[1]唐宁,樊金宇,邢利娜,等.基于图论的视网膜自动分层方法[J].生物医学工程研究,2022,02:137-142.
TANG Ning,FAN Jinyu,XING Lina,et al.Automatic retinal layers segmentation based on graph theory[J].Journal of Biomedical Engineering Research,2022,02:137-142.
点击复制
《生物医学工程研究》[ISSN:1006-6977/CN:61-1281/TN]
- 期数:
-
2022年02期
- 页码:
-
137-142
- 栏目:
-
- 出版日期:
-
2022-06-25
文章信息/Info
- Title:
-
Automatic retinal layers segmentation based on graph theory
- 文章编号:
-
1672-6278 (2022)02-0137-06
- 作者:
-
唐宁; 樊金宇; 邢利娜; 王晶; 蒋天亮; 李云耀; 史国华△
-
中国科学院苏州生物医学工程技术研究所,苏州 215163
- Author(s):
-
TANG Ning; FAN Jinyu; XING Lina; WANG Jing; JIANG Tianliang; LI Yunyao; SHI Guohua
-
Suzhou Institute of Biomedical Engineering and Technology Chinese Academy of Sciences, Suzhou 215163, China
-
- 关键词:
-
图像处理; 光学相干层析成像; 视网膜图像; 自动分层; 权值分配; 端点初始化; Dijkstra算法
- Keywords:
-
Image processing; Optical coherence tomography; Retinal images; Automatic segmentation; Weight assignment; Endpoint initialization; Dijkstra algorithm
- 分类号:
-
R318;TP391.41;R774.1;R770.4
- DOI:
-
10.19529/j.cnki.1672-6278.2022.02.06
- 文献标识码:
-
A
- 摘要:
-
针对现阶段大部分眼底光学相干层析成像(optical coherence tomography,OCT)依赖于医生的主观分析,工作量大且人为分割精度差,本研究提出了一种基于图论的视网膜自动分层方法。首先对眼底OCT图像进行权值分配,然后选取分割路径的端点,最后利用Dijkstra算法在限定区域内搜索最低权值路径,实现对视网膜的自动分层。实验结果表明,本研究方法能够精确地对视网膜OCT图像进行七层自动分层,且分层偏差小于两个像素。该方法为病变视网膜的诊断提供了关键信息。
- Abstract:
-
In view of most optical coherence tomography (OCT) images rely on the subjective analysis of doctors, with heavy workload and poor artificial segmentation accuracy at present,we proposed an automatic retinal layering method based on graph theory. First, weights were assigned to fundus OCT images, then the endpoints of the segmentation paths were selected to achieve automatic layering of the retina. The experimental results showed that this method could accurately stratify retinal OCT images with seven layers automatically, and the stratification deviation was less than two pixels.It provides key information for the diagnosis of diseased retina.
备注/Memo
- 备注/Memo:
-
(收稿日期:2022-03-28)△通信作者Email:ghshi_lab@126.com
更新日期/Last Update:
2022-07-21