Hi Will
I'm not aware of any outstanding bugs in melodic that would cause it to
silently hang. Are you sure it's not just that you've run out of RAM and are
swapping?
Yes it's better to do group-ICA on grayordinates. You can still dual-regress
that (step one into subject CIFTI, step two back
I think the issue, as I read it, is that Will’s data is only NIFTI currently,
so he doesn’t have any subject CIFTI that he could use for the stage 1 of dual
reg.
Our suggestion of course to remedy that would be that you process your data
into CIFTI, using the HCP Pipelines. ☺
Cheers,
-MH
--