基于EMD-ICA的心冲击信号降噪研究

心冲击信号(BCG)是反映心脏机械运动的生理信号,能实现无电极束缚条件下的连续采集测量。但BCG信号微弱,易受干扰,测量时经常会淹没在噪声中。为了消除噪声,有效识别BCG信号特征,提出一种基于经验模态分解(EMD)联合独立分量分析的BCG信号降噪方法。首先,将含噪BCG信号进行EMD分解,获得一系列按频率从高到低的固有模态分量(IMF),采用模态相关准则进行信号层与噪声层的判定;其次,将分界之上的IMF分量构建虚拟噪声通道,基于ICA算法对原始BCG信号进行盲源分离,从而得到降噪后的BCG信号。采集10名健康受试者的BCG信号进行降噪处理。量化评价结果表明,与小波方法和EMD方法相比,降噪后信噪比均显著提高(小波方法11.01±1.58,EMD方法5.19±1.29,所提出方法14.87±3.04,P<0.05),能量百分比也均显著提高(小波方法88.81%±2.81%,EMD方法96.15%±2.96%,所提出方法96.64%±2.92%,P<0.05),从而证明所提出方法降噪效果明显,能够有效还原BCG信号特征。

BCG Signal De-Noising Method ResearchBased on EMD-ICA

Ballistocardiogram (BCG) signal is a physiological signal reflecting heart mechanical status. It can achieve continuous acquisition measurement without electrodes constraint. However, BCG signal is so weak that it would often be interfered by superimposed noises. Aiming to eliminate the noise and recognize BCG signal characteristics effectively, this paper proposed a de-noising method of BCG signal based on empirical mode decomposition (EMD) and independent component analysis (ICA). Firstly, the noisy BCG signal was decomposed by EMD to obtain a series of intrinsic mode components (IMF) ranked by frequency in descending order, and the EMD mode was used to distinguish the boundary of noise and useful signal and remove the maximum noise. Secondly, the IMF components of above the boundary were employed to construct a virtual noise channel, and the blind source was separated with the original BCG signal based on ICA algorithm to extract the de-noising BCG signal. Acquisition of 10 healthy subjects BCG signals for noise reduction processing. Quantitative evaluation results indicated that the proposed method significantly increased SNR (14.87±3.04,P<0.05) compared with wavelet method (11.01±1.58) and EMD method (5.19±1.29), significantly increasing energy percentage (96.64%±2.92%,P<0.05) compared with wavelet method (88.81%±2.81%) and EMD method (96.15%±2.96%), which proved that the proposed method was effective in the reconstruction of the characteristics of BCG.