Hi - there's no one single approach for this kind of thing - but the kind of solutions you suggest are reasonable.  Note that that kind of analysis is very easy to do on the basis of the outputs from dual-regression.

Cheers, Steve.

On 4 Feb 2013, at 23:39, Matthew Hyett wrote:

Dear FSL users,

I've recently completed a MELODIC analysis across three separate groups (total N = 48) with the intention to use the resultant maps (as matched according to the Beckmann RSN8 nii file with fslcc) as a guide for selection of regions for a higher-level analysis.  There are a few questions, however, that I needed a little guidance on:

-- Given ICA maps cover quite a large area, I was wondering if there was a way to choose peak voxels of a given map (say, executive control) for subsequent analyses (e.g. DCM)?  Would Featquery or FSLView be useful here?  Would it be best to first choose peak voxels from the group-wise maps (e.g. melodic_IC.nii.gz) and then select voxels from each individual map (which I assume are the files named "filtered_func_data.nii.gz") using a function such as fslmeants?

Any help would be greatly appreciated!

Cheers,
Matt

---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------