Voxel Volume Overlap: Voxel-Size Sensitive Indicators of Subject Motion in Functional MRI

Abstract

Subject motion is a significant problem for the analysis of functional MRI data and is usually described by “total displacement” or “scan-to-scan displacement”. Neither parameter, however, takes into account voxel size, which clearly is relevant for the actual effects of motion on the data. Consequently, it is hitherto impossible to compare motion between subjects/studies acquired using different voxel dimensions, precluding the development of generally applicable recommendations for fMRI quality control procedures. Here, a new set of “voxel volume overlap” (VVO) parameters is developed and explored, assessing the actual volumetric effects of subject motion on the voxel-level. Further, the extent of out-of-plane movement (particularly detrimental to image quality) can be quantified. Analyses show the new parameters to be valid and sensitive to voxel sizes. Their relation to existing parameters is explored, and defaults are suggested. The algorithm is freely available as a toolbox for a common image processing software solution (SPM).