基于卷积神经网络的心律失常分类方法

Arrhythmia Classification Based on Convolutional Neural Network

使用一种组合式心拍分割方法,利用带通滤波对原始心电数据进行降噪处理,实现QRS波群定位和心拍截取;设计7层的一维卷积神经网络(convolutional neural network,CNN)模型,对正常搏动(N)、左束支传导阻滞(L)、右束支传导阻滞(R)、室性早搏(V)4类心拍数据自动分类检测,从而完成4类心律失常的分类。以MIT-BIH心律失常数据库47条数据进行训练,结果显示,其准确度为99.00%,召回率为99.08%;与相关文献的研究方法对比,本方法具有较高识别精度,能有效解决人工对心电图识别的误诊、错诊问题。

In this paper,a combined beat segmentation method is used to denoise the original ECG data by band-pass filtering to realize QRS complex location and beat interception.A 7-layer one-dimensional convolutional neural network(CNN)is designed for automatic classification and detection of four types of cardiac beat data—normal beat(N),left bundle branch block(L),right bundle branch block(R)and ventricular premature beat(V).47 pieces of data from MIT-BIH arrhythmia database are trained.The results show that the accuracy is 99.00%and the recall rate is 99.08%,making a high recognition accuracy compared with the relevant literature,which can effectively improve the missed diagnosis and misdiagnosis of manual ECG recognition.