首页 | 本学科首页   官方微博 | 高级检索  
     

基于人工神经网络及心音小波分析的冠心病诊断方法的研究
引用本文:叶学松.基于人工神经网络及心音小波分析的冠心病诊断方法的研究[J].浙江大学学报(自然科学版 ),1999,33(2):123-128.
作者姓名:叶学松
作者单位:浙江大学生命科学和医学工程系
摘    要:通过对冠状动脉动力学及湍流诱发声学的理论研究后,首次提出用整周期心音信号小波分析来提取冠状动脉疾病(CAD)心音特征的方法,诊断系统将CAD病人组及非冠心病对照组提取的心音特征结合人体的个体特征参数输入到神经网络进行学习训练后,最后达到自动诊断冠状动脉疾病,为了检测这种较为微弱的生理信号,本文通过对人体体表声传播模型研究后设计对外界噪声有较强抗干扰能力而又适合于拾取人体体表心音的高灵敏度音腔,实验

关 键 词:冠心病  心音  小波分析  RBF网络  诊断  神经网络

The study of the noninvasive detection system of early coronary artery disease based on neural networks and the application of wavelet transform to heart sounds
YE Xue song.The study of the noninvasive detection system of early coronary artery disease based on neural networks and the application of wavelet transform to heart sounds[J].Journal of Zhejiang University(Engineering Science),1999,33(2):123-128.
Authors:YE Xue song
Abstract:This paper studied theories of the blood flow dynamies in coronary artery and acoustics induced by turbulence and then provided a new method to detect coronary artery disease(CAD) which was based on the application of wavelet transform to whole cycle's heart sounds,the features obtained form the wavelet transform of the two groups' heart sounds of the CAD and the contrast were input into the neural networks as well as their physical parameters,after training,this diagnosis system can diagnose the CAD automatically.In order to detect these weak physiology signals,this paper studied the transfer model sounds in the chest and developed a highly sensitive cavity having a great ability of anti noises to catch the heart sounds.The detection system adopted a highly sensitive sensor to detect signals from various parts of thorax of the CAD and the contrast.After pretreated,the high frequency signals in heart sounds were sampled by computer.It also developed a method of wavelet transform of heart sounds which was fit for the analysis of heart sounds because of its localization analysis both in frequency domain and in time domain.The average power ratios of the whole cycle's heart sounds to the diastolic period's heart sounds at different frequency segments were input into the RBF neural networks as well as the their physical parameters such as age,the condition of smoking and blood pressure.It has been proved by clinical practice that this diagnosis system provided a potential possibility for the detection of CAD,the analysis of the medicines effect on coonary artery and the detection of the conditions of coronary artery after its surgical operation.
Keywords:coronary artery disease  heart sounds  wavelet transform  RBF neural networks  model classification  
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号