Ideal observer and human observer analyses of visual acuity with wavefront aberration level

Abstract

The discrepancy between subjective and objective refraction is known to be large for eyes with large optical aberrations. 22 eyes with moderate keratoconus were refracted subjectively and objectively. Objective data (KR9000PW autorefracter and wavefront sensor, Topcon, Japan) were regressed against subjective data, as subjective refractive results are considered the gold standard. The regression for equivalent sphere was y = 0.531x − 2.43, far from the expected y = x. Average difference of equivalent sphere between subjective and objective refraction was more than two diopters. This defocus value would represent significant blur for a normally sighted subject. LogMAR visual acuity of these keratoconic eyes was 0.024 ± 0.15 SD, almost equal to 20/20. We investigated the relationship between the severity of optical aberrations and the discrepancy between refractive measures. Aberration data from two keratoconic eyes, two normal eyes, and computational model eyes with simple wavefront aberrations were used to simulate retinal images of Landolt rings with added amounts of defocus ranging from −2D to +2D. Visual acuity for the simulated retinal images was evaluated with ideal observer analysis and with human psychophysical experimental analysis. For images simulated from a normal eye, visual acuity clearly peaked at low introduced defocus levels, while for images simulated from keratoconic, largely aberrated eyes, visual acuity was relatively unaffected by the amount of introduced defocus. This result was found for both ideal and human analyses, though efficiency of the ideal observer was much better than that of human observers. The data pattern may explain why patients with highly aberrated eyes are not sensitive to the defocus blur caused by the discrepancy between objective and subjective refractions.