Computational neuroscience: Simulating brain dynamics and generative modelling of brain networks
Project Leader: Andrew Zalesky
Staff: Dr Caio Seguin
Collaborators: Professor Olaf Sporns (Indiana University)
Primary Contact: Andrew Zalesky (email@example.com)
Keywords: complex systems; computational neuroscience; connectome; network science; neuroimaging
Disciplines: Biomedical Engineering,Electrical & Electronic Engineering
Simulate a person’s brain activity based on their connectome and develop models to grow brain networks in silico.
Two projects are available under this research theme, both of which aim to use cutting-edge network models of large-scale brain activity to gain fundamental insight into brain function in both health and psychiatric disorders.
This project will use computational modelling to investigate the impact of focal perturbations on network-wide brain activity. Focal perturbations can represent the acute effects of brain stimulation, as well as long-term changes in structural brain connectivity due to disease, pharmacological interventions and training. A key advantage of this in silico approach relative to empirical experiments is that the characteristics and location of the perturbation can be systematically changed and investigated. A key application is in determining the optimal brain location to administer focal stimulation (e.g. deep brain stimulation) in order to elicit a desired change in network-wide brain activity. In this case, brain regions can be modelled as populations of inhibitory (I) and excitatory (E) neurons. These populations (nodes) are interconnected with each other according to an individual's connectome. A simple approach to model the impact of brain stimulation is to change the E-to-I ratio of the targeted node. Check out some of the seminal work by Alstott et al, 2009 and Gollo et al, 2017.
Brain networks are astoundingly complex! One approach to reduce this complexity and understand brain network organisation is with generative network modelling. It turns out that networks showing several of the key organisational properties of the human connectome can be grown in silico using a few simple topological rules. These include probabilistic rules to encourage connections between nodes in close spatial proximity and nodes connected to similar neighbours (homophily). The extent to which these rules are enforced can be tuned to grow networks that resemble empirically measured connectomes. The tuning parameters can be compared between populations and provide new insight into brain network organisation. Check out the seminal research on generative modelling in network neuroscience by Vertes et al, 2012 and Betzel et al, 2016.
Further information: Check out our lab website for further details: www.sysneuro.org