An automated framework for 3D serous pigment epithelium detachment segmentation in SD-OCT images

Zhuli Sun, Haoyu Chen, Fei Shi, Lirong Wang, Weifang Zhu, Dehui Xiang, Chenglin Yan, Liang Li, Xinjian Chen
2016 Scientific Reports  
Pigment epithelium detachment (PED) is an important clinical manifestation of multiple chorioretinal diseases, which can cause loss of central vision. In this paper, an automated framework is proposed to segment serous PED in SD-OCT images. The proposed framework consists of four main steps: first, a multi-scale graph search method is applied to segment abnormal retinal layers; second, an effective AdaBoost method is applied to refine the initial segmented regions based on 62 extracted
more » ... third, a shape-constrained graph cut method is applied to segment serous PED, in which the foreground and background seeds are obtained automatically; finally, an adaptive structure elements based morphology method is applied to remove false positive segmented regions. The proposed framework was tested on 25 SD-OCT volumes from 25 patients diagnosed with serous PED. The average true positive volume fraction (TPVF), false positive volume fraction (FPVF), dice similarity coefficient (DSC) and positive predictive value (PPV) are 90.08%, 0.22%, 91.20% and 92.62%, respectively. The proposed framework can provide clinicians with accurate quantitative information, including shape, size and position of the PED region, which can assist clinical diagnosis and treatment. Optical coherence tomography (OCT) was first introduced in 1991 by Huang et al. 1 . Recently, spectral domain (SD) OCT has been used in diagnosis of many ocular diseases, including age-related macular degeneration (AMD), glaucoma and diabetic macular edema 2 . SD-OCT has many advantages comparing to traditional OCT, such as high resolution, real 3D volumetric image of retina and manifestation of more comprehensive anatomical structures. PED is an important feature of several chorioretinal diseases, such as, AMD, central serous chorioretinopathy and polypoidal choroidal vasculopathy 3,4 . PED can cause damage to central vision finally 5,6 . Generally, PED can be classified into three types: serous, drusenoid and fibrovascular. Though the three types of PED share several basic similarities, there are many distinct differences in clinical and prognostic aspects. For example, the serous PED tends to be a smooth, arch-like shape region with retinal pigment epithelium (RPE) deformation 3 . As shown in Fig. 1 , the serous PED region is located between RPE floor and Bruch's membrane (BM). Quantitative information of serous PED, including accurate boundary, size, position and total number, is important for diagnosis and treatment of the relevant retinal diseases. Therefore, automatic segmentation for serous PED objects in SD-OCT is of great clinical significance. However, automatic segmentation for abnormal retinal structures still remains a challenging task. There are two critical problems for this task. First, the retinal morphology and intensity may have changed severely resulting from the abnormal structures. Therefore, the prior knowledge about morphological and optical features used for normal retinal image segmentation may not be valid. Second, the segmentation performance is affected by the blurred boundary, various shapes and random position of the abnormal structures. For retinal images analysis, an important pre-processing step for region segmentation is retinal layers segmentation. The retinal layers segmentation result can serve as constraints for automatic detection or segmentation of the abnormalities.
doi:10.1038/srep21739 pmid:26899236 pmcid:PMC4761989 fatcat:mmy76ghj5jelrh6lxrxt5jq2l4