Comparisons of Joint Moments Estimated by Markerless and Marker-based Motion Capture Systems During Treadmill Running
International Journal of Exercise Science: Conference Proceedings
BACKGROUND: Markerless (ML) motion capture may transform biomechanics. The feasibility of an ML system to analyze lower extremity kinematics has been reported. However, no literature was found that compares inverse dynamics estimates of joint moments between ML and marker-based (MB) systems. PURPOSE: To compare lower extremity joint moments estimated by ML and MB systems during treadmill running. METHODS: 16 healthy recreational active young adults were recruited. Participants ran on an instrumented treadmill at 3.58 m/s for 2 min. Kinematic data were recorded simultaneously by 8 infrared and 8 high-resolution video cameras. An instrumented treadmill recorded the force data. Sagittal plane moments at the right hip, knee, and ankle were calculated. The dependent variables were extracted from the last 10 strides from both ML and MB systems. Local peak moment values (Nm) and the relative time (% gait cycle) to peak values were identified within each stride. Peak moments of the hip were: extension in the early stance phase (HM1), flexion in the stance-swing transition phase (HM2), and extension at the end of the swing phase (HM3); for the knee, extension moment in the early stance phase (KM1), and flexion at the end of the swing phase (KM2); for the ankle, extension moment in the stance phase (AM1). Paired t-test with Bonferroni correction was used for statistical analysis. α level was set at .008 (.05/7). RESULTS: Compared to the MB system, the ML system had similar patterns but showed significantly (p<.008) greater peak joint moment magnitudes at HM2 (ML: -1.73±0.27, MB: -1.38±0.29), HM3 (ML: 2.27±0.45, MB: 1.42±0.29), KM2 (ML: -1.17±0.24, MB: -0.74±0.13), and AM1 (ML: 3.32±0.55, MB: 3.14±0.51), but less peak magnitude at KM1 (ML: 1.28±0.32, MB: 1.40±0.42). In addition, relative timing to the peak was significantly (p<.008) different between MB and ML systems. To be specific, ML took longer than MB to reach HM1 (ML: 6.74±3.40, MB: 5.16±1.27), HM2 (ML: 43.59±7.00, MB: 40.26±6.90), HM3 (ML: 92.73±3.00, MB: 90.58±3.39), KM1 (ML: 13.53±3.89, MB: 12.98±2.18), and KM2 (ML: 92.19±2.51, MB: 90.93±2.21). No difference in HM1 and time to AM1 was detected. CONCLUSIONS: ML was characterized by greater joint moment magnitudes and temporal differences than the MB system. While differences were present, the kinetic and temporal results from an ML system make it promising for biomechanical applications.
Tang, Hui, Jiahao Pan, Barry Munkasy, Li Li.
"Comparisons of Joint Moments Estimated by Markerless and Marker-based Motion Capture Systems During Treadmill Running."
International Journal of Exercise Science: Conference Proceedings, 16 (2): Western Kentucky University Press.