《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (3): 825-832.DOI: 10.11772/j.issn.1001-9081.2021040856

• 2021年中国计算机学会人工智能会议(CCFAI 2021) • 上一篇    

U-Net与自适应阈值脉冲耦合神经网络相结合的眼底血管分割方法

徐光柱1,2, 林文杰1, 陈莎1, 匡婉1, 雷帮军1,2(), 周军3   

  1. 1.三峡大学 计算机与信息学院, 湖北 宜昌 443002
    2.湖北省水电工程智能视觉监测重点实验室(三峡大学), 湖北 宜昌 443002
    3.宜昌市中心人民医院 超声科, 湖北 宜昌 443003
  • 收稿日期:2021-05-25 修回日期:2021-06-29 接受日期:2021-06-30 发布日期:2021-11-09 出版日期:2022-03-10
  • 通讯作者: 雷帮军
  • 作者简介:徐光柱(1979—),男,山东单县人,教授,博士,CCF会员,主要研究方向:人工神经网络、计算机视觉
    林文杰(1996—),男,湖北宜昌人,硕士研究生,主要研究方向:深度学习、医学图像处理
    陈莎(1995—),女,湖北宜昌人,硕士,主要研究方向:深度学习、医学图像处理
    匡婉(1996—),女,湖北宜昌人,硕士研究生,主要研究方向:深度学习、计算机视觉
    周军(1973—),男,湖北宜昌人,副教授,博士,主要研究方向:超声医学影像处理、核医学影像处理。
  • 基金资助:
    国家自然科学基金资助项目(61402259);宜昌市科技局项目(A19?302?13)

Fundus vessel segmentation method based on U-Net and pulse coupled neural network with adaptive threshold

Guangzhu XU1,2, Wenjie LIN1, Sha CHEN1, Wan KUANG1, Bangjun LEI1,2(), Jun ZHOU3   

  1. 1.College of Computer and Information Technology,China Three Gorges University,Yichang Hubei 443002,China
    2.Hubei Key Laboratory of Intelligent Vision Based Monitoring for Hydroelectric Engineering (China Three Gorges University),Yichang Hubei 443002,China
    3.Ultrasound Department,Yichang Central People’s Hospital,Yichang Hubei 443003,China
  • Received:2021-05-25 Revised:2021-06-29 Accepted:2021-06-30 Online:2021-11-09 Published:2022-03-10
  • Contact: Bangjun LEI
  • About author:XU Guangzhu,born in 1979, Ph. D., professor. His research interests include artificial neural network, computer vision.
    LIN Wenjie, born in 1996, M. S. candidate. His research interests include deep learning, medical image processing.
    CHEN Sha, born in 1995, M. S. Her research interests include deep learning, medical image processing.
    KUANG Wan, born 1996, M. S. candidate. Her research interests include deep learning, computer vision.
    ZHOU Jun, born in 1973, Ph. D., associate professor. His research interests include ultrasound medical image processing, nuclear medical image processing.
  • Supported by:
    National Natural Science Foundation of China(61402259);Yichang Science and Technology Bureau(A19-302-13)

摘要:

由于眼底血管结构复杂多变,且图像中血管与背景对比度低,眼底血管分割存在巨大困难,尤其是微小型血管难以分割。基于深层全卷积神经网络的U-Net能够有效提取血管图像全局及局部信息,但由于其输出为灰度图像,并采用硬阈值实现二值化,这会导致血管区域丢失、血管过细等问题。针对这些问题,提出一种结合U-Net与脉冲耦合神经网络(PCNN)各自优势的眼底血管分割方法。首先使用迭代式U-Net模型凸显血管,即将U-Net模型初次提取的特征与原图融合的结果再次输入改进的U-Net模型进行血管增强;然后,将U-Net输出结果视为灰度图像,利用自适应阈值PCNN对其进行精准血管分割;在U-Net模型中引入Batch Normalization和Dropout,提高训练速度,有效缓解过拟合问题。实验结果表明,所提方法的AUC在DRVIE、STARE和CHASE_DB1数据集上分别为0.979 6,0.980 9和0.982 7。该方法可以提取更多的血管细节,且具有较强的泛化能力和良好的应用前景。

关键词: 全卷积神经网络, 眼底血管分割, 脉冲耦合神经网络, U-Net, 医学图像分割

Abstract:

Due to the complex and variable structure of fundus vessels, and the low contrast between the fundus vessel and the background, there are huge difficulties in segmentation of fundus vessels, especially small fundus vessels. U-Net based on deep fully convolutional neural network can effectively extract the global and local information of fundus vessel images,but its output is grayscale image binarized by a hard threshold, which will cause the loss of vessel area, too thin vessel and other problems. To solve these problems, U-Net and Pulse Coupled Neural Network (PCNN) were combined to give play to their respective advantages and design a fundus vessel segmentation method. First, the iterative U-Net model was used to highlight the vessels, the fusion results of the features extracted by the U-Net model and the original image were input again into the improved U-Net model to enhance the vessel image. Then, the U-Net output result was viewed as a gray image, and the PCNN with adaptive threshold was utilized to perform accurate vessel segmentation. The experimental results show that the AUC (Area Under the Curve) of the proposed method was 0.979 6,0.980 9 and 0.982 7 on the DRVIE, STARE and CHASE_DB1 datasets, respectively. The method can extract more vessel details, and has strong generalization ability and good application prospects.

Key words: fully convolutional neural network, fundus vessel segmentation, Pulse Coupled Neural Network (PCNN), U-Net, medical image segmentation

中图分类号: