摘要
对心电信号的实时、准确识别在临床上具有重要意义。研究基于小波变换自动识别室性早搏(PVC)和房性早搏(APB)的方法,首先对信号进行Marr小波变换并提取信号在小波域上的特征参数,构建时频域特征向量,然后使用径向基核SVM进行训练,研究模型参数的选取对训练结果的影响。使用MIT-BIH心电数据库中的数据进行测试,结果表明:在小样本的情况下,建立的模型对正常心电、房性早搏和室性早搏的识别具有较高的准确率。
The real-time and accurate examination of ECG has important clinical significance. In this paper,we study the method of wavelet transform-based automatic identification of premature ventricular contractions（PVC） and atrial premature beats（APB）. First,Marr wavelet transform is applied to the signal,its characteristic parameters on wavelet domain are extracted,and the time-frequency domains eigenvectors are constructed; then the support vector machine（ SVM） with RBF kernel function is employed for training,the influence of model parameters selection on training performance is explored. The data of MIT-BIH arrhythmia database is used for the test,results show that the model proposed in this paper has high accuracy rate in identifying PVC,APB and normal ECG in the case of small sample size.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第8期182-185,共4页
Computer Applications and Software
关键词
ECG
房性早搏
室性早搏
小波变换
支持向量机
ECG Premature ventricular contraction Atrial premature beat Wavelet transform SVM
