摘要
目的对以深度学习为代表的人工智能在肿瘤放射治疗全程中的应用现状进行归纳,并对目前其存在的问题和发展前景进行阐述。方法以“放射治疗、人工智能、深度学习、自动勾画、质量保证、图像配准”等为中文关键词,“radiotherapy,artificial intelligence,deep learning,automatic contouring,quality assurance,image registration”等为英文关键词,在PubMed、CNKI数据库中检索2012-2023发表的相关文献。纳入标准:(1)人工智能在放疗图像配准和自动勾画中的应用;(2)人工智能在放疗计划制定中的应用;(3)人工智能在放疗质量保证及疗效预测中的应用。排除标准:与放射治疗相关性较低。结果人工智能中的深度学习技术在放射治疗的多个环节中扮演重要角色,尤其在医学图像处理方面,现有研究证明深度学习在图像合成与配准、靶区自动勾画领域可以减少放射科医生的工作量并提高结果的一致性。然而在放射治疗计划和质量保证中的应用仍需要进一步开发和临床验证以充分发挥其潜力。结论人工智能与放射治疗的结合已经初步取得了一些成果,但仍面临着数据数量与质量、算法可靠性、道德伦理等诸多问题。因此人工智能应用到临床还需要不断优化算法以加速该领域发展。
Objective To review the current status of artificial intelligence,represented by deep learning,in radiotherapy,and elaborate on the current problems and development prospects.Methods A literature search for the period from 2012 to 2023 was conducted in the PubMed and CNKI databases using the keywords"radiotherapy,artificial intelligence,deep learning,automatic contouring,quality assurance,and image registration".Inclusion criteria:(1)Application of artificial intelligence in radiotherapy image registration and automatic delineation;(2)The application of artificial intelligence in the formulation of radiotherapy plans;(3)The application of artificial intelligence in quality assurance and efficacy prediction of radiotherapy.Exclusion criteria:Literature with low relevance and credibility to radiation therapy.Finally,85 relevant articles were selected for analysis.Results Deep learning technologies within the realm of artificial intelligence have assumed a critical role across various stages of radiotherapy,particularly in the field of medical image processing.Current research demonstrates that deep learning can significantly reduce the workload of radiological physicians in the areas of image synthesis and registration,as well as in the automatic delineation of target areas,thereby enhancing the consistency of the outcomes.However,the application of these technologies in the development of radiotherapy plans and in quality assurance still requires further refinement and clinical validation to fully realize their potential.Conclusions The combination of artificial intelligence and radiotherapy has made some initial achievements but still faces many challenges such as the quantity and quality of data,algorithm reliability,ethical and moral issues.Therefore,continuous algorithm optimization is necessary for the application of artificial intelligence to clinical practice to accelerate the development of this field.
作者
张琪月
李鸿岩
张红
徐度玲
柳佳娣
ZHANG Qiyue;LI Hongyan;ZHANG Hong;XU Duling;LIU Jiadi(Department of Medical Physics,Institute of Modern Physics,Chinese Academy of Sciences,Lanzhou,GanSu 730000,China;School of Nuclear Science and Technology,Lanzhou University,Lanzhou,GanSu 730000,China;Key Laboratory of Heavy Ion Radiation Biology and Medicine,Chinese Academy of Sciences,Lanzhou,GanSu 730000,China;Gansu Provincial Isotope Laboratory,Lanzhou,GanSu 730300,China;School of Nuclear Science and Technology,University of Chinese Academy of Sciences,Beijing 100039,China;Advanced Energy Science and Technology Guangdong Laboratory,Huizhou,Guangdong 516006,China)
出处
《中华肿瘤防治杂志》
CAS
北大核心
2024年第3期173-180,186,共9页
Chinese Journal of Cancer Prevention and Treatment
关键词
放疗
人工智能
深度学习
自动勾画
质量保证
图像配准
radiotherapy
artificial intelligence
deep learning
automatic contouring
quality assurance
image registration
