Related Experiment Videos
Individualized FAZ morphometry and standardized peri-FAZ vascular density: a pilot AngioVue OCTA study in diabetic
Erhan Özyol1, Pelin Özyol2, Nazlı Dilbaz2
1MD, Ophthalmology Department, SANKO University Faculty of Medicine, Gaziantep, Türkiye. ozyolerhan@gmail.com.
Aim:
Foveal avascular zone (FAZ) area, perimeter, acircularity, and peri-FAZ vascular density capture complementary aspects of diabetic retinopathy (DR) pathophysiology, yet their anatomical variability, geometric interdependencies, and sampling bias inherent to FD-300 measurements pose methodological challenges. This study developed AI-Plus and geometry-normalized FD-300 (nFD-300) to subsume these interdependencies and evaluated their capacity to reduce anatomical noise and improve diagnostic performance in DR staging.
Methods:
FAZ geometry derived from OCTA images was projected onto the FD-300 midline, integrating FAZ area (A-FAZ), perimeter (P-FAZ), and acircularity index (AI-FAZ) into a single morphometric parameter, AI-Plus, which normalized FD-300 into nFD-300. Noise-attenuation capacity was assessed via regression modelling and the delta-method. In this cross-sectional study, 256-eyes from 256-participants were evaluated: 168-treatment-naïve diabetic patients (51-NoDR, 51-mild NPDR, 66-moderate NPDR) and 88-controls. Diagnostic performance was evaluated by ANOVA, ROC analysis, and multivariable logistic regression.
Results:
AI-Plus demonstrated high geometric fidelity (R²=0.962-0.984) and attenuated FAZ size variability (CV reduction: 54-78%). Controls and NoDR did not differ on any parameter. nFD-300 detected the NoDR-to-mild NPDR transition (p = 0.041), whereas FD-300 did not. nFD-300 achieved the highest diagnostic performance for controls versus mild NPDR (AUC = 0.727) and controls versus moderate NPDR (AUC = 0.833); effect size was largest for nFD-300 (η²=0.267). In multivariable models, AI-Plus absorbed P-FAZ variance and nFD-300 subsumed AI-Plus contributions (NagelkerkeR²≈0.22).
Conclusions:
AI-Plus provides a geometrically stable, noise-filtered, device-independent summary of FAZ morphology that substantially reduces the inter-individual variability limiting conventional metrics. By relating peri-FAZ vascular density to individual FAZ geometry, nFD-300 demonstrates discriminative sensitivity at early DR staging transitions where established parameters fall short. Neither metric requires additional imaging or acquisition time beyond routine OCTA, supporting their feasibility for prospective multicenter evaluation.