摘要
针对心音信号的非线性、非平稳特征和心音识别准确率不高且分类速度较慢的实际情况,提出一种经验模式分解(Empirical Mode Decomposition,EMD)近似熵(Approximate Entropy,ApEn)结合支持向量机(Support Vector Machine,SVM)的心音分类识别方法。通过EMD方法将非平稳的心音振动信号分解成若干个平稳的固有模态函数(Intrinsic Mode Function,IMF);利用互相关系数准则对IMF进行筛选,计算所筛选IMF的近似熵构成特征向量;将特征向量输入SVM分类器进行分类识别。对临床采集的心音样本按该方法进行测试,结果表明,该方法能有效地用于心音识别。
Aiming at the non-stationary and non-linear characteristics of a heart sound and the difficulty to gain higher accuracy and classification speed, a new pathological diagnosis method based on empirical mode decomposition （EMD） approximate entropy （ApEn） and support vector machine （SVM） was proposed. Firstly, the heart sound signals were decomposed into a finite number of intrinsic mode function （IMF）. Then, the ApEns of five IMFs were used to form eigenvectors. Finally, the eigenvectors were put into a support vector machine categorizcr for automatic discrimination between normal and abnormal signals. The clinical data experimental diagnosis and test results showed that the approach proposed can identify the pathological heart sound effectively.
出处
《振动与冲击》
EI
CSCD
北大核心
2012年第19期21-25,共5页
Journal of Vibration and Shock
基金
国家自然科学基金资助项目(30770551)
重庆市新型医疗器械重大科技专项(CSTC
2008AC5103)
关键词
经验模式分解
心音
近似熵
支持向量机
empirical mode decomposition （ EMD ）
heart sound
approximate entropy （ApEn）
support vector machine （SVM）
