Dear Christian, thank you very much for your reply. I checked the log but I could not find the fslmath call you mentioned. As I understand, the group intersection mask it is created /usr/share/fsl/4.1/bin/fslmaths mask -Tmin -bin mask -odt char but it doesn't seem to get applied anywhere in the log if I read it correctly (Melodic v3.10) I should mention that we actually do the preprocessing using Feat (motion correction, unwarping, betting, smoothing, high-pass filtering) and not Melodic GUI. These files are than added in Melodic GUI to do the registration to MNI space, downsampling and ICA decomposition (concat mode) I have attached the report_log.html for a sample dataset with 2 input files. Kind regards, andi -----Original Message----- From: Christian F. Beckmann <[log in to unmask]> To: FSL <[log in to unmask]> Sent: Mon, Sep 24, 2012 8:59 am Subject: Re: [FSL] melodic mask Hi Very observant ;) When you call melodic from the GUI, the masking is done in preprocessing, you should find a line $FSLDIR/bin/fslmaths prefiltered_func_data -mas mask prefiltered_func_data_thresh in the log file. Therefore, once the melodic command line gets active it no longer needs bet (--nobet) and can simply threshold based on intensity (--bgthreshold=10) hth Christian On 23 Sep 2012, at 14:31, Andi Heckel <[log in to unmask]> wrote: > Dear FSL experts, > > I have a question regarding masking under Melodic (Multi-Session temporal concatenation). > > In our case, the FoVs of the bold images differ a bit across subjects (with e.g. parts of the cerebellum cut away in some subjects). > > According to "report_log.html" as produced by the Melodic GUI, an intersection mask seems to be created applying "fslmaths -Tmin" on the merged > individual brain-masks. > > However, this group mask does not appear in the subsequent melodic call: > > /usr/local/fsl/bin/melodic -i .filelist -o groupmelodic.ica -v --nobet --bgthreshold=10 --tr=3.330 --report --guireport=../../report.html --bgimage=bg_image -d 0 --mmthresh=0.5 -a concat > > Doesn't it make sense to add the intersection mask to this command line using melodic's --mask switch in order to exclude anatomy that is not present in all input volumes ? ( ... --mask=mask.nii.gz ... ) > > Moreover, if the --nobet option is omitted, would melodic create such a group (intersection) mask in "-a concat" mode ? > > Kind regards, > andi > > >