Spike Tools:用于神经元网络锋电位分类的软件  

Spike Tools: Software Applying to Neuronal Network Spike Sorting

作　　者：孙倪[1] 周炜[1] 李向宁[1] 曾绍群[1] 

机构地区：[1]华中科技大学生物医学光子学教育部重点实验室,武汉430074

出　　处：《计算机与数字工程》2005年第1期13-16,29,共5页Computer & Digital Engineering

基　　金：国家杰出青年科学基金项目 (6 0 0 2 5 5 14 );教育部科学技术重大项目 (重大 10 4 2 0 )。

摘　　要：使用多电极阵列可无损的记录神经元网络的电活动 ,但是单个电极往往会同时记录到多个神经元的活动 ,因而需要将所记录的信号进行分类。本文通过幅度、形状、主成分等分类方法与聚类方法实现了对锋电位的分类 ,并在MATLAB平台下编写了相应的可视化软件SpikeTools。对模拟数据的分类结果表明 :均有 90 %以上的锋电位得以正确归类。对实验数据的分类结果表明 :阈值分类方法能对幅值有明显区别的锋电位进行合理分类 ,形状分类方法能对幅值接近的锋电位进行分类 ,而主成分分类方法区分波形的整体差异 ,能分离前两种方法难以区别的类 ;聚类方法可实现自动分类。开发的Spike Tools软件能对各种锋电位群数据进行分类处理 ,为锋电位信号的分析提供了有力的工具。The technology of micro-electrode arrays is getting more and more important in the research of brain neural networks because of the ability to stimulate and record more neurons' activities simultaneously. When the signals of many neurons are picked up with a microelectrode, a neurophysiologist may wish to sort these signals by assigning particular spikes to putative neurons. In this article, We have accomplished spike sorting through analyzing the amplitude, shape and principal components of the spikes by SpikeTools, a visualization software programed on MATLAB system. The sorting result to artifical data shows that over 90% of spikes are classed correctly. The sorting results to experimental data indicate that, the sorting method by amplitude can reasonably divide the spikes whose amplitudes are remarkably different; the sorting method by shape can divide the spikes based on spikes' shape whose amplitudes are contiguous; and the sorting method by principal components can distinguish the whole diversity of the wave more precisely; cluster analysis can sort spikes automatically without the manual boundary. So SpikeTools is a powerful tool for spike sorting.

关 键 词：多电极阵列 神经元网络 分类 主成分分析 聚类分析 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]