To determine likely errors in estimating retinal shape using partial coherence interferometric instruments when no allowance is made for optical distortion.
Errors were estimated using Gullstrand no. 1 schematic eye
and variants which included a 10 diopter (D) axial myopic eye, an emmetropic eye with a gradient-index lens, and a 10.9 D accommodating eye with a gradient-index lens. Performance was simulated for two commercial instruments, the IOLMaster (Carl Zeiss Meditec) and the Lenstar LS 900 (Haag-Streit AG). The incident beam was directed toward either the center of curvature of the anterior cornea (corneal-direction method) or the center of the entrance pupil (pupil-direction method). Simple trigonometry was used with the corneal intercept and the incident beam angle to estimate retinal contour
. Conics were fitted to the estimated contours.
The pupil-direction method gave estimates of retinal contour
that were much too flat. The cornea-direction method gave similar results for IOLMaster and Lenstar approaches. The steepness of the retinal contour
was slightly overestimated, the exact effects varying with the refractive error, gradient index
, and accommodation.
These theoretical results suggest that, for field angles ≤30°, partial coherence interferometric instruments are of use in estimating retinal shape by the corneal-direction method with the assumptions of a regular retinal shape and no optical distortion. It may be possible to improve on these estimates out to larger field angles by using optical modeling to correct for distortion.