Brain: towards new tools to estimate, quantify and compare brain dynamics

The brain - and in particular the human brain - is probably the most complex organ in existence.

Understanding how it works remains a challenge for modern science. If we understand fairly well what happens at the level of the neuron, the macroscopic behavior of this system, which contains some 100 billion neurons, remains - all in all - rather poorly understood.

Over the last ten or twenty years, the vision of the organization and activity of the brain has evolved. We have gone from a rather static vision where cerebral areas were associated with specific tasks to a much more complex vision where different areas "network" to perform a task in a dynamic way (the networks fluctuate according to time).

One of the challenges of current science is to have the means to better understand macroscopically how our brains function or dysfunction.

Providing tools to estimate, quantify and compare brain dynamics at the macroscopic level is the central topic of Cronos.

We aim to develop models and algorithms that allow us to estimate, understand and quantify the whole brain in a non-invasive way.

For this, we want to use the unifying concept of dynamic network to integrate and analyze information from various modalities of observation of brain function in coherent global models. With this in mind, we want to develop methods for data analysis and processing at different levels of abstraction.

First and foremost, there are clinical applications.

We hope that our tools will enable doctors to better diagnose, understand or treat various brain-related diseases.  More playful applications (but which also have clinical applications) are the brain-computer interfaces which allow to "launch commands" directly from the measured signals without going through a muscular activity.

We work mainly with hospitals, which - fortunately - are not quite "industries", but are often assimilated as such when we talk about industrial relations.

Our closest partners are the Pasteur hospital in Nice and La Timone in Marseille. They allow us to have access to data and help us define objectives that will ultimately be useful to patients, which is crucial for our activity. We also have some more typical industrial relationships with equipment suppliers (either as customers or as partners in contracts) that allow us to use and study new sensors that will improve the quality of the signals we work on or simplify their use.

Imaging modalities (in the broadest sense) that allow us to measure brain activity have multiplied and greatly improved over the last ten or twenty years. At the same time, some of them have become - relatively - more democratic. As in many other fields, there is an explosion in the amount of data available.

A major challenge remains that of merging these data, which can be very heterogeneous, both in terms of the nature of the quantities measured and their spatial or temporal characteristics. Moreover, we have moved from a static vision of brain activity (we were looking at what was happening at a given moment), to a much more dynamic vision.

Our research topics are at the border between signal and image processing (computer science), neuroscience - since we are interested in data processing that measures the activity of the brain and the central or peripheral nervous system - with the use of mathematical tools or numerical analysis. Finally, we regularly work with biologists or physicians for the more medical aspects.

In addition to my participation in internal Inria committees, I am an elected member of the academic council of the Université Côte d'Azur.

I am also a member of the scientific council of the Neuromod Institute, which I also represent on the Scientific and Pedagogical Orientation Council (COSP) of the University Research School (EUR) Healthy.

I also participate in a few editorial boards for various journals and conferences.