Dear Sarvy,
> I have several dozen T1-weighted MRI scans from subjects with focal lesions.
> I wish to do VBM analysis on these. One problem is that sometimes tissue
> that is within the lesion variably gets classified as GM or WM. To deal
> with this, I would like to turn all voxels within the lesion to a zero (by
> masking with a binary mask). When I try this, I see that absolutely no
> voxels within the lesion are classified as GM or WM, instead they are
> ignored when they are zeros; furthermore, the non-lesion GM and WM appear to
> be properly classified with this mask approach.
>
> This seems fair and sensible but I have not seen this done before in prior
> reports of VBM in subjects with a brain lesion. Are there any caveats or
> suggestions regarding this approach?
Yes, your approach seems to be reasonable. Note, that a mask - depending on how you perform the normalisation - would also be helpful/necessary during the normalisation procedure in order to improve the accuracy.
You can, perhaps, simplify your procedure in the way that you use your mask only for the analysis, i.e. you create a mask which excludes all voxels, which are representing a lesion in one of your subjects, and use this as an explicite mask during the statistic. Than you don't need to touch the single images at all.
Good luck,
Karsten
--
-------------------------------------------------------------
Karsten Specht, PhD
Department of Biological and Medical Psychology
& National Competence Centre for functional MRI
University of Bergen
Jonas Lies vei 91
5009 Bergen
Norway
Tel.: +47-555-86279
Fax: +47-555-89872
[log in to unmask]
http://fmri.uib.no/
|