On 4 April 2017 at 12:33, Robert Mok <[log in to unmask]> wrote:
I just wanted to check if there is an easier way to find the cluster of interest? I have been trying to find some clusters and sometimes it's hard (e.g. because it joins with another larger cluster). Or some semi-automised way to find the clusters?

Well, if it joins another cluster it's technically one cluster!  

 

e.g. in an ideal world, i can find approx location of the peak of the cluster i want, then algorithm tells me what threshold to use. or more likely possible, adjust the threshold, then select the cluster in order to 'find' it (rather than try to find it again on the cluster table..).


I think people sometimes write ad-hoc things for this sort of thing, but I think there's always going to be cases when it's unclear which is the right approach, and it could be hard to justify.  I guess a simple solution if you want one is to take your peak voxels, and include voxels within a certain radius that are also in the same cluster (look at https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;de8e9a62.0802 for a how to create a spherical mask ).

Eugene
 
Thanks!
Rob

From: Robert Mok
Sent: 01 April 2017 22:52
To: FSL - FMRIB's Software Library
Subject: RE: [FSL] Creating mask from localiser scan - selecting a cluster / exclude others

Ah - I should've went back to the manual page.. For some reason I had the impression cluster couldnt do this. thank you!

From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Eugene Duff [[log in to unmask]]
Sent: 01 April 2017 22:47
To: [log in to unmask]
Subject: Re: [FSL] Creating mask from localiser scan - selecting a cluster / exclude others

Hi Robert,

Check out the cluster tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Cluster) - this should help you get want you want.

Best

Eugene

On 31 March 2017 at 18:19, Rob Mok <[log in to unmask]> wrote:
Dear Experts,

I'd like to make ROI masks based on a localiser scan. Using the uncorrected zstat images, I have clear clusters for the areas of interest, but there are also a lot of other clusters (sometimes large sometimes small) that I would like to exclude from this mask.

I have done this before using the following command which takes all values above a threshold to make a mask (but the localiser mask I was using did not have other clusters at this threshold so was not a problem):

fslmaths zstat1.nii.gz -thr 3.5 -bin output_mask.nii.gz

I also know I can find the info about the clusters in an image by using the Cluster function. Could I select one of the clusters, either in FSL view or using some function (e.g. cluster) and create an ROI mask?

Thank you
Rob