Offre en lien avec l’Action/le Réseau : RoCED/– — –

Laboratoire/Entreprise : GREYC UMR 6072
Durée : One year
Contact : luc.brun@ensicaen.fr
Date limite de publication : 2022-04-29

Contexte :
We are seeking an outstanding postdoctoral research fellow with experience in deep learning / machine learning to work with us at Caen University, France on a project investigating the analysis of multimodal time series for the characterization of brain functional connectivity in different levels of wakefulness.

The postdoctoral position is funded under the research project LOR supported by the Region Normandy (France). The LOR project gathers engineering schools and universities.

Sujet :
Background

Brain activity can be recorded in humans either by techniques based on the
metabolic functioning of the neuron, such as Positron Emission Tomography (PET) or Magnetic Resonance Imaging (MRI), or by techniques based on the electrical functioning of the neuron, such as electroencephalography (EEG) or magnetoencephalography (MEG). If the first type of measurement allows to obtain recordings with an interesting spatial resolution, the second type allows a higher temporal resolution. To summarize, these two approaches are both imperfect but complementary.

In this project, we are interested by this double approach in the context of
the human characterization of different levels of wakefulness (from full awake to deep sleep). Data in animals indicate that the transition from wakefulness to sleep causes changes in the relationships between different brain structures. Sensory inputs via the thalamus are inhibited leading to a decrease in thalamo-cortical links in favor of intra-cortical relations. These modifications progressively isolate the cortex in order to facilitate the descent into sleep. In terms of connectivity, these modifications are reflected in EEG by a clear decrease in global long-distance connections which are progressively replaced by an intensification of local cortico-cortical connectivity [1]. On the other hand, MRI connectivity analyses tend to show that the spatial extent of the networks is preserved during the early phases of sleep [2]. It is necessary to explore this apparent paradox in order to better understand the cerebral mechanisms at play during the descent into sleep but also more broadly to explore the networks of human consciousness. . .

Objectives and challenges

The project is based on data from a cohort being currently acquired. It includes EEG and MRI acquisitions performed while the subjects are falling asleep for a nap. In a first step, the candidate will study the dynamic evolution of the functional connectivity measured in EEG as a function of the correlation metric (spectral coherence, synchronization probability, phase synchronization method, etc.). In a second step, these results will have to be compared to those obtained in MRI by taking into account the physical characteristics of the different signals.

Work plan

In both cases (EEG/MRI) the correlations between the different areas will be
measured using positive defined matrices measuring the correlation of the signals. For the EEG, these correlations can be measured directly from the temporal signals or from time-frequency analyses. In a first step, and for the EEG, we will have to characterize the matrices corresponding to the different phases of sleep using the calculation of averages on the variety of positive defined matrices [3]. In a second step, we will try (in EEG as well as in fMRI) to design recurrent networks on such matrices [4] in order to automatically classify the sleep phases as the acquisition progresses.

Profil du candidat :
* Interpersonal skills and the ability to work well individually or as a member of a project team are recommended.
* Good written and verbal communication skills are required, the candidate
has to be fluent in spoken French or English and written English. Working
language can be English or French.

Formation et compétences requises :
* The candidate must have a recent Ph.D. (within 5 years) in Computer
Science (or Applied Mathematics) in the field of Machine Learning.
* Knowledge and experience within Deep Learning frameworks is highly
recommended.
* The candidate will perform research and algorithmic developments and
solid programming skills are required.

Adresse d’emploi :
Interested candidates should submit their application to

• luc.brun@ensicaen.fr and
• olivier.etard@unicaen.fr

Please include in your application email one Curriculum Vitae, one statement of research letter explaining your interest and your skills for this position, and 2 reference letters (all in a single pdf file). Applications will be admitted until the position is filled.

Document attaché : 202111161052_postdoc_en.pdf