Evaluating Lower Limb Alignment in Knee Osteoarthritis Patients Using Markerless Motion Capture

Abstract

Knee osteoarthritis (OA) is one of the most prevalent musculoskeletal diseases in the world, where patients diagnosed with this disease report pain, limited mobility, and an overall compromised functional status. Marked abnormality in lower limb alignment is often associated with the progression of knee OA. Varus or valgus malalignment is assessed clinically from a static radiograph. Incorporating a dynamic assessment into the clinical workflow would provide clinicians an opportunity to examine the knee joint under natural loading conditions. Markerless motion capture using Theia3D is a potential technique to examine lower limb alignment that can be used to augment current healthcare practices. Contrary to other motion capture technology, markerless motion capture does not require palpating patients with infra-red markers or require a specialized laboratory for data collection. The first study in this dissertation evaluated the ability of markerless motion capture to quantify frontal plane lower limb alignment in an orthopedic population. The knee adduction angle was computed to assess both static and dynamic lower limb alignment and to determine if the measurable outcomes were correlated. Ninety-two patients completed a double limb support stand task (static) and a gait task (dynamic) during the same visit. Statistically significant differences were found between predominantly advanced medial and lateral knee OA groups from static and dynamic alignment. The static and dynamic alignment measures were also highly correlated, consistent with previous work. The second study in this dissertation investigated the level of agreement between markerless motion capture and long leg radiology to measure static frontal plane alignment. A mobile camera system was created to collect markerless data in the radiology room. Twelve patients underwent a long leg radiograph immediately followed by markerless motion capture data collection. The static hip-knee-ankle angle measurements were found to be highly correlated between systems. However, notable measurement variance existed between systems prompting additional exploration into inter-joint distances, segment lengths, influence of patient BMI, and camera configuration. Overall, markerless motion capture using Theia3D shows promise to act as a supplementary technique for healthcare practitioners to deploy in their clinical assessment of those diagnosed with musculoskeletal diseases such as knee OA.