基于人工智能测量指标的新型冠状病毒肺炎CT影像鉴别诊断  

Application of Artificial Intelligence Measurement Index in CT Differential Diagnosis of Coronavirus Disease 2019

作　　者：李大胜[1] 王娜娜[1] 董建平[2] 霍志毅[1] 裴丽君[3] 刘曦 LI Dasheng;WANG Nana;DONG Jianping;HUO Zhiyi;PEI Lijun;LIU Xi(Departments of Radiology,Beijing Haidian Section of Peking University Third Hospital,Beijing 100080,China;Departments of Infection,Beijing Haidian Hospital,Beijing Haidian Section of Peking University Third Hospital,Beijing 100080,China;Institute of Population Research,Peking University,Beijing 100871,China;United Imaging Intelligence(Beijing)Co.,Ltd.,Beijing 100036,China)

机构地区：[1]北京市海淀医院(北京大学第三医院海淀院区)放射科,北京100080 [2]北京市海淀医院(北京大学第三医院海淀院区)感染科,北京100080 [3]北京大学人口研究所,北京100871 [4]联影智能医疗科技(北京)有限公司,北京100036

出　　处：《中国医疗设备》2020年第6期70-74,共5页China Medical Devices

摘　　要：目的探索基于CT影像的人工智能(Artificial Intelligence,AI)辅诊系统在新型冠状病毒肺炎(Coronavirus Disease2019,COVID-19)与其他社区获得性肺炎(Community Acquired Pneumonia,CAP)鉴别诊断中的应用价值。方法收集2019年1月27日至2020年2月20日103例肺炎患者,其中29例COVID-19患者,74例为排除COVID-19的CAP患者。两组患者均行MSCT扫描,应用常规阅片+AI辅诊系统辅助阅片方式对CT图像进行评价,并使用AI辅诊系统对两组患者的感染区域进行定量测量,比较两组诊断结果差异。结果 COVID-19组的AI测量值密度HU(-750,-300)感染体积和感染占比平均水平均显著高于CAP组(P<0.05)。其他测量值结果在两组间无统计学差异(P>0.05)。结论基于AI测量的结果可以作为鉴别COVID-19和CAP的补充依据,结合AI定量测量结果和人工判断可以更方便和快捷地为COVID-19提供可靠的鉴别诊断依据。Objective To explore the application value of CT images-based artificial intelligence(AI)auxiliary diagnosis system in the differential diagnosis of coronavirus disease 2019(COVID-19)and community acquired pneumonia(CAP).Methods A total of 103 patients with pneumonia from January 27,2019 to February 20,2020 were collected,including 29 COVID-19 patients and 74 CAP patients without COVID-19.MSCT scanning was performed in both groups.CT images were evaluated by the methods of conventional film reading and AI auxiliary diagnosis system,and the infected areas in the two groups were quantitatively measured by AI auxiliary diagnosis system.The difference of diagnosis results between the two groups were compared.Results The infection volume and infection percentage under AI measured density of HU(-750,-300)in the COVID-19 group were significantly higher than that in the CAP groups(P<0.05).There was no significant difference in other measurements beteen the two groups(P>0.05).Conclusion The results based on AI measurement can be used as a supplementary basis for identifying COVID-19 and CAP.Combining the quantitative measurement results of AI with the manual judgment can provide a more convenient and rapid identification of COVID-19.

关 键 词：新型冠状病毒肺炎 核酸检测 计算机断层扫描 人工智能 定量评估 

分 类 号：R563.1[医药卫生—呼吸系统] R318[医药卫生—内科学]