The Study of Electrocardiograph Based on Radial Basis Function Neural Network

In this paper we introduce a set of adaptive signal procedure techniques which could be used. Firstly, we introduce discrete wavelet transform and extract the characteristics of Electrocardiogram (ECG) optimization. Then, we make use of Radial Basis Function (RBF) neural network to achieve the classification of ECG and to compare the performance of their respectively. Among which two types of ECG arrhythmias which obtained from MIT-BIH ECG Arrhythmias database are normal beat and paced beat respectively. The experimental results show that the use of RBF neural network algorithm to classify the 8-dimensional feature vector training under the db2 wavelet performs as the optimal classifier, and the overall classification accuracy rate is more than 90%.