Page 203 - 《软件学报》2021年第5期

软件学报 ISSN 1000-9825, CODEN RUXUEW E-mail: jos@iscas.ac.cn
Journal of Software,2021,32(5):1427−1460 [doi: 10.13328/j.cnki.jos.006205] http://www.jos.org.cn
©中国科学院软件研究所版权所有. Tel: +86-10-62562563

∗
基于深度学习的数字病理图像分割综述与展望

2
1
1
2
宋 杰 , 肖 亮 , 练智超 , 蔡子贇 , 蒋国平 1
1
(南京邮电大学 自动化学院、人工智能学院,江苏 南京 210023)
2
(南京理工大学 计算机科学与工程学院,江苏 南京 210094)
通讯作者: 肖亮, E-mail: xiaoliang@mail.njust.edu.cn

摘 要: 数字病理图像分析对于乳腺癌、前列腺癌等良恶性分级诊断具有重要意义,其中,组织基元的形态和目标
测量是量化分析的重要依据.然而,由于病理数据多样性和复杂性等新特点,其分割任务面临着特征提取困难、实例
分割困难等挑战.人工智能辅助病理量化分析将复杂病理数据转化为可挖掘的图像特征,使得自动提取组织基元的
定量化信息成为可能.特别是随着计算机计算能力的快速发展,深度学习技术凭借其强大的特征学习、设计灵活等
特性在数字病理量化分析领域取得了突破性成果.系统概述目前代表性深度学习方法,包括卷积神经网络、全卷积
网络、编码器-解码器模型、循环神经网络、生成对抗网络等方法体系,总结深度学习在病理图像分割等任务中的
建模机理和应用,并梳理了现有方法的方法理论、关键技术、优缺点和性能分析.最后讨论了未来数字病理图像分
割深度学习建模的开放性挑战和新趋势.
关键词: 数字病理;组织基元;实例分割;特征表示学习;深度模型
中图法分类号: TP391

中文引用格式: 宋杰,肖亮,练智超,蔡子贇,蒋国平.基于深度学习的数字病理图像分割综述与展望.软件学报,2021,32(5):
1427−1460. http://www.jos.org.cn/1000-9825/6205.htm
英文引用格式: Song J, Xiao L, Lian ZC, Cai ZY, Jiang GP. Overview and prospect of deep learning for image segmentation in
digital pathology. Ruan Jian Xue Bao/Journal of Software, 2021,32(5):1427−1460 (in Chinese). http://www.jos.org.cn/1000-9825/
6205.htm

Overview and Prospect of Deep Learning for Image Segmentation in Digital Pathology

1
2
2
1
SONG Jie , XIAO Liang , LIAN Zhi-Chao , CAI Zi-Yun , JIANG Guo-Ping 1
1 (College of Automation & College of Artificial Intelligence, Nanjing University of Posts and Telecommunications, Nanjing 210023,
China)
2 (School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China)
Abstract: The quantitative analysis of digital pathology images plays a significant role in the diagnosis of benign and malignant
diseases such as breast cancer and prostate cancer, in which the morphology measurements of histologic primitives serve as a basis of
quantitative analyses. However, the complex nature of digital pathology data, such as diversity, present significant challenges for such
segmentation task, which might lead to difficulties in feature extraction and instance segmentation. By converting complex pathology data
into minable image features using artificial intelligence assisted pathologist's analysis, it becomes possible to automatically extract

∗ 基金项目: 国家自然科学基金(62001247, 61871226, 61571230, 62006127, 61873326, 61672298); 江苏省社会发展重点研发计
划(BE2018727); 江苏省自然科学基金(BK20190728); 江苏省高等学校自然科学研究面上项目(20KJB520005); 南京邮电大学引进
人才科研启动基金(NY219152, NY218120)
Foundation item: National Natural Science Foundation of China (62001247, 61871226, 61571230, 62006127, 61873326,
61672298); Jiangsu Provincial Social Developing Project (BE2018727); Natural Science Foundation of Jiangsu Province (BK20190728);
Natural Science Foundation for Colleges and Universities in Jiangsu Province (20KJB520005); Introduction of Talent Research Start-up
Fund of Nanjing University of Posts and Telecommunications (NY219152, NY218120)
收稿时间: 2020-08-15; 修改时间: 2020-09-27; 采用时间: 2020-11-18; jos 在线出版时间: 2020-12-02