摘要
目的探讨人工智能(artificial intelligence,AI)质量控制系统在提高胎儿上腹部水平横切面超声图像标准率中的应用价值。方法应用“产前超声AI智慧云平台”对深圳市60家医院三个季度在产科超声检查中存储的胎儿上腹部水平横切面图像共18114张进行智能质控,受质控医生通过查阅质控结果及图像存在的不足之处针对性地加以改进。应用χ^(2)检验对三个季度图像的标准率、不足原因及申诉情况进行两两比较,评估智能质控对于提高胎儿上腹部水平横切面标准程度的价值。结果各医院第一、二、三季度胎儿上腹部水平横切面图像标准率分别为80.15%(5649/7048)、86.2%(4391/5096)、90.55%(5406/5970),各季度图像标准率逐步提高,两两比较差异有统计学意义(P<0.01);每个季度的不标准图像中,图像不足的主要原因是切面中出现了大片肺脏,分别为8.9%(631/7048)、7.0%(358/5096)、4.9%(294/5970),其次是脐静脉和门静脉汇合部显示不清,分别为4.3%(305/7048)、3.5%(181/5096)、2.6%(155/5970),这两类图像均显著减少,两两季度比较差异有统计学意义(P<0.01);医生申诉后维持智能评价结果的图像分别为66.4%(79/119)、49%(25/51)、33.3%(10/30),两两比较差异有统计学意义(P<0.01),提示各季度医生对图像认识错误率逐渐降低。结论人工智能质量控制可有效提高胎儿上腹部水平横切面超声图像的标准率,提升医生对标准切面的认识。
Objective To explore the application value of the artificial intelligence(AI)quality control system in improving the standard rate of transverse section of fetal abdomen.Methods The“Intelligent ultrasonic quality control system”was applied to performquality control in 18,114 transverse sections of fetal abdomen imagesstored in 60 hospitals in Shenzhen in three quarters.The doctors subjected to quality control can make targeted improvements by referring to the quality control results and the shortcomings of the images.Image standard rate,reasons for the insufficiency and the instancesof the complaint in three quarters were compared by χ^(2) test.Results The standard rates of transverse section of abdomen in the first,second and third quarters were 80.15%(5649/7048),86.2%(4391/5096)and 90.55%(5406/5970),respectively.The image standard rate has been gradually increased in each quarter,and the comparative difference between every two quarters was statistically significant(P<0.01).The main reason for the non-standard images in each quarter was the presence of lung in the section(8.9%(631/7048),7.0%(358/5096),and 4.9%(294/5970),respectively).Another reason for the non-standard images was that umbilical vein and portal vein were not shownin the section(4.3%(305/7048),3.5%(181/5096)、2.6%(155/5970)).Bothreasons for the non-standard imageswere significantly reduced(P<0.01).The images that maintained the results of AI evaluation after doctors’complaints were 66.4%(79/119),49%(25/51)and 33.3%(10/30),respectively,and the difference was statistically significant(pairwise comparison between groups,P<0.01),indicating that the error rate of doctors’understanding of images decreased gradually.Conclusion The standard rates of transverse section of the fetal abdomen could be improved effectively by AI quality control,and doctors’understanding of the standard section is also improved.
作者
彭桂艳
谭莹
曾晴
罗丹丹
黄文兰
江瑶
温昕
李胜利
Peng Guiyan;Tan Ying;Zeng Qing;Luo Dandan;Huang Wenlan;Jiang Yao;Wen Xin;Li Shengli(Depart ment of Ultrasownd,Affiliated Shenzhen Maternity&Child Healthcare Hospital,Southern Medical University,Shenzhen 518028,China)
出处
《中国产前诊断杂志（电子版）》
2022年第4期6-10,31,共6页
Chinese Journal of Prenatal Diagnosis(Electronic Version)
基金
国家重点研发计划资助(2022YFF0606301)
深圳市科技计划项目(JCYJ20210324130812035)。
关键词
人工智能
上腹部水平横切面
胎儿
标准切面
Artificial intelligence
Transverse section of the abdomen
Fetals
Standard section
