Hi,
We your loyal opposition over at the nipy project have been struggling
to do good conversions of some Philips data, and ran into a problem
that started to seem as if it was a specificity to FSL / fslview - can
we ask for advice?
The problem is described here:
https://github.com/nipy/nibabel/issues/285
When we do a default conversion of the data, the image array is in
Right Posterior Superior orientation. The converted bvectors appear
to be correct for that orientation, but when we run `dtifit` and
display the tensor directions, the tensors look flipped in Y. If we
reorient the data and bvecs to Left Anterior Superior, then the
tensors look correct from dtifit / fslview.
When dtifit calculates the tensor directions, are these in the
coordinate space of the bvecs (the voxel axes)? Or some other
coordinate space? Is is easy to explain why reorienting the image
and bvecs together would change the relationship of the voxel axes to
the bvecs?
Cheers,
Matthew
|