Hello,
when censoring in xcp_d is performed on the basis of filtered FD, default FD<0.2 mm thresholding does not seem to be appropriate anymore (e.g., Gratton et al., Neuroimage 2020).
Leaving aside the general limitations of applying cut-offs from the literature (Power et al., Neuroimage 2015): Are there any rules of thumb how much adaptation of FD thresholds is necessary?
The abovementioned study (Gratton et al., 2020) suggested that an FD<0.1 mm threshold may be reasonable, and this seems to be adopted in many of their follow-up studies, at least for fast-TR multi-band data. Yet, not everyone may have the same practical experience.
Best wishes,
Marcel