Neurodynamics in fMRI and its clinical applications


Antoine Bernas

PhD candidate

Sveta Zinger


Bert Aldenkamp


Why neurodynamics?

Magnetic resonance imaging (MRI) is one of the most powerful brain imaging technique. It allows, in vivo, a visualization at submillimeter scale of brain tissues, and hence, represents a great tool for diagnoses of brain diseases, disorders or impairments. With the emergence of functional MRI (fMRI), not only structures, but also brain activity is assessable. FMRI images can provide insights on the functional brain and its network organization (functional connectivity), and hence deliver new descriptors for brain disorders when brain structures are not impaired. This is particularly interesting in case of cognitive impairments or behavioral and developmental brain disorders. However, in some pathologies, neither brain morphometry nor its functional connectivity seems to differ when compared with a ‘healthy’ population. This is the case for individuals with autism spectrum disorder (ASD). That is why, over the past decade, neuroscientists have started to investigate the effectiveness of the functional connectivity between nodes of a given functional networks (within the network), or between the networks. But more interestingly, it is the changes in these effective connectivity values that can provide new information over the development of a pathology (such as autism). That is what we describe here as neurodynamics.

A concrete case

In order to prove the concept that brain dynamics is insightful in understanding a brain disorder and may be helpful for diagnosing it, a framework to assess and compare brain dynamics between two cohorts was built, and showed interesting and clinically relevant results in case of adolescents with ASD. The figure below shows the directed brain dynamic pattern between the ventral stream and executive networks that has a weaker effective connectivity in autism (as compared to controls).


Limitations and new direction

The study on autism uses Granger causality for comparing neurodynamic patterns, and therefore, assumes stationarity in brain signals – which is quite not the brain dynamics realm. Also Granger causality gives us only one value to describe causal-effect between two networks over the full scan length, and hence, loses the time properties of the effective connectivity - when is that causal-effect within or between networks strong or weak?- Therefore, a new direction using wavelet coherence, that allows to describe localized (in time and frequency spaces) correlations between two signals, is currently investigated. Wavelet coherence maps and their localized phase information between two functional networks have already produced great results in case of autism – creation of a wavelet coherence-based classifier for autism is on-going.

A multidisciplinary project Finally, this project is part of Neu3-ca, a multidisciplinary and international research program. Neu3-ca stands for neurodegeneration, neuronal network, and neuromodulation in epilepsy-induced cognitive ageing. This program aims at looking at the accelerated cognitive ageing phenomenon observed in certain types of epilepsy, its network organization, and the hypothetic cognitive improvement by neuromodulation (brain stimulation). Therefore, brain dynamics parameters could represent great descriptors of the cognitive decline and the origin of the accelerating process. Neurodynamics are also insightful to help describe or predict possible outcomes of brain stimulation. Finally such fundamental research on brain dynamics could help building a cognitive ageing model, not only in case of brain pathology, but also for the ‘healthy’ population.