Yes, that all sounds perfectly correct. Simply define your band-pass filter in the Denoising tab, and that will be applied separately to the voxel-level BOLD timeseries as well as to the ROI-level BOLD timeseries  computed during the Setup step (e.g. resulting from the computation of the mean BOLD timeseries across all voxels within an ROI)

Hope this helps

Alfonso

Originally posted by jtanne98:

I am working with two data sets (one with TR 0.8 and one with TR 2.01). 

To calculate correlation between ROIs one each data set, I want to be consistent with temporal dynamics as much as possible and avoid the high frequency components the TR 0.8 may acquire that are not present in the TR 2.01. 

Discussing with team and colleagues, we decided to bandpass the mean ROI BOLD signal. While there are options in preprocessing to do this (listed below), I believe doing this before the denoising would be an error. So I have a few questions:

(1) does this process seem sound?

(2) does de-noising filter on voxel or on ROI?

(3) if de-noising filters on voxel, is there a way to do it again after ROI means? 

 

Pre-processing steps mentioned: 

functional ROI extraction (compute BOLD timeseries within ROI) 

functional Band-pass filtering (temporal filtering of BOLD data)