Retinal flow density by optical coherence tomography angiography is useful for detection of nonperfused areas in diabetic retinopathy

Purpose: Fluorescein angiography (FA) has been conventionally used for detection of retinal nonperfused area (NPA) in diabetic retinopathy (DR) in spite of its qualitative evaluation. Optical coherence tomography angiography (OCTA) has been recently reported to be useful for the quantification of retinal vascular disorder in DR. In this study, we examined whether retinal flow density (FD) measurement in OCTA was useful for NPA detection in DR. Methods: The study included 41 eyes from 29 patients with DR who underwent FA and OCTA. Regions surrounded by arteries or veins were extracted in the OCTA image, and the FDs in each region were measured by Image J. Furthermore, each region was classified as NPA or perfused area (PA) in FA. The receiver operating characteristic (ROC) curve was prepared by logistic regression analysis of the FD. The AUC (area under the ROC curve) and cutoff value of FD were also calculated. Results: Two hundred fifty-two regions were analyzed and classified into 38 NPA regions and 214 PA regions using FA. FD of each capillary plexus in NPA was significantly smaller than in PA (p < 0.0001). The AUC of total capillary plexus layers (TCP), superficial capillary plexus layer (SCP), and deep capillary plexus layer (DCP) was 0.975, 0.974, and 0.971, respectively. All areas, where the FD was more than the cutoff value (0.07 in TCP), were diagnosed with PA. Three areas with intraretinal microvascular abnormalities (IRMA) were diagnosed as PA despite being below the cutoff value. Conclusions: FD measurement in OCTA is useful for NPA detection in DR.