What explains the difference in scale for DVARS between fmriprep confounds and Power et al 2012?

In Power et al 2012, the range of DVARS x10% is ~0-15 whereas the range of the DVARS column in the fmriprep confounds we’re seeing ~20-80.

Is this explainable by dataset variability or is there some other reason the ranges are different (eg different calculation, different preprocessing prior to dvars calculation, etc…)?


What kind of population are you working with (children/clinical populations move more → have higher DVARS). Is this task or resting state (tend to move more during tasks → have higher DVARS)? Both are some considerations on the data side as to what could explain this difference. I believe, but could be wrong, that the fMRIPrep DVARS calculation should match or closely match the Power 2012 definition.


1 Like

Hm… maybe its because “In this report, DVARS was calculated upon final functional connectivity images. Note that DVARS could, in principle, be calculated at other points of the processing stream (e.g., prior to functional connectivity processing).”

In my task-fmri data my nonstandardized dvars is also in that range with a median of 46. Note that fmriprep uses the standardized dvars for flagging outlier volumes.

Ooo good catch. I think this makes sense. Thanks!

Also the fmriprep/mriqc implementation controls for volume (1/N) that the original paper doesn’t include - IQMs for functional images — mriqc 0.16.1+101.g78518dc documentation

1 Like