I think their after rotating the diffusion gradient vectors (i.e. the bvecs file) rather than something like the data_V1.nii.gz file. This isn’t something we do in the HCP Pipelines either, but the code to do it for a single transformation matrix (e.g. raw diffusion space to T1w native space) exists and could be adapted for this purpose:
I don’t have a good sense of how much difference this makes. Perhaps some enterprising soul will try this on HCP data and show how much it reduces dispersion or otherwise improves bedpostX estimates…
Matt.
Hi - see the manual:
Cheers
Hi all,
When correcting for eddy currents, I understand that FSL affine registers the diffusion volumes to the reference volume with no diffusion weighting, which involves some sort of transformation/rotation. Several papers have noted that if a transformation/rotation has been applied to the image, then the same must be applied to the encoding vectors. My understanding is that these "encoding vectors" are simply the gradient directions applied during acquisition of the diffusion-weighted volumes (bvecs). However, I can't seem to find any info on whether or not this has been factored into FSL. If it hasn't, how do people go about applying the same transformation to the bvecs file prior to fitting diffusion tensors or running tractography?
Thanks in advance!
Manaal
