Vous êtes ici : Accueil » Kiosque » Annonce

Identification

Identifiant: 
Mot de passe : 

Mot de passe oublié ?
Détails d'identification oubliés ?

Annonce

22 décembre 2017

Deep learning for ElectroCorticography (ECoG) neuronal signal decoding for motor BCI


Catégorie : Stagiaire


Accurate neural signal decoding is a key point of BCI. In CLINATEC BCI project movement, information is currently decoded from space-time-frequency features extracted from the ECoG recording from sensory-motor cortex. Nevertheless, end-to-end decoding of movement information from raw neural signal (EEG and ECoG) using Deep Learning (DL) approach was reported recently. This approach optimizes feature extraction in parallel to nonlinear decoding. The objective of internship will be the test and the comparison of DL to conventional ECOG decoding approaches for both classification and continuous 3D trajectory reconstruction tasks. To test DL for continuous 3D upper limb movement decoding (regression) and/or state (rest/active/multi-limb) classification from ECoG recordings, student will study existing deep learning software tools as well as state of art of applications of such approaches for BCI. Then he/she will try one or more Deep Neural Networks (CNN, RNN etc.) to solve the problems of continuous decoding and classification of preclinical and clinical data sets.

 

Context

The French Alternative Energies and Atomic Energy Commission (CEA) is a key player in research, development and innovation in four main areas: defense and security, nuclear and renewable energies, technological research for industry, fundamental research in the physical sciences and life sciences. Clinatec is a biomedical research center of CEA based at the ‘Polygone Scientifique’ in Grenoble. Opened in 2012, Clinatec assembles medical doctors, biologists and specialists in micro- and nanotechnologies and electronics to promote innovative treatments and diagnostic methods and physiopathological research. The priorities are cancer, neurodegenerative diseases and handicap.

Clinatec develops a number of projects, including the Brain Computer Interface (BCI) project (http://www.leti-cea.com/cea-tech/leti/english/Pages/Industrial-Innovation/Demos/BCI.aspx). The internship will be performed in the context of this project. The overall goal of project is the clinical application of BCI, allowing a quadriplegic subject to control complex effectors with multiple degrees of freedom such as exoskeleton. To do this, a fully implantable chronic recording device WIMAGINE® record the electrical activity (Electrocorticography, ECoG) of the sensory-motor cortex. The data collected with an electrodes array are sent to an external receiving station to process the information and to detect the changes of the brain activity induced by the patient’s intention to execute a movement. After the signal decoding the commands is sent to activate the effectors. Clinical trial is in progress.

Internship project

Accurate neural signal decoding is a key point of BCI. In CLINATEC BCI project movement, information is currently decoded from space-time-frequency features extracted from the ECoG recording from sensory-motor cortex. Nevertheless, end-to-end decoding of movement information from raw neural signal (EEG and ECoG) using Deep Learning (DL) approach was reported recently. This approach optimizes feature extraction in parallel to nonlinear decoding. The objective of internship will be the test and the comparison of DL to conventional ECOG decoding approaches for both classification and continuous 3D trajectory reconstruction tasks.

To test DL for continuous 3D upper limb movement decoding (regression) and/or state (rest/active/multi-limb) classification from ECoG recordings, student will study existing deep learning software tools as well as state of art of applications of such approaches for BCI. Then he/she will try one or more Deep Neural Networks (CNN, RNN etc.) to solve the problems of continuous decoding and classification of preclinical and clinical data sets.

Profile

The internship is addressed to Master students of Engineering Universities (Ecole d’ingénieurs) (electrical engineering / computer science) with knowledge in machine learning, Artificial Intelligence (AI), signal-image processing. Experience in MATLAB and Python is required. Knowledge of TensorFlow is an advantage.

Unit

Department/Service/Laboratory

CEA – LETI – Clinatec

Address

CEA/GRENOBLE 17 rue des Martyrs 38054 Grenoble CEDEX 9

Responsible persons

Dans cette rubrique

(c) GdR 720 ISIS - CNRS - 2011-2018.