Prediction of Gait Kinetics After ACL Reconstruction Using a Depth Sensor-Driven Musculoskeletal Gait Model

Gait impairment is commonly observed following ACL reconstruction. Many individuals experience persistent deficits in lower extremity neuromuscular function and alterations in lower extremity movement patterns after ACL injury. Gait analysis using a laboratory-based motion capture system provides a quantitative assessment of the primary deficiencies during gait in ACL patients; however, due to its high cost, technical difficulty, lack of portability, considerable space requirements, and the protracted setup time, it is not feasible to utilize this technology in clinical settings. According to the literature, measurement of kinetics is vital following ACL reconstruction. Therefore, the purpose of this study was to validate ground reaction forces using a full-body musculoskeletal gait model driven by an RGB-D sensor during over-ground gait and stair ascent. In addition, ground reaction forces and lower extremity joint moments between ACL patients and healthy individuals were compared during both tasks to assess the capacity of the RGB-D sensor to detect group differences. Fifteen ACL patients who had undergone BTB (Bone-Patellar Tendon-Bone Autograft) ACL reconstruction surgery at least one year prior to surgery and 15 age-matched healthy control participants were recruited for this study. Over-ground gait trials were collected along a 5-meter walkway equipped with two force platforms, and a custom-built 3-step staircase, with force platforms embedded in the first two steps, was utilized for the stair ascent test. The subjects performed three walking trials barefoot at their normal walking speeds. The RGB-D sensor data were analyzed by subtracting the background depth information and tracking the subjects’ movements using anthropometric models in order to extract 26 joint trajectories. The musculoskeletal AnyBodyTM GaitFullBody model generated the ground reaction forces and lower extremity joint moments using a musculoskeletal model attained by 25 artificial muscle-like actuators that were attached to the soles of each foot. Based on our findings, the ability of the RGB-D sensor-driven musculoskeletal model to effectively estimate peak ground reaction forces is highly dependent on the component force being evaluated. Additionally, injured limb with the ACL group showed less hip flexion, knee extension moment, and ankle plantarflexion moment during braking phase, due to their inability to generate sufficient braking impulse followed by a decline in the propulsion impulse, than those generated by the healthy control group. The ability of the model to effectively estimate gait kinetics was apparent in its ability to effectively assess gait abnormalities in patient with ACL injury. The gait-related kinetic outcomes obtained using our RGB-D sensor-driven musculoskeletal model, proves that this approach has the potential to be an effective, accurate gait analysis tool following ACL reconstruction