Hi all,
In our research we are trying to analyze FA values inside the tracts, that we get from MNI atlases (White-matter probability atlas in FSL).
we first coregister FA to T1 linearly, then T1 to MNI with FLIRT and FNIRT and then take the inverse transform and apply it to tract masks drawn in MNI.
in our study we have stroke lesion patients, so we use FNIRT with cost function masking in the lesion area.
The point is that in FSL tutorials it's said that transfrom made by FNIRT isn't diffeomorphic by construction, and I'm trying to realize how it would affect our masks when we transform it from MNI to T1 space. Is there any recommendations on the range of Jacobian or its histogram to preserve diffeomorphism as possible? What advice can you give on analyzing the jacobian or warp fields to preserve correctness of the results.
thank you
Dmitry Isaev
Neurolinguistics lab
Higher School of Economics
Moscow
|