定量递归分析和近似熵在心音分类识别中的应用

Application of Recurrence Quantification Analysis and Approximate Entropy in Classification and Recognition of Heart Sound

为了有效利用心音信号的非线性特征信息对心音信号进行分类识别,提出一种基于定量递归分析和近似熵的心音特征提取方法。首先利用递归图对心音信号进行定性分析;然后,定量提取心音的非线性特征参数:递归率、确定率、近似熵构成特征矢量;最后将特征矢量输入二叉树支持向量机,对采集到的正常以及5类心脏瓣膜性心音信号进行分类识别。对于文中提取的非线性特征参数,通过统计学分析证明了其有效性。结果表明,该方法能有效识别心音信号。

In order to recognize different types of heart sounds effectively by nonlinear characteristics,a method of feature extraction is proposed based on recurrence quantification analysis and approximate entropy.Firstly,recurrence plots are applied to the qualitative analysis of heart sound signals.Then recurrence rate and determination rate are extracted and combined with approximate entropy to form eigenvectors.Finally,the eigenvectors are put into binary tree support vector machine （BT-SVM） for classifying and recognizing different types of heart sounds.The testing results show that the proposed approach can classify and recognize the pathological heart sound effectively.