基于改进多导联残差网络的广泛前壁心肌梗死自动诊断  

Automatic Diagnosis of Extensive Anterior Wall Myocardial Infarction Based on Improved Multi-lead Residual Network

作　　者：赵滢琳 庞春颖[1] 李爽 ZHAO Yinglin;PANG Chunying;LI Shuang(School of Life Science and Technology,Changchun University of Science and Technology,Changchun 130022)

机构地区：[1]长春理工大学生命科学技术学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2022年第3期111-117,共7页Journal of Changchun University of Science and Technology(Natural Science Edition)

基　　金：吉林省科技厅项目(20200404212YY)。

摘　　要：广泛前壁心肌梗死是心梗中危害较大的一种类型,其发病迅速,致死率高,堵塞面积大,对人体造成不可逆的损伤。心电图是检测心梗的重要手段,医生要凭经验判断其发生部位,过程繁琐。针对以上情况,提出一种基于深度学习算法的广泛前壁心肌梗死的自动诊断。算法先将前壁心梗涉及到的V_(1)~V_(5)导联信号进行去噪滤波,分割心拍处理,然后搭建改进多导联残差网络模型,将信号并行输入,实现健康与前壁心梗的分类。实验采用PTB心梗数据库进行验证,准确率达到97.85%,高于经典神经网络。该研究能辅助医生进行广泛前壁心梗的诊断,具有一定的临床意义。Extensive anterior myocardial infarction is a more harmful type of myocardial infarction. It has a rapid onset,high fatality rate,large blockage area,and irreversible damage to the human body. The electrocardiogram is an important method to detect myocardial infarction. Doctors have to judge its occurrence and location based on experience. The process is cumbersome. In view of the above situation,an automatic diagnosis of extensive anterior myocardial infarction based on deep learning algorithm is proposed. At first,the V_(1)~V_(5)lead signals involved in the anterior wall myocardial infarction are denoised and filtered by the algorithm;the heartbeat processing is split and then an improved multi-lead residual network model is built;and the signals are input in parallel to realize the classification of healthy and anterior wall myocardial infarction. The experiment uses the PTB myocardial infarction database for verification;and the accuracy rate reaches 97.85%,which is higher than the classic neural network. This study can assist doctors in the diagnosis of extensive anterior wall myocardial infarction,and has certain clinical significance.

关 键 词：广泛前壁心肌梗死 深度学习 改进多导联残差网络 心电图 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]