Abstract
Purpose :
Routine confrontational visual perimetry testing is used to assess peripheral vision, but requires dedicated practitioner assistance and is limited by low sensitivity and negative predictive values. We propose and evaluate a method to conduct a quick assessment of peripheral vision using a VR system that uses only eye telemetry data and offers quantifiable and reliable results.
Methods :
We designed a test for assessment of peripheral vision using VS 2020 (Twenty/Twenty Therapeutics LLC), an eye gaze-based ophthalmic device with custom software that presents a series of time and response-dependent stimuli. Three stimuli are presented in each quadrant at ~22° from center at an intensity detectable by 95% of the population matched to the subject’s age. The system tracks the user’s gaze response to stimuli, discerning seen and unseen stimuli without requiring additional user input. Subjects were also examined using a Humphrey Visual Field Analyser (HFA; Carl Zeiss Meditec, Inc.) with a 24-2 central pattern (which includes all locations presented by the headset) using SITA Fast to measure total deviation values. Total deviation values in corresponding individual locations were compared between those with a positive response detected by the VS 2020 versus those without. Additionally, the average total deviation in the periphery locations and the number of positive responses detected by VS 2020 were compared for each quadrant. Comparison between groups was done by analyzing each group's total deviation mean and standard deviation and computing the F-score between groups using ANOVA.
Results :
Data collected from 35 eyes from 18 subjects (ages 27-84, 7 normal, 11 glaucoma) in routine clinical care were analyzed. Comparison of individual peripheral locations (N=396) yielded a mean (std.) total deviation of -2.5 dB (5.4 dB) for those where a positive response was recorded versus -7.1 dB (5.8 dB) for those with no response, and a F-score of 38.3 (p<0.01). Quadrants with 2-3 recorded detections had mean (std.) total deviation of -2.4 dB (4.9 dB), N=115 versus those with 0-1 detections had -8.8 dB (5.1 dB), N=17, with an F-score of 24.3 (p<0.01). Average test time for patients in the VS2020 was 25.1 s (8.4 s) per eye.
Conclusions :
We demonstrate a rapid confrontational visual perimetry test using a VR system that is an automated, quantifiable and reliable alternative to manual confrontation visual field testing.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.