Repeatability of retinal curvature estimation on wide field OCT systems

Abstract

Purpose : Wide field optical coherence tomography (OCT) enables imaging large regions of the retina with the potential to make accurate measurements of the retinal surface profile. However, the apparent curvature of the retina in conventional OCT images does not represent the actual anatomical shape. Here we present a technique for estimating the retinal curvature from data acquired with a Swept-Source OCT system (PLEX^® Elite 9000, ZEISS, Dublin, CA) and an extensive tolerance analysis demonstrating its capabilities.

Methods : In a B-scan image, each line orthogonal to an A-scan corresponds to a surface in the sample with constant group delay (GD). By simulating the system imaging the Arizona eye model in Zemax (Fig. 1), we calculated the A-scan paths (for different angular orientations of the scan mirrors) and the surface of constant GD for the PLEX® Elite 9000 with 80° field-of-view add-on lens. B-scan images acquired with the system were transformed by repositioning the displayed A-scans in angle and space to match the GD surface profile calculated in the optical model. After segmentation of retinal layers, a circular fit was applied to the retinal pigment epithelium to quantify the retinal curvature. To validate the performance of the system, we conducted a tolerance analysis that determined the error in retinal curvature measurements as a function of various parameters and refractive error (Fig. 2).

Results : The axial length of the eye has the biggest effect on errors in the retinal curvature estimation (Fig. 2(a)). The eye position and tilt relative to the system had a negligible effect, regardless of refractive error (Fig. 2(c)-(f)).

Conclusions : The GD technique presented can potentially estimate the curvature of the retinal surface. Our simulations demonstrate that the axial length of the eye is critical for estimating the retinal curvature and the working distance should be considered for hyperopic patients (Fig. 2(b)).

This abstract was presented at the 2019 ARVO Imaging in the Eye Conference, held in Vancouver, Canada, April 26-27, 2019.

 

Figure 1. (a) Eye model displaying difference in retinal surface and surface of constant group delay (b) OCT image of retina collected with PLEX^® Elite 9000 (c) Corrected OCT image.

 

Figure 2. (a) Percent error in estimated retinal curvature as a function of axial length and refractive error ranging from -15 to 5 Diopter. Results are from simulations conducted in Zemax. (b-f) Same as a, but for other parameters in the model.