Effects and Correction of Patient Bulk Motion in Cranial DENSE MRI

Applications of DENSE to measure cardiac driven brain tissue pulsations are highly sensitive to bulk patient motion due to the sub-millimeter displacement encoding required, limiting its accuracy, reproducibility, and use in pediatric and aging populations. This study aims to assess the impact of induced bulk motion on DENSE scans and to what extent these motion effects can be mitigated with existing and newly proposed methods.

Participants (N = 10) underwent test-retest 2D DENSE scans at three slice locations with and without induced motion using a 3.0 T system. Brain displacement fields were calculated using pipelines without and with motion correction based on polynomial fitting to an outer ring of brain tissue. Subsequently, voxel-wise comparisons were made between scans and pipelines to evaluate scan repeatability and measure biases in displacement measures.

In comparing scans with and without induced motion, motion significantly impacted displacement measures, resulting in intensity variations and phase wrap artifacts, as well as increased the mean peak-to-peak displacements. Motion-correction removed the intensity variations and phase wrap observed in phase images, and reduced variations between scans taken with and without induced motion ( $R_{\text{corr}.}^2$ = 0.45 ± 0.29, $R_{\text{corr}.}^2$ = 0.96 ± 0.05; ${\text{RMSD}}_{\text{uncorr}.}$ = 0.089 ± 0.005 mm, ${\text{RMSD}}_{\text{corr}.}$ = 0.00788 ± 0.00004 mm). Test-retest reproducibility increased after motion correction with induced motion ( ${\rho }_{\text{uncorr}.}$ = 0.46 ± 0.38, ${\rho }_{\text{corr}.}$ = 0.98 ± 0.01), and in the absence of induced motion ( ${\rho }_{\text{uncorr}.}$ = 0.76 ± 0.35, ${\rho }_{\text{corr}.}$ = 0.98 ± 0.02).

Motion correction significantly improved the correspondence of DENSE measures acquired during induced motion and improved test-retest reproducibility, even in the absence of induced motion.