《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (1): 265-272.DOI: 10.11772/j.issn.1001-9081.2021111882

所属专题: 多媒体计算与计算机仿真

• 多媒体计算与计算机仿真 • 上一篇    下一篇

融合滤波增强和反转注意力网络用于息肉分割

林荐壮1, 杨文忠1, 谭思翔1, 周乐鑫2, 陈丹妮1   

  1. 1.新疆大学 信息科学与工程学院,乌鲁木齐 830046
    2.新疆大学 软件学院,乌鲁木齐 830091
  • 收稿日期:2021-11-09 修回日期:2022-05-08 发布日期:2023-01-12
  • 作者简介:林荐壮(1995—),男,福建莆田人,硕士研究生,主要研究方向:计算机视觉、深度学习、医学图像分割;杨文忠(1971—),男,河南南阳人,教授,博士,CCF会员,主要研究方向:网络空间安全、算法设计与分析、人工智能 email:ywz_xy@163.com;谭思翔(1996—),男,湖南卲阳人,硕士研究生,主要研究方向:深度学习、图像语义分割;周乐鑫(1996—),男,重庆人,硕士研究生,主要研究方向:自然语言处理、文本方面级情感分析;陈丹妮(1995—),女,重庆人,硕士研究生,主要研究方向:计算机视觉、深度学习、医学图像分割;
  • 基金资助:
    国家自然科学基金资助项目(U1603115)。

Fusing filter enhancement and reverse attention network for polyp segmentation

LIN Jianzhuang1, YANG Wenzhong1, TAN Sixiang1, ZHOU Lexin2, CHEN Danni1   

  1. 1.School of Information Science and Engineering, Xinjiang University, Urumqi Xinjiang 830046, China
    2.School of Software, Xinjiang University, Urumqi Xinjiang 830091, China
  • Received:2021-11-09 Revised:2022-05-08 Online:2023-01-12
  • Contact: YANG Wenzhong, born in 1971, Ph. D., professor. His research interests include cyberspace security, algorithm design and analysis, artificial intelligence.
  • About author:LIN Jianzhuang, born in 1995, M. S.candidate. His research interests include computer vision, deep learning, medical image segmentation;ZHOU Lexin, born in 1996, M. S. candidate. His research interests include natural language processing, text aspect-level sentiment analysis;CHEN Danni, born in 1995, M. S. candidate. Her research interests include computer vision, deep learning, medical image segmentation;
  • Supported by:
    This work is partially supported by National Natural Science Foundation of China (U1603115).

摘要: 准确分割结肠镜获取图像中的息肉区域,可辅助医生诊断肠道疾病,但下采样过程中息肉区域结构信息缺失,现有方法存在过度分割和欠分割的问题。为解决以上问题,提出融合滤波增强和反转注意力分割网络(FFRNet)。首先,在跳级连接中加入了滤波增强模块(FEM),以增强下采样特征中局部病灶区域的结构信息;其次,通过聚合浅层特征来获取全局特征;最后,在上采样过程中采用多尺度反转注意力融合机制(MAFM),结合全局特征和上采样特征生成反转注意力权重,逐层挖掘特征中的息肉区域信息,引导网络建立目标区域与边界之间的关系,以提高模型对息肉区域分割的完整性。在Kvasir和CVC-ClinicDB数据集上,与不确定性增强上下文注意力网络(UACANet)相比,FFRNet的Dice相似系数(DSC)分别提升了0.22%和0.54%。实验结果表明,FFRNet能够有效提高息肉图像分割精度,同时具有较好的泛化能力。

关键词: 医学图像分割, U-Net, 息肉分割, 注意力, 卷积神经网络

Abstract: Accurate segmentation of the polyp region in the colonoscopic images can assist doctors in diagnosing intestinal diseases. However, the structure information of polyp region is missing in the down sampling process, and the existing methods have the problems of over segmentation and under segmentation.Aiming at the problems above, a Fusing Filter enhancement and Reverse attention segmentation Network (FFRNet) was proposed. Firstly, Filter Enhancement Module (FEM) was added to the skip-connection to enhance the structure information of local lesion region in the down-sampling features. Secondly, the global features were obtained by aggregating the shallow features. Finally, Multiscale reverse Attention Fusion Mechanism (MAFM) was adopted in the up-sampling process, by combining the global features and up-sampling features to generate the reverse attention weight, the polyp region information was mined in the features layer by layer, and the relationship between the target region and the boundary was established by the guidance network to improve the integrity of the model on polyp region segmentation. On Kvasir and CVC-ClinicDB datasets, compared with Uncertainty Augmented Context Attention Network (UACANet), FFRNet has Dice Similarity Coefficient (DSC) increased by 0.22% and 0.54% respectively. Experimental results show that FFRNet can effectively improve the accuracy of polyp image segmentation and has good generalization ability.

Key words: medical image segmentation, U-Net, polyp segmentation, attention, Convolutional Neural Network (CNN)

中图分类号: