【摘要】 目的：本文研究一种利用微机神经网络识别冠心病心电信号的模式识别方法。方法：本项研究利用心电电极、12位模／数转换器和奔腾－II／233多媒体微机系统等硬件采集心电信号，采用改进的反向传播人工神经网络方法进行冠心病心电信号模式识别。该神经网络采用三层结构：输入层、隐含层和输出层。其中输入层的节点数为4，隐含层的节点数为3，输出层的节点数为2。从心电信号中提取如下四项特征参数作为输入层的输入量。这四项特征参数是：R－R间期、QRX波间期、XT段斜率和XT段电平。输出层的节点输出模式识别后的结果。结果：本项研究所采用的BP神经网络能对16例训练样本正确地进行模式识别，准确地区分冠心病和非冠心病病例。经训练后的神经网络也能对10例未参加训练的样本正确地进行模式识别，准确地区分冠心病和非冠心病病例。神经网络模式识别的结果和临床专家诊断结果完全一致。结论：它是一种新的有应用前景的冠心病心电信号模式识别方法。更多还原

【Abstract】 ： Objective： The paper researches a method of pattern recognition for Coronary Heart Disease＇s ECG using a microcomputer－based neural network． Method： It samples ECG signal using a set of ECG electrodes, a 12－bit A／D converter and a P－II／233 microcomputer,et al,in this research,an improved BP neural network arithmetic for Coronary Heart Disease＇s ECG recognition has been developed．The neural network has three layers, input layer, hidden layer and output layer．There are 4 nodes in the input layer 3 nodes in the hidden layer and 2 nodes in the output layer． It picks up four parameters from ECG as input parameters of the input layer． These four parameters are interval of R－R segment, span of QRX, slop of X－T segment, altitude of X－T segment． The nodes in the output layer output the results after pattern recognition． Result： The BP neural networks in the paper can correctly recognize for 16 trained samples． It accurately distinguishes samples of Coronary Heart Disease from samples of non－Coronary Heart Disease． The trained BP neural network can also correctly recognize for 10 non－trained samples． The result of pattern－recognition with neural network method is identical with diagnostic result by clinical experts． Conclusion： It is a new and great benefit method for pattern recognition for ECG signals of Coronary Heart Disease．更多还原