Hi Suzanna,
> HI,
>
> I was wondering if I could check the exact procedure
> for
> investigating a small volume in randomise
> (specifically the
> cerebellum and peduncles).
>
> Would the following pipeline be correct?
>
> 1) Run TBSS normally with the normal bet binary
> mask.
Well, if you're looking for voxel-wise analysis of the
whole-brain, yes. Otherwise, you're running TBSS on
the skeleton mask so you want to go till the 4D image
all_FA_skeletonised is created, not only all_FA.
> 2) Generate a ROI mask and use that with the -m
> function in
> randomise, with or without the TFCE option.
Right. With TFCE or cluster-based correction, you're
still looking at "voxel-skeleton wise" results and not
at the significance of the global change of FA within
your ROI...
> Regarding generating a binary mask, is the following
> procedure okay?
>
> 1) load the mean_FA_skeleton into FSLview (or should
> I use the MNI152?)
I would recommend the mean_FA_skeleton on top of
mean_FA so you can draw *grossly* a mask of the
cerebellum and/or the brainstem white matter, and then
mask it with the mean_FA_skeleton (fslmaths
mycerebellum_mask -mas mean_FA_skeleton_mask
myfinal_mask).
> 2) create new mask as a 4D mask. I assume this will
> then project the
> ROI onto all the volumes with the 4D data of each
> individual or the
> mean_/all_ files
No, you only need a 3D mask both for randomise and
fslmeants if you want some absolute values in your
ROI.
> 3) draw around cerebellum in every slice in one
> orientation (eg.
> coronal)
> 4) fill the cerebellum regions with 1s, fill
> everything outside that
> region with 0s
> 5) run randomise with this mask
> Does this sound acceptable?
Yes
Hope this helps,
Gwenaelle
_____________________________________________________________________________
Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail http://mail.yahoo.fr
|