Detection Method Research For T Wave Alternans In Electrocardiogram

As the highest morbidity and mortality of diseases, cardiovascular disease has been widely concerned. The related analysis and treatment has also been paid great attention. As a weak ECG alternans, TWA has aroused attention during the inspection and analysis of ECG. TWA is a non-stationary variability in ECG phenomenon, that T wave amplitude changes in shape-by-cycle alternating. TWA related to ventricular arrhythmias and sudden cardiac death, and TWA has become a non-invasive and independence predictor of cardiovascular disease.In this dissertation, the automatic detection and analysis of TWA are studied, and the work accomplished is listed as following.First of all, the preprocessing of ECG is studied. In this dissertation, moving average filter and 50Hz notch filter are used to remove high frequency noise and power line interference of ECG, respectively. Adaptive filter has dynamic adaptability and good effect of filtering; and it would choose different filter parameters according to the characteristics and noise level of signal. So the adaptive filter which combines the advantages of high-pass filtering and cubic spline used to remove baseline drift of ECGThe fiducial point positioning in ECG is studied. Dynamic adaptive differential threshold method is used to detect the position of R-wave peak. R-wave peak is considered as a reference point, and a window is used to search and position the T-wave peak. An effective approach is proposed to identify T-wave inversion. A new algorithm is put forward to construct T-wave analysis matrix. The experience value is applied to choose an accurate location of T-wave window. The ECG waveform segments are extracted with scanning window. Mean square deviation and template are used to select the final location of T-wave. Finally, the T-wave analysis matrix is built as the accomplishment of the fiducial position work.Systematic analysis and TWA detection are studied in the dissertation. First, the characteristics and detection difficult of TWA is discussed, and the detection methods of TWA are compared, the principles of various methods are analyzed in detail, these methods are simulated and the strengths and weaknesses of variety detection methods are illustrated based on simulation results. A method which combines the spectral analysis method and correlation method is designed to detect TWA. Singular value decomposition is used to calculate the relevant index, effective avoid the interference of noise on the test results, and improve the accuracy of detection.The ECG signal processing are compared with the traditional methods, there are a lot of the characteristics and advantages. At the process filtering, for the requirements of high-quality ECG, reasonable method is designed to filter out the baseline drift and maintain the quality of ECG effectively. At the part of the fiducial point positioning, improve differential threshold method can position R wave peak accurately; finally, a comprehensive and effective algorithm is put forward for TWA detection.