Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (2): 1-14.DOI: 10.3778/j.issn.1002-8331.2104-0408

• Research Hotspots and Reviews • Previous Articles     Next Articles

Review of U-Net and Its Application in Segmentation of Liver and Liver Tumors

ZHANG Huan, LIU Jing, FENG Yibo, QIU Dawei   

  1. College of Intelligence and Information Engineering, Shandong University of Traditional Chinese Medicine, Jinan 250355, China
  • Online:2022-01-15 Published:2022-01-18



  1. 山东中医药大学 智能与信息工程学院,济南 250355

Abstract: It is of great clinical significance to realize automatic and precise segmentation of liver and liver tumor regions. With rapid developments of deep learning, deep neural networks have been gradually applied to medical fields, and computer-aided diagnosis has become a research hotspot. U-Net has been widely used in the fields of medical image segmentation because of its good performance in small sample datasets. Firstly, commonly used liver and liver tumor segmentation datasets and relevant evaluation criteria are introduced, the U-Net model and its improvements concerning encoder-decoder, skip-connection, and the overall structure are summarized. Then, the applications of U-Net in the segmentation of liver and liver tumors are analyzed from two aspects of single network structure and multi-network structure. Finally, the deficiencies of related works and the future development trends are discussed.

Key words: deep learning, U-Net, medical imaging processing, liver tumor segmentation

摘要: 医学上实现肝脏及肝脏肿瘤区域自动精准分割具有十分重要的临床意义,随着深度学习技术的迅速发展,深度神经网络逐步应用于医学领域,计算机辅助诊断成为研究热点。U-Net网络由于其在小样本数据集上的良好表现,在医学图像分割领域得到了广泛应用。基于此,介绍了肝脏和肝脏肿瘤分割中常用的数据集和评价指标,归纳了U-Net网络模型及围绕编解码器、跳跃连接和整体结构的改进。从单网络结构和多网络结构改进两个方面对U-Net模型在肝脏及肝脏肿瘤分割的相关应用加以论述。对相关研究工作的不足进行总结,并对未来发展予以展望。

关键词: 深度学习, U-Net, 医学图像处理, 肝脏肿瘤分割