Journal of Computer Applications ›› 2021, Vol. 41 ›› Issue (8): 2273-2287.DOI: 10.11772/j.issn.1001-9081.2020101638

Special Issue: 综述 多媒体计算与计算机仿真

• Multimedia computing and computer simulation • Previous Articles     Next Articles

Review of deep learning-based medical image segmentation

CAO Yuhong1, XU Hai2, LIU Sun'ao2, WANG Zixiao2, LI Hongliang3   

  1. 1. Chinese Institute of Electronics, Beijing 100036, China;
    2. School of Information Science and Technology, University of Science and Technology of China, Hefei Anhui 230026, China;
    3. School of Engineering Sciences, University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2020-10-21 Revised:2021-01-13 Online:2021-01-27 Published:2021-08-10
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (62022076, U19A2057, 61976008).

基于深度学习的医学影像分割研究综述

曹玉红1, 徐海2, 刘荪傲2, 王紫霄2, 李宏亮3   

  1. 1. 中国电子学会, 北京 100036;
    2. 中国科学技术大学 信息科学技术学院, 合肥 230026;
    3. 中国科学院大学 工程科学学院, 北京 100049
  • 通讯作者: 李宏亮
  • 作者简介:曹玉红(1968-),女,河北平山人,高级工程师,博士,主要研究方向:计算机视觉、人工智能;徐海(1993-),男,江西乐平人,博士研究生,主要研究方向:计算机视觉、人工智能、语义分割;刘荪傲(1997-),男,山东青岛人,硕士研究生,主要研究方向:计算机视觉、人工智能、语义分割;王紫霄(1999-),男,河北石家庄人,硕士研究生,主要研究方向:计算机视觉、人工智能、半监督学习;李宏亮(1987-),男,河南安阳人,工程师,博士,主要研究方向:计算机视觉、计算成像。
  • 基金资助:
    国家自然科学基金资助项目(62022076,U19A2057,61976008)。

Abstract: As a fundamental and key task in computer-aided diagnosis, medical image segmentation aims to accurately recognize the target regions such as organs, tissues and lesions at pixel level. Different from natural images, medical images show high complexity in texture and have the boundaries difficult to judge caused by ambiguity, which is the fault of much noise due to the limitations of the imaging technology and equipment. Furthermore, annotating medical images highly depends on expertise and experience of the experts, thereby leading to limited available annotations in the training and potential annotation errors. For medical images suffer from ambiguous boundary, limited annotated data and large errors in the annotations, which makes it is a great challenge for the auxiliary diagnosis systems based on traditional image segmentation algorithms to meet the demands of clinical applications. Recently, with the wide application of Convolutional Neural Network (CNN) in computer vision and natural language processing, deep learning-based medical segmentation algorithms have achieved tremendous success. Firstly the latest research progresses of deep learning-based medical image segmentation were summarized, including the basic architecture, loss function, and optimization method of the medical image segmentation algorithms. Then, for the limitation of medical image annotated data, the mainstream semi-supervised researches on medical image segmentation were summed up and analyzed. Besides, the studies related to measuring uncertainty of the annotation errors were introduced. Finally, the characteristics summary and analysis as well as the potential future trends of medical image segmentation were listed.

Key words: medical image segmentation, deep learning, Convolutional Neural Network (CNN), semi-supervised learning, uncertainty estimation

摘要: 医学影像分割是计算机辅助诊断中的一项基础且关键的任务,目的在于从像素级别准确识别出目标器官、组织或病变区域。不同于自然场景下的图像,医学影像往往纹理复杂,同时受限于成像技术和成像设备,医学影像噪声大,边界模糊而不易判断。除此之外,对医学影像进行标注极大依赖于医疗专家的认知和经验,因此可用于训练中的标注数据少且存在标注误差。由于上述的医学影像边缘模糊不清、训练数据较少和标注误差较大等特点,基于传统图像分割算法搭建的辅助诊断系统难以满足临床应用的要求。近年来随着卷积神经网络(CNN)在计算机视觉和自然语言处理领域的广泛应用,基于深度学习的医学影像分割算法取得了极大的成功。首先概述了近几年基于深度学习的医学影像分割的研究进展,包括这些医学影像分割算法的基本结构、目标函数和优化方法。随后针对医学影像标注数据有限的问题,对目前半监督条件下医学影像分割的主流工作进行了整理归纳和分析。此外,还介绍了针对标注误差进行不确定度分析的相关工作。最后,总结分析了深度学习医学影像分割的特点并展望了未来的研究趋势。

关键词: 医学影像分割, 深度学习, 卷积神经网络, 半监督学习, 不确定性估计

CLC Number: