Agreement Between Two Markerless Motion Capture Systems in Running Biomechanics : A Comparative Study of Theia3D and Motionmetrix

Sammanfattning

Background: Markerless motion capture has emerged as a practical alternative to traditional marker-based systems for running gait analysis, offering faster setup and improved ecological validity. However, the level of agreement between these consumer-oriented systems such as Motionmetrix and Theia3D remain underexplored. Aim: This study evaluated the agreement between Motionmetrix and Theia3D across spatiotemporal, kinematic, and kinetic parameters during treadmill running. Methods: Fourteen recreational runners (five men, nine women) completed treadmill running trials at 10, 12, 14, and 16 km/h. Data were collected simultaneously using both systems, synchronized via a hardware trigger. Theia3D data were processed in Visual3D, while Motionmetrix outputs were directly exported. Comparisons across 56 trials included descriptive statistics, systematic bias (t-test), Pearson’s correlations, intraclass correlation coefficients (ICC [2,1]), and Bland-Altman analyses. Results: Spatiotemporal variables showed the strongest agreement, with cadence demonstrating perfect correlation (r = 1.00, ICC = 0.999) and vertical oscillation showing excellent reliability (r = 0.98, ICC = 0.989). Hip kinematics also demonstrated strong agreement (ICC ≥ 0.90). In contrast, knee angles at heel strike and midstance showed moderate agreement with systematic underestimation by Theia3D (bias ≈ –15°). Foot angles exhibited poor correlations (r = -0.26 to 0.33) with large biases. Joint moment outputs showed weak correlations and poor reliability. Conclusion: Level of agreement between systems was high for spatiotemporal and some proximal kinematic parameters but demonstrated reduced level of agreement for distal joint kinematics and poor agreement for joint moments. These findings underscore the significance of high-frequency motion capture for accurately capturing running kinematics, which is essential for producing subsequent joint moment estimations.