一种提取近红外光谱信号中诱发血液动力学反应的方法  

A Method for Extracting Evoked Hemodynamic Response from NIRS Signal

作　　者：蔡云丽 高琳[1] 王逸飞[1] 闫相国[1] 

机构地区：[1]西安交通大学生物医学信息工程教育部重点实验室,西安710049

出　　处：《西安交通大学学报》2016年第7期152-158,共7页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目(61471291);国家"863计划"资助项目(2012AA02A604)

摘　　要：为了更好地提取与大脑功能活化有关的诱发血液动力学反应(EHR),综合比较了不同经验模态分解算法的优缺点,提出了一种提取近红外光谱信号中EHR的方法——ICEEMDAN-RLS。利用5种经验模态分解算法对近端通道信号进行分解,根据分解结果对远端通道信号进行自适应滤波,借助皮尔森相关系数和相对均方误差评估不同经验模态分解算法的EHR提取性能,在此基础上分析经验模态分解和传统块平均方法的块平均次数与EHR信号质量之间的关系。结果表明,基于5种经验模态分解算法的自适应滤波方法都能有效提取远端通道信号中包含的EHR,ICEEMDAN-RLS具有更大的皮尔森相关系数和更小的相对均方误差,仅需要进行10次块平均便可获得稳定的EHR信号,比传统方法的块平均次数减少了75%,且具有较高的EHR信号质量,可以更有效地提取淹没在全局干扰中的EHR。该结果可为大脑功能活化的研究提供参考。To extract evoked hemodynamic response（EHR）relating to functional activation in human brain,a method named improved complete ensemble empirical mode decomposition with adaptive noise and recursive least square（ICEEMDAN-RLS）was proposed to pick up EHR from NIRS signal after comparing the advantages and disadvantages of different empirical mode decomposition（EMD）algorithms.The near-channel signal was firstly decomposed by five EMD algorithms,then EHR was estimated from far-channel signal by adaptive filter.The performance of different EMD algorithms was assessed by Pearson correlation coefficient and relative mean square error.The relationship between block average times and the quality of EHR was analyzed with EMD and traditional block average method,respectively.The results show that EHR can be effectively extracted from far-channel signal by all the above methods,while the biggest Pearson correlation coefficient and the smallest mean square error are obtained by ICEEMDAN-RLS.Moreover,10 times of block average are enough for ICEEMDAN-RLS to obtain stable EHR,which is reduced by 75% compared with the traditional method,and better signal quality is gained.Therefore,ICEEMDAN-RLS is more effective in extracting EHR contaminated byphysiological components and performs better than block average method.

关 键 词：近红外光谱技术 诱发血液动力学反应 经验模态分解 块平均 

分 类 号：R318[医药卫生—生物医学工程]