Measure of the regularity of events in stochastic point processes, application to neuron activity analysis

Numerous researches aim at understanding the high brain functions such as memory or decision making by analysing the activity of brain neurons. This activity corresponds to sequences of electrical potentials and thus can be viewed as a point processes. In this paper, we propose a method to measure the regularity level of event occurrences in point processes. Based on the analysis of the so-called density histogram, the proposed approach has the advantage of providing a decision to classify the process into one of the three following distinct classes: the "regular" processes, the "irregular" processes and the "bursting" processes. To illustrate the efficiency of the method, we first carry out a comparative study based on synthetic data. Then, the algorithm is tested in the framework of neurosciences for the classification of neurons according to their activity.