Graph-Theory-Based Multilevel Cortical Functional Connectivity Developmental Analysis

Abstract

Functional connectivity (FC) is an efficient measurement to describe brain’s traits in information processing. But FCs at developmental stages from infacy to adolescence usually have severe individual variance, may affect the FC characterization analysis. Yet few past studies try to address this problem. In this paper, we select quiet sleep (QS) and non-rapid eye movement (NREM) period electroencephalogram (EEG) of 42 healthy subjects from 0 to 17 years old for study. Random network combined with stability measurement using Shannon entropy (SE) is used to construct the individual-level functional connectivity (ILFC), which can describe the individual brain’s interaction. Majority voting is applied to construct the group-level functional connectivity (GLFC), which can describe the core part of ILFCs within the same age. Based on FCs, graph theory and statistical analysis are further applied, where the following conclusions are observed: (1) The β band of EEG is most important in showing the age related variance of functional separation and integration. (2) 3 months 3 years old is the transition period from intra-subnetwork interacting mode to inter-subnetwork interacting mode. (3) The phenomenon of complementation and overlay among GLFCs in different bands can be found. (4) The GLFC’s network centrality becomes not obvious as age increases during 1 14 years old.