Flexible selection of confounds in addition to AROMA Components

[LuRoe7]

What would you like to see added in fMRIPost-AROMA?

How does fmripost-aroma deal with white matter signal, CSF signal and global signal?

I was wondering if it's planned to include an option to flexibly select other confounds to regress from the BOLD time series in addition to the AROMA noise components?

For instance, choosing between the following different strategies would be great:

1. white matter signal, csf signal + AROMA ICs
2. white matter signal, csf signal, global signal + AROMA ICs
3. white matter + derivatives, csf + derivatives, global signal + derivatives + AROMA ICs
4. ...

My suggestion arises from Parkes et al. (2018), Neuroimage (https://doi.org/10.1016/j.neuroimage.2017.12.073) who compared various denoising strategies including those specified above...

Do you have any interest in helping implement the feature?

No

Additional information / screenshots

No response

[tsalo]

I would recommend combining your AROMA regressors with other confounds in another tool, like Nilearn or XCP-D (which is actually set up to ingress fMRIPrep + fMRIPost-AROMA derivatives).