小波包与混沌集成的心音特征提取及分类识别  

Heart sound feature extraction and classification based on integration of wavelet packet analysis and chaos theory

作　　者：郭兴明[1] 丁晓蓉[1] 钟丽莎[2] 雷鸣[1] 翁渐[1] 

机构地区：[1]重庆大学生物工程学院,重庆400044 [2]泸州医学院,泸州646000

出　　处：《仪器仪表学报》2012年第9期1938-1944,共7页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(30770551);中央高校基本科研业务费资助(CDXS102300);重庆市新型医疗器械重大科技专项(CSTC,2008AC5103)资助项目

摘　　要：针对心脏疾病诊断过程中心音识别的难点,提出了一种结合小波包分析及混沌的特征提取的心音识别方法。首先分析统计了心音信号的小波包能量特征,然后选取小波包分解中能表征心音信号特征的分量进行混沌分析,计算了最大Lyapunov指数和关联维数;最后以这些参数构成特征矢量作为支持向量机的输入,对临床采集到的65例正常及有心脏疾病的心音信号进行识别分类。结果表明,结合小波包分析和混沌的特征参量,较传统的分类识别方法具有更高的识别精度,说明非线性混沌特征能够较有效地表征心音信号的特征,为下一步临床心脏疾病的更准确诊断奠定了基础。Aiming at the difficult point of heart sound recognition in heart disease diagnosis process, a new method of heart sound feature extraction and classification is presented based on integration of wavelet packet analysis and chaos theory. Firstly, the wavelet packet energy features of heart sound signal are acquired. Then the main components that can represent the features of the heart sound signal in the wavelet packet decomposition are analyzed with chaos theory, and the largest Lyapunov exponent and number of correlation dimensions are calculated. Finally, these parameters constitute the feature vectors, which are used as the inputs of support vector machine to classify the healthy heart sound signals and those with heart disease of 65 cases. The results show that using the proposed method the recognition rate is higher than that of traditional classification method, which indicates that the nonlinear chaotic feature can characterize the heart sound signal effectively. This method lays a foundation for accurate diagnosis of heart disease in the future clinic applications.

关 键 词：心音信号 小波包 混沌理论 能量特征 分类 

分 类 号：TN911.72[电子电信—通信与信息系统]