Abstract
Purpose :
To extract vascular features in the superficial (SVC) and deep (DVC) vascular complex using optical coherence tomography angiography (OCTA), and to compare them among subjects with mild to moderate diabetic retinopathy without macular edema (DR), diabetics without retinopathy (DnoR), and normal controls (NRL)
Methods :
En face OCTA images of the SVC and DVC were generated after scanning the 3 × 3 mm2 macular regions. Twenty subjects were imaged, including 5 NRL, 10 DnoR, and 5 DR, and only the right eye was included in the analysis. The quantification defines the region of interests (ROIs) as 45° equal-area sectors namely NS, SN, ST, TS, TI, IT, IN, and NI (N = nasal, S = superior, T = temporal, I = inferior), in addition to the whole circular zone centered at the macula. The microvascular morphologic features were analyzed using fractal dimension (FD) and vascular orientation pattern curve (Ma et al, Sci Rep, 2021). FD is a metric to characterize global vessel anatomical complexity. Vascular orientation pattern captures local variations in the vessel orientation ranging from 0 to 180°, and the area under the curve indicates the vessel area density (VAD). An analysis of variance (ANOVA) and post hoc Tukey’s test was performed to detect the difference in FD and VAD in each ROI among the three groups.
Results :
For the whole macular zone, significant differences in FD and VAD were observed between NRL and DR (Fig1). Different sectoral ROIs exhibited different abilities to differentiate DR from the other two groups. Sector ST showed no difference among the three groups in FD nor VAD, although n is too low to be conclusive; whereas sector IN showed significant difference not only between NRL and DR, but also between DnoR and NRL in VAD (p=0.0060 for SVC; p=0.0225 for DVC), and between DnoR and DR in FD (p=0.0033 for SVC; p=0.0401 for DVC). In addition, the vascular orientation pattern in DR shows greater difference from NRL or DnoR in sector IN than in the whole macular zone (Fig2).
Conclusions :
Retinal vascular features were extracted from OCTA images for multi-dimensional quantification using layer- and sector-based ROIs. DR-associated microvascular deterioration is not uniform in all sectors. Sector IN detected the microvascular difference between DnoR and DR, and its sectoral analysis holds promise for DR detection at the earliest stage.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.