Analysis Of ECG Signal And Development Of Virtually Automatic ECG Analyzer

Electrocardiogram (ECG) is a synthetic reflection of the heart electricity on body surface. It is very valuable to analyze the ECG signal to diagnose the heart disease. At present there are many analysis methods on ECG signal, yet accurately locating the characteristic waves of ECG is still difficult. So there is room for innovation in theory. As the tools of detecting ECG information, automatic ECG analyzers mainly comes from oversea presently, and the products with complete independent intellectual property are very lacking inland, which conduces its higher price and lower market occupancy. Virtual instrument with predominant performances represents the future mainstream direction of measurement instrument. Developing the virtually automatic ECG analyzer with complete independent intellectual property and pushing them to market are obviously worthy. Based on the demands of academic research and market, author has done research work on signal process, ECG analysis methods and developing the automatic ECG analyzer. In this thesis the currently preprocess and analysis methods of ECG signal and the present status of developing automatic ECG analyzer are first summarized. Then some fast preprocess algorithms usually used by ECG analyzer are analyzed, and their performances are evaluated. The FIR digital filter based on window function, improved adaptive coherent template algorithm, and wavelet transform and LMS adaptive united algorithm are separately researched and designed. They are applicable for different ECG analysis. The FIR digital filter based on window function is faster, real-time and efficient. Baseline drift and power line interference can be filtered by the filter at the same time. It is applicable for ECG monitor to collect and analysis ECG signal immediately for long time. Power line interference can be cancelled by the adaptive coherent template algorithm, yet it can lead the low critical frequency not lower enough and affect the integrality of low frequency information. The algorithm is improved in this thesis through the second cascade filter. According to the frequency relations between 50Hz template signal and 100Hz template signal, a new filter which only cancels 50Hz is structured. And then it is cascaded with the filter used in previous adaptive coherent template algorithm to cancel power line interference. The method can solve the conflict between critical frequency and power line interference. If the ECG signal is transformed by wavelet and the approach information of higher scale is set zero, then the base line drift will be cancelled in the reconstructed ECG signal. But the method will result in the distorted low frequency component like P and T waves. In this thesis an innovative algorithm combined wavelet transform and LMS is presented. The approach information at higher scale is used as the input of LMS algorithm and constant as reference value, the new signal is obtained through the LMS operation, which is used as the new approach information to restructure the ECG signal. Massive experiments show that the method can cancel the base line drift and remarkably reduce the degree of distortion of low frequency component, which provide a better foundation to analyze P and T waves. Then the methods of extracting ECG characteristic information are principally researched. Two algorithms identifying ECG characteristic waves which maturely applied in ECG analyzer are analyzed and improved, and validating experiments are done. An in-depth research on the principle of wavelet transform detecting singular point and some algorithms identifying ECG characteristic waves based on wavelet are also researched and evaluated. Then author innovatively presents the united algorithm based on wavelet to identify and locate the ECG characteristic waves. The principle is: the shapes of transformed ECG characteristic waves are classified first, and then the shapes of detected characteristic waves and peaks are determined by the number of modulus maximum pairs produced by transformed ECG signal at specified scale. Finally the onset and offset of characteristic wave will be determined by amplitude and slope criterion. The differentia with common wavelet algorithm is that the united algorithm first determines wave shape and second locates onset and offset, and then different methods are applied in response to different shapes. The precision to locate ECG characteristic waves, especially to P and T waves, is greatly improved by the algorithm. For a long time it is difficult to identify and locate the low amplitude and multi-shape ECG characteristic waves like P and T wave, Using the algorithm can solve the problem. Massive comparison experiments have been down, which shows that the algorithm is exact and reliable. Arrhythmias and heart rate variability (HRV) have been researched deeply in this thesis. Some familiar arrhythmias have been sum up, and seven criterions have been put forward. HRV is researched from time and frequency domain separately. Power spectrum density of HRV signal is analyzed from time-frequency domain using three-dimensional "color spectrum". The innovative method can directly reflect the energy value of HRV signal at some time and some frequency, which provides fordoctors and researchers more information to analyze ECG signal. The developing technique of virtual instrument (VI) has been research. Component technique is introduced to develop VI, and the developing method of VI based on component technique is presented, which can effectively improve the speed and quality of VI development. Based on above research achievements, author develops the virtually automatic ECG analyzer. The instrument has not only all functions of hardware-based ECG analyzer, but also has richer function, smarter measurement and lower price. Experiments with clinic datum show that compared with results from manual detection the results is validate and steady , and error is within clinic permit. So it can fulfill the clinic requirement. Data-base management function is also developed in the ECG analyzer, which can provide more information and datum for researching and diagnosing cardiovascular disease. Lastly the next research works are presented at the end of the thesis.