Comparison Of Marker-based And Stereo Radiography Knee Kinematics In Assessment Of Activities Of Daily Living

Abstract

Marker-based motion capture is the established standard for the quantification of human movement. Yet, small relative joint rotations and translations are difficult to measure with these systems. Measurement using high speed stereo radiography (HSSR) allows for direct analysis of joint level motion with a high level of accuracy. PURPOSE: Compare knee kinematics measured with marker-based motion capture to kinematics acquired using HSSR and assess the ability of marker-based motion capture to accurately represent true knee kinematics. METHODS: Two healthy subjects (1M/1F, 171.4±3.6cm,78.7±1.0kg) were recruited to perform seated knee extension, step down, and 90° pivot. Stereo radiography and motion capture data were collected simultaneously for each subject. Bone tracking was performed in Autoscoper (Brown University, RI) to obtain the pose of each bone throughout each task. Tibiofemoral rotations and translations were calculated from the pose. For marker-based tracking, commercially available software was used to create a lower-extremity model based on rigid body segment tracking with a 6 degree of freedom (DOF) knee joint. Knee kinematics were compared by mean error reported with standard deviation and total excursion in each DOF. RESULTS: Flexion showed a mean error of 4.75°±3.31° and -8.10°±4.21° for the flexion task over a range of 125.0° and 124.1° for subjects 1 and 2 respectively. Internal rotation showed a mean error of 1.81°±3.42° and 1.79°±1.95° over a range of 21.2° and 18.1° for subjects 1 and 2 for the pivot task. Anterior-posterior showed a mean error of -9.54mm±10.61mm and -5.41mm±4.83mm over a range of 7.0mm and 10.6mm for subjects 1 and 2 for the flexion task. CONCLUSIONS: Few studies have compared knee kinematics recorded using both traditional motion capture and stereo radiography. Results show lowest errors in the rotational axes taking into consideration total excursion during the task for each DOF. Errors in translation often exceed the excursion in that DOF suggesting the inability of motion capture to accurately assess translational DOFs at the knee using rigid body tracking algorithms. Future work will consider 8 healthy subjects and help to further increase the impact of the results.