Study On Recognition Algorithm Of Myocardial Infarction Based On Support Vector Machine And Polynomial Fitting

Cardiovascular disease is one of the most important diseases that threaten the health and lives of all human beings and acute myocardial infarction has become one of the most important causes of death in Chinese cities and rural areas.In addition,the electrocardiogram(ECG)has gone through 100 years of improvement and research,the relialbility of its diagnosis,the simplicity of detection methods,as well as the harmlessness to the patient have made it play an irreplaceable role in the clinical diagnosis of cardiovascular disease and ECG is now an important tool for the diagnosis of myocardial infarction in clinical practice.The performance of Myocardial infarction ECG is mainly abnormal Q wave,ST segment,P wave,etc.,so this article is based on the ECG to distinguish the MI signals and healthy signals.This article is mainly about the following aspects:① Because of the weakness,low frequency characteristics,instability and randomness of ECG signals,noises are inevitable during the ECG signal acquisition process,which mainly contains baseline drift,power frequency interference,EMG interference.So before dealing with ECG signals,it is necessary to take an appropriate way to remove noise interference.In this paper,the method of wavelet decomposition is used to deal with ECG signals at different scales,and a good denoising effect is achieved.② Feature Point Detection of ECG Signals.The first and also the most important is the detection of QRS wave group,and the method used to detect the R wave is the differential method,so as to balance the difficulty of obtaining the recognition rate while minimizing the operation difficulty.As to the detection of QRS wave starting and ending point,P wave,T wave and other band detection,the method is local transform method,based on the detection of R peak,leading to quickly locate it.Apart from the signals with most dramatic changes,the detection accuracy is up to 98%.③ Feature Extraction Method Based on Polynomial Fitting.The principle of this algorithm is to use the polynomial function to fit the ECG after processing,and the number of polynomials of the fitted polynomial function and the scaling parameters involved in the middle are used as the eigenvalues to be processed.In this paper,in order to maximize the fitting curve to approximate the original waveform,the method of two separate fitting curves is used.The main problem to be solved is the determination of the highest order of the polynomial function.④ Classification Based on Support Vector Machine.Because the support vector machine has better generalization and robustness,the classifier used in this paper is support vector machine,which mainly involves the problem of support vector machine kernel function and kernel function parameter selection.