Research On The Technology Of ECG Automatic Analysis And Diagnosis

In recent years, due to the improvement of people’s living standard and change of dietary structure, the disease incidence of heart and circulatory is increased year by year. With the development of the information technology and the computer science, more and more research on the question of heart disease diagnosis using signal processing are carried on. Among these, the research of ECG is particularly prominent. Because it is the direct reflection of heart’s activities, the ECG has great significance to the heart disease diagnosis.This article mainly aims at the automatic diagnosis technology of ECG signals, which can be divided into three parts:ECG preprocessing、feature point extraction and the diagnosis of abnormal ST-T period, meanwhile, in order to solve the problem during the ECG signal acquisition, this paper proposes the method of vectorcardiographicloop reconstruction and the projection of unknown leads.The mainly content of preprocessing is eliminating baseline and high frequency noise. This paper used a variety of methods to solve this problem, and compared the result. Finally we get the best pretreatment effect using smooth filter to remove the baseline, and wavelet threshold value method to remove the high frequency noise.Feature point extraction is mainly aims at the position of QRS complex peak、starting point and terminal point, which also includes the detection of P, T wave corresponding feature point. For QRS complex peak detection, firstly we introduced the two effect traditional detecting method, pointing out the defects after practicing, and then propose the algorithm of this paper, which use mathematical morphology to extract the contour of QRS complex group, for restraining tall T wave interference, and then combined the advantages with wavelet analysis and threshold detection. According to the actual data analysis, the result is great. According to the deficiency of the local transformation method, the new starting and terminal point detection method of QRS complex group is presented. Firstly we use mathematical morphology to integrate QS wave into R wave, and then use the threshold detection algorithm, which also get a good test results.The mainly content of ST-T period diagnosis is ST segment range、form and T wave abnormal high, two-way, for which we defined multiple characteristics, and classified the abnormal type by decision tree.In the last chapter, we use the ECG signal of any two leads to reconstruct the vectorcardiographicloop which in the same plane, with the plane geometry theory and the method of Schimidt orthogonalization, and then get the ECG signal of other unkown leads using the vectorcardiographicloop projection on the direction of the leads. Through the study of this chapter, we can get the vectorcardiographicloop from ECG signal, which provide more information for diagnosis. With the method of leads projection, we can greatly reduce the number of electrode used during ECG signal collection, which has great clinical significance.