Purchase this article with an account.
R.A. Applegate, L.N. Thibos, D.R. Williams; Converting Wavefront Aberration to Metrics Predictive of Visual Performance . Invest. Ophthalmol. Vis. Sci. 2003;44(13):2124.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Background: Wavefront error fundamentally defines the optical quality of the retinal image. Numerous metrics of optical quality can be calculated from wavefront error. Purpose: To determine which of 24 different metrics of optical quality best predicts visual performance. Methods: Metrics of optical quality exist in pupil and image domains. These can include or exclude measures of the neural transfer function. Here we report the correlation of 24 different metrics to data sets of visual performance generated in three independent laboratories. At Houston, visual performance of normal subjects viewing aberrated high contrast acuity charts was measured for different magnitudes and types of wavefront aberrations. Image quality metrics were ranked based on their ability to predict changes in visual acuity. At Bloomington, wavefront aberrations were measured and subjective refractions performed on 200 normal eyes. Image-quality metrics were ranked based on their ability to predict the outcome of subjective refractions. At Rochester, observers determined how much of each Zernike mode, generated with adaptive optics, is required to produce the same subjective blur as that from a standard wave aberration based on 18 Zernike modes. Image quality metrics were ranked according to their ability to predict the magnitudes of two dissimilar Zernike spectra that have the same effect on visual appearance. Ordinal rankings of the metrics were averaged across institutions to determine the best over-all metrics. Results: An image-based metric we call "visual Strehl ratio", defined as volume under CS-weighted OTF normalized by volume under CS-weighted OTF for diffraction limited optics, accounted best for the Houston data (r^2 = 0.75). An image-based metric defined as the area lying below the radially-averaged MTF and above the neural contrast threshold function performed best on the Rochester data (r^2 = 0.94). The wavefront metric defined by the fraction of pupil area for which the aberration function is equal to or less than ¼ wavelength accounts best for the Bloomington data (mean predicted spherical error 0.006 D). Across institutions and experiments, the image-based metric called "neural sharpness" performed best. Conclusions: Many metrics of optical quality are superior to RMS wavefront error for predicting visual performance on specific visual tasks. Robust metrics that work well on a variety of tasks may be enhanced by knowledge of the neural information processing of retinal images.
This PDF is available to Subscribers Only