An eccentric perspective on brain networks

Abstract

The adult human brain comprises an estimated number of 80-100 billion neurons. These neurons do not operate independently, but are interconnected to each other through circa 100-500 trillion neuronal connections, together forming a network of incredible complexity. Although this vast system of neurons and neuronal connections – known as the connectome – can (currently) not be mapped in full detail, the organization of the neural infrastructure which enables us to move, sense, memorize, and think does not have to remain elusive. On a larger and more comprehensible scale, neurons are organized into anatomically and functionally distinguishable brain regions and their neuronal connections form large-scale white matter fibers. Especially during the last two decades, mapping this macroscale connectome of brain regions and white matter fibers has become increasingly feasible and studies have gradually started to unravel its intriguing network architecture. In this thesis, we continue along this path to elucidate the macroscale network organization of our brain a little further. To do so, we adopt a somewhat unusual perspective on brain networks, shifting the conventional focus on brain regions to a focus on the connectome's connections. We first consider the reliability of reconstructed white matter fibers and introduce a model which helps to reduce the number of false positive and false negative connections in macroscale connectome reconstructions. Subdividing the connections of the connectome into different categories, we then demonstrate that connections between highly connected and highly central hub regions are frequently bidirectional and intermodular (i.e., positioned between different subsystems) and tend to be longer and stronger than other types of connections; properties which make them ideal candidates to form the connections of a “global workspace” in the brain. In support of this hypothesis, we propose a framework to obtain a more direct assessment of the network role of connections and show that hub-to-hub connections play a pronounced role in the communication and integration between different functional subsystems. We further show that the white matter fibers of our brain are intrinsically organized into so-called link communities and that hub regions form “hot spots” were connections from a variety of link communities come together. In all, these findings corroborate the idea that hubs and their connections may form a central infrastructure for multimodal and integrative processes in the brain. Moreover, the identified link communities and proposed connection measures may provide interesting targets and tools for studies on the diseased brain, potentially allowing better localization and characterization of connectome differences between patients and controls.