Originally posted by Jose Maximo:

Hi,

I'm currently trying to perform nT2*w analyses using our resting state scans and I apply a minimal preprocessing pipeline to our images: motion correction, co-registration, and MNI normalization. This outputs a wu*.nii image. I do not need to run denoising as the T2*weigthed BOLD image is needed to estimate tissue iron. I still run outlier detection within conn (global signal, z=3 and subject-motion, mm=0.5) and would like to know if I can remove the outlier volumes from my T2*weigthed BOLD image (wu*nii) using the outlier files produced by conn using conn before I estimate tissue iron. 

Thanks. 

You can use the outlier information from CONN, but I would be careful about actually deleting volumes from the nT2* image. In most fMRI workflows, scrubbing/outlier handling is usually done by marking those time points as nuisance regressors rather than physically removing the volumes, because removing frames changes the temporal structure of the dataset.

If your goal is estimating tissue iron from the T2*-weighted signal, I would first check the assumptions of the specific nT2*w analysis pipeline you're using. Some approaches expect the original time series with consistent volume numbers, and you may introduce problems if the number of volumes no longer matches the motion correction or normalization steps.

A safer approach is usually to use the CONN outlier regressors (the ART-derived regressors) during the analysis step if the software allows it, or create a cleaned time series while keeping the volume indexing consistent. If you do decide to remove volumes, make sure the same volumes are removed from every associated file and document exactly which frames were discarded.