一种神经信号压缩感知检测方法  

A Neural Signal Detection Method Base on Compressed Sensing

作　　者：徐声伟[1] 周帅[1,2] 刘欣阳[1,2] 周基阳[1,2] 王蜜霞[1] 宋轶琳[1] 林楠森[1,2] 蔡新霞[1,2] 

机构地区：[1]中国科学院电子学研究所传感技术国家重点实验室(北方基地),北京100190 [2]中国科学院大学,北京100049

出　　处：《网络新媒体技术》2014年第2期21-25,共5页Network New Media Technology

基　　金：中科院先导专项(No.XDA06020101)资助

摘　　要：针对无线传感器在神经信号检测领域需要低功耗设计的需求,设计了一种利用压缩感知方法对含动作电位的神经信号进行分段压缩的方法。首先,通过构造冗余字典,实现了对原始神经信号的稀疏化表征;然后构建了测量矩阵,实现了对原始神经信号的压缩;最后,利用稀疏分解算法,实现了对原始神经信号的重构。利用测量到的大鼠脑部神经信号对设计的方法进行实验测试,结果表明,在压缩比为10∶1及以下的情况下,可以实现对含动作电位的神经信号的分段压缩和可靠重构。This paper presents a new neural signal detection method base on Compressed Sensing. First, a redundant dictionary is con- structed, in which the neural signal has a sparse representation. Second, we make a measurement matrix to compress the original neu- ral signal. Finally, using the Orthoganal Matching Pursuit arithmetic, we reconstruct the neural signal. Using this method, we make some experiments to process neural signals, which are acquired from of a Sprague - Dawley （SD） rat cortex in an experiment. The re- sults indicate that the method can compress and reconstruct the neural signal reliably when compress ratio is set at less than 10times.

关 键 词：压缩感知 动作电位 压缩比 冗余字典 

分 类 号：TN911.23[电子电信—通信与信息系统]