Keywords: Signal and Image Analysis, Sparse Representation, Machine Learning, Inverse problem, Super resolution, Non linear estimation, Real-Time fMRI, EEG, Medical Imaging
Description:Inria is seeking a highly qualified young researcher with strong background in applied mathematics, machine learning and image processing for a post-doctoral research project around inverse problems in neuro-imaging.
Objectives of the project:The post-doc will design a coupling model in the context of multi-sensor (fMRI & EEG) brain imaging. The model will combine a linear inverse problem formulation with sparse regularizations through efficient proximal algorithms.
We primarily envision to learn a coupling model able to extract the useful signals by combining a) a common high-resolution spatio-temporal representation of brain activity, characterized by sparsity in an appropriate dictionary; b) models of the acquisition process in each of the considered sensors. Learning a multi-sensor coupling model will therefore consist in learning both the dictionary in which brain activity is sparse, and in learning the parameters of the acquisition process (this can be seen as a calibration process).
Different learning algorithms and sparsity-enforcing penalties will be evaluated in order to exhibit which brain areas are activated at a certain time. The significance of the methods and the effectiveness of the algorithms will be demonstrated through numerical investigations and in-vivo experiments (normal controls, psychiatric disorders and stroke).
Context and environment:
The proposed position arises in the context of the HEMISFER frontiers project (http://hemisfer.cominlabs.ueb.eu) which aims at making full use of the neurofeedback (NF) paradigm in the context of novel neuro-rehabilitation procedures.
This work will be conducted jointly with the VISAGES and the PANAMA Inria research Teams, and will benefit from a research 3T MRI system equipped with an integrated high resolution EEG recording system provided by the NeurInfo platform.
Application package
Applicants should send their complete application package by email to <Christian.Barillot@irisa.fr; Remi.Gribonval@inria.fr; Pierre.Maurel@irisa.fr> . This includes:
- A Motivation letter
- A Complete CV with publication list
- A PDF of one representative paper (or slideshow) of the candidate in connection with this project
- At least two recommendation letters (preferably directly sent by the mentor)
- Incomplete applications will not be processed.
We refer the candidates to the following websites for more information:
https://www.irisa.fr/visages/positions/index#post-doctoral_fellowship
https://www.irisa.fr/visages/
http://www.neurinfo.org
http://hemisfer.cominlabs.ueb.eu
