用于锋电位分类的二阶差分表示法  

Second-order difference representation applied to spike sorting

作　　者：罗静[1] 戴敏[1] 

机构地区：[1]天津理工大学计算机与通信工程学院天津市智能计算与软件新技术重点实验室,计算机视觉与系统教育部重点实验室,天津300384

出　　处：《天津理工大学学报》2013年第6期21-25,共5页Journal of Tianjin University of Technology

基　　金：天津市自然科学基金(10JCYBJC00700);天津市科技支撑重点项目(108CKFSF00800)

摘　　要：针对锋电位在相似度高及存在大量叠加锋电位时分类结果不理想的问题,提出了一种新的锋电位特征表示方法 -二阶差分表示法.该方法对锋电位波形求取二阶差分,以二阶差分序列作为锋电位的特征信息,形成新的样本向量进行分类.该方法对Wave_clus中不同信噪水平的数据分别进行了实验.实验表明,该方法描述了锋电位波形在各个时刻的波形趋势,在一定程度上能够扩大不同类型波形之间的差异性.将此方法用于锋电位分类,尤其是叠加锋电位分类,可以提高分类准确率,并且可以有效的避免噪声的干扰.It is practically difficult to identify the spikes to different classifications when the units have very similar spike waveforms or lots of overlapped. A novel feature representation of spike waveform is proposed in the paper, called sec- ond - order difference representation. In the method, the second-order difference of every spike is calculated as the al- ternative feature information for spike sorting. The method is tested at various signal-to-noise ratio levels based on simu- lation data coming from the Wave_clus. Experiments show that the method can describe the waveform trend at each time. It can not only reduce the difference between same types of spike waveforms but also enlarge the difference be- tween different types of spike waveforms. The classification accuracy can be improved and the noise can effectively avoi- ded by using the representation.

关 键 词：锋电位分类 二阶差分 相似锋电位 叠加锋电位 K-MEANS聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]