AI-BASED MARKERLESS SINGLE-CAMERA MOTION ANALYSIS FOR ESTIMATING KNEE AND HIP JOINT KINEMATICS DURING GAIT

Authors

Michael Schwenk, University of Konstanz, KonstanzFollow
Philipp Barzyk, University of Konstanz, KonstanzFollow
Alina Boden, University of Konstanz, KonstanzFollow
Daniel Seebacher, Subsequent GmbH, KonstanzFollow
Philip Zimmermann, Subsequent GmbH, KonstanzFollow
Manuel Stein, Subsequent GmbH, KonstanzFollow
Markus Gruber, University of Konstanz, KonstanzFollow

Category

Clinical Biomechanics

Document Type

Paper

Abstract

This study aimed to pilot-test a markerless motion analysis approach (i.e., SMARTGAIT) for estimating knee and hip joint kinematics in the sagittal and frontal plane during overground walking. One person (age 24 years) walked four times over an 8-meter distance. Joint kinematics were measured using SMARTGAIT (single RGB smartphone camera) and Vicon (reference system). Angle trajectories of sixteen gait cycles were used for the analysis. Agreement between SMARTGAIT and Vicon angle trajectories was greater in the sagittal plane (hip: Pearson’s r=0.989; knee: r=0.990; root mean square error [RMSE] ≤2.6 deg) compared to the frontal plane (hip: r=0.789; knee: r=0.793; RMSE ≤3.9 deg). These initial results show the potential of SMARTGAIT for measuring lower extremity joint kinematics. The camera perspective may influence the accuracy of SMARTGAIT.

Recommended Citation

Schwenk, Michael; Barzyk, Philipp; Boden, Alina; Seebacher, Daniel; Zimmermann, Philip; Stein, Manuel; and Gruber, Markus (2024) "AI-BASED MARKERLESS SINGLE-CAMERA MOTION ANALYSIS FOR ESTIMATING KNEE AND HIP JOINT KINEMATICS DURING GAIT," ISBS Proceedings Archive: Vol. 42: Iss. 1, Article 110.
Available at: https://commons.nmu.edu/isbs/vol42/iss1/110