基于BiGRU深度神经网络的心肌梗死检测  

DETECTION OF MYOCARDIAL INFARCTION BASED ON BIGRU DEEP NEURAL NETWORK

作　　者：张行进 李润川[2,3] 赵红领[2] 王宗敏[1,2,3] Zhang Xingjin;Li Runchuan;Zhao Hongling;Wang Zongmin(State Key Laboratory of Mathematical Engineering and Advanced Computing,People's Liberation Army of China Information Engineering University,Zhengzhou 450001,Henan,China;Cooperative Innovation Center of Internet Healthcare,Zhengzhou University,Zhengzhou 450000,Henan,China;Research Institute of Industrial Technology,Zhengzhou University,Zhengzhou 450000,Henan,China)

机构地区：[1]解放军信息工程大学数学工程与先进计算国家重点实验室,河南郑州450001 [2]郑州大学互联网医疗与健康服务河南省协同创新中心,河南郑州450000 [3]郑州大学产业技术研究院,河南郑州450000

出　　处：《计算机应用与软件》2020年第2期48-53,共6页Computer Applications and Software

基　　金：国家重点研发计划项目(2017YFB1401200);兵团重点领域科技攻关计划项目(2018AB017);河南省高等学校重点科研项目(18A520055);河南省科技攻关项目(142102210069)。

摘　　要：针对目前智能医疗诊断领域的研究现状,结合心电信号的时序性和多导联关联性特点,为降低心肌梗死疾病的误诊率,提出一种基于双向门控循环单元神经网络(Bidirectional Gated Recurrent Unit,BiGRU)和多导联心电图(electrocardiogram,ECG)信号的深度神经网络学习算法。对原始心电信号进行去噪处理,分割成心拍序列;将心拍序列送入深度神经网络训练模型学习分类;采用Physikalisch-Technische Bundesanstalt(PTB)心电数据库验证多导联BiGRU算法。算法对心梗检测的灵敏度为99.93%、特异性为99.72%、准确率为99.89%。实验结果表明,该算法的检测效果明显优于其他文献的检测算法,对提高心肌梗死的正确诊断率具有重要意义。In view of the current research status in the field of intelligent medical diagnosis,we combine the timing and multi-lead correlation characteristics of ECG(Electrocardiogram)signals to propose a deep neural network learning algorithm based on bidirectional gated recurrent unit neural network(BiGRU)and multi-lead ECG signals,so as to reduce the misdiagnosis rate of myocardial infarction diseases.The original ECG signal was denoised and segmented into a heartbeat sequence.The heartbeat sequence was sent into the deep neural network training model to learn the classification.The Physikalisch-Technische Bundesanstalt(PTB)ECG database was used to verify the multi-lead BiGRU algorithm.The sensitivity of the algorithm for myocardial infarction detection is 99.93%,the specificity is 99.72%,and the accuracy rate is 99.89%.The experimental results show that the effect is obviously superior to the detection of other literatures,and it is of great significance to improve the correct diagnosis rate of myocardial infarction.

关 键 词：心电图 多导联 BiGRU 心肌梗死 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]