基于小波变换的事件相关电位少次提取方法

Extraction of Event-Related Potential With a Few Trials Based on Wavelet Transformation

本文以反映大脑稀少认知事件的相关电位P300为例,介绍了基于小波变换(wavelettransform,WT)进行事件相关电位(eventrelatedpotential,ERP)少次提取的原理、仿真和实验结果。通过小波变换先去除与ERP混杂的EEG背景干扰,然后对消除干扰后的脑电数据进行重构和ERP波峰检测,在此基础上尝试修正ERP波峰的潜伏期时移,再通过小波重构获得通常需多次相干平均才能得到的ERP波形。仿真计算和实验数据分析结果说明小波分析具有较强的ERP峰值检测和潜伏期模式识别能力,值得进一步在临床工程中深入研究和应用推广。

Based on the wavelet transformation （WT）, the principles, simulations and experimental results of detecting the P300 waveforms which reflect the related potential of the brain＇ s recognizing rare events are presented as the examples of estimating the event-related potential （ERP） only by a few trials. In this algorithm, firstly we remove the interference of EEG background mixed with ERP, then rebuild the denoised data and detect the ERP peaks. After that, with the helps of time-shift correction of ERP latencies and the wavelet reconstruction, we can get a rather satisfied P300 waveform that only can be obtained by the conventional coherent averaging needed over many trials. The results of simulations and the analysis to experimental data show that the wavelet analysis is powerful in the detection and pattern recognition of ERP peaks and latencies, and valuable for further studies and applications in the clinical engineering.