Image-Driven Computational Modeling of Cartilage in Health and Osteoarthritis

Articular cartilage, a load-bearing tissue covering the ends of our long bones, plays an important role in human movement. Unfortunately many adults present osteoarthritis (OA), a cartilage disease resulting in severe pain and a loss of mobility. Current diagnoses and treatments are commonly limited to the late stages of OA, when patients identify significant joint pain and irreversible damage has already occurred. Research on the mechanics of cartilage often involves numerical analyses, (specifically the finite element (FE) method) which, in cooperation with experiments, aims to better understand cartilage and its mechanics during the progression of OA. In this thesis we first reviewed recent studies on the mechanics of cartilage, both computational and experimental. Importantly, patient-specific computational studies often derive from medical images. Thus, image-driven numerical analyses could improve the accuracy of current cartilage and joint models, and contribute to novel computational approaches including multiscale models. Toward these ends, we started by establishing a fitted constitutive model for high concentration agarose gel, a widely used material for mimicking the chondrocyte environment and which has similar mechanical characteristics to cartilage. Specifically we proposed and tested an efficient practice to calibrate this biphasic poro-viscoelastic material using inverse FE analyses and corresponding experimental data. Subsequently, we extended an existing cartilage model (used in our lab) to include osmotic prestress present in most experiments but is often omitted in models of cartilage. Our improved modeling framework enables a more accurate image-driven simulations. Finally, we calibrated our image-driven cartilage model using experimental data from large-strain shear tests on healthy and progressively osteoarthritic cartilage specimens. As a result we provide the evolution of the model parameters during the progression of OA. With our improved cartilage models and the corresponding evolution of parameters we facilitate imaged-driven simulations of cartilage with particular advantages in patient-specific applications. We hope that our model and its further refinements, combined with future studies, facilitate useful tools to inspire new diagnoses and treatments for cartilage in earlystage OA.