Detection of malignant ventricular arrhythmias using multi-scale recurrence quantification analysis
In order to detect malignant ventricular arrhythmias(MVA)in a timely and accurated manner,we proposed a method that integrated multi-scale analysis of electrocardiogram(ECG)signals with recurrence quantification analysis(RQA).Firstly,the method leveraged fixed-frequency range empirical wavelet transform filter banks to decompose ECG signals into two sub-signals(0.5～10 Hz,10～30 Hz).Then,these sub-signals were subsequently mapped into recurrence plot(RP).Moreover,a cross recurrence plot(CRP)was constructed using the sub-signal in the 0.5～10 Hz frequency band with the pre-processed undecomposed ECG signals.Quantitative analysis features were extracted from the three RPs or CRP,and fed into the XGBoost classifier for feature ranking and fil-tering,and finally the selected features were used to classify the ECG signals.The MIT-BIH and the CU ventricular tachyarrhythmia database were used for the experiment.Under 10-fold cross validation,the sensitivity,specificity,and accuracy of the method for de-tecting malignant ventricular arrhythmias in 5 s ECG segments reached 97.40％,99.01％,and 98.69％,respectively.The results dem-onstrate the reliability of the proposed method in detecting malignant ventricular arrhythmias.
万方数据
维普
