摘要
目的:基于CiteSpace软件对我国深度学习在医学影像应用方面的文献进行科学计量分析,找出应用领域及研究趋势,为未来开展人工智能在医学影像应用的研究提供借鉴.方法:检索中国知网论文数据库(CNKI)中以医学影像、深度学习、机器学习和影像组学为主题的文献,应用Microsoft Excel 2019和CiteSpace 6.2.R2进行可视化分析,得出文献的发文时间、作者、机构、期刊、关键词的共现网络及关键词突现情况.结果:共1048篇文献纳入分析,年发文量呈稳步上升趋势.关键词共引聚类分析,高频聚类词为深度学习、人工智能、医学影像、影像组学、图像分割、机器学习、医学图像、迁移学习、神经网络、肺结节.关键词突现分析结果显示,近三年的研究热点为数据库、图像质量、CT图像、残差网络.结论:深度学习在医学影像应用的研究热度呈逐年上升的趋势.深度学习在医学影像中的应用研究,分布在图像处理、目标检测、图像分割和影像组学四个领域.医学影像数据库的构建和运用、残差网络及深度学习在医学影像中的应用为未来的研究趋势.
Abstract
Objective:Based on CiteSpace software,the literature on deep learning in medical im-aging applications in China was analyzed scientifically and metrologically to find out the application ar-eas and research trends to provide a reference for future research on the application of artificial intelli-gence in medical imaging.Methods:We researched the literature of China Knowledge Network Essay Database(CNKI)on medical imaging,deep learning,machine learning,and radiomics.We applied Mi-crosoft Excel 2019 and CiteSpace 6.2.R2 to conduct visualization and analysis,which resulted in the co-occurring network of the time of issuance of the literature,the authors,the institutions,the journals,the keywords,and the keyword emergence.Results:We included 1048 documents in the analysis,with a steady upward trend in annual publications.The keyword co-citation clustering analysis showed that high-frequency clustered terms were as follows:deep learning,artificial intelligence,medical imaging,radiomics,image segmentation,machine learning,medical image,migration learning,neural network,and lung nodule.Keyword emergence analysis showed that the research hotspots in the last three years were database,image quality,CT images,and residual networks.Conclusion:The research of deep learning in medical imaging applications is increased yearly.The research on the application of deep learning in medical imaging is distributed in four areas:image processing,object detection,image seg-mentation,and radiomics.Future research trends include constructing and utilizing medical image data-bases,residual net works,and deep learning applications.
维普
万方数据