Hi Anastasia,
>
> I would like to be able to register a seed mask for probtrackx
> using fnirt.
If I understand you correctly you would like to draw your mask in the
subjects T1 image, and then transform it into the subjects DTI space?
For a given subject the difference between T1 and diffusion space is
due to distortions caused by field inhomogeneities. It is true that
there has been some attempts to estimate these distortions using
fnirt, but it is NOT was it was intended to do. It may or may not
work, and it will almost certainly not work as well as for example
using a fieldmap.
> If I use fnirt to register diffusion data to T1, then use invwarp
> to get the
> transform from T1 to diffusion space and then specify that
> transform to
> register the seed mask drawn on T1 in the probtrackx command, would
> this be
> a plausible way to accomplish the registration?
This would be a plausible way if the probtrackx command can take a
non-linear fnirt transform as input. I doubt it can, but I am not
sure. Saad?
> Or should I register the T1
> mask to the diffusion space first using applywarp and then use this
> diffusion space mask as the seed?
This would certainly always work, so I think I would go for this. It
might then also be a good idea to specify the --super switch (to make
sure you are sampling the T1 space densely enough) and the --
interp=nn to make sure you get a binary mask.
Good luck Jesper
|