基于小波变换和波形特征分析的峰电位分类

Spike Sorting Based on Wavelet Filter and Spike Feature Analysis

峰电位分类是细胞外记录中一个重要的技术。提出了一个基于离散小波变换和波形特征分析的新的峰电位分类方法(DWT-SFA),同时定义了3个不同的波形特征。首先,对信号进行小波消噪处理,然后,使用这3个不同的波形特征进行信号分类。通过对大壁虎嗅觉神经信号的分类处理,证明该方法明显优于主成分分析方法。此外,通过对该方法与Offline Sorter软件的分类对比,也验证了它的正确性和精确性。

Spike sorting is an important technology of extracellular recording.In this paper,we present a new method of action potentials classification based on discrete wavelet transform and spike feature analysis（DWT-SFA）.Here,we develop three different waveform features.First we apply wavelet decomposition and reconstruction as a filter for data.Then the three different features are used to do feature clustering.Comparison between proposed method and PCA（Principal Component Analysis） on olfactory neural signals shows that proposed method is superior to PCA.Comparison between proposed method and Offline Sorter on analytical data verifies the proposed method is right.