I am trying to calculate aCompCor components outside of fMRIPrep, and I want to use the nuisance ROIs (CSF, WM, and combined) produced by fMRIPrep. Are these masks kept in the work directory? Or would they have to be recreated?
It is possible to save those ROIs by using the option --debug compcor in the fmriprep command.
The ROIs will be output in the func folder of the FMRIPREP output with a name like this:
I have a quick question regarding the masks. Are they in the native space? Have they already been eroded? Pretty much, do I need to do anything to these images to replicate the fMRIPrep confounds workflow up to the finished version of the nuisance masks?
Short answer: yes I believe that those masks are in native bold space and already eroded. You should not have anything to do to the masks to replicate the fMRIprep confounds workflow.
Do note however: " fMRIPrep does high-pass filtering before running anatomical or temporal CompCor." More details here.