School of Physics and Optoelectronic Engineering, Foshan University, Foshan 528000, China
National Natural Science Foundation of China(61605026, 61771139, 81601534, 61705036, 61805038);Natural Science Foundation of Guangdong Province (2017A030313386), High-level Construction and Scientific Research Project of Foshan University (CGG07141), and the High-level Talent Scientific Research Startup Project of Foshan University(gg040988).
In this paper, we propose an automatic retinal segmentation method to evaluate the projection area of macular edema (ME) on specific layers of the retina in optical coherence tomography (OCT) images. Ten retinal layer boundaries are segmented using an optimized Shortest-Path Faster Algorithm based on weight matrices first, which effectively reduces the algorithm's sensitivity to vascular shadows. However, the presence of ME will result in an inaccurate segmentation of the edema area. Therefore, we use the intensity threshold method to extract the edematous area in each OCT image, set the values in this area to zero, and ensure that the obtained segmented boundary can automatically cross rather than bypass the edematous area. We use the minimum projection method to calculate the projected area of ME at different layers. To test our method, we used data collected from Topcon's OCT machine. The measured macular area resolution in the axial and B-scan directions was 11.7 microns and 46.8 microns, respectively. The mean absolute and standard deviation difference values of the retinal layer boundary segmentations were 4.5±3.2 microns compared to manual segmentation. The proposed method, thus, provides an automatic, noninvasive, and quantitative tool for the evaluation of edema.
ZhangZhang, TangYanhong, ZengXinghui, XiongHonglian, ZengYaguang, HanDingan. Projection area evaluation of macular edema by optical coherence tomography images with automatic retinal segmentation[J]. Progress in Biochemistry and Biophysics,2020,47(10):1097-1108
Copy® 2024 All Rights Reserved ICP:京ICP备05023138号-1 京公网安备 11010502031771号