Prof. Daniele Marinazzo (UGent) uses supercomputing to model the brain’s function in health and disease.
Marinazzo research lab at Department of Data Analysis, UGent
Prof. Daniele Marinazzo leads a research lab within the Department of Data Analysis, based in the Faculty of Psychological and Pedagogical Sciences at Ghent University.
His group focuses on methodological and computational aspects of neuroscience research. Specifically, they examine how the brain works and how the collective dynamics and interplay of brain regions give rise to function.
Marinazzo’s lab is part of the multi-disciplinary research platform ‘Neuroscience’ at Ghent University.
Computational research and applications
The human brain can be considered as a complex network, a so-called ‘connectome’, from which all function and behaviour can in principle be inferred. A computational model of such a connectome is therefore a very powerful tool, with predictive power and many medical and clinical applications.
Prof. Marinazzo’s group develops new techniques to determine to what extent brain areas are structurally connected or functionally related, based on the data provided by neuroimaging methods such as magnetic resonance imaging (MRI). They employ data-driven and biological dynamical causal modeling, in addition to model-free theoretical frameworks, such as transfer entropy.
Thanks to these developments, and with the aid of supercomputers, Marinazzo’s research group can construct a comprehensive whole-brain map of the structural and functional connections within the brain of an individual patient.
Dynamical network analysis of these ‘brain maps’ in patients with epilepsy has led to an unprecedented characterisation of the epileptic condition, enabling a more objective diagnosis and monitoring of recovery or progression processes over time. Furthermore, it opens up the possibility to predict the onset of epileptic seizures in a patient, and teach patients how to identify the onset and take remedial treatment.
Virtual brain models can also be used to study the mechanisms underlying brain diseases, and carry the potential to directly support treatments. Hannelore Aerts, one of Prof. Marinazzo’s PhD students, recently applied these novel techniques to examine global brain dynamics in brain tumour patients. The final aim of the project is to assist neurosurgeons in pre-surgical planning, by simulating the effects of tumour resection on brain dynamics and functioning using a virtual brain model.
VSC facilitates computational research
“To derive the connectivity architecture of brain regions from recorded neuroimaging data, considerable computing power is required” says Prof. Marinazzo.
“But also our analyses need to run on supercomputers. The availability of several compute clusters at Ghent University and within the VSC network enables us to swiftly test several models at the same time. It really speeds up our research.”
“Finally, the user-specific approach and fast-response helpdesk of the HPC-UGent team is a tremendous help.”
- ‘Brain networks under attack: robustness properties and the impact of lesions’, Hannelore Aerts, Wim Fias, Karen Caeyenberghs and Daniele Marinazzo, Brain, 2016, 139, 3063-3083.
- ‘Modeling global brain dynamics in brain tumor patients using The Virtual Brain’, Hannelore Aerts and Daniele Marinazzo, Front. Neurosci. Conference Abstract: 12th National Congress of the Belgian Society for Neuroscience. doi: 10.3389/conf.fnins.2017.94.00101