DIM: A Diffusion Instability Measure

Motivation: QA in diffusion MRI is challenging and rarely implemented. We provide a new QA measure that is borrowed from fMRI. Goal(s): To test a diffusion instability measure (DIM) on HARDI data from different scanners, with and without concurrent magnetic field monitoring (FM). Approach: We estimate the correlation coefficient matrix between all pairs of diffusion directions across voxels inside the phantom and compute a QA measure from the eigenvalues of the matrix. Results: DIM is sensitive to data quality affected by b-values, TE, acquisition/reconstruction types. Data acquired with FM showed a DIM value that was, on average, 10 times smaller than those without FM. Impact: We developed and tested a diffusion instability measure (DIM) that is sensitive to factors affecting image quality, such as magnetic field monitoring, f0 stabilization, TE, gradient strengths, making it a promising tool for routine diffusion MRI quality assurance.