Peripherally optimised corneal ray tracing models

Session Details

Session Title: Ocular Surface & Quality of Vision

Session Date/Time: Monday 24/09/2018 | 14:00-16:00

Paper Time: 15:06

Venue: Room A3, Podium 1

First Author: : L.van Vught THE NETHERLANDS

Co Author(s): :    Y. Cheng   G. Luyten   J. Beenakker              

Abstract Details

Purpose:

Until now, ray-tracing modelling has mainly been used for central vision. As these ray-tracing models provide the unique opportunity to visualize the intraocular path of light-rays, they have the potential to provide patient-specific insight in peripheral vision and visual complaints. To provide an accurate, patient-specific, analysis, the individual elements of the eye-model have to be personalized. In this study we evaluated various methods to model the corneal for peripheral ray-tracing analyses.

Setting:

Department of Ophthalmology, Leiden University Medical Center, Leiden, the The Netherlands.

Methods:

Ocular ray-tracing models, based on the Navarro eye-model, were built for 23 pseudophakic eyes and personalized using biometry measurements (axial length, anterior chamber depth, Lenstar LS900) and an approximated IOL geometry. The cornea was personalized based on the corneal topography (Pentacam), using different models with increasing complexity: a biconic cornea, based on central corneal radii of curvature, and 2nd, 4th and 8th order Zernike models, fitted over 8.0 mm. The accuracy of these models was compared with the full topography maps in terms of surface elevation and resulting central and peripheral refraction expressed as power vectors M and J.

Results:

The central 2.0mm of the cornea was approximated with a root-mean-square error of less than 3.0µm. For the central 10.0mm, the average error was 28µm with the biconic model and decreased with the increasing complexity of the Zernike models to below 10µm. Although the 4th order Zernike models better describe the surface elevation, a significant difference in simulated refraction was not apparent. They showed errors in M and J of 0.6 and 0.3Diopter (0° eccentricity) and 0.6 and 0.6Diopter (45° eccentricity). In the biconic models, these errors were 0.36 and 0.3Diopter and 0.5 and 0.65Diopter respectively.

Conclusions:

Overall, the 4th order Zernike models provide the best trade-off between model complexity and simulation accuracy, resulting in clinically acceptable difference to study peripheral refractive complaints. A significant advantage of the Zernike description over the original topography maps is that they allow for ray-tracing analyses at higher eccentricities. By combining this model with measured lens and retinal geometries in ray-tracing models, they can create unique opportunity to objectify peripheral vision.

Financial Disclosure:

-