Maureen Clerc, listening to the neural symphony

Date :
Changed on 25/03/2020
As a senior researcher in the Athena project team, Maureen Clerc seeks to measure and interpret brain activity, combining applied mathematics and computer science with neuroscience. By contributing her numerical analysis skills, she has made her group a pioneer in the modelling of brain wave propagation. Portrait of a pragmatic mathematician who believes in interdisciplinarity, advocates open science and dreams of seeing her work lead to e-health applications.

After a first glimpse of an Inria project team during a master's internship, Maureen Clerc did her thesis at the École Polytechnique, where she had entered in 1990. She studied wavelets, a signal processing technique derived from the Fourier transform*. As a pianist, she is fascinated by this "time-frequency" formalism, so close to that of music. Very free in her thesis subject, she goes in search of new applications for this tool that she is enthusiastic about. This quest brings her to the field of computer vision, which is completely foreign to her, but which her thesis director, Stéphane Mallat, knows well. His doctorate also provided the opportunity to start a collaboration with the Department of Statistics at Stanford University, which is developing a library of signal processing algorithms. Pursuing a post-doctorate in this Stanford team is a logical continuation. All the more so as its then director, David Donoho, was a pioneer in the reproducible research she began during her thesis. For him, it is essential to publish not only the results, but also the entire method by which they were obtained; not only the algorithms, but also their implementation. You have to "do open science, open the hood so that people can see how it's done inside". A vision of science that fits well with Maureen Clerc's vision, for whom research is all about teams, collaborations and sharing.

Modeling for understanding...

Maureen Clerc's shift to the neurosciences was due to a combination of circumstances. A member of the Corps of Bridges, she quite naturally joined the School of Bridges on her return from Stanford. She then embarked on the School's project to create Odyssey, a joint team with Inria and the École Normale Supérieure. The latter is interested in the brain, to understand how human vision works, and thus improve computer vision. With her thesis in the field, Maureen Clerc has found her place in the project, to which she brings her knowledge from other research communities, particularly numerical analysis. What will interest her is brain activity. How does the electricity produced by the functioning of neurons travel through the skull and the skin of the head? How do we link the signal measured at the scalp surface to the signal emitted by the neurons? These questions lead him to develop, with another permanent researcher of the team, Théo Papadopoulo, a research axis on the modelling of the propagation of electromagnetic waves in the head. Because if we can calculate the degradation of information between the signal emitted by the brain and the signal picked up externally, this makes it possible to "de-noise" the recordings, and to go back to the actual brain activity: this is what we call the "inverse problem".

... and usefulness

Resulting from his work, theses and post-doctorates supervised on the subject, a software for the simulation of electromagnetic propagation in biological tissues, OpenMEEG, was created in 2010. Very quickly, from conferences to international benchmarks, it appears to be the best tool in its category, the one that best predicts information degradation. An advance that has not yet been denied to this day and which has enabled its code to be integrated into several functional brain imaging software packages, thus spreading to the entire scientific community. This mastery of the measurement of brain activity will give rise to the idea of exploiting this information and transforming it into commands. And thus to open up to brain-machine interfaces, a rapidly expanding field of research. The challenge here is to analyse signals in real time, to make sense of them very quickly, without the benefit of long recordings to facilitate the denoising of information. But, even before these technical questions, for Maureen Clerc, this research must have "somewhere, in the line of sight, a horizon with people for whom it would be useful".

