窦性心律与心室纤颤信号分类的研究

Study on Classification of Sinus Rhythm and Ventricular Fibrillation Signals

目的实现正常的窦性心律与心室纤颤信号的分类,从而检测出心室纤颤信号。方法该算法基于支持向量机技术,Hurst指数和相空间重构算法。待检测信号取自BIH-MIT和CU数据库,首先对待检测信号进行预处理,然后取滑动窗长度为3s计算出心电信号段的动力学指标值Hurst指数与相空间重构算法中的d值,最后把这两个参数作为特征向量输入到事先设计好的二分类支持向量机中,从而实现分类。结果成功实现了心室纤颤信号的分类,并通过计算该算法的灵敏度、特异性、预测性和准确度且与其他算法比较,可得新算法总体准确率优于其他算法。结论该算法可用于心电信号的检测,进行算法优化之后可嵌入到心电检测仪器中实现应用。

Objective To classify normal sinus rhythm （SR） and ventricular fibrillation （VF）,accordingly detection of VF. Methods A new algorithm was proposed,based on support vector machine （SVM）,Hurst index,and time-delay algorithm [phase space reconstruction （PSR）]. The data sets were taken from the BIH-MIT database and the CU database. After data preprocessing,the Hurst index of dynamics and d value of space reconstruction algorithm of ECG segment from 3 s length slide window were calculated. Two parameters and as feature vectors, were put into SVM, which had been designed for two classification. Finally, VF and no VF were classified. Results VF and no VF classification was realized. Besides the quality parameters sensitivity, specificity, the positive predictability and accuracy of this new algorithm were calculated and compared with others. It was showed that the proposed algorithm had a high detection quality and outperformed all other investigated algorithms. Conclusion This new algorithm can be used for detecting VF, after optimization it can be embedded in portable electrocardiograph （ECG） monitor to find application.