Quantitative Evaluation of the Microsoft Kinect for Use in an Upper Extremity Virtual Rehabilitation Environment

Abstract

Low cost depth sensors could potentially allow for home-based care and rehabilitation using virtual systems. Currently, no publicly available and peer-reviewed assessment has been made on the accuracy of joint position data determined by the Microsoft KinectTM for assessment of upper extremity movements. We devised and validated clinically-based angle classifications for random arm movements in 3D-space, specifically, the shoulder joint flexion/extension angle, shoulder joint abduction/adduction angle, and 3-dimensional shoulder joint angle of 19 subjects at a distance of 2.0m using an eight camera Vicon Motion Capture system. Results show an average absolute error of these angle measurements not exceeding 10.0%.