Dear Ricky,
>
> I have a data set of 84 subjs with 2 DTI scans each in which I would like to perform TBSS on between two groups. On a large subset of the subjects (say 75%), GRE fieldmaps were also scanned, most of the time just before the two DTI scans.
>
> My question is would it be kosher to use the fieldmaps to perform the distortion correction for the 75% of subjects who do have the EPI scans? And the for the ~25% of subjects who do not, just don't perform and fieldmap based distortion correction?
>
> My intuition says that this should be OK and would only help, but I wanted to double check to make sure that this would not bias in favor of subjects with the fieldmap scans.
It would only be kosher if the probability of having a fieldmap is independent of the group. For example if 85% had fieldmaps in one group and 65% in the other it would NOT be kosher.
Kosherness aside, it is not clear it would be a good idea anyway. Even though there is individual differences in distortions the general pattern tend to be the same. That means that even though your TBSS results may not be in the correct anatomical space you would at least compare like with like (roughly). By correcting some subjects and others not there is a risk that you do worse in terms of “like with like”.
Also, since TBSS performs its non-linear registration on the (distorted) EPI images part of the distortions will be (roughly) corrected for by that registration step.
Having said all that my advice for future studies remains: ALWAYS acquire some means of correcting for susceptibility distortions.
Good luck Jesper
>
> Thanks!
>
> -Ricky
|