Monte Carlo simulation of diabetic macular edema changes on optical coherence tomography

Abstract

Purpose: To identify the origin, at the cellular level, of changes seen in optical coherence tomography (OCT) scans of diabetic macular edema (DME) patients.

Methods: High-definition OCT (Cirrus, Carl Zeiss Meditec, Dublin, CA, USA) scans of eyes diagnosed with DME (12 eyes of 11 patients) and eyes of healthy controls (12 eyes of 8 controls) were collected from our institutional database to be processed and analyzed. Furthermore, the DME group was divided into those without significant visible changes and those with a noticeable increase in ONL volume. The outer nuclear layer (ONL) was manually segmented by a human grader (AC). From this retinal layer, an average A-scan profile was computed from both the DME and healthy control eyes. These average A-scan profiles were then normalized by the intensity at the retinal pigment epithelium, thus being independent of media opacities. Significant differences in the backscattering properties within the ONL were found between A-scan profiles from one group to the other. Aiming to determine the relative influence of the changes in the extra-cellular and nuclear space, a simulation of light interaction with the ONL structures was conducted. To this end, variations in the extra-cellular space’s refractive index and the nucleus’s refractive properties, size and density within the ONL were considered and tested to justify the differences to the healthy status. To perform the simulation, we resorted to the Monte Carlo method and Mie theory, based on the rationale that the ONL may be regarded as a homogeneous space filled with spherical nuclei.

Results: With the simulation process, it was possible to mimic both the DME and healthy cases. Moreover, the parameters for the simulations were found to match those of a healthy retina. Interestingly, the volume change matching the increase in ONL thickness is solely responsible for the OCT signal change (backscattering difference) for the DME cases where the volume was noticeably altered. Additionally, for the DME cases where the increase in retinal thickness is less apparent, only by increasing the nucleus's size is it possible to match the simulation with the scanned data.

Conclusions: In this work, we were able to identify cellular changes that can reproduce the differences found in OCT scans of DME eyes. Moreover, two distinct forms of DME seem to exist with fundamental differences at the cellular level.

Keywords: 688 retina • 473 computational modeling • 505 edema