Concurrent validity and reliability of the Microsoft kinect™ device in cervical spine range of motion assessment

Abstract

This study assessed the cervical spine range of motion in asymptomatic individuals using the universal goniometer and the Microsoft Kinect™ device, respectively and also determined the validity and test-retest reliability of the Microsoft Kinect™ device with a view to establishing the accuracy and reproducibility of an alternative but a valid tool for the assessment of cervical spine range of motion. were evaluated using the universal goniometer and the Kinect™ device. The Microsoft Software Development Kit (SDK), version 1.8 and the Markus Bader Software solution were used to calculate the cervical motion based on Kinect skeleton tracking data. Anthropometric characteristics (age, height and weight) were recorded and body mass index was calculated. Every measurement was made twice and the average value used for statistical analysis. The intraclass correlation coefficient was used to investigate correlation and reliability between measurements obtained from the two techniques. An alpha level was set at p<0.05. The cervical spine range of motion for flexion, extension, right and left lateral rotation and right and left lateral flexion for the goniometer were 36.71 ± 6.34⁰, 43.11 ± 5.54⁰, 49.15 ± 5.88⁰, 47.69 ± 5.11⁰, 23.22 ± 3.95⁰ and 22.67 ± 3.83⁰ while for the Microsoft Kinect™ were 48.80 ± 6.76⁰, 15.34 ± 2.94⁰, 15.73 ± 3.04⁰, 15.09 ± 3.09⁰, 23.49 ± 4.59⁰ and 24.49 ± 4.72⁰ , respectively. There was no significant correlation in all of the cervical spine range of motion measurements obtained using the Microsoft Kinect (extension r = 0.09, p = 0.571; right lateral rotation r = 0.01, p = 0.614; left lateral rotation r = 0.008, p = 0.437; right lateral flexion r = 0.09, p = 0.571; left lateral flexion r = 0.01 p = 0.591) with the exception of cervical flexion (r = 0.198, p = 0.001). The intra-rater reliability of both the Microsoft Kinect™ and goniometer in the assessment of cervical spine range of motion was excellent (r > 0.75). The intra-rater reliability values for flexion, extension, right and left lateral rotation and right and left lateral flexion for the Microsoft Kinect™ were 0.98, 0.93, 0.95, 0.96, 0.97, and 0.98 while for the goniometer were 0.98, 0.98, 0.98, 0.98, 0.96, and 0.96, respectively. The Microsoft Kinect™ was found to be a reliable tool yet showing weak concurrent validity when compared with the universal goniometer in the cervical spine range of motion assessment, except for cervical flexion, among apparently healthy undergraduates.