计算机应用 ›› 2021, Vol. 41 ›› Issue (2): 556-562.DOI: 10.11772/j.issn.1001-9081.2020060809

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

基于注意力机制的两阶段纵膈淋巴结自动分割算法

徐少伟1,2, 秦品乐1,2, 曾建朝1,2, 赵致楷3, 高媛1,2   

  1. 1. 山西省医学影像人工智能工程技术研究中心(中北大学), 太原 030051;
    2. 中北大学 大数据学院, 太原 030051;
    3. 山西医科大学第一医院, 太原 030001
  • 收稿日期:2020-06-15 修回日期:2020-08-13 出版日期:2021-02-10 发布日期:2021-02-27
  • 通讯作者: 曾建朝
  • 作者简介:徐少伟(1995-),男,山西太原人,硕士研究生,CCF会员,主要研究方向:机器学习、计算机视觉、数字图像处理;秦品乐(1978-),男,山西长治人,教授,博士,CCF会员,主要研究方向:大数据、机器视觉、三维重建;曾建朝(1963-),男,山西太原人,教授,博士,博士生导师,CCF会员,主要研究方向:复杂系统的维护决策和健康管理;赵致楷(1982-),男,山西太原人,硕士,主治医师,主要研究方向:CT及MRI诊断;高媛(1972-),女,山西太原人,副教授,硕士,主要研究方向:机器视觉、大数据处理、三维重建。
  • 基金资助:
    山西省研究生教育创新项目(2020SY381)。

Automatic segmentation algorithm of two-stage mediastinal lymph nodes using attention mechanism

XU Shaowei1,2, QIN Pinle1,2, ZENG Jianchao1,2, ZHAO Zhikai3, GAO Yuan1,2   

  1. 1. Shanxi Medical Imaging and Data Analysis Engineering Research Center(North University of China), Taiyuan Shanxi 030051, China;
    2. School of Computer Science and Technology, North University of China, Taiyuan Shanxi 030051, China;
    3. First Hospital of Shanxi Medical University, Taiyuan Shanxi 030001, China
  • Received:2020-06-15 Revised:2020-08-13 Online:2021-02-10 Published:2021-02-27
  • Supported by:
    This work is partially supported by the Innovation Project of Graduate Education of Shanxi Province (2020SY381).

摘要: 判断淋巴结分区是否存在淋巴结转移以及准确分割恶性淋巴结对于肺癌诊断以及治疗意义重大。针对纵膈淋巴结尺寸差异大、正负样本不平衡、与周边软组织和肺肿瘤特征相似等问题,提出了一个新颖的用于纵膈淋巴结分割的基于注意力机制的级联算法。首先,根据医学先验设计了两阶段分割算法剔除纵膈干扰组织后对疑似淋巴结进行分割,减少负样本的影响和训练难度,同时增强对纵膈淋巴结的分割能力;然后,引入全局聚合模块和双注意力模块以提升网络对多尺度目标和背景的分类能力。实验结果表明,提出的算法在纵膈淋巴结数据集上的准确率达到0.707 9,召回率达到0.726 9,Dice score达到 0.701 1,在准确率和Dice score上均明显优于当前其他纵膈淋巴结分割算法,能较好地解决淋巴结尺寸差异大、样本不平衡、特征易混淆等问题。

关键词: 纵膈淋巴结分割, 注意力机制, 计算机辅助诊断, 三维卷积神经网络, 三维医学影像

Abstract: Judging weather there exists mediastinal lymph node metastasis in the location of mediastinal lymph node region and correctly segmenting malignant lymph nodes have great significance to the diagnosis and treatment of lung cancer. In view of the large difference in mediastinal lymph node size, the imbalance of positive and negative samples and the feature similarity between surrounding soft tissues and lung tumors, a new cascaded two-stage mediastinal lymph node segmentation algorithm based on attention was proposed. First, a two-stage segmentation algorithm was designed based on the medical prior to remove mediastinal interference tissues and then segment the suspicious lymph nodes, so as to reduce the interference of the negative samples and the difficulty of training, while enhancing the ability to segment mediastinal lymph nodes. Second, a global aggregation module and a dual attention module were introduced to improve the network's ability to classify multi-scale targets and backgrounds. Experimental results showed that the proposed algorithm achieved an accuracy of 0.707 9, a recall of 0.726 9, and a Dice score of 0.701 1 on the mediastinal lymph node dataset. It can be seen that the proposed algorithm is significantly better than other current mediastinal lymph node segmentation algorithms in terms of accuracy and Dice score, and can solve problems such as the big difference in size, sample imbalance and easily confused features of lymph nodes.

Key words: mediastinal lymph node segmentation, attention mechanism, computer-aided diagnosis, 3D convolutional neural network, 3D medical image

中图分类号: