Using Multi Network Alignment for Analysis of Connectomes

The human brain is a complex organ that may be represented by a complex network called connectome. An important first step to understand the function of the brain connectome is to model and to analyze its nodes and connections, in order to achieve a comprehensive description of the brain. In this work we apply the graph theory formalisms to represent the connectomes. The human brain connectomes are usually derived from Magnetic resonance imaging (MRI); then an atlas-free random parcellation is used to define network nodes of individual brain networks. In this network space, the question of comparison of the structure of networks arises. Such issue may be modeled as a network alignment (NA) problem. The use of different NA approaches, widely applied in molecular biology, has not been explored in relation to MRI connectomics. In this paper, we first defined the problem formally, then we applied three existing state of the art multiple alignment algorithms (MNA) on diffusion MRI-derived brain networks and we compared the performances. The results confirm that MNA algorithms may be applied in cases of atlas-free parcellation for a fully network-driven comparison of connectomes.