Noninvasive quantification of mechanical stress in soft biological tissue

The physical properties of biological tissues fundamentally shape homeostasis and disease progression. Tissue mechanics are governed by stress, strain, and its intrinsic elastic properties, all of which undergo significant changes during disease and interact in complex and often nonlinear ways. For example, during liver fibrosis, extracellular matrix deposition increases stiffness which in turn impedes blood flow and leads to elevated portal pressure. Elastography enables in vivo mapping of tissue stiffness and has become a central tool for measuring the stiffness of biological materials. Yet, from a physical standpoint, it alone cannot fully characterize the mechanical state of a material: the in vivo quantification of mechanical stress and pressure, remains an open challenge. This dissertation therefore poses two central questions: Can stiffness and deformation measurements be combined to noninvasively infer mechanical stress and pressure in living tissue? And can such methods be generalized across distinct biological systems? To address these questions, this work develops a framework that integrates elastography with deformation fields derived from image registration. First, a U-Net–based pipeline for automated organ segmentation on magnetic resonance (MR) elastography images is developed. The network produced volumetric masks and mean shear wave speed estimates with accuracy comparable to expert manual segmentation. After training, the network reduced processing time from nearly one hour to seconds, and enabled standardized, high-throughput quantification of volumetric organ changes and stiffness across large datasets. Building on this development, a noninvasive imaging method for estimating portal pressure was developed by combining stiffness measurements with volumetric strain obtained from deformable registration of MR images. Controlled physiological maneuvers such as water ingestion and respiration induced changes in portal pressure, during which stiffness and volumetric strain were measured. Analysis revealed a linear stiffness–strain relationship consistent with compression stiffening. When calibrated against ex vivo data, this relationship enabled estimation of portal pressure, marking the first noninvasive measurement in healthy humans. To evaluate the generality of the approach, the framework was applied to brain tumors, in which growth generates measurable mechanical stress. By combining stiffness maps with deformation fields a heterogeneous peritumoral region characterized by elevated compressive strain and softening was characterized. Solid stress was measured, and a new biomarker termed “excess solid stress” was introduced which correlated strongly with patient survival, establishing a direct link between tumor biomechanics and clinical outcome. Together, these findings demonstrate that noninvasive mechanical stress and pressure quantification is possible through the combined analysis of stiffness and deformation. Beyond establishing proof of concept, this work offers a generalizable methodological foundation with broad potential applications across biological systems, while oncology serves as a compelling example of how biophysics can inform diagnostic and mechanobiological understanding.