I have extracted WM and CSF signal prior to scaling my data, which I plan to use in 3dDeconvolve. Currently, the CSF units are in the 4400 units range, and the WM is in the 5500 units range. Given that the signal of my scaled data is around 100 units, would the large discrepancy in signals cause any potential issues? I noticed this thread (https://afni.nimh.nih.gov/afni/community/board/read.php?1,142689,142689#msg-142689), which had the same issue, but since my data has larger signal differences, I’m unsure if that will be a problem.
Out of curiosity, are those tissue-based regressors applied per run, or across all runs? Per run is preferable.
Assuming there is no per-run scaling variance, then in theory there would be no difference at all. Any applied betas would scale inversely. In practice, there might be tiny truncation differences, but even those are almost certainly not worth thinking about.
This would be easy to test. Use 1deval to scale the ROI regressors as you see fit, and compare the results in a new regression. There should be little to see.
Yes, the tissue-based regressors are applied per run.
Thanks for the help!
The
National Institute of Mental Health (NIMH) is part of the National Institutes of
Health (NIH), a component of the U.S. Department of Health and Human
Services.