PURPOSE: To determine the ability of nonperfusion, vessel density, and morphologic measurements using projection-resolved optical coherence tomography angiography to detect early retinal microvasculature impairments in diabetes mellitus. METHODS: A retrospective review was performed on Type 2 diabetes mellitus patients with no diabetic retinopathy (DR) or mild nonproliferative DR and age-matched controls imaged with optical coherence tomography angiography. Foveal avascular zone-related metrics and extrafoveal avascular area were measured in optical coherence tomography angiography images. Vessel density and fractal dimension were calculated with and without a skeletonization process. The vessel diameter index and vessel tortuosity were computed. The area under the receiver operating characteristic curve (AUC) estimated diagnostic performances. RESULTS: Dilated capillary diameter was observed in the deep capillary plexus in the diabetic groups. Vessel density and fractal dimension of skeletonized deep capillary plexus significantly and progressively decreased in the no DR and mild nonproliferative DR groups compared with controls. Superficial extrafoveal avascular area, vessel density, and fractal dimension of the skeletonized deep capillary plexus had the highest diagnostic performance to differentiate mild nonproliferative DR from control eyes, with AUCs of 0.885, 0.876, and 0.876, respectively. CONCLUSION: Vessel density and fractal dimension from the skeletonized deep capillary network may be the most sensitive for detecting early retinal capillary loss in diabetes mellitus.
PURPOSE: To determine the ability of nonperfusion, vessel density, and morphologic measurements using projection-resolved optical coherence tomography angiography to detect early retinal microvasculature impairmentsin diabetes mellitus. METHODS: A retrospective review was performed on Type 2 diabetes mellituspatients with no diabetic retinopathy (DR) or mild nonproliferative DR and age-matched controls imaged with optical coherence tomography angiography. Foveal avascular zone-related metrics and extrafoveal avascular area were measured in optical coherence tomography angiography images. Vessel density and fractal dimension were calculated with and without a skeletonization process. The vessel diameter index and vessel tortuosity were computed. The area under the receiver operating characteristic curve (AUC) estimated diagnostic performances. RESULTS: Dilated capillary diameter was observed in the deep capillary plexus in the diabetic groups. Vessel density and fractal dimension of skeletonized deep capillary plexus significantly and progressively decreased in the no DR and mild nonproliferative DR groups compared with controls. Superficial extrafoveal avascular area, vessel density, and fractal dimension of the skeletonized deep capillary plexus had the highest diagnostic performance to differentiate mild nonproliferative DR from control eyes, with AUCs of 0.885, 0.876, and 0.876, respectively. CONCLUSION: Vessel density and fractal dimension from the skeletonized deep capillary network may be the most sensitive for detecting early retinal capillary lossin diabetes mellitus.