Discovering Network Patterns in Microelectrode Array Data

D. Patnaik*, P. S. Sastry*, and K. P. Unnikrishnan#

*Department of Electrical Engineering, Indian Institute of Science, Bangalore 560012
#General Motors R&D Center, Warren, MI 48090

Microelectrode array(MEA) recording is a relatively new experimental technique in neurobiology for studying simultaneous activity of groups of neurons. The objective of analyzing the MEA recordings is to discover different types of temporal correlations between the neurons in an ensemble and hence infer the functional connectivity of the neural tissue. To discover such relationships from multi-neuronal data, there is a need for analysis techniques which are efficient and which can unearth interesting regularities that involve more than pairs of neurons. In this article, a novel application of frequent episode discovery framework to microelectrode array data analysis is presented. It is shown, through simulations, that by combining discovery of different types of episodes with suitable temporal constraints, one can discover the network structures and connectivity patterns of the neurons constituting the network.