The Development and Application of Automated Tools for Evaluating Tibiofemoral Cartilage Health in Populations at Risk of Developing Osteoarthritis

Abstract

Osteoarthritis (OA) is a debilitating musculoskeletal disease that impacts approximately 33 million people in the United States alone. OA is characterized by cartilage deterioration and is frequently associated with joint pain, stiffness, and reduced mobility. Further, joint replacement is a common long-term sequela. Despite the prevalence and impact of OA on quality of life, there are currently no disease modifying treatments. Additionally, the precise mechanisms leading to the onset and development of OA are still unknown. Given the societal burden associated with OA, it is critical to better understand the mechanisms leading to OA development in order to devise strategies to prevent or mitigate disease progression. Magnetic resonance imaging (MRI) is a medical imaging modality particularly well-suited for studying cartilage health, as it is capable of probing both cartilage composition and biomechanical function in vivo. Analysis techniques utilizing MRI scans typically involve manual delineation of regions of interest, commonly referred to as segmentation. However, manual segmentation can be time-intensive and potentially introduce both intra- and inter-segmenter variability. Thus, Aim 1 of this dissertation developed and validated a suite of machine learning tools to automate segmentation of tibiofemoral bone and cartilage from knee MRI scans. These models proved capable of measuring cartilage morphology and composition within a resolution similar to that of expertly trained manual segmenters. Further, utilization of these models significantly improves research throughput and enables larger studies powered to answer clinically relevant questions. Aim 2 of this dissertation utilized the tools developed in Aim 1 to evaluate cartilage biomechanical function in two populations particularly at risk of developing OA. Namely, obesity and injury to the anterior cruciate ligament (ACL) are both conditions that significantly increase the likelihood of developing OA compared to the general population. Thus, Aim 2 evaluated walking-induced tibiofemoral cartilage strain and recovery, first in a study of individuals who have underwent ACL-reconstruction surgery, and second in a pilot study of individuals with obesity. Additionally, cohorts of participants without injury and with a BMI in the normal range were collected for reference. In these analyses, we discovered altered loading-induced cartilage strains as well as a decreased capacity to recover from loading. Further, for the cohort with an ACL-reconstruction, these changes were most pronounced on the medial femoral condyle near the intercondylar notch, potentially indicating where cartilage degeneration may begin and stem from. These findings demonstrate clear evidence of biomechanical changes that may disrupt cartilage homeostasis, potentially contributing to OA development.