[Mathematical model of bursting spike train and its spectrum features].

Bursting is an important firing mode of neurons. To propose a stochastic model of bursting spike train, the interspike interval (ISI) characteristics of single-spiking train and bursting spike train were analyzed and compared. In contrast with the exponential distribution of ISI in single-spiking train, normal distribution is supposed to be the ISI model of bursting spike train. Simulated neural spike trains were produced to investigate the spectrum features of the ISI model. The results showed that: (1) bursting spike train with normally distributed ISI held a low-pass spectrum while the spectrum of single-spiking train was flat; (2) the coefficient of variation of ISI in bursting train decided the bandwidth of its low-pass spectrum. Then neural activities from anesthetized rodent were used to check the validity of the model. 10 simultaneously recorded bursting spike trains and 10 single-spiking trains were selected during anesthesia and after pure-oxygen-washout period respectively. The spectrograms of these neural spike trains were analyzed and the results were matched with our mathematical model. It is believed that the bursting spike train model established in this paper will help to theoretically study the statistical characters of neural spike train and to add mathematical foundation in neural coding schemes.