Analysis of the Choroid and Its Relationship with the Outer Retina in Patients with Diabetes Mellitus Using Binarization Techniques Based on Spectral-Domain Optical Coherence Tomography

Abstract

(1) Background: We aimed to reveal the relationship between the choroid and the outer retina with optical coherence tomography (OCT) in patients with diabetes mellitus (DM) with mild or no diabetic retinopathy (DR) in order to find early biomarkers for progressing retinopathy. (2) Methods: We performed a prospective study including 61 eyes of patients with type 1 or type 2 DM and 36 eyes of healthy controls. All subjects were imaged with Spectralis OCT. The choroid was assesseed using enhanced depth imaging OCT (EDI-OCT). Binarization of subfoveal choroidal images was done with public domain software, ImageJ (version 1.53a; National Institutes of Health, Bethesda, Maryland, USA). (3) Results: Luminal area, stromal area and total choroidal area were significantly decreased in diabetic patients compared to control: 0.23 ± 0.07 vs. 0.28 ± 0.08, p = 0.012; 0.08 ± 0.03 vs. 0.10 ± 0.04, p = 0.026; 0.31 ± 0.09 vs. 0.38 ± 0.11, p = 0.008. The thickness of retinal pigment epithelium (RPE) correlated positively with the choroidal vascularity index (CVI). The correlations between outer nuclear layer (ONL), photoreceptors (PR) and foveal choroidal thickness (FChT) were moderately negative. (4) Conclusion: Thicker RPE and a thinner PR layer may be assigned the role of early biomarkers signaling the conversion time to progressing retinopathy.

Keywords: OCT; binarization; choroid parameters; choroidal thickness; choroidal vascularity index; diabetic retinopathy; outer retina.

Figure 1

Flow diagram illustrating the study selection process. n, number; DM, diabetus mellitus; DR, diabetic retinopathy; AMD, age-related macular degeneration; OCT, optical coherence tomography; DME, diabetic macular edema; ERM, epiretinal membrane.

Figure 2

Retinal layer segmentation. MZ, myoid zone; EZ, ellipsoid zone; OS, outer segment; IZ, interdigitation zone.

Figure 3

Image binarization for the choroid with normal choroidal thickness. (A) Original choroidal enhanced depth imaging (EDI) Spectral Domain-OCT (SD-OCT) 1 × 1 pixel * scan showing black area (luminal area, LA) and white area (stromal area, SA). (B) A 1 mm segment block of the subfoveal choroidal area. (C) The image was converted to 8-bit and Niblack auto local threshold tool was applied. (D) Color threshold was applied to highlight the luminal Area. (E) Composite image representing luminal area. (F) Overlay of region of interest created after image binarization was performed on the EDI-OCT image. * We used stretched 1 × 1 pixel images for illustrative purposes only, to help visualize the details of the retinal and choroidal structures.

Figure 4

Foveal choroidal thickness. OCT 1 × 1 pixel * original image was uploaded in ImageJ software and 3 measurements were performed. * We used stretched 1 × 1 pixel images for illustrative purposes only, to help visualize the details of the retinal and choroidal structures.

Figure 5

Bland–Altman plots of TCA and LA. (A) Intra-grader reliability for TCA and (B) intra-grader reliability for LA. The difference was calculated by the 1st measurement minus the 2nd measurement. Pink line represents regression line of difference between 1st and 2nd measurements.

Figure 6

Blox plots. Differences among groups regarding CVI (left) and FChT (right).

Figure 7

(A) Correlation between foveal choroidal thickness and retinal pigment epithelium thickness in NDR. (B) Correlation between choroidal vascularity index and retinal pigment epithelium thickness in NDR. (C) Correlation between foveal choroidal thickness and photoreceptors’ thickness in DR. (D) Correlation between choroidal vascularity index and retinal pigment epithelium thickness in DR.