Hi Anderson,

Thank you for your suggestion. After I get the ${peaks}, I ran the 'cluster' command to get coordinates for the peaks, but they do not correspond to where I expect to have maxima from the ${img}. In fact, most of the coordinates from 'cluster' correspond to locations with zero intensity. Am I misinterpreting something here, or did I do something incorrectly?

Also, what do you mean by "masked by clusters"? Do I use the ${img} as a mask in fslmaths?

Thanks in advance,

Ye Yuan

On Wed, Oct 29, 2014 at 10:10 AM, Anderson M. Winkler <[log in to unmask]> wrote:

Hi Ye Yuan,

Have you considered gray scale dilation in fslmaths? Say the image with the peaks you'd like to find is ${img}, and you'd like local maxima that are at least ${k} voxels apart. Then it would be something like this:

fslmaths ${img} -kernel sphere ${k} -dilF dilated
fslmaths dilated -sub ${img} -bin tmp
fslmaths dilated -bin -sub tmp -mul ${img} peaks

The output can then be further masked by clusters so as to include only the maxima within each cluster. And it can also be given to the command "cluster" to get their coordinates.

Hope this helps.

All the best,

Anderson

On 28 October 2014 18:43, Ye Yuan <[log in to unmask]> wrote:
Hi FSL experts,

I am trying to look for peak voxel coordinates in fMRI activation using the 'cluster' command, but I'm finding that it is missing some of the possible peaks within a large cluster. I'm wondering if any of you might know how to perform sub-cluster analysis in FSL. I suppose I could iteratively adjust the thresholding until I got all the peaks I wanted, but is there an easier way to do this?

Thank you all in advance,

Ye Yuan