Evaluating the Performance and Repeatability of Poroelastic and Poroviscoelastic Models in Intrinsic MR Elastography

• Published in: NMR in biomedicine Vol. 38; no. 7; pp. e70073 - n/a
• Main Authors: Burman Ingeberg, Marius, Van Houten, Elijah, Zwanenburg, Jaco J. M.
• Format: Journal Article
• Language: English
• Published: England Wiley Subscription Services, Inc 01.07.2025; John Wiley and Sons Inc
• Subjects: Actuation; Adult; Boundary conditions; Brain; Brain - diagnostic imaging; Brain - physiology; brain mechanics; cerebral vascular pulsations; Consistency; DENSE; Displacement measurement; Elasticity; Elasticity Imaging Techniques - methods; elastography; Female; Fluid flow; Humans; iMRE; Magnetic Resonance Imaging; Male; Material properties; Mechanical properties; Models, Biological; Performance evaluation; poroelastic; Poroelasticity; Porosity; Porous media; poroviscoelastic; Reproducibility; Reproducibility of Results; Spatial distribution; Stiffening; Storage modulus; Symmetry; Viscoelasticity; Viscosity; DENSE; poroelastic; iMRE; brain mechanics; elastography; poroviscoelastic; cerebral vascular pulsations
• ISSN: 0952-3480; 1099-1492; 1099-1492
• DOI: 10.1002/nbm.70073

ABSTRACT Intrinsic MR elastography (iMRE) leverages brain pulsations that arise from cerebral arterial pulsations to reconstruct the mechanical properties of the brain. While iMRE has shown much potential recently, the technique was demonstrated for a viscoelastic brain model only, which suffered from data‐model mismatch at the low actuation frequencies of the arterial pulsations. This work aims to address those limitations by considering the porous nature of brain tissue, where both a poroelastic and a poroviscoelastic model are assessed and compared. As a secondary goal, the influence of two driving frequencies on the material properties is investigated by looking at the 1 Hz and 2 Hz components of the motion data. The poroelastic and poroviscoelastic properties of the brain were reconstructed using a subzone‐based nonlinear inversion scheme, using displacement measurements of eight healthy subjects from a previous study at 7 T MRI. The performance of each model was evaluated by assessing consistency of spatial distributions, repeatability through repeated scans, and left–right symmetry. The poroelastic model yielded mean storage moduli of 6.08 ± 0.87 and 32.01 ± 11.92 Pa, and the poroviscoelastic model yielded 5.32 ± 0.87 and 26.15 ± 8.02 Pa for the 1‐ and 2‐Hz motion components, respectively. Among the mechanical properties of interest, the storage modulus was the most consistent, with low limits of agreement of (upper/lower) 15.0%/−22.2% for the poroelastic model and 10.9%/−18.5% for the poroviscoelastic model, relative to the whole‐brain mean. It was also highly symmetric, with a mean whole‐brain symmetry ratio of 0.99 across subjects for both models. Mechanical properties related to fluid flow demonstrated less consistency. The 2‐Hz motion component was found to contain considerable information as it reflected the frequency‐related stiffening associated with porous media, highlighting its relevance for use in multifrequency mechanical characterization. Both models demonstrated good performance, with the poroviscoelastic model in general showing the highest consistency in terms of test–retest repeatability. Future work aims to improve the models by addressing current assumptions on the boundary conditions of the pressure field. Poroelastic and poroviscoelastic property reconstruction is found to be feasible using intrinsic MR elastography. The poroviscoelastic model was found to be more suitable for intrinsic magnetic resonance elastography (MRE) elastography in the current implementation. The brain exhibits ultra‐soft properties at the low driving frequencies given by the cardiac pulse.