多通道局部场电位θ分量与锋电位的互信息对大鼠工作记忆的协同编码

摘要

目的 研究大鼠在Y迷宫工作记忆过程中,责任脑区前额叶皮层记录的2类不同模态的神经信号:局部场电位(LFPs)和锋电位(Spikes)的协同编码,为研究工作记忆的神经编码机制提供计算支持.方法 实验数据:4只SD大鼠在Y迷宫工作记忆过程和静息状态下前额叶皮层的多通道LFPs及Spikes数据由天津医科大学神经工程实验室提供;LFPs的数据预处理:对场电位去除工频干扰和基线漂移;Spikes的连续化:选取生理窗口为500ms,窗口移动步长为125 ms,把Spikes通过发放频率的方式转换为连续信号;场电位的时频分析与特征频段的提取:通过短时傅里叶变换,获取能量分布的主要频段,并通过带通滤波提取特征频段信号,从而得到2种不同模态信号的数据对;特征频段的场电位与锋电位的互信息编码:对单通道的LFPs-Spikes数据序列加窗,计算每个滑动窗口内的互信息值,进而得到多通道的互信息值的分布.结果 场电位的时频分布表明工作记忆期间其能量主要集中在θ频段(4～12 Hz);4只大鼠θ频段LFPs与Spikes在10次工作记忆实验中平均值为0.49±0.04、0.39±-0.03、0.41±0.03、0.48±-0.02,显著大于各自静息状态的值,其差异具有统计学意义(P＜0.05).结论 θ频段(4～12 Hz)是工作记忆的特征频段,θ频段的LFPs与Spikes的互信息有效编码了工作记忆事件.%Objective This paper aimed to investigate the collaborative coding of two different modes of neural signals including local field potentials and spiking activity (multi-unit activity) which recorded in medial prefrontal cortex of sprague-dawley (SD) rats in Y maze working memory (WM) task,to provide computing support for neural coding mechanism of WM.Methods 1.Experiment data was multi-channel neural signals (local field potentials (LFPs)-Spikes) recorded from prefrontal cortex of 4 SD rats during WM process and resting state,provided by the lab of Neuro-engineering,Tianjin Medical University.2.LFPs preprocessing included baseline drift removing and power-line noise eliminating.3.Physiology window width 500 ms,step 125 ms were selected and average rate per channel was calculated to turn the discrete point signal spikes to continuous signal.4.LFPs characteristic frequency band was obtained by using short time Fourier transform and signals of the characteristic frequency band were extracted by band pass filter.5.Mutual information (MI) of LFPs-spikes was computed to get the distribution of multi-channel MI values.Results 1.LFPs distribution demonstrated that the power focused at θ band (4-12 Hz)during the WM tasks.2.The average MI value of θ band LFPs and spikes (4 rats,over 10 trials) were 0.49±0.04,0.39±0.03,0.41±0.03,0.48±0.02,respectively,which were significantly larger than those in the resting state (t test,P＜0.05).Conclusions These findings indicate that θ band LFPs represents behavior correlated to WM and its synergy with spiking activity plays an important role in encoding WM task.