Identification of choriocapillaris in choroideremia using optical coherence tomography angiography

Abstract

Purpose : To develop a method of identifying choriocapillaris from choroidal vessels and projection artifacts in optical coherence tomography (OCT) angiography.

Methods : Macular scans of participants with a clinical diagnosis of choroideremia were taken using a spectral OCT system (RTVue-XR). The split-spectrum amplitude-decorrelation angiography (SSADA) algorithm was used to detect flow. Semi-automated segmentation was used to identify inner limiting membrane (ILM), outer boundary of outer plexiform layer (OPL), and Bruch’s membrane (BM). Maximum flow projection between ILM and OPL, 10 µm below BM to 20 µm below, and 10 µm below BM to the bottom of the image produced en face angiograms of the inner retinal circulation, choriocapillaris, and choroid, respectively (Fig. 1A-C). We sought to identify choriocapillaris by assessing the local standard deviation (SD) of the flow signal (decorrelation) in a 9×9 pixel matrix on the en face choroidal angiogram (Fig.1D). We hypothesize that the local SD of choriocapillaris would be low due to its near-confluent appearance, in contrast to larger choroidal vessels and projected retinal vessels.

Results : Using the SSADA derived flow signal and local SD image, a rough binary mask of intact choriocapillaris was generated. Morphological operations of erosion, removal of small elements, and dilation were used to refine the mask (Fig. 1E). Overlaying the outline of the mask onto the choroidal image showed good agreement whereby choriocapillaris was included while choroidal vessels and projected vessels were excluded (Fig. 1F). Comparison of the intact choriocapillaris to en face structural images of the ellipsoid zone revealed choriocapillaris beneath preserved photoreceptors but not outer retinal tubulations.

Conclusions : We were able to distinguish choriocapillaris from choroidal vessels and projection artifacts using the flow signal, local SD of the flow signal, and morphological operations. This approach may be useful in the assessment of diseases such as choroideremia and age-related macular degeneration.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.

 

Figure 1. En face OCT angiograms of the (A) retinal circulation, (B) choriocapillaris, and (C) choroid in a participant with choroideremia. (D) Local standard deviation image of (C). (E) Mask of intact choriocapillaris derived from (C), (D), and morphological operations. (F) Outline of (E) on a grayscale version of (C).