总体经验模态分解能量向量用于ECG能量分布的研究

总体经验模态分解(EEMD)改进了经验模态分解(EMD)存在的模态混叠问题, 依据信号自身的波动特点将信号分解, 特别适合非线性非平稳信号的分析处理. ECG信号能量分布有一定的规律, 疾病会引起能量分布的变化, 研究ECG能量分布的改变对心脏疾病的研究和临床诊断有重要意义. 本文将ECG信号通过EEMD方法分解为多个本征模态函数(IMF)分量, 观察IMF分量的波动规律, 指出了ECG信号在不同时间尺度上的波动特点和物理意义. 将IMF分量分别计算能量, 得到ECG的能量向量, 并对健康人和三种心脏疾病患者能量向量进行对比分析. 结果表明心脏疾病导致EEMD能量向量的高频分量显著降低, 尤其是p₁分量具有较好的区分度, 可以作为心脏疾病诊断的参考依据. 相比较传统的频域分析方法单纯关注频率而忽略信号自身特点和信号成分之间的相互作用, EEMD的分解结果依赖于ECG信号本身, 因此更能够反映ECG信号的真实情况, 揭示年龄和疾病对ECG能量分布的影响.

ECG energy distribution analysis using ensemble empirical mode decomposition energy vector

Ensemble empirical mode decomposition (EEMD) method eliminates mode mixing phenomenon which is an inherent problem in empirical mode decomposition (EMD), and decomposes signals according to their intrinsic characteristics. It is suitable for analyzing nonlinear and non-stationary signals. Electrocardiogram (ECG) energy distribution exhibits a certain regularity which may vary with heart diseases. Researches on ECG energy distribution change are important for heart disease clinical diagnosis. In this paper, we use EEMD method to analyze ECG and find out how ECG energy distribution varies with age and heart diseases. We decompose the ECG signal into several intrinsic mode function (IMF) components by EEMD, and find that these IMFs can reveal the fluctuation rhythm and physical significance of ECG on different time scales. After IMFs have been decomposed, we calculate their energy and obtain an energy vector. By comparing the energy vectors among healthy young subjects, healthy old subjects, and three types of patients suffering from different heart diseases, we find that there is a significant decrease of high-frequency components of energy vector in heart disease patients as compared to healthy subjects, and a slight decrease of healthy old subjects as compared to healthy young subjects. T-test is performed to compare heart disease subjects with healthy subjects. Results show that there are significant differences between certain energy vector components, especially the first component p₁ which could be used as heart disease auxiliary diagnosis. Compared to traditional frequency-domain analysis methods which simply concern about the frequency of a signal and ignore its own characteristics and interactions between signal components, EEMD method depends on ECG signal itself, therefore can reflect its real characteristics, and reveals the way how age and illness influence ECG energy distribution accurately.