数据与计算发展前沿 ›› 2019, Vol. 1 ›› Issue (2): 37-52.doi: 10.11871/jfdc.issn.2096-742X.2019.02.004

所属专题: “人工智能”专刊

• 人工智能专刊 • 上一篇    下一篇

深度学习在医学影像分析中的应用综述

俞益洲,马杰超*(),石德君,周振   

  1. 深睿医疗人工智能研究院,北京 100080
  • 收稿日期:2019-09-25 出版日期:2019-12-20 发布日期:2020-01-15
  • 通讯作者: 马杰超 E-mail:majch7@mail2.sysu.edu.cn
  • 作者简介:俞益洲,1970年生,深睿医疗首席科学家,曾任美国伊利诺依大学香槟分校终身教授。加州大学伯克利分校计算机博士,美国计算机协会杰出科学家,IEEE Fellow。已在具有影响力的国际会议和期刊发表学术论文150余篇,多次获最佳论文奖。担任图像、视觉计算领域重要国际会议程序委员会委员或主席,并在多个国际学术期刊担任副主编。
    本文承担工作为:AI技术及其在医学影像中的应用分析与讨论以及全文统筹。
    Yu Yizhou, born in 1970, is a full professor at the University of Hong Kong. He was a tenured professor at University of Illinois, Urbana-Champaign (UIUC). He received his PhD degree in computer science at University of California, Berkeley. He is an IEEE fellow and ACM distinguished member. He has published more than 150 academic papers in influential international conferences and journals and has won many best paper awards. He serves as a member or chairman at important international conferences in the field of image processing and computer vision and serves as deputy editor of several international academic journals.
    Role in this paper: conceptualized and organized the review on the application of deep learning in medical imaging analysis.
    E-mail: yizhouy@acm.org|马杰超,1991年生,深睿医疗AI研究院机器学习研究员,中山大学计算机硕士。主要研究方向为AI技术在医学图像上的应用。
    本文承担工作为:调研、分析与讨论DL在脑卒中、肺结节、肺栓塞、乳腺疾病中的应用研究。
    Ma Jiechao, born in 1991, is a machine learning researcher of Deepwise AI Lab and a Master of Computer Science from Sun Yat-sen University. His main research direction is the application of artificial intelligence technology in medical image.
    Role in this paper: reviewed and analyzed the research applications of deep learning in stroke, pulmonary nodules, pulmonary embolism and breast diseases.|石德君,1990年生,深睿医疗AI研究院机器学习研究员,北京大学医学部理学学士。主要研究方向为DL和医学影像分析。
    本文承担工作为:调研、分析与讨论DL在糖网、肺结核、骨龄估计中的应用研究。
    Shi Dejun, born in 1990, he graduated from Peking University Health Science Center, and is a machine learning researcher at Deepwise AI Lab. His research interests are deep learning and medical imaging analysis.
    Role in this paper: reviewed and analyzed the research applications of deep learning in diabetic retinopathy, pulmonary tuberculosis, and bone age assessment.
    E-mail:shidejun@deepwise.com|周振,1990年生,中国科学院自动化研究所博士,深睿医疗高级算法研究员。主要研究方向为DL和医学影像分析。
    本文承担工作为:制定论文结构,撰写前言、DL简介和结论与展望。
    Zhou Zhen, born in 1990, holds a PhD degree from Institute of Automation, Chinese Academy of Sciences. His research focuses on deep learning and medical imaging analysis.
    Role in this paper: outlined the review and drafted the sections including overview, introduction to deep learning and conclusion.
    E-mail:zhouzhen@deepwise.com
  • 基金资助:
    国家自然科学基金面上项目“基于预后的肺亚实性结节AI辅助影像决策系统的建立”(81971616);国家自然科学基金青年项目“基于CT影像组学预测转移性膀胱癌PD-1/PD-L1抑制剂治疗疗效的研究”(81901742);上海市卫生系统先进适宜技术推广项目“智能云平台在三级医疗体系中的推广应用”(2019SY063)

Application of Deep Learning in Medical Imaging Analysis: A Survey

Yu Yizhou,Ma Jiechao*(),Shi Dejun,Zhou Zhen   

  1. Deepwise AI Lab, Beijing 100080, China
  • Received:2019-09-25 Online:2019-12-20 Published:2020-01-15
  • Contact: Ma Jiechao E-mail:majch7@mail2.sysu.edu.cn

摘要:

【目标】综述近年来深度学习(Deep Learning, DL)在医学影像分析领域的研究和应用进展。【文献范围】采用关键词检索和引文二次检索的方法初步收集相关论文。【方法】首先简要介绍基于卷积神经网络的DL模型,然后按病症介绍近年来DL在医学影像辅助诊断中的表现,病症包括脑卒中、肺结节、骨龄测量等。【结果】DL在多种疾病的影像辅助诊断中展现出优势,包括精度高、速度快、结果稳定、可规模化等。同时,很多问题阻碍了DL从实验走向临床,如依赖大量数据、标注标准不统一、模型泛化能力欠佳、可解释性不足等。【局限】检索文献仅覆盖最近几年的工作,对于更久之前的可能存在遗漏。【结论】深度学习可提高放射医师解读影像的效率和精度,但DL还不完美,在广泛渗透医学影像解读之前,还需要经历长时间的研究和验证。

关键词: 深度学习, 医学影像, 综述

Abstract:

[Objective] This paper reviews the recent progress of deep learning researches and applications in medical image analysis. [Coverage] Relevant papers were first retrieved by keyword search and then by citation screening. [Methods] Deep learning based on convolutional neural networks is briefly introduced. Then, we review the diagnostic performance of deep learning on medical images in recent years with respect to different types of diseases, such as stroke, pulmonary nodules and bone age estimation. [Results] Deep learning for medical image interpretation has demonstrated advantages in many aspects, including accuracy, speed, stability and scalability. Meanwhile, existing problems may hinder clinical adoption of deep learning, such as dependence on a large amount of labelled data, inconsistent labeling standards, poor generalizability and interpretability of deep learning methods. [Limitations] There may be omissions of the retrieved literature, and it is impossible to compare the performance of the same deep learning model across different diseases. [Conclusions] Powerful artificial intelligence can improve the efficiency and accuracy of image interpretations for radiologists, but artificial intelligence is not perfect. Before being widely adopted in medical image interpretation, deep learning methods need more verification in real applications.

Key words: deep learning, medical imaging, survey