Research On The Algorithm Of Analysis And Diagnosis Of Electrocardiogram

It can help us to understand some physiological phenomena and pathological mechanism of the disease through the study of ECG data.Currently, analysis and diagnosis of ECG is primarily related to two aspects:the detection of characteristic wave and automatic diagnosis.Recently, an increasing number of methods have been applied to detect characteristic wave, such as wavelet analysis, neural networks. The concept of chaos has been applied in automated diagnosis of the disease more and more, because the heart is a chaotic system in nature. We can get better results than other methods if we start from the concept of chaos.This article does research mainly on these two aspects: the detection of characteristic wave and automatic diagnosis of disease.In the detection of characteristic wave, the article focuses on the detection of R-wave. First of all, the effects of noise,P-wave and T-wave are filtered using wavelet filting. And then, we calculate the Hilbert transform after calculating first-order differential of ECG .Considering with the original ECG signal,we set the rules of the variable threshold to detect R-wave.At last, a recall algothim according to the physiological character of ECG is carried on for the undetected R-wave and errors.And we use MIT-BIH arrhythmia database to verify the algothim. And using detected R-wave position as an auxiliary, we make use of permute entropy applied to the detection of ventricular arrhythmias in achieving a high detection rate.In automated diagnosis of ECG, this article extracts multiple features from ECG thought the use of non-linear signals : a divergent path from the phase space, extent of the distribution of points in phase space density, phase space trajectory of the movement itself, as well as power the complexity of the system and so on. We combine recurrence quantification analysis and chaos parameters to form a Multi-feature vector.At last, we use SVM to classify four kinds of ECG: normal sinus rhythm,ventricular tachycardia,ventricular fibrillation and atrial fibrillation and Achieved a high accuracy rate.