基于CNN和ET的智能ECG识别方法

ECG identification method based on CNN and extreme random tree

心电图(ECG)是检测心血管疾病的重要依据之一,通过对各类心电图的实时分析,可以达到检测被测者房颤及心脏健康情况的目的。采用基于卷积神经网络(CNN)和极端随机树(ET)混合模型的心电信号分类方法,通过连续小波变换对数据进行滤波处理,在此基础上通过CNN-ET混合模型,实现了心电信号的分类。方法结合了CNN对一维数据的强大表征能力,通过ET降低了异常值影响,预防了过拟合问题,具有较强的泛化能力。将所提出的方法在MIT-BIH数据集上进行了测试,在5类心电心拍次数不平衡问题检测中准确率达到99.95%,与现有方法相比,该改进方法进一步提高了ECG信号分类的精确度。

Electrocardiogram(ECG)is one of the important basis for detection of cardiovascular disease.Through real-time analysis of ECG,the purpose of detecting atrial fibrillation and heart health condition can be achieved.The ECG signal classification method based on the convolutional neural network(CNN)and extreme random tree(ET)hybrid model is adopted in this paper.Firstly,the data is filtered through continuous wavelet transform technique,then the ECG signal classification is realized through the CNN-ET hybrid model.This method combines a powerful representation of the CNN for one-dimensional data,reduces influence of abnormal data through ET,and prevents overfitting problems,thus it has a good generalization capability.The proposed method has been tested on the MIT-BIH data set and the accuracy rate reaches 99.95%in the five types of ECG heart beat frequency imbalance detection problems.Compared with existing methods,this algorithm further improves the accuracy of ECG signal classification.