Vous êtes ici : Accueil » Kiosque » Annonce


Mot de passe : 

Mot de passe oublié ?
Détails d'identification oubliés ?


5 mars 2017

Characterizing Brain Networks in Real Time Across Frequencies Based on a Combination of Time Frequency and Source Localization Methods

Catégorie : Doctorant

A 3 years PhD fellowship is available at Aix-Marseille University for a research project entitled

Characterizing Brain Networks in Real Time Across Frequencies Based on a Combination of Time Frequency and Source Localization Methods

under the joint supervision of Christian Benar (Institut de Neurosciences des Systèmes) and Bruno Torrésani (Institut de Mathématiques de Marseille).

The fellowship will start at the fall 2017 (october 1st); applicants must be at the time of recruitment, in possession/or finalizing the master degree or equivalent degree, and must at the time of recruitment, not to have resided or carried out their main activity (work, studies, etc.) in France for more than 12 months in the 3 years immediately prior to the reference date (October 1st, 2017). Complete eligibility criteria and details on the research project can be found at the url

Detailed Project and eligibility criteria

The application form can be found at Online application form

For more details, please contact



Key concepts and acronyms

Magnetoencephalography (MEG): state-of-the art electrophysiological technique that records non-invasively the very small magnetic fields produced by brain signals, based on sensors in a supraconductive state. Application of advanced source localization methods make MEG an imaging tool with both high spatial and high temporal resolution.

Intracerebral EEG (sterotaxic EEG, SEEG): invasive method that consists in placing depth electrodes within the brain. This technique is routinely used during presurgical evaluation of epilepsy, and permits recording directly within brain structures, including deep regions such as hippocampus and amygdala. Presurgical evaluation of epilepsy: patients with epilepsy that are not responding to pharmaceutical treatment can be evaluated for a surgical approach, where the brain region responsible for seizure is removed by resective surgery or neutralized by radiosurgery.

State of the art

It is now commonly accepted that the substrate of brain function is the activation of networks at different spatial and temporal scales (Nikolic et al., 2013). Thus, cerebral activity relies on the dynamical interaction between brain areas at different spatial and temporal scales. The dysfunction of brain networks underlies several pathologies, one of the most common being epilepsy (Bartolomei et al., 2010). Functional imaging techniques permit to map the regions involved in brain networks as well as their patterns of interaction, the most common methods being magnetic resonance imaging (fMRI) and electrophysiology techniques (electroencephalography, EEG and magnetoencephalography, MEG). The great advantage of electrophysiological tools is their ability to track neuronal activity at its temporal scale of activity, i.e. on the order of the millisecond (Sergent et al., 2005). In particular this permits characterizing activity across different frequency range, from the classical low frequency bands (between 1 and 30 Hz), towards the gamma band (between 35 and 80 Hz) that have been shown to be a marker of regional neuronal activation, related to fMRI results (Lachaux et al., 2007). Some recent work push the limit to very high frequencies, which are thought to be highly involved in memory processes and to be a marker of epileptic tissues (Urrestarazu et al., 2007). An area of much interest is that of cross frequency interactions, that are believed to be a key player in brain network functioning (Canolty et al., 2006; Florin and Baillet, 2015).

However, the difficulty of noninvasive electrophysiology tools is the fact that they rely on surface measurements. In order to be a full imaging method, they thus require solving a difficult inverse problem that projects the surface data towards the cortex (Baillet et al., 2001). This results in the estimation of the time course of each brain region, a sort of 'virtual electrode', on which methods of analysis of brain networks can be performed (see Figure 1). The inverse problem is ill-posed, which requires the use of mathematical constraints that are not always physiologically justified. Many methods have been proposed to solve the inverse problem, but several other difficulties remain: (i) The sensitivity of methods to the signal to noise ratio, which is a key issue when studying spontaneous activity (ii) the 'source leakage' arising from the blurred reconstruction arising from the inverse problem methods, resulting in spurious correlations between brain regions (Brookes et al., 2012) (iii) the mixture of transient and oscillatory activity that overlap in the frequency domain (Jmail et al., 2011), with varying amplitudes resulting from the 1/f nature of electrophysiological signals (Dehghani et al., 2010). So far, all these issues have been tackled mostly separately, with progress in inverse problem, wavelet analysis, orthogonalization methods, convex and non-convex optimization. However, a sensible strategy would be to combine all these approaches into a common multi-dimensional framework, that would take advantage of their added values while avoiding the pitfalls that have been attributed to the separate approaches.

The optimization of cross-frequency characterization of brain networks in low signal to noise conditions would be beneficial both to clinical practice and fundamental research. One particularly interesting application would be in the field of epilepsy, which is a disease of brain networks. For pharmacoresistant patients, the proposed solution is mostly based on surgery, consisting in removing a part of the brain. A new promising approach would be to help patients reshaping their brain networks based on neurofeedback (Kubik and Biedron, 2013). Within this approach, the synchrony of the pathological network, reconstructed thanks to the tools developed within this project, would be shown to the patient, who could thereby learn how to actively desynchronize such pathological activity. This technique has been used with electrodermal skin response (Micoulaud-Franchi et al., 2014), with encouraging results, and is expected to be even more efficient when based on actual brain network activity.


