Summary of what happened:
According to my knowledge, fMRIprep generates an unbiased estimation (merge T1 images from 2 sessions) as the template to normalize data with --longitudinal flag. Without this flag, fMRIprep uses the first template (e.g. baseline T1 weighted images) as the template to normalize the data.
Thus, I thought that (1) baseline data preprocessed separately and (2) baseline data preprocessed with follow-up data (2 sessions for each participant) without --longitudinal flag would generate the same outputs for the baseline data since (1) and (2) use baseline T1-weighted images as the template.
However, I noticed that fMRIprep generates different outputs (for both preprocessed data and preprocessed data denoised by ICA-AROMA). Why does fMRIprep generate different outputs while using baseline T1-weighted images as the template? Does fMRIprep have additional processing for (2)?
Version:
fMRIprep 22.0.2