摘要
针对高刺激率模式下运用连续循环去卷积的方法提取的听觉诱发电位(AEP)可能需要比常规方法更长的记录时间,受试者要忍受大量的声音刺激的问题,本文提出运用希尔伯特-黄变换的去噪方法来减少刺激个数。首先对采集到的每个扫程的脑电信号进行经验模态分解,得到一组固有模态函数(IMFs),然后将这些IMFs分为有用信号层和噪声层,采取非线性阈值方法分层进行去噪处理,以提高信号去卷积前的信噪比。处理结果的评估以1 014个扫程的脑电数据按照常规方法处理的提取AEP为基准,计算去噪后AEP和基准AEP的信噪比与相关系数。结果表明该方法能够有效减少约2/3的刺激个数,是一种十分有前景的高刺激率下AEP信号的提取方法。
The high rate stimulation paradigm using continuous loop averaging deconvolution（CLAD） for extracting the transient auditory evoked potential（AEP） usually requires more recording time than conventional one in order to obtain a clear AEP with equivalent signal-to-noise ratios（SNRs）,which means subjects have to suffer a much larger amount of stimulus sound delivered to the their ears.This study aims at reducing the number of stimulus sweeps through signal denoising technique developed by Hilber-Huang transform（HHT）.The EEG sweeps are decomposed into layers of intrinsic mode functions（IMFs） which can then be classified as noise layers and signal layers accordingly.The nonlinear thresholdfilters are applied to these layers separately to improve the corresponding SNRs before deconvolution.The proposed method is validated by 1 014 sweeps of AEP data examined with measurements of SNR,i.e.Fmpand correlation coefficients CCs,which shows that more than two thirds of sweeps can be saved to reach equivalent signal quality.
出处
《数据采集与处理》
CSCD
北大核心
2011年第3期280-285,共6页
Journal of Data Acquisition and Processing
基金
国家自然科学基金(60771035)资助项目
广东省2009年高等学校人才引进资助项目
