Hi i would like to preprocess a large fMRI and ASL dataset using fmri/ASLprep, so i’m trying to figure out what is the most efficient way to do so.
I understand that both fmriprep and ASLprep can make use of precomputed smriprep derivatives. I have a couple of questions:
if i’ve previously run fmriprep, do i set the -d flag to the fmriprep output directory when running ASLprep?
Currently fmriprep uses smriprep 0.18, whereas ASLprep uses smriprep 0.19.1 will this be an issue when running fmriprep first then ASLprep, or vice versa?
Do I need the freesurfer subject directory i.e., <output_dir>/sourcedata/freesurfer/ if the smriprep derivatives are already provided before running fmri/ASLprep ? if the freesurfer subject directory is deleted prior to running fmri/ASLprep, will fmri/ASLprep run recon-all despite having the precomputed smriprep derivatives?
If you are precomputing smriprep, then would want it to go to the precomputed smriprep directory, e.g. –derivatives smriprep=/path/to/smriprep. If you are first doing anatomical+fMRI with fmriprep, then yes, you can point it to fmriprep directory.
For most use cases, I think it should be fine unless you want to do session-wise processing within individuals (e.g. not create subject averages, if indeed you do have multiple sessions per subject). Session-wise processing was added in smriprep 0.19.
For all of the great surface steps that are enabled by FreeSurfer, yes you would.
They should, at least I’m sure fMRIPrep would, unless you were to enable --fs-no-reconall which is not recommended.