Reliability and validity of a method to measure trunk rotation angle from images using a camera and posture mirror

Abstract

This study aimed to measure trunk rotation angle representations from images using a single camera combined with a posture mirror and to examine its reliability and validity. We applied a trunk rotation angle model using a tripod and markers simulating trunk rotation. We compared two methods of trunk rotation angle measurement: the conventional method from the superior aspect using a manual goniometer and a novel measurement method using images from a digital camera and a posture mirror. Measurement error was calculated as the average absolute error between the angle measured by the goniometer and that calculated from the camera and mirror image. The intraclass correlation coefficient (ICC 1, 1) and ICC (2, 1) were calculated as the intra-rater reliability and agreement between the measurement angles of the two methods, respectively. Systematic errors of the angles measured by the two methods were examined by a Bland‒Altman analysis. The mean (SD) of the mean absolute error was 1.17° (0.71°). ICC (1, 1) was 0.978, and ICC (2, 1) was 0.991. The Bland‒Altman analysis showed no systematic errors. The results suggest the validity and accuracy of our novel method to measure the angle of trunk rotation, which does not require high-cost equipment or a special environment.