Marius Burman Ingeberg

In recent years, scientists have discovered that the mechanical properties of the brain (for example, how stiff or viscous they are) play an important role in how the brain functions and how it changes in disease. Because the brain is protected by the skull and consists of very fragile tissue, its mechanical condition has long been difficult to study. The development of advanced MRI techniques has made it possible to characterize these properties safely and non-invasively in humans. Despite this progress, much remains unknown - both regarding the normal mechanical behavior of the brain and how this changes in various diseases. Studying brain mechanics therefore offers great opportunities to improve our understanding of brain disorders and, ultimately, hopefully also their diagnosis and treatment.

The general objective of this project was two-fold. The first goal was the development of intrinsic Magnetic Resonance Elastography (iMRE), a technique that utilizes the subtle pulsations of the brain caused by the swelling of blood vessels with each heartbeat. These pulsations are captured using a highly sensitive MRI method called Displacement Encoding with Stimulated Echoes (DENSE). The resulting displacement fields are used to derive the stiffness of brain tissue, as described in Chapters 2-4. The second objective was to investigate the biological and physiological factors influencing pulsatile tissue deformations, as described in Chapters 5 and 6. Greater insight into the biological factors behind the measured tissue deformations is needed. This helps to better understand how these relate to brain function and disease. The following section summarizes each chapter.

At the start of this project, intrinsic Magnetic Resonance Elastography (iMRE) was largely unexplored territory. The first attempt, published in 2012, demonstrated the feasibility of the technique, but the resulting parameter maps contained little detailed anatomical information and much noise. Consequently, iMRE was hardly developed further. The limited accuracy was likely due to the low sensitivity of the MR techniques used at the time, as they were not optimized for measuring brain pulsations. In Chapter 2, we addressed this by applying an optimized DENSE method, which can measure the small tissue displacements caused by brain pulsations with high sensitivity. This increased sensitivity for tissue pulsations proved crucial: it allowed us to map the stiffness of the brain accurately and with high resolution. The results showed that iMRE can map mechanical properties without external actuation, with spatial accuracy and reliability comparable to traditional MRE, which does depend on external mechanical activation. Importantly, iMRE captures the brain in its natural, untouched state, providing a more physiological view of brain mechanics. This chapter showed that challenges remained—such as noise from fluid motion, a mismatch between data and model, and measurements that were arbitrarily scalable (known as non-uniqueness). Nevertheless, this work successfully established iMRE as a promising, fully non-invasive instrument for studying brain tissue mechanics in humans.

Chapter 3 built on the findings of Chapter 2. Using poroelastic and poro-viscoelastic tissue models within the non-linear inversion (NLI) scheme, important limitations of viscoelastic iMRE were addressed to improve the estimation of mechanical properties. Poro-(visco)-elastic models treat the brain as a combination of a porous solid tissue filled with fluid, allowing for more physiologically realistic mechanical properties to be estimated. This approach made it possible to map the compression modulus (lambda-modulus) and hydraulic permeability per voxel, while reducing artifacts from fluid flow. Additionally, the mismatch between data and model was reduced, and these models yielded unique, reliable solutions (i.e., values comparable between subjects rather than values showing only relative, regional differences within a single subject). Overall, poroelastic and poro-viscoelastic properties showed comparable or improved repeatability compared to the viscoelastic model. The poro-viscoelastic model particularly achieved higher repeatability and symmetry than the purely poroelastic model, while also providing extra insight into the viscous behavior of the solid brain tissue. Together, these advancements offered a more consistent and comprehensive picture of brain mechanics. Some challenges persisted, however. Notably, the need to assume pressure boundary conditions introduces uncertainty, especially in the lambda-modulus and hydraulic permeability. Poro(visco)elastic modeling proved to be a good step forward for iMRE and, by enabling unique solutions, provided new evidence of the intrinsically ultra-soft nature of the brain.

