第107届北美放射学会年会亮点:人工智能  

Highlights of the 107;scientific assembly and annual meeting of Radiological Society of North America: Artificial intelligence

作　　者：林洁琼 黄燕琪[3] 梁长虹[3] 曾洪武[1] LIN Jieqiong;HUANG Yanqi;LIANG Changhong;ZENG Hongwu(Department of Radiology,Shenzhen Children's Hospital,Shenzhen 518038,China;Shantou University Medical College,Shantou 515041,China;Department of Radiology,Guangdong General Hospital,Guangzhou 510080,China)

机构地区：[1]深圳市儿童医院放射科,深圳518038 [2]汕头大学医学院,汕头515041 [3]广东省人民医院放射科,广州510080

出　　处：《磁共振成像》2022年第3期111-114,121,共5页Chinese Journal of Magnetic Resonance Imaging

基　　金：国家自然科学基金青年项目(编号:81701782);深圳市医疗卫生三名工程项目(编号:SZSM202011005)。

摘　　要：第107届北美放射学会(Radiological Society of North America,RSNA)年会人工智能亮点聚焦于:(1)新技术新算法:联邦学习破解数据孤岛难题,迁移学习应用于多中心数据;(2)真实年龄新概念:“影像-生理年龄”;(3) AI赋能影像,从实验室走向临床应用,包括早期诊断、风险评估、预后预测、临床决策辅助、自动化智能测量等;(4) AI应用的挑战:数据“黑盒”、模型适用性,数据管理及法律责任等。结合近年文献,本文对2021 RSNA年会AI研究进行概述。The highlights of artificial intelligence(AI) at the 107;scientific assembly and annual meeting of Radiological Society of North America(RSNA) were:(1) Advanced technologies and algorithms: federated learning aimed to solve the problem of data island.Transfer learning has been applied to multicenter studies;(2) As the new concept of real age, ’Image-based physiological Age’ was first time raised;(3) AI empowers imaging, from laboratory to clinical applications, including early diagnosis, risk assessment, prognostic prediction, clinical decision support and automatic intelligent measurement, etc;(4) Application of AI also meets challenges such as data’black box’, model applicability, data management and legal liability. AI related studies published in recent years and 2021 RSNA were reviewed in this article.

关 键 词：人工智能 深度学习 卷积神经网络 联邦学习 迁移学习 影像组学 

分 类 号：R445[医药卫生—影像医学与核医学]