基于WICA的事件相关电位P300少次提取研究

发布时间：2018-05-07 06:51

本文选题：事件相关电位 + WICA　；参考：《天津师范大学》2012年硕士论文

【摘要】：事件相关电位(Event-related Potential, ERP),是与实际刺激或预期刺激(声、光、电等)有固定时间关系的脑反应所形成的一系列脑电波。有广泛应用的P300是事件相关电位的重要成分,但其信号非常微弱,并容易受噪声干扰,通常提取事件相关电位要经过40～100次的叠加平均,这就需要大量的重复刺激实验,于是人们力求通过少次叠加的方法来实现对事件相关电位的提取。因此,如何实现事件相关电位信号的少次提取成了脑电信号处理的重点及难点问题。由于ERP混叠在自发脑电之中,而近年发展起来的独立分量分析(Independent Component Analysis, ICA)能够从多通道的观测数据中分离出相互独立的信号,不受信号频谱混叠的限制,适用于脑电信号中各种噪声的消除。小波变换(Wavelet Transform)具有时频特性并且有分析非平稳信号的优势,在一定程度上能够精确地分析脑电信号的瞬态特性。因此运用小波变换与独立分量分析相结合(WICA)的方法,可以对脑电信号进行高精度子频带分析,同时也可以弥补小波变换算法不能有效解决噪声和信号的频谱混叠的缺陷。本文将WICA方法应用于P300的提取过程中,主要做了以下研究工作： (1)介绍脑电信号、事件相关电位以及P300的产生原理、特点、分类以及应用等相关知识。 (2)深入研究独立分量分析和小波变换算法,利用两种算法结合即WICA对脑电信号进行处理。首先,利用db5小波对脑电信号进行去噪处理,然后把处理过的信号用ICA算法分解,获得各个信号的独立分量,然后根据事件相关电位的先验知识对各独立分量进行优选,最后重构脑电信号。 (3)利用事件相关电位的固定锁时关系将处理过的脑电信号进行少次叠加,获取事件相关电位信号,并通过P300的先验知识,成功地提取出了P300复合波中的亚成分。这些亚成分对人类认知等高级神经活动有重要意义。 (4)实验对比,将WICA方法获得的P300成分与传统叠加平均方法获取的P300成分进行对比,说明本实验的优越性。本实验对传统的基于叠加平均的P300的提取研究提供了新思路,同时对基于P300的BCI研究有重要的参考价值。
[Abstract]:Event - related potential ( ERP ) is a series of brain waves formed by brain reactions with fixed - time relationship with real stimulus or expected stimulus ( acoustic , light , electricity , etc . ) . The widely used event - related potential ( ERP ) is an important component of event - related potential , but its signal is very weak , and it is easy to be disturbed by noise .

Because of ERP aliasing in spontaneous EEG , independent component analysis ( ICA ) developed in recent years can separate independent signals from multi - channel observation data . It is suitable for the elimination of various noises in EEG signals . Wavelet Transform has the advantage of analyzing the transient characteristics of EEG signals .

In this paper , the WICA method is applied to the extraction process , and the following research work is done :

( 1 ) Introduce the relevant knowledge of EEG signal , event - related potential and the generation principle , characteristics , classification and application .

( 2 ) In - depth study of the independent component analysis and wavelet transform algorithm , the EEG signal is processed by combining two algorithms , namely , WICA . First , using db5 wavelet to de - noising the EEG signal , then the processed signal is decomposed by ICA algorithm to obtain the independent component of each signal , and then the independent component is optimized according to the prior knowledge of the event - related potential , and then the EEG signal is reconstructed .

( 3 ) Using the fixed lock - time relation of event - related potential , the processed EEG signal is superimposed on a few times , the event - related potential signal is acquired , and the sub - component in the composite wave of the P 300 is successfully extracted by a priori knowledge of the event , and the sub - components have important significance for the high - grade nerve activity such as human cognition .

( 4 ) Compared with that obtained by the traditional superposition mean method , the experimental results show the superiority of this experiment . This experiment provides a new idea for the traditional extraction study based on the superposition average method , and has important reference value for the study of the BCI study based on P 300 .

【学位授予单位】：天津师范大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TN911.7;R33

【参考文献】