论文部分内容阅读
目的研究一种基于多尺度化基本尺度熵(multiscale base-scale entropy,MBE)和希尔伯特-黄变换(Hilbert-Huang transform,HHT)的心电信号自动分类方法。方法首先利用离散小波变换对心电信号进行去噪预处理;其次利用多尺度化基本尺度熵进行分析,提取多个尺度下的基本尺度熵值;然后利用希尔伯特-黄变换得到希尔伯特边际谱,并求取边际谱的信息熵;最后将这两部分特征参数输入到支持向量机中,实现心电信号的自动分类。结果健康人、心律不齐患者、呼吸暂停患者和房颤患者的心电信号分类准确率分别为87.5%、93.75%、90.63%和90.63%。结论本文提出的基于多尺度化基本尺度熵和希尔伯特-黄变换的分类方法,可以有效实现心电信号的自动分类。
Objective To study an automatic ECG classification method based on multiscale base-scale entropy (MBE) and Hilbert-Huang transform (HHT). Firstly, the signal is denoised and pre-processed by discrete wavelet transform. Secondly, the basic scale entropy is used to analyze the basic scale entropy, then the basic scale entropy is extracted at multiple scales. Then, the Hilbert- Burt marginal spectrum, and obtain the information entropy of the marginal spectrum; Finally, the two parts of the characteristic parameters are input to the support vector machine to achieve automatic classification of ECG signals. Results The classification accuracy of ECG in healthy subjects, patients with arrhythmia, apnea patients and patients with AF was 87.5%, 93.75%, 90.63% and 90.63%, respectively. Conclusion The classification method based on multi-scale basic scale entropy and Hilbert-Huang transform proposed in this paper can effectively classify ECG signals automatically.