I’m using nipype’s ACompCor module for denoising.
In fmriprep, decompositions are done on a highpass filtered version of the data, which is why the recommendation is to use cosine regressors in addition to compcor regressors for denoising.
However, this is the case when pre_filter is set to “cosine” in ACompCor right?
My question is, if I choose “polynomial” instead, will it have a similar effect? Will it address the issue mentioned by Chris here?