The Study Of Spike Detection And Burst Analysis Of Hippocampal Neural Network

Now studying the mechanism of information processing of neural network has become a frontier in the neuroscience research field. Here we make use of a technique named Multi-electrode arrays (MEA) system. This is a useful tool in neuroscience research that makes it possible to study the simultaneous activity of several neurons in a network in vitro. There are two difficulties in the spike detection: spike signals are frequently superimposed on local field potential (LFP) that reflect the activity of populations of cells near the electrode tip, moreover, the recorded extracellular signal represent a time derivative of the intracellular signal. In order to detect spikes accurately, we have compared the method of threshold-detection and peak-detection. Then we bring up a revised version of the peak-detection scheme. This advanced approach not only works out an unwanted detection problem of the peak-detection caused by the position of sliding window boundary, but also extremely speeds up spike detection. Its inclusion errors and exclusion errors is lower than the threshold detection. We have used several methods, such as inter spike interval (ISI) histogram, burst duration (BD), inter burst interval (IBI), ISI serial correlation coefficients (SCCs) and so on. They have performed usefully in the analysis of temporal structure of burst. The results are: in each refractory period, the ISI will be never less than several millisecond; there is a quiescent period about several hundred milliseconds between two bursts; bursting is periodical, and the periods are all the same when the neurons are in the same network. After analyzing the bursting procedure of several cells, we conclude that there are two types of bursting cells. The first type experiences three procedures: single spikes, dense spikes and burst. The second type merely experiences two procedures: single spikes and burst. By researching the trains before and after bursting, we understand three points. First, the mean of ISI is gradually minishing. Second, there are two new characteristics: consecutive spikes and stabled IBI. Third, negative ISI correlations at short lags have been observed, which is produced by consecutive spikes and bursting quiescent period appearing alternatively.