Seismology of the brain: characterizing brain mechanics using MRI

Abstract

Recent advances have shown that the brain’s mechanical properties, such as stiffness and viscosity, are crucial for understanding its function and changes in disease. However, studying these properties has historically been difficult due to the brain’s delicate structure and protection by the skull. The development of advanced MRI techniques has enabled non-invasive investigation, yet many aspects of brain mechanics remain poorly understood. This project aimed to address these gaps through two objectives: (1) developing intrinsic Magnetic Resonance Elastography (iMRE), and (2) investigating physiological factors influencing brain tissue deformation. iMRE leverages natural brain pulsations caused by vascular swelling during the cardiac cycle, measured using Displacement Encoding with Stimulated Echoes (DENSE). These displacement fields enable estimation of tissue stiffness without external mechanical stimulation. Early implementations of iMRE suffered from low sensitivity and noisy results. In Chapter 2, this limitation was addressed by optimizing DENSE for detecting small displacements, resulting in high-resolution and accurate stiffness maps. This demonstrated that iMRE can achieve reliability comparable to conventional MRE while capturing the brain in its natural, unperturbed state. Remaining challenges included fluid-related noise, model mismatch, and non-unique solutions. Chapter 3 improved mechanical property estimation by introducing poroelastic and poroviscoelastic models within a nonlinear inversion framework. These models account for both solid tissue and interstitial fluid, enabling estimation of physiologically meaningful parameters such as compressional stiffness and hydraulic permeability. Compared to traditional viscoelastic models, these approaches reduced artifacts, improved repeatability, and produced unique, comparable solutions across subjects. The poroviscoelastic model further enhanced performance and provided insight into viscous behavior. Despite some uncertainty due to assumed boundary conditions, this work established poro(visco)elastic modeling as a key advancement and revealed the brain’s intrinsically ultra-soft nature. Chapter 4 demonstrated that the brain is far softer than previously believed, with stiffness values around 6 Pa - nearly two orders of magnitude lower than earlier estimates. Prior studies likely overestimated stiffness due to external actuation or post-mortem changes. These findings suggest that the brain behaves in a fluid-like manner under physiological conditions, driven by fluid redistribution. Revisiting earlier assumptions showed that viscoelastic models can capture this ultra-soft behavior when constraints are relaxed, emphasizing the importance of intrinsic measurements and appropriate modeling at low frequencies. Beyond stiffness, iMRE enables analysis of brain deformation through strain tensor imaging (STI). Chapter 5 examined drivers of volumetric and octahedral shear strain, finding that both are influenced by pulse pressure, with volumetric strain linked to cerebral blood volume and shear strain to tissue stiffness. These metrics arise from a complex interplay of physiological factors. Chapter 6 investigated directional strain patterns, showing that first principal strains are largely shaped by boundary conditions such as the skull, while third principal strains align with underlying microstructure. This suggests the presence of mechanical anisotropy in the brain, where properties depend on direction. Overall, this work opens new possibilities for obtaining insights into brain mechanics in a non-invasive way, and for improving the detection and understanding of brain disease.