Repeatability and reproducibility of vessel density measurements on optical coherence tomography angiography in diabetic retinopathy.

PURPOSE: Understanding the precision of measurements on and across optical coherence tomography angiography (OCTA) devices is critical for tracking meaningful change in disease. The purpose of this study is to investigate the repeatability and reproducibility of vessel area density and vessel skeleton density measurements from various commercial OCTA devices in diabetic eyes.
METHODS: Patients were imaged three consecutive times each on three different OCTA devices. En face OCTA images of the superficial capillary plexus, deep capillary plexus, and full retinal layer were exported for analysis. Vessel area density and vessel skeleton density were calculated. The coefficient of repeatability (CoR) was calculated to assess the repeatability of these measurements, and linear mixed models were utilized to assess the reproducibility of these measurements.
RESULTS: Forty-four eyes from 27 diabetic patients were imaged. Normalized CoR values ranged between 3.44 and 6.65% when calculated for vessel area density and between 1.35 and 23.39% when calculated for vessel skeleton density. When stratified by disease severity, the swept-source OCTA device consistently produced the smallest CoR values for vessel area density in the full retinal layer. Vessel area density measurements were repeatable across the two spectral-domain devices in the full retinal layer when all severities were combined, as well as in diabetic patients without retinopathy, mild nonproliferative diabetic retinopathy (NPDR), and moderate NPDR.
CONCLUSION: Vessel area density measured in the full retinal layer may be a more precise measure than vessel skeleton density to follow diabetic retinopathy patients both on the same device and across devices.