DPARSF head movement

Q3. Methods, line 181. 
To control for confounds due to head movement effects during fMRI scanning, “subjects whose mean framewise displacement (FD) is greater than 1 mm was also excluded”. Is this average or maximal FD? If this is average, single functional volumes could still be affected by motion artefacts, whereas if this is maximal, 1mm is very permissive. Why not eliminate functional volumes with high FD (eg >0.5 mm or even less if possible) from the fMRI time series (scrubbing or censoring). In addition, given that control for motion concerns an extremely important issue when studying functional connectivity and when studying children and patients in fMRI, it also would be useful to also test for differences between the groups in motion variability, not only FD (eg. RMSE). 
For motion correction issues, see for examples: 
doi: 10.1016/j.neuroimage.2017.12.073 
doi: 10.1016/j.neuroimage.2015.02.063 
doi: 10.1016/j.neuroimage.2014.10.044 
我想请问下您 我怎么回复比较好呢?