应用经验模式分解去除近红外光的运动伪迹

发布时间：2018-07-04 10:05

本文选题：运动伪迹 + 近红外光谱　；参考：《西安交通大学学报》2014年02期

【摘要】：针对近红外光谱(NIRS)技术对运动十分敏感和运动伪迹会严重影响后续分析处理结果这一问题,提出了一种利用经验模式分解去除运动伪迹的方法(EMD-MAR)。首先计算NIRS信号的移动标准偏差,根据其概率分布特性自动选取阈值和确定运动伪迹范围,然后利用经验模式分解(EMD)将信号分解成多个固有模式,对在运动伪迹范围具有明显异常的固有模式置0来消除运动伪迹,最后将处理后的固有模式加和得到校正后的信号。仿真与实测结果表明,EMD-MAR方法可有效检测和消除NIRS测量中常见的基线漂移、瞬时脉冲以及短暂低频振荡3种运动伪迹,并且极大提高了运动伪迹的自动检测程度,利用经验模式分解可以在去除运动伪迹的同时有效保留NIRS信号中的生理信息,为NIRS信号预处理提供了一种有效手段。
[Abstract]:Aiming at the problem that near infrared spectroscopy (NIRS) is very sensitive to motion and motion artifacts will seriously affect the results of subsequent analysis, a method of removing motion artifacts by empirical mode decomposition (EMD-MAR) is proposed. First, the moving standard deviation of NIRS signal is calculated, the threshold value is automatically selected and the range of motion artifacts is determined according to its probability distribution characteristics, and then the signal is decomposed into multiple inherent modes by empirical mode decomposition (EMD). In order to eliminate the motion artifacts, the original modes with obvious anomalies in the range of motion artifacts are set 0 to eliminate the motion artifacts. Finally, the processed intrinsic modes are added and the corrected signals are obtained. The simulation and experimental results show that the EMD-MAR method can effectively detect and eliminate three kinds of moving artifacts, such as baseline drift, instantaneous pulse and transient low frequency oscillation, and greatly improve the automatic detection of motion artifacts. The empirical mode decomposition (EMD) can effectively preserve the physiological information of NIRS signals while removing motion artifacts, which provides an effective means for NIRS signal preprocessing.
【作者单位】：西安交通大学生物医学信息工程教育部重点实验室;
【基金】：国家“863计划”资助重大项目(2012AA02A604) 陕西省自然科学基础研究资助项目(2011JM4049)
【分类号】：R318.51

【参考文献】