心电图检测与辨识的人工神经网络方法

Detection and Identification of Electrocardiogram(ECG) with ANN

应用人工神经网络(Artificial Neural Network,ANN)算法对MIT-BIH心电数据库中的数据进行检测,对波形辨识算法做初步研究。设计三层神经网络结构:输入层、隐含层和输出层。从心电信号中提取4项特征参数作为输入层的输入量,并对MIT-BIH心电数据库中的15例心电数据进行了检测。表明该算法对QRS波总体检测灵敏度为98.96%,检测真阳性率为99.93%,对室性异位博动检测灵敏度为94.97%,检测真阳性率为98.72%。实验证实该神经几乎络算法对心电波形辨识非常有效。

This article analyzed the data in the MIT-BIH ECG database with the method of ANN and did a primary study of the algorithm of waveform idenfication. It designed a network that includes input layer, hidden layer and output layer, and distilled four characteristic parameters as the input data, and evaluated this algorithm with 15 data from MIT-BIH EGG database. The result shows that the sensitive percentage of QRS detecting is 98.96%, the positive percentage value is 99. 93 %, the gross ventricular entopic beat（VEB） sensitive percentage is 94.97 %, and the VEB positive percentage value is 98. 72%. The experiment confirms that the algorithm of ANN is very effective to identify ECG waveforms.