Denoising for rs-fmri


I have been reading up in the archives here to sort out a denoising strategy for my resting state dataset which will be used for dual regression and reho analysis later on. I am still confused about certain things.

@ChrisGorgolewski offered a good starting point in this post, but I did not find the AROMAAggrComp in my confound.tsv file. Are they the same as the aroma_motion columns in the confound.tsv file?

I understand that aroma_motion regressors are meant to aggressively denoise _preproc.nii.gz, and should not be used in conjunction with rot, and trans regressors and already non-aggressively denoised data, is that right? How about the motion_outliers regressors, what are they and how should they be used?

Thank you!

Hi @Heechberri,

Yes, those are the same. Earlier versions of fmriprep named those columns AROMAAggrComp, but now they go by aroma_motion

The general consensus is that when using the aroma denoising strategy (non-aggressive or aggressive), you should avoid including the trans and rot regressors, as they can potentially reintroduce noise.

This post explains the motion_outliers and how they can be applied post-fmriprep.

Hope this helps a bit.

1 Like

Thank! This helped alot!