Abstract
Purpose :
Measuring visual acuity (VA) is highly challenging in non-verbal adults and young children. We developed an automated pipeline to objectively determine VA by optokinetic nystagmus (VA-OKN). The purpose of this study was to compare our VA-OKN to VA-ETDRS measured using the Electronic Visual Acuity system (EVA, Emmes, US) in adults with and without refractive error.
Methods :
VA-OKN and VA-ETDRS were measured monocularly without optical correction for each eye in healthy participants (n=23, age 30±12) with or without refractive error at 3m. VA-ETDRS classified each eye into reduced VA (n=22,≥0.2 logMAR) or no VA deficit (n=24,<0.2 logMAR) groups.
The OKN system presented an array of vanishing disks (drift speed 5 deg/sec) on a consumer grade monitor (Dell S2417DG, US) to induce OKN. The stimulus was swept 10 times with the spatial scale of the disks stepped between 1.0 and 0.0 logMAR in ascending/descending order (0.1/2 logMAR/sec). Eye tracking data (Pupil Invisible, Pupil Labs, DE) was processed online to find the times when OKN activity ceased (descending sweep) or initiated (ascending sweep). These times were converted to a per sweep VA and were averaged per eye. Sweep traces were randomized and assessed by a reviewer blinded to VA-ETDRS to give a final per sweep VA and final VA-OKN.
Results :
A single randomly selected eye was used for analysis. VA deficit group: final VA-OKN and VA-ETDRS were 0.54±0.36 and 0.54±0.45 logMAR (mean±2SD). There was no significant difference between results (p>0.05,paired t-test). The r2 statistic between final VA-OKN and E-ETDRS was 0.84. The 95% limits of agreement (LoA) were 0.19 logMAR. No VA deficit group: final VA-OKN and VA-ETDRS were 0.05±0.1 and -0.19±0.13 logMAR. There was a significant difference between results (p<0.05, paired t-test). The bias was 0.24 logMAR and no correlation was found (r2 = 0.06). VA-ETDRS was outside the range of the OKN device. However, the sensitivity/specificity for detecting a deficit of logMAR 0.2 in an eye was 1.0. The expert review contributed a minor improvement to the overall results.
Conclusions :
We have developed a pipeline for assessing VA by OKN. A robust correlation and positive agreement was found between VA-ETDRS and VA-OKN in adults with VA deficit. The system had high sensitivity and specificity for detecting a VA deficit. The results suggest that fully objective estimation of VA by OKN is highly feasible.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.