Dear Alfonso and CONN community,

I am currently analyzing resting-state fMRI data that were preprocessed using fMRIPrep and imported manually into CONN (since the automatic fMRIPrep import did not work in my case).

In the preprocessing options within CONN, I initially selected only:

Structural segmentation (grey/white/CSF tissue estimation)
Functional masking

However, I noticed that in the denoising step, only white matter and CSF confounds are included, but no motion parameters (realignment) or scrubbing regressors appear.

This raises the following question:

👉 When using fMRIPrep-preprocessed data in CONN (without automatic import), is it recommended to also enable:

Realignment (to estimate motion parameters), and
ART-based outlier detection (scrubbing)?

Or would this constitute redundant preprocessing, given that motion correction has already been performed in fMRIPrep?

In addition, I am unsure how to correctly handle the fMRIPrep confounds file (desc-confounds_timeseries.tsv):

I tried importing the timeseries file as a 1st-level covariate, but it does not appear in the denoising confounds and I am not sure whether this is appropriate.

Is there a recommended way to use the fMRIPrep confounds within CONN?

Or is it preferable to ignore the TSV file and rely entirely on CONN’s internal denoising (WM/CSF, motion, ART)?

I would like to ensure that my denoising pipeline is methodologically sound and consistent with best practices.

As I am currently finalizing analyses for my PhD, I would greatly appreciate any guidance on this, as it is somewhat time-sensitive.

Thank you very much for your help!

Best regards,
Ricarda