ABSTRACT
PURPOSE: Studies have shown that marker-less motion detection systems, such as the first generation Kinect (Kinect 1), have good reliability and potential for clinical application. Studies of the second generation Kinect (Kinect 2) have shown a large range of accuracy relative to balance and joint localization; however, few studies have investigated the validity and reliability of the Kinect 2 for upper extremity motion. This investigation compared reliability and validity among the Kinect 1, Kinect 2 and a video motion capture (VMC) system for upper extremity movements. DESIGN: One healthy, adult male performed six upper extremity movements during two separate sessions. All movements were recorded on the Kinect 1, Kinect 2 and VMC simultaneously. Data were analyzed using MATLAB (Natick, MA), Microsoft Excel (Redmond, WA), and SPSS (Armonk, NY). RESULTS: Results indicated good reliability for both Kinects within a day; results between days were inconclusive for both devices due to the inability to exactly repeat the desired movements. Range of motion (ROM) magnitudes for both Kinects were different from the VMC, yet patterns of motion were very highly correlated for both devices. CONCLUSION: Simple transformations of Kinect data could bring magnitudes in line with those of the VMC, allowing the Kinects to be used in a clinical setting. Implications for Rehabilitation The clinical implications of the investigation support the notion that the Kinects could be used in the clinical setting if an understanding of their limitations exists. Using the Kinects to make assessments with a given data collection session is acceptable. Using the Kinects to make comparisons across different days such as before or after an intervention should be approached with caution. The Kinect 2 provides a more cost effective option compared to the VMC. Additionally, the Kinect is more portable, requires less time to set-up, and takes up less space, thus increasing its overall usability compared to the VMC.
