBSS style
analysis it will be based on diffusion data.
Good luck Jesper
|