Validation of a kinect-based system to quantify proximal arm non-use after a stroke

Abstract

Proximal arm non-use is when an individual with a weaker arm can use it, however spontaneously persists in under-using shoulder and elbow joints. The score of proximal arm non-use (PANU) is computed as the difference between the spontaneous use and the maximal use of shoulder and elbow joints. We previously quantified PANU using an ultrasound and marker-based movement analysis system (Zebris, CMS20s ® ). The aim of this study is to validate a low-cost markerless system (Microsoft Kinect ® -based) against the reference system (CMS20s ® ) to determine PANU during seated reaching. In 19 hemiparetic stroke individuals the PANU score, reach length, trunk length, proximal arm use were measured during seated reaching simultaneously by the Kinect ® (v2) and CMS20s ® over two testing sessions separated by 2 h. Intraclass correlation coefficients (ICC) and linear regression analysis showed that the PANU score (ICC = 0.95, r 2 = 0.90), reach length (ICC = 0.82, r 2 = 0,70), trunk length (ICC = 0.97, r 2 = 0.94) and PAU (ICC = 0.97, r 2 = 0.93) measured using the Kinect were strongly related to those using the CMS20s. The PANU scores showed good test-retest reliability for both the Kinect (ICC = 0.76) and CMS20s (ICC = 0.72) with Bland & Altman plots showing slightly reduced PANU scores (i.e., improved performance) in the re-test session for both systems (Kinect: −4,25 ± 6.76; CMS20s: −4,71 ± 7.88), which suggests a practice effect. We conclude that the Kinect ® can offer a low-cost and widely available solution to clinically assess PANU to monitor the progress of paretic arm recovery and to better individualise rehabilitation.