Automated segmentation of intraretinal cystoid macular edema for retinal 3D OCT images with macular hole

An automated method is proposed to segment and quantify the volume of cystoid macular edema (CME) for the abnormal retina with macular hole (MH) in 3D OCT images. The proposed framework consists of three parts: (1) preprocessing, which includes denoising, intraretinal layers segmentation and flattening, MH and vessel silhouettes exclusion; (2) coarse segmentation, in which an AdaBoost classifier is used to get the seeds and constrained regions for Graph Cut; (3) fine segmentation, in which a graph cut algorithm is used to get the refine segmentation result. The proposed method was evaluated in 3D OCT images from 18 typical patients with CMEs and MH. The true positive volume fraction (TPVF), false positive volume fraction (FPVF) and accuracy rate (ACC) for CME volume segmentation are 84.6%, 1.7% and 99.7%, respectively.