基于改进残差网络的心脏杂音检测方法  

A heart murmur detection method based on improved residual network

作　　者：李世龙 何培宇[1] 黄昭涵 李莉[2] 赵启军[3] 潘帆[1] LI Shilong;HE Peiyu;HUANG Zhaohan;LI Li;ZHAO Qijun;PAN Fan(School of Electronics and Information Engineering,Sichuan University,Chengdu 610065,China;Division of Pediatric Cardiology,West China Second University Hospital,Sichuan University,Chengdu 610041;School of Computer Science,Sichuan University,Chengdu 610065)

机构地区：[1]四川大学电子信息学院,成都610065 [2]四川大学华西第二医院儿童心血管科,成都610041 [3]四川大学计算机学院,成都610065

出　　处：《生物医学工程研究》2024年第5期369-376,共8页Journal Of Biomedical Engineering Research

基　　金：国家自然科学基金项目(62066042);四川省重点研发项目(2022YFG0045);中央高校基本科研业务费专项资金(2022SCU12008)。

摘　　要：针对目前缺乏识别心音收缩期杂音和舒张期杂音的方法,本研究提出了一种基于改进残差网络的杂音检测方法,通过检测多个听诊区的收缩期和舒张期杂音,判断患者是否存在心脏杂音。首先,按照心音时相将心音数据分割成心音信号片段;然后,提取心音片段样本的对数梅尔谱特征;最后,使用嵌入通道注意力机制的残差神经网络模型进行心脏杂音检测。本研究在CirCor Digiscope dataset 2022数据集上进行了五折交叉验证,心脏杂音检测平均准确率、平均召回率、平均精确率与平均F1分数分别为90.05%、63.74%、84.20%和72.28%。实验结果表明,本研究方法在基于时相切割的心音数据的杂音检测任务中准确率较好,可为心血管疾病的自动分析提供重要依据。In view of the fact that there is no recognition method for systolic and diastolic murmurs of heart sounds,we proposed a murmur detection method based on improved residual network to determine whether a heart murmur exists in a patient by detecting systolic and diastolic murmurs in multiple auscultation zones.Firstly,the heart sound data were segmented into heart sound signal segments according to the heart sound time phase.Then,the log-Mel spectral features of the heart sound segment samples were extracted.Finally,the residual neural network model embedded in the channel attention mechanism was used to detect heart murmurs.We performed 5 cross-validation on the CirCor Digiscope dataset 2022,and the average accuracy,average recall,average precision and average F1 score of heart murmur detection reached 90.05%,63.74%,84.20%and 72.28%,respectively.The experimental results show that the proposed method has a good accuracy in detecting noise based on time-phase cut heart sound data,and can provide an important basis for automatic analysis of cardiovascular diseases.

关 键 词：心血管疾病 心脏杂音 杂音检测 对数梅尔谱 注意力机制 残差神经网络 

分 类 号：R318[医药卫生—生物医学工程] TN912.3[医药卫生—基础医学]