Self-similarity of human brain connectomes

Abstract

The human brain is extremely complex, with multiple spatial scales interacting simultaneously The connectome is a comprehensive map of the neural connections between brain regions that is useful to delve deeper into the human brain. We studied the multiscale organization of human brain connectomes and found two symmetries using two datasets from healthy subjects. First, we measured the multiscale properties at five different hierarchical resolutions and recovered the reviously reported result that they remain self-similar as the resolution is decreased. Second, we performed a quantitative analysis that shows that the structural features of the connectomes remain self-similar when using a degree-thresholding renormalization method on the highest resolution network layer. Our results suggest that the human brain may be working near a critical point, and that brain regions could be mapped to a hyperbolic hidden metric space, defining distances that explain the structure of brain connectomes and helping in the development of simulation and brain reconstruction tools.