I'm having an issue with the melodic_IC.hdr volume output from melodic. 

I'm currently running melodic 3.1 to find group level independent components in a cohort of 700 subjects. It's running on preprocessed resting state data, via command line on an SGE cluster.

The issue I'm having is that the components in the melodic_IC.hdr volume are all the same, i.e. it just contains repeated volumes of the first component but with slightly different values.  This is strange because in the stats folder, each of the zthresh maps for each component do appear to be correct and do show different components. I've run it with batches of fewer subjects with the same issue. 

I'm also not sure why it is giving me analyze .hdr/img output rather than nifti; all my input files are nii.gz and I don't see an option to change output type. 

Has anyone run into a similar issue or does anyone have any advice?