Markerless motion capture for running: validity and reliability of whole-body, joint, and muscle kinematics.

Abstract

Markerless motion capture can greatly enhance the accessibility and efficiency of running-kinematics analysis. In this study we aimed to establish the validity and reliability of a markerless motion-capture system (OpenCap) for measuring kinematics of the whole body, individual joints, and lower-limb muscle-tendon units, during running at various speeds. Markerless and marker-based motion-capture results were compared for 30 runners running on a treadmill at five speeds (8-16 km/hr). OpenCap demonstrated similar validity for joint angle measurements compared to other markerless systems that have previously been compared to marker-based motion capture. However, we found systematic differences between OpenCap and marker-based motion capture throughout the stride cycle, and across running speeds and kinematic variables. Between-system disagreements were most evident around peak centre-of-mass displacements and joint angles, and during the swing phase, with smaller differences observed at touchdown, take-off, and during the stance phase. We thus recommend caution when quantitatively comparing kinematic outcomes from OpenCap to other motion-capture systems. Despite these differences, high agreement in inter-stride and inter-trial variability was observed between systems across all kinematic variables. Overall, OpenCap is a powerful tool to enable accessible, efficient, and reliable 3D measurements of whole-body, joint, and muscle-tendon unit kinematics during running.