Are confounds from fmriprep necessary for GLM if data run through Denoiser nuisance removal tool?

, ,
1

Hello experts,

I have a dataset that has been preprocessed using fmriprep. Using the six motion confounds specified by fmriprep (among other confounds), I used the Denoiser tool (https://github.com/arielletambini/denoiser) to remove these sources of noise from the data.

I am now preparing to enter these data to a first level GLM in SPM12. As an SPM user, I am used to including the .txt file containing motion regressors from SPM in first level GLMs. Having preprocessed the at a using fmriprep and denoiser, however, I do not have that file.

My question is whether I should include the motion regressors from fmriprep in first level analyses. Is it redundant given that the sources of noise from the data have been removed via the Denoiser tool? If anyone could explain the logic of including the regressors or not, it would really help me better understand this process.

Thanks so much!

2

If you have regressed out the motion parameters separately, you should not include them in your GLM.