A comparative analysis of an AI-assisted virtual reality platform with technician-based examination for visual field, visual acuity, and color vision in neuro-ophthalmology.

PURPOSE: This study aimed to compare the AI-assisted Virtual Reality Platform (VRP) with technician-assisted exams for assessing visual acuity, color vision (CV), and visual field (VF) in a neuro-ophthalmology clinic.

METHODS

A comparative, cross-sectional study including 59 adult patients was performed in the neuro-ophthalmology clinic at Bascom Palmer Eye Institute. The Snellen chart for visual acuity, Ishihara testing, and the SITA or AVA Fast strategy for VF testing were used in both the VRP and technician-based examinations. Best corrected visual acuity (BCVA), CV results, mean deviation (MD), pattern standard deviation (PSD), and reliability indices were collected.

RESULTS

The study included 118 eyes with a mean age of 47.2 ± 18.8 (71.19% female). Significant positive correlations were found between the VRP and technician-based examination for BCVA (r = .641, P<.001) and CV (r = .824, P<.001). Bland-Altman analysis showed good agreement between VRP and the traditional exams for BCVA (mean bias 0.09 ± 0.25, limits of agreement (LOA): [0.58] to [-0.40]) and CV (mean bias 0.07 ± 0.11. LOA: [0.29] to [-0.16]). The MD and PSD of the VRP were significantly correlated with the standard Humphrey Visual Field (HVF) exam (MD: r = .91, P<.001; PSD: r = .88, P<.001). Bland-Altman analysis revealed a mean bias of -0.71 ± 2.81 for MD and - 0.90 ± 2.08 for PSD. The VRP reduced test duration (124.31 ± 0.35 s) compared to the standard exam (256.8 ± 7.1 s, P<.0001).

CONCLUSIONS

The VisuALL VRP, a compact portable device, effectively provides results comparable to technician-based assessments in neuro-ophthalmological patients while reducing test duration and improving clinical workflow.