Affiliations: [a] Computational Neuroscience Laboratory, Department of Applied Mathematics and Computational Sciences, PSG College of Technology, Coimbatore, Tamil Nadu, India | [b] Department of Computer Science and Software Engineering, Miami University, Oxford, Ohio, USA | [c] Cognitive Neuroengineering Laboratory, School of Information Technology and Mathematical Sciences, Division of IT, Engineering and the Environments, University of South Australia, Adelaide, South Australia
Abstract: Understanding and analyzing the dynamic interactions among the millions of spatially distributed and functionally connected regions in the human brain constituting a massively parallel communication system is one of the major challenges in computational neuroscience. Many studies in the recent past have employed graph theory to efficiently model, quantitatively analyze and understand the brain’s electrical activity. Since, the human brain is believed to broadcast information with reduced material and metabolic costs, identifying various brain regions in the shortest pathways of information dissemination becomes essential to understand the intricacies of brain functioning. This paper proposes a graph theoretic approach using the concept of shortest communication paths between various brain regions (electrode sites) to identify the significant communication pathways of information exchange between various nodes in the functional brain networks constructed using multi-channel EEG data. A special weighted network called Shortest Path Network is constructed from the functional brain network where the edge weight is computed as the sum of frequency of occurrence of that edge in all possible shortest paths between every pair of electrodes. The weighted Shortest Path Networks thus constructed portray the information on the number of times the edges are used in information propagation during different cognitive states. This network is further analyzed by computing the influential communication paths to characterize the information dissemination among brain regions during different cognitive load conditions. The experimental results presented demonstrate the efficacy of the novel graph theoretic approach in identifying, quantifying, and comparing the significant shortest pathways of information exchange during mild and heavy cognitive load conditions. The experimental analysis also provides an element of future research to consider the biological inferences of the ability of the human brain to use reduced material and metabolic cost during instantaneous transmission of information.