Interfacing the human with the machine

In order to achieve this interfacing, it is necessary to work on the adaptation of the algorithms to the signal variability. Because even over a short period of time, the characteristics of the same brain activity, in the same person, are not constant: they vary in intensity and time-frequency. It is therefore also a challenge to exploit information in real time. Fortunately, some signals are much more robust, and have led to concrete applications. One example is the P300 speller, a virtual keyboard that allows paralyzed patients to communicate with the outside world (see box).  It was developed in close collaboration with the doctors at the Nice University Hospital so that the development of this application will eventually lead to products that are comfortable, easy to use and truly useful to patients. Maureen Clerc would like to extend her work to the field of epilepsy, whose high prevalence and the difficulty of treating it leave many patients bereft in the face of a very disabling disease. Her new line of research focuses on home monitoring of certain physiological parameters. The idea: a simple device, portable for daily tasks, could make it possible to better diagnose epilepsy. But the researcher is going even further by wishing to give the patient feedback on his brain activity in the hope that this neurofeedback will enable him to learn how to regulate it.

Spelling with your mind with the P300 speller Imagine a virtual computer keyboard that you can operate by focusing your attention on the keys you want to type on. There is such a tool, the P300 speller. How does it work? By monitoring brain activity, and more specifically the P300 wave, so called because it arrives 300 milliseconds after the visual stimulus. In practice, the patient is equipped with an electroencephalogram helmet. The letters on the virtual keyboard are programmed to flash in changing groups, and the patient must count the number of times the one he wants to select lights up. By comparing the activation of his P300 signal and the sequence of flashing letters, it is possible, after a few repetitions, to determine which one he has focused on and display it on the screen. A study by the Nice University Hospital has shown that it is very easy to learn how to use the device without a long learning phase. And this even for people who are not used to computer interfaces.

Interdisciplinarity as a working principle

Beyond the motivation linked to the needs of doctors and patients, these advances and research potentialities are also made possible by a very rich working and exchange environment. First of all, locally since, thanks to the efforts of numerous researchers from Inria Sophia Antipolis, the CNRS and the University of Nice, the NeuroMod [1] institute was recently set up, supported by the grouping of "Université Côte d'Azur" establishments. The institute has confirmed that neurosciences need to benefit from funding and dedicated events and intends to encourage future local collaborations in this field. Inria's support does not date back to the creation of this structure. As early as 2014, Maureen Clerc coordinated BCI-LIFT, an Inria Project Lab, a sort of super-team-project that enables different centres to work together for four years on a common objective. This brought together six Inria teams, Inserm in Lyon and the University of Rouen. This success contributed to the creation of a French association for research on brain-computer interfaces [2]. This adventure has enabled the development of skills and the exchange of good practices between researchers from different but highly complementary specialties, from electrophysiology to mathematical modeling. This double culture between high-flying theory and concrete applications is found in the very structure of the Athena project-team. Young researchers and engineers find there an excellent mastery of the mathematical foundations and a great wealth of software production. They also have the opportunity to see the concrete results of their research directly, thanks to the neurophysiological measurement laboratory at their disposal. This experimental approach is built in a true spirit of mutual aid and sharing, so dear to Maureen Clerc.

1. outil central de l’analyse du signal qui permet d’exprimer une fonction complexe en la somme d’éléments plus simples. NeuroMod, institut de Modélisation en Neuroscience et Cognition 2. Cortico, Collectif pour la Recherche Transdisciplinaire sur les Interfaces Cerveau Ordinateur,

Maureen Clerc en 5 major dates

Maureen Clerc
Photo Aurélie Macarri - Université Côte d'Azur
  • 1999 : Elle soutient sa thèse de doctorat sur le traitement du signal au Centre de Mathématiques Appliquées de l'École polytechnique.
  • 2007 : Elle obtient son Habilitation à Diriger des Recherches.
  • 2010 : Sortie officielle du code OpenMEEG, qui simule la propagation des ondes cérébrales dans les tissus biologiques.
  • 2014 : Elle reçoit le prix Pierre Faurre de l’Académie des sciences pour ses travaux sur la modélisation et l’interprétation des signaux électriques du cerveau.
  • 2016 : Sortie de l’ouvrage pluridisciplinaire Les interfaces cerveau-ordinateur (Iste éditions), qu’elle coordonne avec Fabien Lotte (Inria Bordeaux) et Laurent Bougrain (Inria Nancy).