基于深度学习在医学图像的分类与诊断综述  

A Review of the Researches on the Classification and Diagnosis of Medical Images Based on Deep Learning

作　　者：阚明阳 KAN Ming-yang(College of Information Engineering,Hebei University of Geosciences,Hebei Shijiazhuang 050031,China)

机构地区：[1]河北地质大学信息工程学院,河北石家庄050031

出　　处：《新一代信息技术》2021年第20期17-22,共6页New Generation of Information Technology

基　　金：国家自然科学基金资助项目(项目编号:61402481);河北省重点研发计划项目(项目编号:18212005);河北省教育厅自然科学基金重点项目(项目编号:ZD2018083);河北青年拔尖人才支持计划项目(项目编号:冀字[2013]17号)。

摘　　要：随着人工智能技术和互联网医疗的发展,利用卷积神经网络实现医学图像病变部位的定位,分割与分类于一体的计算机医疗辅助诊断系统已经成为新型医疗模式发展的必然趋势。如何将庞大的图像数据进行有效的分类与分割,发现和挖掘其中有价值的信息是当前分类和分割算法的研究热点。本文介绍了深度学习的概念和发展,以及深度学习模型,卷积神经网络深度置信网络模型的发展过程,并对其在医学图像分析中的应用研究现状进行综述。通过阅读文献对深度学习模型在医学图像分析领域的应用进行整理,并在此基础上突出一些关键的应用领域,进而讨论深度学习基本模型存在的问题,了解到深度学习模型目前所遇到的问题以及深度学习在医学领域的前景与展望。With the development of artificial intelligence technology and Internet medical treatment,it has be-come an inevitable trend of the development of new medical mode to use convolutional neural network to re-alize the localization of pathological parts of medical images,and the computer medical assisted diagnosis system which integrates segmentation and classification.How to classify and segment the huge image data ef-fectively and discover and mine the valuable information is the research hotspot of the current classification and segmentation algorithm.This paper introduces the concept and development of deep learning,as well as the development of deep learning model and deep confidence network model of convolutional neural network.This paper also reviews the application and research status of medical image analysis.[Methods/Process]The application of deep learning model in the field of medical image analysis was sorted out by reading literature,and some key application fields were highlighted on this basis,and then the problems existing in the basic model of deep learning were discussed.[Results/Conclusions]The current problems of deep learning model and the prospect of deep learning in the medical field were understood.

关 键 词：深度学习 医学图像分类 卷积神经网络 

分 类 号：TM351[电气工程—电机]