Retinal optical coherence tomography angiography imaging in population studies

Abstract Background Widespread use of retinal optical coherence tomography angiography (OCT-A) imaging in population-based studies requires methodological and analytical consensus to ensure reproducible and accurate results. Objective Development of a consensus framework for assessment of OCT-A image quality and working with OCT-A-derived variables in population studies. Methods Criteria for image quality were developed for fovea-centered, 3x3-mm OCT-A images acquired with three commercially available devices. Inter- and intragrader agreements for overall image quality, vessel density (VD) and foveal avascular zone (FAZ) area were evaluated among five graders (% agreement). Intergrader agreement was validated on 6x6-mm OCT-A images. Recommendations for grading and use of OCT-A images in epidemiological studies were developed. Results Mean intergrader agreements for overall Unusable, Usable, and Excellent 3x3-mm OCT-A image quality (52 images) were 82.3%, 70.8%, and 87.1%; for Unusable VD and FAZ area agreements were 80.0% and 81.9%, respectively. Mean intragrader agreements for overall Unusable, Usable and Excellent 3x3-mm OCT-A image quality (27 images) were 91.1%, 82.2%, and 81.9%; agreements for Unusable VD and FAZ area were 89.6%; and 92.6%, respectively. Mean intergrader agreements for assessment of overall Unusable, Usable and Excellent 6x6-mm OCT-A images (21 images) were 73.3 %, 57.1% and 83.8%, agreements for Unusable VD and FAZ area were 60.0 % and 84.8 %, respectively. Three analytic scenarios were developed to account for common types of bias related to suboptimal OCT-A image quality and ocular comorbidities. Conclusion These recommendations provide a framework for working with OCT-A imaging in population studies.