Sampling density and configuration of stimulator points in determination of astigmatism, coma and higher order aberrations of the posterior surface of the cornea using Purkinje imaging reflection technology

Abstract

Purpose : Research on improving intraocular lens outcomes is shifting focus onto accurate measurement of the posterior corneal surface (Koch DD. Ophthalmology. 2015 Jun;122(6):1070-1.). The Cassini Corneal Shape Analyzer (Cassini BV, The Hague, Netherlands) is at the frontier of anterior and posterior surface measurement technology (Ch. 5 of Refractive Lens Exchange, ed. Wang MX, Slack Inc © 2016). In this paper we investigate the sampling density required for this technique to reconstruct a Zernike model of the posterior surface of the cornea.

Methods : Source and image points are simulated for a test eye with an 8 mm radius spherical anterior surface and a 6.5 mm base radius posterior surface, with corneal aberrations described by Zernike coefficients of up to 6^th order. Gaussian noise (25 μm s.d.) is added to the image points to simulate experimental noise. Source points are arranged in 9 configurations of concentric rings (3, 4 and 5 rings containing 7, 14 or 21 points each) providing even angular coverage up to a 6 mm corneal zone diameter (CZD). A novel raytracing algorithm that considers sequential surface interactions is used to calculate a posterior surface height map. This map is used to calculate a Zernike fit of arbitrary order of the posterior surface. Fits of up to 3^rd, 4^th and 5^th order are made. The precision of the resulting shape contribution to astigmatism (2^nd and 4^th order), coma (3^rd and 5^th order), trefoil, and quadrafoil (where possible) is compared. The threshold of acceptable corneal height precision is 10 μm at 6 mm CZD, relating to around 0.25 eq. Dioptre.

Results : A minimum of 37 stimulator points covering the 6 mm CZD is found to be sufficient to achieve a precision of better than 10 μm for astigmatism, using an order 4 Zernike fit. Third order coma is found to have a precision of better than 5 μm with this configuration. Attempting a fit of up to order 5 is not sufficiently stable for clinically relevant determination of 5^th order coma or higher order aberrations with this level of experimental noise, up to and including 105 points split between five rings.

Conclusions : The results show this algorithm reconstructs up to 4^nd order astigmatism and coma to a clinically relevant precision when using more than 37 stimulator points as input.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.