【作者】 胡健； 张凯；

【Author】 HU Jian;ZHANG Kai;North China University of Technology;

【机构】 北方工业大学；

【摘要】 文中提出了一种基于局部均值分解与样本熵相结合的频域处理方法用于心电信号的特征提取,并研究了其在心电信号异常类型分类中的实施与应用。对于按心拍分割的心电信号,利用局部均值分解方法将原始信号分解为若干个PF分量,计算前3个PF分量的样本熵值,并将这3个样本熵值与心电信号的RR间期、QRS波群宽度、R波峰值一同作为输入元素构造3个六维的特征向量。使用支持向量机模型进行预测分类,并通过启发式算法调取最优参数来提高识别心律失常类型的准确率。实验结果表明,该方法能够对心律异常类型进行分类,准确率可达98.67%。更多还原

【Abstract】 This paper proposed a frequency domain processing method based on local mean decomposition combined with sample entropy for ECG signal feature extraction,and its implementation and application in the abnormal classification of ECG signals were studied.For ECG signals segmented by heart beat,the original signal is decomposed into several PF components using a local mean decomposition method,the sample entropy values of the first three PF components are calculated,a 6-dimensional feature vector is constructed with the RR interval of the ECG,the QRS complex width,and the R-wave peak as input elements.The support vector machine model is used to predict classification and heuristic algorithm to obtain the optimal parameters to improve the accuracy of recognition of arrhythmia type.The experimental results show that this method can classify the type of abnormal heart rhythm with an accuracy of 98.67%.更多还原