A Study Of The Technology In The Detection Of Transient T-wave Alternans

Recently, the incidence of cardiovascular diseases is higher and higher. The sudden cardiacdeath has been one of the most common reasons in fatal diseases, which is a serious threat tohuman life. By researching, it is found that malignant arrhythmia is the main cause of suddencardiac death. T-wave alternans(TWA) , consisting of beat-to-beat alternation in the morphologyof the T wave, is a index of malignant arrhythmia and sudden cardiac death. As most of TWA areat microvolt level, we had to use some digital signal processing technology.Several famous algorithm of detection of TWA were discussed in this paper, including thefast-Fourier-transform spectral method, complex-demodulation method, correlation method,modified-moving-average method and so on. The time-domain correlation method was themain emphasis of this study. In this paper we have completed three aspects as follows:The first is preprocessing the ECG signals as well as calibrating the ECG feature point. Weuse simple integer coefficient method, digital filtering and double-orthogonal wavelet functionbior2.2 for ECG filtering process, in order to remove the power interference, EMG interferenceand baseline drift. Then we can get ECG signals with clear characteristics. And using the wavelettransform to the de-nosing ECG, QRS complex waves were located in 2-3 scales by modulusmaxima algorithm.Then the T wave interval was recognized from the ECG, using the modulus maximaalgorithm. According to the R-peak as well as the empirical values of T-wave beginning andT-wave end, the temporary T-wave intervals were identified on the 2-4 scale. The T-wavemodulus maxima pairs of every beat were found in those temporary T-wave intervals. And thenthe T-wave morphology was determined on the basis of modulus maxima pairs’quantity andplus-minus. As different T-wave morphologies correspond with different detection methods ofthe T-wave beginning and T-wave end, classification detection of T-wave interval was used toimprove the detection accuracy. Matlab was used to simulate data files from QT database.At last, the detection of TWA was based on correlational method in time domain.Time-domain correlation method (CM) detects TWA by computing, for each consecutive T wave, an alternans correlation index based on a crosscorrelation technique. Firstly, the method of Twave interval alignment was put forward based on correlational analysis. Then, calculate themean of several continuous beats. And this average T wave was considered as a template fordetecting continuous beats, as well as the maximum cross-correlation index was figure out.Finally, this index was compared with the maximum self-correlation of the template, and theresult was used to decide whether there were TWA or not. Matlab was used to simulate data filesfrom American MIT/BIH standard arrhythmia database and European ST-T ECG database.In this paper, we focused on the time-domain correlation method, one method of TWAdetection. And before this detection, a lot of work had been done to identify the T wave interval.Correlation method can track the transient TWA episode. It can detect and quantify stationaryand nonstationary episodes of TWA in ECG recordings acquired during sinus rhythm. However,this method is easy to be affected by noise, which will largely reduce its clinical applicationvalue. We are favor to reduce the fallout ratio of the correlation method by improving theaccuracy of the T wave interval.