I have a question regarding the deletion of dummy scans in fMRI preprocessing. My dataset includes both resting-state and task-based fMRI. To avoid the effects of scanner instability at the beginning of the acquisition, I have reserved 6 seconds between the scanner trigger and the official start of the task.
Now, I am planning to preprocess the data using FMRIPREP, but I am unsure about the correct way to handle these initial 6 seconds. Should I:
Remove the first 6 seconds of data before running FMRIPREP,
Use the -dummy-scans option within FMRIPREP, or
Remove the first 6 seconds after FMRIPREP preprocessing but before performing the main analysis?
I would appreciate any guidance or best practices on this.
I would not remove the data before running fMRIPrep. People looking at your dataset will expect the raw data to be untouched, straight from the scanner, and removing those volumes would violate that. Instead, I would use the --dummy-scans option in fMRIPrep, and be sure to include the one-hot regressors from that in your first level analyses. This recommendation could change in special circumstances, for example there is so bad motion in the beginning that brain registration is affected.
I have a similar dataset in which I plan to preprocess both resting-state and task-based fMRI data using fMRIPrep. For the task-based runs only, three dummy scans were acquired before the task started, whereas the resting-state runs do not include any dummy scans.
I was wondering: if I specify the --dummy-scans option in fMRIPrep, will this setting also be applied to the resting-state data, even though no dummy scans were acquired for those runs?