Structural And Network-Based Analysis Of a Human Connectome

The article deals with neurobiological networks - connectomies that describe the neural network of the human brain. Connectomes have significant differences from other graphs, such as structure in the form of hemispheres or mapping groups of vertices to different functional parts of the brain. Therefore, the study of connectoms is a task in its own right. Their topological properties, such as vertex degree and spectrum, played a key role in this work. Various versions of averaged graphs were created to identify distinctive and common features of different connectomes of the same ensemble. Such networks allowed comparisons to be made with individual instances, both within and outside the same scale group. The paper analyzed the distribution of link fractions in graphs associated with specific instances and in the general case. In particular, evidence for the importance of interhemispheric connections is presented. The second part of the paper is devoted to spectral characteristics of connectomes, which allow to numerically estimate the degree of similarity between different networks. The study showed that the connectomes differ significantly from the randomly generated graphs constructed from the baseline measures belonging to the data network under study. The data were also tested for robustness by calculating the Spearman correlation of the different parts.