Hi Paul,
First, thanks so much for the link to that incredible paper! I just finished reading it all through and it is so helpful.
Second, for our study we are looking at an often high-motion group (ADHD) and a (typically) lower-motion control group. We've deliberated extensively regarding the censor thresholds we should use, given the ADHD group has disproportionately higher levels of motion. Using the default of 0.3mm and 0.1 outlier fraction often brings the DOF below 50-60% for these subjects, hence why we adjusted them to 1mm and 0.15 (based on what we could find from other studies and looking at our own data). This does have the downside of missing some motion in subjects that don't reach the higher threshold, but at least the outlier fraction tends to capture some of the motion still.
As far as GCOR, I tested some subjects by lowering the censor thresholds but the visual GCOR image tends to not change very much. And the radcor images look exactly the same regardless of the censor threshold. Even in some subjects with low motion and otherwise clean data there can be fairly high visual GCOR activity. Should this be a concern for our data? Following the lead of the paper you shared, would you suggest using a specific GCOR cutoff percentage for exclusionary purposes?
Sincerely,
Ian