Chapter 4 built on the findings of Chapter 3 by examining the implications of poroelastic results for viscoelastic iMRE. The ability of poroelastic models to produce unique solutions revealed that the brain is much softer than previously assumed, with a global stiffness around 6 Pa—nearly two orders of magnitude lower than the kilopascal values reported in most previous studies. Previous estimates were based on external actuation, which artificially increases stiffness, or on post-mortem tissue, in which mechanical properties have changed after death. The finding that the brain is ultra-soft suggests that it exhibits fluid-like behavior at physiologically relevant, low frequencies and small deformations. This behavior likely reflects the redistribution of interstitial and vascular fluids in response to deformation. These findings challenged the assumptions previously made in viscoelastic iMRE. In Chapter 2, a lower limit of 100 Pa was imposed during the calculation of mechanical properties, based on the assumption that the brain was much stiffer. This limitation prevented the model from determining the ultra-soft behavior of the brain. Chapter 4 showed that the viscoelastic model actually yields unique, physically meaningful values consistent with the frequency-dependent behavior reported in the literature. These results not only confirm the mechanically ultra-soft nature of the human brain but also emphasize the importance of intrinsic measurements. Furthermore, the results show that understanding tissue mechanics requires accounting for the specific characteristics of (porous) tissue dynamics at low frequencies.

Based on DENSE MRI measurements of small movements in the brain, iMRE provides detailed insight into how the mechanical properties of the brain under normal, natural conditions. By calculating strain tensor images (STBs), the same measurements can also be used to study how the brain stretches and deforms during the heartbeat. STBs offer a different perspective on brain mechanics, for example, through measurements called relative volume change and octahedral shear, two metrics that may have clinical value. However, as accurate measurement of STBs is a recent development, the biological and physiological processes behind these measures are not yet fully understood. Chapter 5 investigated possible underlying factors contributing to volumetric strain and octahedral shear deformation. Three complementary categories of factors were considered: pulse pressure effects, vascular and hemodynamic factors, and tissue properties. The results indicated that both measures were influenced by pulse pressure, with volumetric strain more strongly linked to cerebral blood volume, while octahedral shear deformation was more associated with tissue stiffness. The findings also suggested that both deformation measures arise from a complex interplay of factors, underscoring the importance of further research for a better understanding of these relationships.

In addition to relative volume change and octahedral shear, STBs can also show the principal directions in which the brain stretches and contracts, known as the first and third principal strain (EH and DH). As with the aforementioned deformation measurements, the factors determining spatial patterns in these principal directions are not yet fully understood. Chapter 6 aimed to clarify this by investigating whether EH and DH are primarily determined by the general boundaries of the brain, such as the skull, or by the microstructure of the brain. The results showed that EH patterns are largely determined by boundary effects: during each heartbeat, the swelling of the brain is limited by the rigid skull, and part of the total expansion occurs via the foramen magnum (the large opening at the base of the skull). This suggests that the skull strongly determines how the brain deforms. In contrast, the DH showed a preference for aligning perpendicularly to the main direction of the brain's microstructure, suggesting that local tissue structure influences the DH in the brain. This points to an important idea: the brain can exhibit mechanical anisotropy; that is, its stiffness and deformation depend on the direction in which it is loaded. Because STB measures the brain in its natural state, these findings offer a unique perspective compared to traditional tests in animals or post-mortem tissue and can help resolve ongoing discussions about direction-dependent differences in the mechanical properties of the brain.

This thesis demonstrates that iMRE, particularly in combination with poro-viscoelastic models, provides a powerful approach for studying the mechanical behavior of the living human brain. By measuring the brain in its natural, untouched state, this approach yields insights that were previously out of reach and reveals the intrinsic mechanical properties of the brain. The work provides new evidence that the brain is fluid-like and ultra-soft at low frequencies and small deformations, and also suggests the presence of mechanical anisotropy. These findings improve our understanding of brain mechanics and open new possibilities for investigating how interactions between tissue and fluid, blood flow, and microstructure work in healthy brains and what role they play in the development of diseases.