Dear FSLers,
I’ve been analyzing FA using TBSS to compare 2 groups (controls vs patients)
and got expected results. I then began wondering whether groups differences
in FA were simply due to excess motion during the scan since our patients
were diagnosed with a movement disorder, and head movement is inevitable in
our study.
So I had an idea of extracting rotation matrix from “eddy***.ecclog” file
and enter them as additional covariate in TBSS.
The problem is that the eddy correction is affine transformation and
decomposed rotation matrix by “avscale” is not unique.
I’m kind of stuck with this idea, and am wondering whether I should abandon
this idea of having additional covariate. I appreciate if someone has an
opinion about this kind of approach.
Also, if there is any paper written about effect of motion induced
false-positives in voxel-wise DTI analysis, please let me know.
Thanks in advance,
Hedok
|