Thanks Stephen - I haven't tried the masking option in melodic yet, so very useful advice.  As an aside in our case, would the intersection be best taken from the resting state functional images (after using BET) or the brain extracted T1 images?

Cheers,
Matt

On 20 May 2013 16:11, Stephen Smith <[log in to unmask]> wrote:

Hi - sounds like the data hasn't been masked (brain-extracted) correctly - probably best if you create a mask that is the intersection of all valid data from all inputs and pass that in using melodic's mask option.

Cheers.

On 20 May 2013, at 04:54, Matthew Hyett <[log in to unmask]> wrote:

Dear FSL users,

Apologies if this is a re-post - I tried to upload an example image of the problem below but it was over 200k.

Our group recently ran melodic on previously pre-processed (using SPM) resting state fMRI data (48 subjects).  We used the following command (both with and without BET subject-specific main structural image registration):

melodic -i .filelist -o groupmelodic.ica -v --nobet --bgthreshold=3 --tr=2.00 --report --guireport=../../report.html --bgimage=bg_image -d 100 --mmthresh=0.75 --Ostats -a concat

We have encountered an issue with the derived components in that many of the thresholded image (on either a mean high-res or standard space background) voxels appear to extend beyond the background image.  I am using the obtained melodic components in another program (DCM) that uses standard space so I would like to get ensure the melodic voxels are also reliably in standard space.

Is this simply a reporting issue or something more problematic that should be fixed?

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

---------------------------------------------------------------------------

Stop the cultural destruction of Tibet

--
Matthew Hyett | Research Assistant | Queensland Institute of Medical Research

[log in to unmask] | Mob: 0416 206 908

Systems Neuroscience Group
http://www.sng.org.au/people/matthew-hyett