Purchase this article with an account.
Jos J Rozema, Pablo Rodriguez Perez, Rafael Navarro, Marie-Jose B R Tassignon; Higher order statistical eye model for normal eyes. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):6020.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
This work presents a stochastic model capable of generating an unlimited number of random, but realistic biometry sets, including the corneal elevation, intraocular distances and wavefronts, with the same statistical and epidemiological properties as the original data it is based on.
One cohort of 312 eyes of 312 healthy Caucasian subjects (aged 20 - 60 years) was measured with an autorefractometer, Scheimpflug imaging (Oculus Pentacam), optical biometer (Haag-Streit Lenstar) and an aberrometer (Tracey iTrace). The corneal elevation maps, represented by Zernike coefficients, were compressed using Principal Component Analysis, leaving a total of 17 parameters to describe the variability of the ocular biometry. These data were then fitted with a linear combination of three multivariate Gaussians through an Expectation Maximization algorithm, which has been shown to give a good representation of the ocular biometry in a population. Based on this fit a stochastic model was built that generates an unlimited number of random biometry sets, from which total wavefronts and other ocular parameters can be calculated. Equality between the original and the synthetic data was assessed using non-parametric "two one-sided" tests.
The wavefronts calculated using the measured biometry were significantly equal to the originally measured wavefronts (two one-sided Wilcoxon test, p < 0.05), confirming the accuracy of the ray tracing algorithm. Subsequently, the stochastic model was used to randomly generate the biometry of 1000 eyes and calculate the associated wavefronts by ray tracing. For both the biometry and the wavefront this synthetic data were significantly equal to the originally measured data (two one-sided Mann-Witney test, p < 0.05), thus making them statistically indistinguishable.
The statistical eye model is able to produce synthetic biometry data that is indistinguishable from actual biometry. As such this model may be an interesting alternative to static eye models for researchers in visual optics that do not have access to biometry data.
This PDF is available to Subscribers Only