A network-based approach to discovering functional connectivity in neuronal data

Background and Aim : Fast development of recording technology provides a great chance of access to large amount of data for neuroscientist. Recording simultaneous individual neurons of large group in different part of the brain necessitates variety of quantitative techniques in order to analyses spiking activity and methods for addressing functional activity across neuronal population.Methods : In this work we aim to model the functional connectivity with statistical based modeling. We consider a special multivariate Point Processes, Hawkes process on a Bayesian framework. The advantages of our work are, but not limited to, the following: a) estimation causality with directed property of the fitted graph and b) estimation the strength of the connection with weighted property of the fitted graph.Results : We applied our method on the recorded neurons from FEF area of the macaque brain in a memory guided saccade task. By estimating our network based on spike counts in the three epochs: visual, memory and saccade, we evaluated functional connectivity graph and extracted network measures such as average shortest path, clustering coefficient, betweenness centrality and small world network (SWN) property on each epoch.Conclusion : By comparing this network measures across epochs, we monitored the dynamic of the functional connectivity within cognitive states. Furthermore, we investigated the hypothesis that neural networks are arranged in functional groups, recalling the concept of cell assembly.

کلیدواژه ها:

Point Processes ، Hawkes process ، Bayesian framework ، average shortest path ، clustering coefficient