On the prediction of optical aberrations by personalized eye models

The purpose of this study is to develop and analyze a method to obtain optical schematic models of individual eyes. Each model should be able to reproduce the measured monochromatic wave aberration with high fidelity.

First, we choose a generic eye model as the input guess and then apply a two-stage customization procedure. Stage 1 consists of replacing, in the initial generic model, those anatomic and optical parameters with experimental data measured on the eye under analysis. The set of experimental data was that provided by a standard clinical preoperative examination, namely lens topography, ultrasound biometry, and total wave aberration. Then, the second stage is to find the unknown lens structure that would reproduce the measured wave aberration through optical optimization. Two totally different initial eye models have been compared; one considers a simpler constant refractive index for the lens, whereas the second model has a gradient-index (GRIN) lens.

This automatic customization method has been applied to 19 eyes with different degrees of spherical ametropia (from +0.4 D to -8 D). Two models have been obtained for each eye (constant and gradient index lens). The results were highly satisfactory, with 100% convergence, and with average RMS prediction errors approximately lambda/100. This is one order of magnitude lower than typical measurement errors. The models with a constant refractive index lens tended to overestimate surface curvatures, whereas for the GRIN model, lens surfaces were too flat.

The proposed method is highly efficient and robust giving a high-fidelity reproduction of the wavefront in all cases attempted so far. Limitations found in reproducing the geometry of the lens seem to be associated with the use of inaccurate models of its refractive index.