Abstract
Our study evaluated the diagnostic capability of flow density (FD) in OCT angiography (OCTA) for diabetic retinopathy (DR) detection in diabetic patients. We studied 93 eyes of 68 diabetic patients who underwent OCTA (36 and 57 eyes without and with DR, respectively). Retinal capillary FD of a 2.6 × 2.6 mm(2) area and four divided areas at the superficial (SCP) and deep capillary plexus (DCP) were measured. Predictions were evaluated using the area under the receiver operating characteristic curve (AUC). The diagnostic capabilities of the FDs in discriminating between eyes without DR and eyes with total or early DR were compared. Furthermore, predictions with foveal avascular zone (FAZ) area, hemoglobin A1c (HbA1c), and DM duration were also compared with FD. Prediction using FD AUC in the temporal side in the DCP (0.83) was the highest and significantly better than all other AUCs examined (P < 0.05), including discriminating between eyes without DR and with early DR (P < 0.01). Prediction using this particular AUC was also significantly better than that by FAZ area and HbA1c (P < 0.001 and <0.001, respectively). Area-divided FD in OCTA may be valuable for diagnosing retinopathy in diabetic patients.