I ran fMRIprep on the fMRI data without removing dummy scans or indicating the number of dummy scans.
I would like to know whether it might cause some problems (e.g. wrong alignment or coregistration…).
I’m planning to remove the dummy scans before running the first level GLM with SPM12!
For any dummy scans you remove, you’ll need to remove their corresponding regressors. Alterrnately, you can use the
non_steady_state_XX regressors to simply censor those scans.
If you know the number of dummy scans, I would check whether fMRIPrep detected them. This will be present in the reports, and will also show up as
non_steady_state_XX columns in the confound regressors. If the non-steady-state scans don’t match your dummy scans, or vary from subject to subject or run to run, you’re introducing some needless variance into the preprocessing. If that’s the case, I would suggest re-running with
--dummy-scans N . The calculation of some confound regressors will be affected (however slightly) by the number of scans declared or detected.
Thank you so much for your help!