基于深度学习的新型冠状病毒肺炎CT征象检测研究  

Study of COVID-19 CT Imaging Detection Based on Deep Learning

作　　者：祖莅惠 胡博奇[1] 王平[1] 张忠 刘景鑫[1] ZU Lihui;HU Boqi;WANG Ping;ZHANG Zhong;LIU Jingxin(China-Japan Union Hospital of Jilin University,Jilin University,Changchun Jilin 130033,China;WX Medical Technology Co.,Ltd.,Shenyang Liaoning 110000,China)

机构地区：[1]吉林大学中日联谊医院,吉林长春130033 [2]辽宁万象联合医疗科技有限公司,辽宁沈阳110000

出　　处：《中国医疗设备》2020年第6期89-92,共4页China Medical Devices

基　　金：国家重点研发计划项目(2018YFC1315604,2018YFC0116900);吉林省科技发展计划项目“突发疫情远程智能防控诊断平台体系研究与应用”。

摘　　要：放射诊断是新型冠状病毒肺炎(Coronavirus Disease 2019,COVID-19)诊疗过程中的重要环节,然而CT影像数据量较大,单个患者阅片耗时较长,为医生诊断带来巨大压力。本研究基于不同医院COVID-19患者的数据脱敏CT影像,通过深度学习的方法学习样本病灶纹理,提出了一种基于时间空间序列卷积的图像检测模型。该模型能快速定位CT影像中病灶区域,并关联同一患者不同阶段CT影像,综合得到更准确的检测结果。本文的研究可以提高COVID-19的初步诊断及鉴别诊断效率,可用于辅助临床诊断,为疾病控制做出贡献。Radiological diagnosis is an important part of the diagnosis and treatment of coronavirus disease 2019(COVID-19).However,the amount of CT image data is large and the time taken by a single patient is long,which brings great pressure to the diagnosis of doctors.Based on the desensitized CT images of patients with COVID-19 in different hospitals,this study proposed an image detection model based on convolution of time-space series by learning the texture of the sample lesions through depth learning.This model could quickly locate the focus area in the CT image,and associate with the different stages of CT image of the same patient,so as to get more accurate detection results.This study helpful to improve the efficiency of preliminary diagnosis and differential diagnosis of COVID-19,which can be used to assist clinical diagnosis and contribute to disease control.

关 键 词：新型冠状病毒肺炎 深度学习 目标检测 放射诊断 CT征象检测 

分 类 号：R197.39[医药卫生—卫生事业管理] TP391[医药卫生—公共卫生与预防医学]