1. INTRODUCTION Traditional motion capture uses markers to measure movement Markers are prone to soft tissue artifact and experimental error [1] and may possibly impede natural motion Optical motion capture systems cannot be used in a variety of natural settings such as a patient’s home, on the sports field, or in public. Has led to the development of markerless motion capture [2] The accuracy of marker-based systems has been analyzed extensively [3-6] whereas markerless motion capture has not System accuracy is critical to determine so the best clinical decisions on treatment can be made Marker-based and markerless motion capture have been compared [7], but this only provides a measure of relative accuracy Markerless systems have been evaluated but they used an array of cameras and were not systems such as the Microsoft Kinect [8], which relies on just two cameras: a depth map and a color camera. Goal of study: Is the Kinect as accurate as a marker-based motion capture system? Can Human Movement be Measured Accurately with a Microsoft Kinect? 1 Brian Noehren, 1 Anne Schmitz, 2 Mao Ye, 3 Robert Shapiro, 2 Ruigang Yang, 1 Division of Physical Therapy 2 Department of Computer Science 3 Department of Kinesiology and Health Promotion METHODS Jig built to simulate leg (Fig. 1) Jig placed in 6 different poses: flexed, hyperextended, abducted, adducted, internally rotated, externally rotated Accuracy of joint angles assessed and compared between systems (Fig. 2) UNIVERSITY OF KENTUCKY RESULTS REFERENCES ACKNOWLEDGEMENTS: funded by NSF IIS 1231545 1.Leardini, A., et al. Gait & Posture 21, 212, 2005. 2.Mundermann, L., et al. J Neuroengr and Rehab 3, 2006. 3.Richards, J. G. Human Move Sci 18, 589-602, 1999. 4.Miranda, D. L., et al. J Biomech 46, 567-573, 2013. 5.Holden, J. P., et al. Gait & Posture 5, 217-227, 1997. 6.Kiran, D., et al. Phys Ther Sport 11, 81-5, 2010. 7.Mündermann, L., et al. Proc of ASB, Cleveland, OH, 2005. 8.Microsoft. http://www.microsoft.com/en-us/kinectforwindows/ 9. Noehren, B., et al. Clin Biomech 27, 366-371, 2012. 10. Noehren, B., et al. Med Sci Sports Exerc in press, 2013. 11. Butler, R. J., et al. Physical Therapy 91, 1235-1243, 2011. DISCUSSION This is one of the first studyies to quantify the performance of a single markerlerss motion capture system in evaluating joint angles Is this accuracy good enough for clinical applications? Patients with patellofemoral pain show 3 deg difference in hip adduction [9] Patients with a reconstructed anterior cruciate ligament show 4 deg difference in hip flexion [10] Patients with knee osteoarthritis show 4-5 deg difference in knee adduction [11] Yes. We expect the Kinect to be able to discern clinically relevant differences between patients and healthy controls. Next step is to evaluate the accuracy of the Kinect to measure joint angles in healthy subjects Figure 2: Method used to compare a marker-based motion capture system with a Kinect KinectFusion Alignment Articulated Iterative Closest Point Algorithm for Alignment Surface model with marker positions Surface model Kinect global marker positions for pose of interest Inverse Kinematics using Visual3D Global marker positions from MotionAnalysis cameras Kinect Joint Angles Marker-based Joint Angles Kinect image and depth map for pose of interest Kinect image and depth map for a neutral pose thigh shank Ball and socket Figure 1: Jig built to simulate a leg For flexion-extension and ab-adduction: Angles typically altered in rehabilitation programs Accuracy of the Microsoft Kinect is less than 0.5 deg different than the marker- based system (Fig. 3) For axial rotation, the systems agree within 2 deg (Fig. 3) Figure 3: Accuracy results