Evaluation of cerebral cortex viscoelastic property estimation with nonlinear inversion magnetic resonance elastography

Abstract

Objective: Magnetic resonance elastography (MRE) of the brain has shown promise as a sensitive neuroimaging biomarker for neurodegenerative disorders; however, the accuracy of performing MRE of the cerebral cortex warrants investigation due to the unique challenges of studying thinner and more complex geometries. Approach: A series of realistic, whole-brain simulation experiments are performed to examine the accuracy of MRE to measure the viscoelasticity (shear stiffness, μ, and damping ratio, ξ) of cortical structures predominantly effected in aging and neurodegeneration. Variations to MRE spatial resolution and the regularization of a nonlinear inversion (NLI) approach are examined. Main results: Higher-resolution MRE displacement data (1.25 mm isotropic resolution) and NLI with a low soft prior regularization weighting provided minimal measurement error compared to other studied protocols. With the optimized protocol, an average error in μ and ξ was 3% and 11%, respectively, when compared with the prescribed ground truth. Mid-line structures, as opposed to those on the cortical surface, generally display greater error. Varying model boundary conditions and reducing the thickness of the cortex by up to 0.67 mm (which is a realistic portrayal of neurodegenerative pathology) results in no loss in reconstruction accuracy. Significance: These experiments establish quantitative guidelines for the accuracy expected of in vivo MRE of the cortex, with the proposed method providing valid MRE measures for future investigations into cortical viscoelasticity and relationships with health, cognition, and behavior.

Mapping brain mechanical property maturation from childhood to adulthood

“…Only damping ratio of the vmPFC ( r = −0.28; p = 0.001), superior parietal lobe ( r = −0.28, p = 0.001), and inferior parietal lobe ( p = −0.28; p = 0.001) were significantly correlated with volume after correction for multiple comparisons. Some correlation between MRE properties and volume is expected as they both will vary with age, but the general lack of significant relationships is expected given previous studies that have shown independence in these parameters and that MRE outcomes are not significantly biased by volume ( Schwarb et al, 2016 ; Hiscox et al, 2022 ; Scott et al 2022 ; Hiscox et al, 2020 ). Full results are found in Supplemental Information Table S9 .…”

“…Cortical maturation had not previously been investigated using MRE, with only a few studies reporting cortical mechanical property in the adult brain. However, using high-resolution MRE imaging ( Hiscox et al, 2022 ), here we are able to identify several cortical mechanical property developmental trends. Of note is the difference in stiffness maturation between the adjacent regions of the motor cortex and the somatosensory cortex.…”

OSCILLATE: A low‐rank approach for accelerated magnetic resonance elastography

“…This requirement for repeated measurements has imposed limits on several aspects of MRE research, such as use on populations who have trouble laying still for extended periods of time including children 9 . Furthermore, the long acquisition times limit achievable spatial resolution necessary to study small anatomical areas, such as the subfields of the hippocampus 10 and thin sections of the cerebral cortex 11 . And recent techniques involving multiple MRE datasets, such as multiple actuations to study brain anisotropic properties 12–15 or multiple frequencies to study brain frequency‐dependent material properties, 16,17 are limited in number of scans possible or useable spatial resolution in order to keep exams at reasonable length.…”