脉搏波信号预处理方法研究

Research on Preprocessing Method of Wrist Pulse Signal

为克服脉搏波信号采集时受到的高频噪声干扰和低频噪声干扰,脉搏波信号的预处理成为心脉信号处理中的关键环节。使用零相位滤波法处理脉搏波信号,改善传统数字滤波器直接滤波的输出失真问题;通过建立评价参数,对去除高频噪声的数字滤波器和小波阈值滤波器的滤波效果进行评价,获得最佳滤波方法;对比常用去基线漂移方法在处理脉搏波信号时的特点,获得最佳去基线漂移算法。使用新研发的脉搏波采集手环采集多名受试者的脉搏波原始信息,将采集到的信号按上述方法去噪和去基线漂移后,实现脉搏波信号的预处理过程。实验结果表明,采用sym4小波基、固定阈值、软阈值函数等小波阈值去噪方法去除高频噪声并使用三次样条插值拟合曲线去除基线漂移后,所获得的脉搏波信号平滑无毛刺,每个周期起始点和终点都在同一水平基线上,满足后续脉搏波信号的医学分析和疾病诊断需要。

To overcome the high-frequency noise interference and low-frequency noise interference during pulse wave signal acquisition,the preprocessing of wrist pulse signal becomes a vital part of the heart pulse signal processing.The zero-phase filtering method is used to process the wrist pulse signal to improve the output distortion of the traditional digital filter direct filtering.By establishing evaluation parameters,the filtering effects of digital filter and wavelet threshold filter for removing high-frequency noise are evaluated,and the best filtering method is obtained.By comparing commonly used methods for processing wrist pulse signals,the most suitable method for removing baseline wander is obtained.The newly developed wrist pulse acquisition bracelet is used to collect the original pulse signal on many subjects,and the collected signal is de-noised and de-baseline drifted according to the above methods to realize the wrist pulse signal preprocessing process.The experimental results show that the obtained wrist pulse signal is smooth and burr-free after the high-frequency noise is removed by wavelet threshold denoising methods such as sym4 wavelet basis,sqtwolog threshold,and soft threshold function,and the baseline drift is removed by cubic spline interpolation fitting curve,and the start-point and end-point of each cycle are on the same horizontal baseline,which can meet the needs of medical analysis and disease diagnosis of subsequent pulse wave signals.