Association of Subregional Quantitative Ultra-widefield Fluorescence Angiography Characteristics With the Occurrence of Diabetic Macular Edema and Proliferative Diabetic Retinopathy

Purpose: To explore the relationships of region-specific properties of ultra-widefield fluorescence angiography (UWFFA) images with two adverse outcomes, diabetic macular edema (DME) and proliferative diabetic retinopathy (PDR), and also the severity of diabetic retinopathy (DR). Methods: A cross-sectional observational study was performed to retrospectively analyze UWFFA images of patients with DR. All patients underwent UWFFA and optical coherence tomography examination. Leakage index and microaneurysm (MA) count were measured using Trainable Weka Segmentation, a machine learning algorithm, and ischemic index (ISI) was measured manually. The correlation between UWFFA parameters and severity of DR was analyzed, and receiver operating characteristic curves were used to estimate their diagnostic value for DME and PDR. Results: A total of 108 eyes from 108 patients with DR (mean age of 56.04 ± 8.85 years) were analyzed. As the severity of DR increased, the ISI and leakage index of the panretina and all subregions increased. Panretinal MA count and leakage index were significantly higher in eyes with DME than those without DME (p = 0.044 and 0.001, respectively). Leakage index and ISI were significantly higher in eyes with PDR than those without PDR in both panretinal and subregion-specific measurements (all p < 0.05). Throughout the retina and specifically in the posterior area (PoA), the leakage index had a higher diagnostic value for DME than ISI or MA count (all p < 0.05). The diagnostic value of MA count for PDR was lower than that of ISI and leakage index (all p < 0.05). Conclusion: The ISI, leakage index, and MA count in the PoA and panretina correlated with the severity of DR, especially the posterior parameter. The leakage index was more valuable than ISI and MA count in determining the occurrence of DME. ISI and leakage index were better predictors of PDR.