A Study Of Poincare Scatter And Spectral Method In The Detection Of T-wave Alternans

In recent years, a large amount of material indicates that ventricular arrhythmia is one of themain reasons leading to sudden cardiac death, and the microvolt T-wave alternans (MTWA) is animportant index for ventricular arrhythmias prediction. MTWA is a phenomenon of electrocardio variation that beat to beat variation of T-wave morphology and polarity at constant heartrate is embodied in neat cardiac rhythm.In accordance with the statistical method difference of TWA detection, the methods withpathologic significance of detecting MTWA are divided into three kinds: Short Time FourierTransform (STFT), symbol transform and nonlinear methods [2]. The Spectral method(SM)which is one kind of STFT is the most mature.Nonlinear methods include Poincare mapping(PM), moving average (MMA) method, etc.This paper mainly adopts Spectral method and Poincare mapping method for ECG signalanalysis, and uses the United States of America MIT/BIH arrhythmia database and the EuropeanST-T ECG database for simulation. The process of the TWA detection is as follows:First is demarcating the ECG characteristics. First of all, in the ECG signal pretreatment, thesubject used the simple integral coefficient method to remove the50HZ interference, employedzero phase digital filter for the removal of ECG baseline drift and the threshold denoisingmethod of bior2.2wavelet transform to remove muscle power interference, therefore ECG signalof the obvious characteristics were obtained. Then, for QRS wave group characteristicscalibration, this paper adopts Marr wavelet and in accordance with the a’trous algorithm, ECGsignal filtered was processed for certain scales of the wavelet transform. At that moment, eachpeak of Q wave, R wave and S wave was calibrated in the appropriate scale, and according to Rpeak location, we determined the T wave location.Secondly, T wave alternans detection. On electrocardiogram research, this paper adopts twomethods: Spectral method and Poincare mapping method. To begin with, using the ECGcharacteristic quantity extracted, we chose a improved T wave window for each heartbeat cyclein the spectrum analysis.In addition,7points were averagely collected in a window andcontinuous128heartbeat cycles was chosen, thus a7*128sequence was obtained. Subsequently,the7*128sample-point amplitudes were changed into a power spectrum by fast Fouriertransform.The power spectrum were superimposed and normalized. Subtracting the backgroundnoise, the value of spectrum at the0.5cycle per beat was the index to judge whether the TWA existed. If there were TWA, the alternate amplitude was the square root of the value of spectrum.Then,7*128sequence with subtraction between adjacent beats of initial samples was used todraw a Poincare scatter and we calculated the vector angle index by the Poincare scatter.Through a large number of experimental studies, the vector angle index range was obtained tojudge the existence of TWA.Thirdly, the simulation results of two methods were analyzed. Through a large number ofsimulation tests, the data attained by the Spectral method and the Poincare mapping method wasanalyzed and using MTALBE7.0software, we got the second curve fitting of the VAI and Vtwadata, the fitting degree and the cross correlation coefficient between them. Thus, the reliabilityof Vector Angle Index was evaluated.The article presents the Poincare mapping (PM) based on nonlinear dynamic theory todetect MTWA and compares the result with that of Spectral method. Vector Angle Index andVtwa were used to judge whether the TWA existed.