Dear Jake,
This is easy - just take the cluster mask image and start by selecting
the cluster you want (you can do this by taking the index image where
each cluster has a unique number, with the biggest being the biggest
cluster, and using fslmaths with the -thr and -uthr options to zero out
any numbers below and above the value you are interested in).
Once you've done this, just multiply this cluster by the individual
labels or probability images in the atlas images (contained in
$FSLDIR/data/atlases/). You can then threshold these so either
select out the desired labels or only keep voxels above a certain
label probability. Note that your original image needs to be in the
MNI space for this to work (the atlases are in this space) and of the
same resolution as the atlas. If that is not the case then you should
resample and rethreshold your cluster image with applywarp.
Hope this is clear.
All the best,
Mark
On 25 Jul 2011, at 18:11, Jake Kuster wrote:
> Dear FSLers!
>
> We would like to know if there is a way to take a mask image of a large cluster from our analysis (~30k voxels) and use an atlas to get a list of regions that the cluster intersects with. (Ie, all regions the cluster runs through). Is this possible?
>
> We have been looking for documentation on how this could be done and have not been able to find this yet!
>
>
> Thank you in advance,
>
> Jake
>
|