基于人工智能的辅助诊断在先天性房间隔缺损筛查中的应用研究  

Application of artificial intelligence-based auxiliary diagnosis in screening of congenital atrial septal defect

作　　者：杨宏波[1] 潘家华[1] 王威廉[2] 郭涛[1] 张戈军[3] 唐永研 许虹莉 YANG Hongbo;PAN Jiahua;WANG Weilian;GUO Tao;ZHANG Gejun;TANG Yongyan;XU Hongli(Yunnan Cardiovascular Hospital,Kunming Medical University,Kunming 650000,China)

机构地区：[1]昆明医科大学附属心血管病医院,昆明650000 [2]云南大学,昆明650000 [3]国家心血管病中心,北京100000

出　　处：《实用医学杂志》2019年第16期2637-2640,共4页The Journal of Practical Medicine

基　　金：云南省基础研究计划基金项目[编号：2018FE001)-105)]

摘　　要：目的评估基于人工智能的辅助诊断在先天性房间隔缺损(atrial septal defect,ASD)筛查中的应用价值。方法2014年9月至2018年9月,在先心病筛查中对10 142名0~14岁儿童进行人工听诊及基于人工智能的辅助诊断。人工智能对已确诊ASD的儿童采集标准部位心音,通过去噪和提取特征信息学习,实现心音和ASD对应的辅助诊断;在筛查中随机分为人工听诊组(n=6 280)和人工智能组(n=3762),比较ASD的发现率;在确诊的162例患儿中比较人工听诊和人工智能诊断的准确率。结果人工智能共学习6 253个ASD患儿和6 544个正常儿童心音周期;采用学习所得辅助诊断技术听诊3 762名儿童,诊断率(4.5‰)与人工听诊(1.9‰)相比差异无统计学意义(P>0.05);对162例确诊患儿进行人工智能辅助诊断,准确率为69.1%。结论基于人工智能的辅助诊断在先天性ASD筛查中是一个有效的辅助手段。Objective To evaluate the value of artificial intelligence(AI)-based auxiliary diagnosis in the screening of congenital atrial septal defect(ASD).Methods From September 2014 to September 2018,10,142 children aged 0~14 years were enrolled in a diagnosis of congenital heart disease(CHD)by artificial auscultation and AI-based diagnosis.Heart sounds of standard parts in children with confirmed atrial septal defect were collected by AI,and the denoising and extracting characteristic information were conducted to realize the auxiliary diagnosis corresponding to heart sound and atrial septal defect.During the screening,patients were randomly divided into artificial auscultation group(n=6,280)and AI group(n=3,762).The detection rate of atrial septal defect was compared in the two groups and the accuracy of artificial auscultation and AI diagnosis was compared among the162 diagnosed children.Results A total of 6,253 children with atrial septal defect and 6,544 normal children′s heart sound cycle were detected by AI.Auscultation of 3,762 children was conducted with AI and there was no significant difference between the detection rate of auscultation with AI(4.5‰)and that with manual stethoscope(1.9‰)(P>0.05).AI-assisted diagnosis was performed on 162 confirmed patients with an accuracy rate of69.1%.Conclusion AI-based assisted diagnosis is an effective adjunct in the screening of congenital atrial septal defect.

关 键 词：先天性房间隔缺损 先心病筛查 人工智能 

分 类 号：R725.4[医药卫生—儿科] TP18[医药卫生—临床医学]