摘要
肝脏肿瘤严重危害着全人类的生命健康。近年来,随着深度学习的快速发展,涌现了许多使用腹部计算机断层扫描(computed tomography,CT)图像进行肝脏肿瘤分割的深度学习方法,这些方法的应用对于临床上实现肝脏肿瘤的计算机辅助诊断具有十分重要的意义。为此,本文对深度学习方法在肝脏肿瘤CT图像分割中的应用进行了归纳,将各种卷积神经网络(convolutional neural networks,CNN)分为二维(two-dimensional,2D)、三维(three-dimensional,3D)和2.5维(2.5-dimensional,2.5D)卷积神经网络。此外,本文总结了各类网络的优缺点以及改进方法,为深度学习在肝脏肿瘤分割中的应用提供了有益参考。
Liver tumors seriously endanger the life and health of all mankind.In recent years,with the rapid development of deep learning,many deep learning methods for liver tumor segmentation using abdominal computed tomography(CT)scans have emerged.The application of these methods is of great significance for computer-aided diagnosis of liver tumors in clinic.Therefore,the applications of deep learning methods in liver tumor segmentation from CT scans are summarized in this paper.Various convolutional neural networks(CNN)are divided into two-dimensional(2D),three-dimensional(3D)and 2.5-dimensional(2.5D)CNNs.In addition,this paper summarizes the advantages,disadvantages and improved methods of various CNNs,which provides a useful reference for the applications of deep learning in liver tumor segmentation.
作者
夏栋
张义
巫彤宁
陈新华
李从胜
XIA Dong;ZHANG Yi;WU Tongning;CHEN Xinhua;LI Congsheng(China Academy of Information and Communications Technology,Beijing 100191;The First Affiliated Hospital,Zhejiang University School of Medicine,Hangzhou 310003)
出处
《北京生物医学工程》
2023年第3期308-314,共7页
Beijing Biomedical Engineering
基金
国家自然科学基金(61971445)
国家科技重大专项(2018ZX10301201)资助。
关键词
肝脏肿瘤
图像分割
深度学习
卷积神经网络
liver tumor
image segmentation
deep learning
convolutional neural networks
