Hi,
If you have only done linear registration then this command is fine. The masks are
likely to larger simply because you have not thresholded the output. If you run it
through fslmaths with -thr 0.5 -bin then you should get a mask of similar size to the
original.
If you have done non-linear registration then you need to use applywarp rather than
flirt with applyxfm, but the thresholding issues above still hold.
All the best,
Mark
On 16 Dec 2010, at 16:01, Hilary Watson wrote:
> Hi All
>
> I originally posted the message below and didn't hear back - any help would be really good...
>
> "Just having a little trouble making some ROI masks for some group level analysis.
>
> So far I have drawn three MTL rois (e.g. hippocampus, parahippocampal cortex and perirhinal cortex) for each of my subjects based on anatomical landmarks on their highres structural. As such, my masks are all in the same space as the structural scan.
>
> I now want to use some combination of these masks (across subjects, not structures) to do some group analysis and was wondering the best way to go about this. I initially tried to covert each mask for each subject to standard space using FLIRT:
>
> - e.g. flirt -in ${j}_rightphc.nii.gz -ref standard.nii.gz -applyxfm -init highres2standard.mat -out ${j}_rightphc_standard
>
> but this made new masks that were much much bigger than the originals, and which were highly overlapping at the individual subject level. The plan (if this had worked) was to combine the standard space masks using fslmaths to create two group level masks for each structure: one that comprised all voxels that the combined masks covered and another that only comprised voxels that overlapped across all subjects.
>
> Any help with where I am going wrong with the registration would be greatly appreciated.
>
> Cheers
>
> Hilary"
>
|