Algorithm of Foveal Avascular Zone Detection with Denoising Mechanism Learning on Optical Coherence Tomography Angiography Images

Abstract

The foveal avascular zone (FAZ) is a region on a retinal image where blood vessels distribute sparsely. It helps identify diabetic retinopathy. In this study, we investigate the optical coherence tomography angiography (OCTA) to analyze the extent of the FAZ more precisely. Moreover, a learning-based denoising architecture is applied to well distinguish the vascular vessel and the noise. With these techniques, the FAZ in an OCTA image can be well extracted and its area can be estimated accurately.