Accuracy Of A Portable Motion Capture System For Kinematic Analysis Of Movement Skills In Children

Abstract

PURPOSE: With recent advances in motion capture technology, there remains to be an accessible field-based measurement system for kinematic analysis in physical education. The purpose of this research is to assess the accuracy of a portable, cost-effective motion capture system to assess movement skills in children. METHODS: Eighteen healthy children ages 5-17 years performed the vertical jump and land (JL) and leap while being recorded by the Microsoft Kinect V2 with AssessLink software and Vicon MX13 with Nexus 2.0. Data was analyzed using Pearson correlation coefficient and a Bland-Altman plot to determine accuracy of the Kinect V2 to collect kinematic data for nine body segments during five phases of the JL and eight body segments during three phases of a leap. RESULTS: Overall, the correlation and mean bias between the two devices varied by body segment, movement type, and movement phase. For JL, the Pearson r for knees and shoulder ranged from -.239 (p = .109) to .521 (p < .001) and -.743 (p < .001) to .410 (p < .05), respectively. For JL, bias and level of agreement showed better agreement in the spine, hips, and knees. Sagittal assessment of leap often produced results for the appendicular skeleton that did not appear valid. Pearson r for axial segments including the sagittal spine ranged from .345 (p = .032) to .553 (p < .001) across the three movement phases. CONCLUSIONS: With limited measurements that are considered acceptable, our results indicate the Kinect V2 with AssessLink should not be used as a replacement for laboratory-based kinematic assessment. If used, the intrinsic limitations of the Kinect V2 promote a necessity to be selective of body segments, movement skills, and phases to be quantitatively assessed.