基于自组织联想小波神经网络的QRS波聚类方法

A Method for Clustering of QRS WavesUsing a Selforganized Associative Neural Network

介绍一个由小波变换和神经网络结合的自组织联想神经网络模型，它可用在心电信号的自动分析中对ＱＲＳ波进行聚类．还对算法实现中的一些具体问题及详细的编程步骤进行了讨论．用ＭＩＴ心电数据库对此算法的性能进行评估的结果表明，此神经网络对ＱＲＳ波具有较高的分类精度，尤其适合于室性早搏等异位心搏的检测．

This paper presents a self-organized associative network model which is constructed by combining the wavelet transfer with the neural network. This network is capable of clustering the QRS waves in the automatic analysis of ECG signals. Several problems in the implementation of this algorithm and the detailed steps to develop the programs are also discussed. The results of evaluation of this algorithm by using the MIT ECG database have shown that this neural network has higher clustering accuracy, and that it is specially useful the detection of such ectopic beats as premature ventricular contraction (PVC). 