人工智能质控在提高孕11~13^(+6)胎儿颈后透明层厚度超声检查图像合格率中的应用价值  

Application value of artificial intelligence quality control in improving the qualified rate of ultrasonic examination image of the thickness of nuchal translucency at 11^(+0)to 13^(+6)weeks

作　　者：陈美池 刘静华[2] 柯华玲 宋金爽 徐金毛 董宇萌 郑洪平[2] Chen Meichi;Liu Jinghua;Ke Hualing;Song Jinshuang;Xu Jinmao;Dong Yumeng;Zheng Hongping(Jiamusi University,Jiamusi,Heilongjiang,154400,China;Longgang District Maternal and Child Health Hospital,Shenzhen,Guangdong,518172,China)

机构地区：[1]佳木斯大学,黑龙江佳木斯154400 [2]深圳市龙岗区妇幼保健院,广东深圳518172

出　　处：《中国产前诊断杂志（电子版）》2024年第1期10-15,共6页Chinese Journal of Prenatal Diagnosis(Electronic Version)

摘　　要：目的 探讨人工智能(artificialintelligence, AI)质量控制系统在提高孕11~13~(+6)周胎儿颈后透明层厚度(nuchal translucency, NT)超声检查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”对深圳市龙岗区妇幼保健院2021年1月至2022年9月共7个季度进行11~13^(+6)周NT超声检查的1315例检查者超声图像质量进行智能质控,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7个季度图像的标准率进行比较,评估智能质控对于提高图像合格率的价值。结果 深圳市龙岗区妇幼保健院七个季度胎儿NT图像标准率分别为51.87%(250/482)、65.02%(197/303)、54.85%(407/742)、68.72%(279/406)、98.25%(336/342)、94.74%(324/342),95.60%(391/409),各季度图像标准率逐步提高直至趋于稳定,两两比较差异有统计学意义(P<0.01);提示各季度医生留存图像的合格率逐渐升高。结论 人工智能质量控制可有效提升11~13~(+6)周NT超声检查图像达标率,提升医生对标准切面的认识。Objective To explore the application value of artificial intelligence(AI)quality control system in improving the standard rate of ultrasound images of nuchal translucency(NT)of fetus at ll~13^(+6) weeks gestation.Methods Using the"prenatal ultrasound AI smart cloud platform",the ultrasonic image quality of 1315 examiners who underwent NT ultrasound examination for 11~13^(+6) weeks in seven quarters from January 2021 to September 2022 in Longgang District Maternal and Child Health Hospital of Shenzhen was intelligently controlled,and the quality-controlled doctors made targeted improvements by consulting the quality control results and the shortcomings of the images.χ^(2) test was used to compare the standard rate of images in seven quarters and evaluate the value of intelligent quality control in improving the qualified rate of images.Results The standard rates of fetal NT images in Longgang Maternal and Child Health Hospital of Shenzhen in seven quarters were 51.87%(250/482),65.02%(197/303),54.85%(407/742),68.72%(279/406),98.25%(336/342),94.74%(324/342)and 95.60%(391/409),respectively.It is suggested that the qualified rate of doctors'retained images increases gradually in each quarter.Conclusions The quality control of artificial intelligence can effectively improve the standard rate of NT ultrasound images in 11~13+6 weeks,and improve doctors'understanding of standard sections.

关 键 词：人工智能 质量控制 胎儿 颈后透明层测量 超声检查 

分 类 号：R445[医药卫生—影像医学与核医学]