Hi, I’m relatively new to fMRIPrep and pretty rusty with FSL, but I’m trying to conduct a group ICA on resting state data and I’m trying to clarify how to do this.
I’ve run my BIDS validated functional data through 25.2.4 and have derivatives inboth MNI152NLin6Asym and MNI152NLin2009cAsym. From here, I had planned to run single subject and then group level ICAs with MELODIC, but I’m unsure how to best proceed.
A) do I need to initially run it through the GUI to get a design.fsf file, or can I do everything needed via the melodic command line?
B) Is func/sub-***_ses-M000_task-rest_run-01_space-MNI152NLin6Asym_res-2_desc-preproc_bold.nii.gz the correct 4D input?
C) Given that fMRIPrep already does motion correction, B0 unwarping, slice timing correction, intensity normalisation and highpass temporal filtering, I assume I remove these options? But I’ve read elsewhere that preproc_bold.nii.gz is not skull stripped, so I still need to run BET?
D) if I use M000/anat/sub-***_space-MNI152NLin6Asym_res-2_desc-preproc_T1w as @Main structural image@, can I leave it on BBR, even though it doesn’t follow .nii.gz and _brain.nii.gz convention? Do I even need this included? fMRIPrep already skull strips the T1w and does co-registration with BBR, is there a way to tell that to Melodic?
E) once I’ve run Melodic for the ICAs, I can then train ICA-FIX and then run subsequent group level analyses?
I’ve also seen that ICAs can be run with GIFT (gift/doc/GIFT_Walk_Through.pdf at master · trendscenter/gift · GitHub). Is that more straightforward using fMRIPrep outputs?
Thanks in advance!