基于LDU的视网膜OCT图像分层分割研究  

Layer segmentation of retinal OCT images based on layer dense block U-Net (LDU)

作　　者：谭泰铭 陈林江[2] 蓝公仆 许景江 安林 黄燕平 TAN Tai-ming;CHEN Lin-jiang;LAN Gong-pu;XU Jing-jiang;AN Lin;HUANG Yan-ping(School of Mechatronic Engineering and Automation,Foshan University,Foshan 528000,Guangdong Province,China;Department of Ophthalmology,Nanfang Hospital,Southern Medical University,Guangzhou 510515,China;School of Physics and Optoelectronic Engineering,Foshan University,Foshan 528000,Guangdong Province,China;Guangdong-Hong Kong-Macao Joint Laboratory for Intelligent Micro-Nano Optoelectronic Technology,Foshan University,Foshan 528000,Guangdong Province,China;Innovation and Entrepreneurship Teams Project of Guangdong Pearl River Talents Program,Guangdong Weiren Meditech Co.,Ltd.,Foshan 528000,Guangdong Province,China)

机构地区：[1]佛山科学技术学院机电工程与自动化学院,广东佛山528000 [2]南方医科大学南方医院眼科,广州510515 [3]佛山科学技术学院物理与光电工程学院,广东佛山528000 [4]佛山科学技术学院粤港澳智能微纳光电技术联合实验室,广东佛山528000 [5]广东省“珠江人才计划”引进创新创业团队,广东唯仁医疗科技有限公司,广东佛山528000

出　　处：《信息技术》2022年第10期31-40,共10页Information Technology

基　　金：国家自然科学基金(61871130);粤港澳智能微纳光电技术联合实验室(2020B1212030010);广东省“珠江人才计划”引进创新创业团队(2019ZT08Y105);佛山科学技术学院高层次人才科研启动项目(Gg07089)。

摘　　要：U-Net是近期流行的视网膜OCT图像分层分割方法,但是在具有复杂纹理信息的图像区域上会出现较大分割误差。提出一种带有层密集块的U型神经网络(LDU)。通过引入图像特征信息重用的层密集块,提取更有效的图像多尺度特征,从而提高视网膜OCT图像的分层分割精度。利用杜克大学公开的两层视网膜OCT数据库对LDU与传统U-Net方法进行对比分析。测试验证结果表明该方法可提升人眼特别是AMD病眼视网膜OCT图像分层分割的性能,提高视网膜各层定量化分析的准确性。U-Net has been a popular method nowadays in the field of layer segmentation of retinal OCT images, but large segmentation errors still exist in image regions with complex texture information. In this paper, we propose a layer dense block-based U-type neural network(LDU) for this purpose. By introducing layer dense blocks for image feature information reuse, more effective multi-scale features of images are extracted, thus improving the accuracy of layered segmentation. The LDU is compared to the conventional U-Net method using the publicly available two-layer retinal OCT database from the Duke University. The test results show that this method could improve the performance of layered segmentation of retinal OCT images in human eyes, especially those with AMD, which consequently improved the accuracy of quantitative analysis for each retinal layer.

关 键 词：视网膜光学相干层析图像 分层分割 层密集块 卷积神经网络 最短路径搜索 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]