ECG Automatic Analysis Based On Wavelet Transforms

Electrocardiograph (abbr. ECG) is the biological reactions in the body surface in the process of heart activity electrical signals generated. It is an important means of clinically diagnosis of heart disease. As a result, the manual analysis of ECG is much subjective, and the workloads are huge, too. With the development of computer technology, technology demand of automatic analysis of ECG starts to become a reality. Automatic ECG analysis depends on the accuracy of ECG parameters. As small ECG signal amplitude, low frequency, the collection process is vulnerable to the external environment and the body's own influence. Collected ECG signals often mixed with a lot of interference, and sometimes the normal ECG signals were disturbed, which made the diagnosis of disease much inconvenience. Meanwhile, different patients also made the collected ECG signal different. How to extract the useful ECG signal out of the clutter and getting the accurate ECG signal parameters are the key points of automated analysis of ECG signal.In my dissertation, the method of lifting wavelet in the ECG signal processing and feature extraction is used. The lifting wavelet relative to the first generation wavelet occupies less system resources, so it is suitable for promotion in practical applications. First, ECG signals are denoised by using wavelet thresholding. Wavelet is used for decomposing ECG signal into layers, in order to removing the significant noise level and retaining the useful signal level. The wavelet reconstructions remove the noise of ECG Second, analyzing the waveform characteristics of denoised ECG signals. Modulus maxima method is used for detecting the exact location of QRS wave group, P wave and T wave, so that the ECG parameters are obtained. Finally, formulating the testing standards, and using the detected ECG signal parameters for automatic analysis. The tests in MIT-BIH database prove that the algorithm is feasible.