Research On Method Of Arrhythmia ECG Signal Based On Series Decomposition And Symbolization

Cardiovascular disease is one of the diseases that threaten human health.In recent years,the incidence of cardiovascular disease gradually increased,seriously endangering people's lives.There are more and more patients suffering cardiovascular disease,and they are more and more young.Sudden cardiac death is the most serious clinical manifestations of cardiovascular disease.If the patient suffers sudden cardiac death without timely and effective treatment,they will lose their lives.For this reason,many countries' medical and health departments and research centers are researching this.Found by research,most sudden cardiac death was caused by ventricular fibrillation or ventricular tachycardia.For VF and VT,the clinical treatments are different.If the patients show ventricular fibrillation,defibrillation must be carried out immediately for them,which is the only treatment currently.If the patients suffer ventricular tachycardia,they need timely and correct drug treatment,in order to reduce the incidence of ventricular fibrillation or sudden cardiac death rate.If VT is diagnosed as VF,the patient will be subjected to unnecessary electric shocks,which can cause heart trauma.If the VF is diagnosed as VT,patients won't get timely defibrillation,and they will suffer the sudden cardiac death.At present,the monitoring algorithm can distinguish sinus rhythm and VT/VF well,but the VT and VF detection algorithm is still under study.According to the current study of the ECG signals,ECG can be regarded as the category of nonlinear signal.Analyzing ECG with nonlinear dynamics has obvious advantages.Based on the theory of nonlinear dynamics,this paper analyzed the symbolic time series analysis method and time series decomposition algorithm.Two new algorithms for VT and VF detection based on sequence decomposition and symbolic time series was proposed.Then proposed algorithms were verified by the experiment with the data set,and the experimental results were analyzed by the symbolic time series theory.The experimental result showed that the recognition rate of VT/VF was 97.82% with the method of EMD combining symbolic time series analysis.The recognition rate of VT/VF based on wavelet analysis combined with symbolic time series analysis method was 99.5%.Besides,in view of the shortcomings of two value coarse graining method for time series analysis,we proposed an improved method.The changes before and after the improved algorithm were compared by experiments.The improved algorithm and the unimproved algorithm are compared by the VT/VF recognition rate and the algorithm execution time.The experimental results showed that the computational complexity of the improved algorithm is reduced by 30 times compared the unimproved algorithm while the VT/VF recognition rate is basically the same as the unimproved algorithm.And the improved algorithm got better classification performance based on small samples.In other words,in the real-time system,the improved algorithm can respond to the input signal more quickly.The improved algorithm greatly reduced the amount of calculation,and the algorithm execution time is shortened,which is suitable for monitor,automatic external defibrillator(AED)implantation and embedded defibrillator(ICD)real-time monitoring system.Finally,we compared the improved algorithm with the single symbol time series decomposition method by two sides of the sample time and the symbolic level.The experiments showed that the improved algorithm has better recognition rate than the reference.It is illustrated that the fusion algorithm based on sequence decomposition algorithm and symbol time series decomposition algorithm is effective.The main innovations of this paper are as follows.First,we proposed the fusion algorithm based on EMD and the symbolic time series analysis method to detect the VT and VF.The results showed that the fusion algorithm has a high recognition rate for VT/VF.Secondly,we proposed the fusion algorithm based on wavelet analysis and the symbolic time series analysis method to detect the VT/VF.The results showed that the appropriate signal sampling rate was beneficial to dig the essential feature of the symbolic time series.The symbolic entropy is obtained by wavelet decomposition to detect the VT/VF which has a higher recognition rate than EMD.Finally,considering of the shortcoming of symbolic time series analysis,the coarse granulation problem is improved.The experiment showed that the recognition rate of VT/VF was basically the same,but the improved symbol entropy calculation time is reduced and the recognition rate of the improved symbol entropy is higher based on small samples.