Abstract
Purpose :
Visual image quality (VIQ) metrics (e.g. logarithm of the visual Strehl ratio; logVSX) weight wavefront aberrations with neural processing of the visual system to serve as analogues for photopic foveal vision but do not consider age-specific pupil sizes or retinal-illuminance-related neural processing. Here we model logVSX for age- and luminance-determined physiological pupil sizes, where the neural weighting function varies with retinal illuminance
Methods :
A model of neural processing (foveal nCSF) as a function of spatial frequency and retinal illuminance (trolands) was derived from literature (Rovamo et al. Vis Res 1994, Xu et al. JOSAA 2017). Wavefront errors previously collected (146 subjects; 20 to 80 years) were scaled to physiological pupil sizes (Watson & Yellott. J Vis 2012) for nine luminances (+4 to –4 log cd/m2). Simulated through-focus experiments determined best sphero-cylindrical corrections to maximize (1) logVSX (Thibos et al. J Vis 2004) and (2) logVSXtd, where the neural weighting function (nCSF) depended on retinal illuminance
Results :
Optical and neural factors govern VIQ in three luminance-dependent ways:
(1) Above 900 td – occurring at high luminances between 1.59 (youngest eyes) and 1.96 log cd/m2 (oldest eyes) – neural processing is constant and VIQ is only influenced by aberrations that decrease as physiological pupil size decreases (improving VIQ). Here logVSX and logVSXtd are similar (unitless differences <0.003).
(2) Between luminances producing 900 td and those causing maximum physiological pupil sizes (all eyes approximately –2 log cd/m2) VIQ is influenced by both optical and neural factors. Here logVSXtd showed better VIQ than logVSX (differences 0.013 to 0.034) because troland-specific nCSFs limit the effect of aberrations (that worsen with increasing pupil size) in the metric numerator and dampen the diffraction-limited denominator more than the generic photopic logVSX nCSF.
(3) Below luminances resulting in maximum physiological pupils, aberrations are constant (pupils are maximum) and VIQ is only influenced by the neural function that worsens as luminance decreases; logVSXtd indicated poorer VIQ (differences –0.053 to –0.216).
Conclusions :
While at high photopic luminances both metrics are similar, at lower luminances the modified metric logVSXtd better resembles mesopic and scotopic vision, which continue to worsen as luminance decreases after maximum physiological pupil size is reached
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.