Analysis Of J-wave Signal Based On HODCS

J-wave is used as a new diagnose index of electrocardiogram (ECG)ventricular repolarization, It has emerged as a marker of risk for heartventricular tachycardia, ventricular fibrillation and even sudden cardiacdeath. For clinical manifestations, there are different results when thedifferent J-wave forms. In recent years, the research about J-wave onidentify danger and benign form has been attention medically. At present,J-wave danger and benign form detection is limited to clinical stage byECG waveform to diagnose, both in China and abroad. it is confined todetect the signal from time domain amplitude, waveform and location,unable to form the judge standard systematically.In order to improve the detection accuracy of the heart disease whichcaused by J-wave signal, detecting and diagnosing J-wave signal timely,we need to make further analysis for J-wave signal. For normal ECGsignal with characteristics of cyclostationary, the method of based on thehigh order cyclostationary theory was used to extract and analyze J-wavesignal in this paper.Firstly, a complete process of signal extraction and analysis requiresthe accurate signal preparation. Currently, the ECG printouts are we can get from domestic hospital, however, the signal analyze software we wereused to experiment can only identify the digital information, therefore wemust convert the format of ECG signal with J-wave and realize of digitalpreparation of signal. Secondly, in the signal processing, we analyze ofthe physiological mechanism of J-wave signal and waveformcharacteristics of ECG signals, combined with the wavelet transform toposition ST-segment of ECG signal with J-wave. Then presented theseparation principles of blind source separation (BSS) based on thehigh-order degree of cyclostationarity (HODCS) to extract the completeand independent J-wave signal.Finally, on the basis of previous work, we improved the algorithmand tried to establish a more accurate BSS model. Using the arbitrarydimension Givens rotation matrix as the separation matrix to improve theextraction accuracy of J wave and the variable step size of learning rate toimprove the convergence speed of algorithm. Through the simulationexperiment results illustrate the effectiveness and superiority of thisalgorithm in the application of extracting J-wave signal.