Validity of Using Automated Two-Dimensional Video Analysis to Measure Continuous Sagittal Plane Running Kinematics

Two-dimensional video analysis is commonly used to assess kinematics when three-dimensional motion capture is unavailable. However, videos are often assessed using manual digitization, which limits the ability to extract outcomes that require continuous data. Here, we introduced a method to collect continuous kinematic data in 2D using an inexpensive camera and an open-source automated marker tracking program. We tested the validity of this method by comparing 2D video analysis to 3D motion capture for measuring sagittal-plane running kinematics. Twenty uninjured participants ran on a treadmill for 1-min while lower extremity kinematics were collected simultaneously in 3D using a motion capture system and in 2D using a single digital camera, both at 120 Hz. Knee, ankle, and foot angle at contact, peak knee flexion, knee flexion excursion, and knee-ankle flexion vector coding variability were computed using both the 3D and 2D kinematic data, and were compared using intraclass correlation coefficients and Bland-Altman plots. The agreement between collection methods was excellent for foot angle at contact and knee flexion excursion, good for ankle and knee angle at contact and knee-ankle vector coding variability, and moderate for peak knee flexion. However, Bland-Altman plots revealed significant differences between the 2D and 3D collection methods, which varied across study participants. These low-cost methods could be useful for collecting continuous sagittal plane running kinematics in non-laboratory settings.