Hi,
> Following are the two approaches that I have tried
>
> 1) Run all the processing as recommended until
> tbss_4_prestats, create a
> mask to mask out regions that are not of any interest to me
> from the
> skeleton, use the masked skeleton when running stats.
> (randomise)
> 2) Mask "FA images" itself and feed in the masked images
> into TBSS pipeline.
> Ofcourse mean_FA image doesnt align well with MNI152 image
> but this option
> helps me to observe more differences between the two
> groups that I am
> trying to compare.
>
> Reading through previous posts I know that there are groups
> that are using
> the first approach for ROI based TBSS. I was just wondering
> if someone could
> help me know if the second approach was correct ?
You can not mask the original FA images and then feed them into TBSS: this would result in a poor (if not completely wrong) registration of the FA maps onto the FA template. On the other hand, you could always apply the warpfields generated by TBSS onto the masked FA images, but I really don't think that there would be any benefit in adopting this approach over masking the skeleton in the standard space...
Cheers,
Gwenaelle
|