Test-Retest Reliability of Virtual Reality Devices in Quantifying for Relative Afferent Pupillary Defect.

Abstract

The swinging flashlight test (SFT) is one of the most prominent clinical tests for detecting the relative afferent pupillary defect (RAPD). A positive RAPD localizes the lesion to the affected afferent pupil pathway and is a critical part of any ophthalmic exam. Testing for an RAPD, however, can be challenging (especially when small), and there is significant intrarater and interrater variability. Prior studies have shown that the pupillometer can improve the detection and measurement of RAPD. In our previous research, we have demonstrated an automatic SFT by utilizing virtual reality (VR), named VR-SFT. We applied our methods to two different brands of VR headsets and achieved comparable results by using a metric, called RAPD score, for differentiating between patients with and without (control) RAPD. We also performed a second VR-SFT on 27 control participants to compare their scores with their first assessments and measure test-retest reliability of VR-SFT. Even in the absence of any RAPD positive data, the intraclass correlation coefficient produces results between 0.44 and 0.83 that are considered of good to moderate reliability. The same results are echoed by the Bland-Altman plots, indicating low bias and high accuracy. The mean of the differences of measurements from test-retest ranges from 0.02 to 0.07 for different protocols and different devices. As variability among various VR devices is an important factor that clinicians should consider, we discuss the test-retest reliability of VR-SFT and the variability among various assessments and between two devices. Our study demonstrates the critical necessity of establishing test-retest reliability measures when bridging virtual reality technology into the clinical setting for relevant afferent pupillary defect.