An electrocardiogram (ECG) classify system based on the features of the ECG and neural network classification, which is the simulation of the real world situation, was present. First, a modified approach of the linear approximation distance thresholding (LADT) algorithm was studied and the features of the ECG were obtained. Then a neural network which can classify the multi-lead ECG data was trained with these features along the theory of the ECG diagnosis and the situation of ECG diagnosis in practice. Thus take a new idea for the ECG automatic analysis. The algorithm was tested using several ECG signals of MIT-BIH, and the performance was good. The correct rate of the trained wave is 100%, untrained is 78.2%.
Algorithms ; Databases, Factual ; Electrocardiography ; classification ; Neural Networks (Computer) ; Signal Processing, Computer-Assisted

In this paper, we using Mexican-hat wavelet transform to detect characteristic points of ECG signal based on the characteristic points corresponding with the extremes of Mexican-hat wavelet transform. It offers a new detection method of ECG signal analysis. This method is simple and it is proved to be accurate and reliable. The correct rate of QRS detection rate examined by the MIT-BIT arrhythmia database rises up to 99.9%.
Algorithms ; Electrocardiography ; Humans ; Signal Processing, Computer-Assisted