Group ICA in FSL melodic with CIFTI surface data

I am trying to perform group ICA in FSL’s melodic but with CIFTI surface data. I’ve had no problems with this in the individual level, but it gives me the following error when performing group analysis. I’ve thought the cause might be that while my first subject’s data has 800 volumes per session, subjects 2 and 3 have approximately 1600 volumes per session, but this should pose no problem as the sessions are temporally concatenated.