Print

Print
Hi Brad,

easiest is to use the command line version of melodic to do this. If all your data is in a common space and all other pre-proc has been done as well (motion correction, filtering, slice timing etc) then simply 

(i) create a text file with all the file names in it, e.g. 

ls /home/blah/study/blahblah/subject_X/data.nii.gz > filenames.txt

(ii) melodic -i filenames.txt -v --report -o outputdir.gica  -a tica --Tdes=tempdesign.mat -Tcon=tempdesign.con ...


if you replace tica by concat you'll get the concatenation approach.

An alternative is to just go ahead with the normal GI, switching the registration to lowest possible DOF and have the data re-registered into the common space (unfortunately also re-resampled ;)

hope this helps
Christian

On 30 Aug 2007, at 01:12, Bradley Buchsbaum wrote:

Hi,

I would like to run group ICA on a number of data sets that have
already been co-registered and normalized outside of FSL. We do this
because EPI --> anatomical registration using flirt does not work well
with our high field data.

Is there a way to run Melodic without any registration?  When group
ICA is selected (either temporal concatentation or tensor ICA), the
registration checkbox cannot be deselected.

I tried setting registration to "0" in the .fsf file but this did not work.


any ideas?

thanks,

Brad Buchsbaum


-- 
Bradley R. Buchsbaum
Helen Wills Neuroscience Institute
210 Barker Hall, UC Berkeley
phone: 510-642-2839
email: [log in to unmask]