The objectives of this PhD project are two-fold:


The dynamical brain mapping (Dynamap) team (head C Bénar): This team is within the Institut de Neuroscience des Systèmes (head V Jirsa) The principal objective of the Dynamap group is to design and optimize signal processing methods for multimodal functional investigation of human cerebral activity (pathological and physiological). Our interests are structured into two research axes: (i) Fusion of recordings from multimodal non-invasive techniques (ii) Confrontation of non-invasive results with depth EEG. Our research is in strong collaboration with the other teams of the laboratory, as well as the Clinical Neurophysiology department and the Stereotactic and Functional Neurosurgery department of the Timone hospital, Marseille (AP-HM).

The magnetoencephalography center is part of the Institut de Neurosciences des systèmes, and is hosted within the clinical neurophysiology department of the Timone hospital (Assistance Publique - Hôpitaux de MArseille, AP-HM). It is an open platform for users from the neuroscience community, and is part of the collaborative Convergence projet Institute language, communication and brain (ILCB). It also takes part in the presurgical evaluation of patients with epilepsy, though the involvement in a day hospital, in link with AP-HM.

Institut de recherche mathématique The "Institut de Mathématiques de Marseille" (I2M, UMR 7373) is a joint research unit Aix-Marseille Université/CNRS/Centrale Marseille. It hosts around 130 teacher-researchers, 30 CNRS researchers, 15 technical and administrative staff, 60 PhD students and 20 postdocs. The I2M arose from the fusion, on janvier 1, 2014, of the LATP (Laboratoire d’Analyse, Topologie et Probabilités) and the IML (Institut de Mathématiques de Luminy).

Other funds: A pre-prosal was submitted to the call from Agence Nationale de la Recherche, "projet de recherche collaborative-internationale" (PRCI). A full proposal will be submitted in April 2017. This proposal includes the Canadian partner (concordia university). This grant will permit funding travels, as well as recruit other members (postdocs). To be noted: the current PhD project as proposed is feasible even without this grant. In particular, all the equipment, infrastructures, support of permanent engineers are already in place.


Baillet S, Mosher J C and Leahy R M 2001 Electromagnetic brain mapping IEEE Signal Processing Magazine 18 14-30

Bartolomei F, Cosandier-Rimele D, McGonigal A, Aubert S, Regis J, Gavaret M, Wendling F and Chauvel P 2010 From mesial temporal lobe to temporoperisylvian seizures: a quantified study of temporal lobe seizure networks Epilepsia 51 2147-58

Brookes M J, Woolrich M W and Barnes G R 2012 Measuring functional connectivity in MEG: a multivariate approach insensitive to linear source leakage Neuroimage 63 910-20

Canolty R T, Edwards E, Dalal S S, Soltani M, Nagarajan S S, Kirsch H E, Berger M S, Barbaro N M and Knight R T 2006 High gamma power is phase-locked to theta oscillations in human neocortex Science 313 1626-8

Dehghani N, Bedard C, Cash S S, Halgren E and Destexhe A 2010 Comparative power spectral analysis of simultaneous elecroencephalographic and magnetoencephalographic recordings in humans suggests non-resistive extracellular media J Comput Neurosci 29 405-21

Florin E and Baillet S 2015 The brain's resting-state activity is shaped by synchronized cross-frequency coupling of neural oscillations Neuroimage 111 26-35

Jmail N, Gavaret M, Wendling F, Kachouri A, Hamadi G, Badier J-M and Bénar C-G 2011 A comparison of methods for separation of transient and oscillatory signals in EEG Journal of neuroscience methods 199 273-89

Kubik A and Biedron A 2013 Neurofeedback therapy in patients with acute and chronic pain syndromes--literature review and own experience Przegl Lek 70 440-2

Lachaux J P, Fonlupt P, Kahane P, Minotti L, Hoffmann D, Bertrand O and Baciu M 2007 Relationship between task-related gamma oscillations and BOLD signal: new insights from combined fMRI and intracranial EEG Hum Brain Mapp 28 1368-75

Lina J M, Chowdhury R, Lemay E, Kobayashi E and Grova C 2014 Wavelet-based localization of oscillatory sources from magnetoencephalography data IEEE Trans Biomed Eng 61 2350-64

Micoulaud-Franchi J A, Kotwas I, Lanteaume L, Berthet C, Bastien M, Vion-Dury J, McGonigal A and Bartolomei F 2014 Skin conductance biofeedback training in adults with drug-resistant temporal lobe epilepsy and stress-triggered seizures: a proof-of-concept study Epilepsy Behav 41 244-50

Nikolic D, Fries P and Singer W 2013 Gamma oscillations: precise temporal coordination without a metronome Trends Cogn Sci 17 54-5

Sergent C, Baillet S and Dehaene S 2005 Timing of the brain events underlying access to consciousness during the attentional blink Nat Neurosci 8 1391-400

Urrestarazu E, Chander R, Dubeau F and Gotman J 2007 Interictal high-frequency oscillations (100-500 Hz) in the intracerebral EEG of epileptic patients Brain 130 2354-66


Dans cette rubrique

(c) GdR 720 ISIS - CNRS - 2011-2015.