After looking through the afni_proc.py script, it appears that the default FD threshold is set to 0.3mm, though from this twitter post (https://twitter.com/AFNIman/status/977622189712211968), reducing it further to 0.2mm is desirable. Fortunately I have a fMRI task study with a cohort of healthy adults, so I wanted to see if setting the FD threshold somewhere in the range of 0.2-0.5mm seems appropriate? I realize that anything higher is probably too lenient, especially since my cohort generally isn’t prone to much motion, but I also want to ensure that I don’t lose too much tDOF from censoring/scrubbing.
afni_proc.py does not actually have a default enorm threshold, as what is reasonable depends on the subjects. One could more accurately say that the default is not to censor at all, though we do not recommend that. It is not reasonable to change defaults often, because that action changes results even on a repeated analysis. People do not appreciate that.
The examples in the help show motion censor limits of 0.3 mm for task and 0.2 mm for rest, assuming the subjects are healthy adults. Using 0.5 seems very high for healthy adults. My guideline is that more censoring is better, as long as there is enough data left for the analysis. Viewing a censored time point as a lost DoF is one perspective, viewing it as keeping bad data from corrupting the results is another. Analysis has shown that more censoring can improve the group results.
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.