Neuronal Spike Sorting In Primary Visual Cortex Of Rats Based On Matched Wavelet Transform

Research on the mapping relationship between visual stimulus and neuronal response is significant to reveal encode and transfer mechanism of visual information in the brain. An embedded multi-electrode array is used to detect neuronal spikes in primary visual cortex (V1) of rats under visual stimulation. Usually each channel would detect spikes coming from several cells. As a result we must devide them apart and determine the corresponding spiking time sequences of each unit in order to analyze the characteristics of the single cell. Although there are many spike classification methods, most of them still have various problems. So it's necessary to make further study on sorting method.Base on the background, an experimental program is planed to collect the spike signals in rats'V1 under the stimulus of grating. A method for spike sorting based on matched wavelet transform is proposed. And this new method is used to determine neuronal orientation selectivity of rats in V1. The main work of this paper is shown as follows.Based on existing experiment platform, the visual stimulus patterns are designed according to visual characteristics and receptive field properties of rats. Then the experimental program is designed through comprehensive analysis and specific experimental verification. Finally the spike signals in rats'V1 under the stimulus of drifting sinusoid grating are collected.Three popular spike sorting methods, such as template matching, principal component analysis and spike sorting based on wavelet transform are systemically studied. The advantages and disadvantages are summed up and a method based on matched wavelet transform is proposed to avoid the limitation of these methods.The principles and computing steps of this new method are introduced in detail. Matched wavelet is constructed according to the spike waveform and specified matching criterion firstly. Then matched wavelet transform is used to extract features of spike waveform and the extracted features are clustered to complete the sorting of spike signals. The method is proved efficient and superior by applying to classify simulation spike signals and recorded neuronal spike signals in rats'VI. It is used to obtain spike sequence and inter spike interval, primary study on neuronal orientation selectivity in V1 of rats.