Neuronal networks

No matter at what level you look at it, the brain is a network. It consists of specialized groups of brain cells (neurons) that are interconnected. Each brain function is mediated by its own set of regions, and these may be visualized as activation maps in task functional MRI.


Even if it is not engaged in an explicit task, the brain is active. This is evident from the fact that although it constitutes about 2% of our body weight, at rest the brain accounts for 20% of our oxygen consumption.

Notably, spontaneous fluctuations in resting-state brain activity may be highly synchronous between distributed regions. These sets of regions are called resting-state networks. By comparison to task activation maps, it has been established that these resting-state networks are related to brain function. They thus inform us about the functional architecture of the brain.

Examples of resting-state networks are the visual networks, the sensori-motor network, and higher level networks such as the default-mode network. Using task-fMRI, it has been demonstrated that the default-mode network increases in activity in absence of a task, and may represent processes of task preparation, introspection and rumination. Higher integrity (interconnectedness) of the default-mode network has been associated with higher IQ, and abnormalities of this network have been found in several neurological diseases. Within the Neu3ca brain research program, we aim to find abnormalities in functional networks related to cognitive impairments in epilepsy.


In addition to studying functional networks, i.e. investigating which brain regions show synchronous activity, it is also interesting to study structural connectivity, i.e. which brain regions are anatomically linked by nerve fibers. To do this, diffusion weighted MRI may be exploited. In this imaging technique, use is made of spontaneous diffusion (movement) of tissue water. In and around nerve fibers, the main direction of diffusion is parallel to the fibers, whereas it is hindered in perpendicular directions. By probing for diffusion in a large number of directions (A in figure above), local fiber directions can be estimated (subfig B), and the course of nerve fibers may be reconstructed in a process called tractography (subfig C).

Both functional and structural connectivity may be impaired in pathology such as cognitive problems. One may test for changes in one or a few connections, but also overall network organization may have changed. To study network topology (instead of separate connections), graph theory can be applied. This is a set of mathematical tools to derive descriptors of networks, such as modularity, clustering coefficient, and path length. In a range of brain diseases, the clustering coefficient is decreased and the path length (i.e. the number of steps needed to get from one brain region to an arbitrary other brain region) is decreased. Also within the Neu3CA brain research program we exploit graph theory to find abnormalities in network organization for patients with cognitive problems, or to assess the effects of neuromodulation.