Optical coherence tomography angiography for quantitative microvascular assessment in diabetic retinopathy: inter-device and intra-device agreement and correlation with clinical staging.

Abstract

To compare acquisitions from single-volume Solix protocol (V1), multi-volume averaged Solix protocol (V4), and AngioVue in patients with diabetic retinopathy (DR) to assess differences in quantitative parameters introduced by high-speed scanning and averaging. Thirty-eight diabetic patients were divided into 4 groups showing either no sign, mild, moderate, or severe DR at fundus examination. For optical coherence tomography angiography (OCTA) acquisitions both AngioVue, Solix V1, and V4 were used on each patient. Outcome measures were macular perfusion density (PD), vessel length density (VLD), and vessel density index (VDI) for both superficial capillary plexus (SCP) and deep capillary plexus (DCP) and flow deficits (FD) in the choriocapillaris (CC). Our study revealed a good agreement in SCP parameters measured with all 3 devices. DCP measurements with Solix V1 showed moderate agreement with V4 and poor agreement with AngioVue measurements. Inter-device agreement in CC-FD assessment was relatively poor considering all 3 devices. The averaging process led to an underestimation of DCP and SCP parameters in all stages of DR (more evident in mild DR). AngioVue measurements compared to Solix V1 led to overestimation of SCP-PD (more pronounced in severe DR patients) and of DCP parameters (more evidently in moderate DR). The regression model derived from Solix V1 parameters was the best for categorization into the different stages. Averaging and high-speed scanning introduce changes in OCTA quantitative parameters in DR. Solix V1 is the most suitable for early diagnosis of DR, while averaged protocol could be the preferred choice in advanced stages of the disease.