Test-retest reliability of gaze precision of a novel virtual reality-based medical device

Abstract

In virtual reality (VR) systems that track eye movements using infra-red cameras, the precision of gaze measurement is crucial for reliable detection of eye movement disorders. To assess gaze measurement ability and gaze precision consistency of an HMD VR-based medical device system, neosTM, under optimal conditions, we used a robotized setup that provides the advantage of mimicking human eye movements with minimal movement variability. We assessed neos™’s gaze examination test twice for thirteen simulated conditions with different noise levels, and then assessed gaze precision and gaze consistency for each by computing the intraclass correlation coefficient (ICC), standard error of measurements (SEM) and Bland-Altman analysis. We found excellent test-retest reliability (ICC > 0.99, SEM = 0.04) for neos™’s gaze precision, with good agreements between first and second gaze precision measurements observed via Bland-Altman analysis. The high ICC and low SEM of neos™ in all nine cardinal directions of gaze demonstrates its eye tracking reliability and measurement consistency. This is a crucial feature for eye-tracking applications for HMD-based VR devices when used in clinical settings. The use of a robotic eye to objectively validate a VR-based eye tracker can be applicable to other devices. Future research will investigate the longitudinal stability of the measurements in different human populations.