Accès direct au contenu

x

Moteur de recherche interne

Grenoble Institut des Neurosciences Grenoble Institut des Neurosciences

  • Youtube
  • Linkedin
  • Twitter

Accueil > Annuaire

Accéder au plan complet du site

Page personnelle de Thomas COUDERT

Membre de

GIN - Équipe "Neuroimagerie fonctionnelle et perfusion cérébrale"

Informations de contact

thomas.Coudert1@univ-grenoble-alpes.fr

Bureau principal

Bureau principal
1/071
Tél. 
+33679546938

Autres coordonnées

Site perso :
https://www.linkedin.com/in/thomas-coudert-91b110193/

Thèmes de recherche et activités

The concept of MRI fingerprint (MRF) proposes to reconstruct images by directly comparing in vivo acquisitions (spatial 3D + temporal 1D = fingerprints) and millions of digital simulations that mimic brain tissue (dictionary). This approach is enough flexible to allow access to several MRI parameters simultaneously and to quantify them. As we have shown in clinical and preclinical studies, when the simulations are based on the physics of NMR (magnetic field, magnetic susceptibility, diffusion, relaxations, etc.) and on biophysics (cellular dimensions and vascular, flow ...), the technique allows to extract information on tissue microstructures. These data are particularly important for the management of patients with acute stroke. Until now, the digital representation of tissues has been based on simple shapes (cells = spheres, blood vessels = cylinders). In addition, the choice of the 4D sequence is currently based on rough nuclear magnetization trajectories and the comparison step is extremely time-consuming. We propose here to develop realistic biophysical simulations coupled with artificial intelligence tools integrated at different levels of the MRF frame to optimize the MRF protocol. We will thus create a tool that can be integrated into a clinical environment and capable of guiding patients more quickly to the appropriate treatments.
I work on the development of digital tools that synthesize realistic 3D structures of brain tissue and that fit into an AI-optimized MRF frame. The objectives of this methodological work are:
  • Develop code that "learns" to generate realistic tissue structures from high resolution microscopy acquisition.
  • To use an existing MRI signal simulations tool, integrate new realistic simulation modules (+ dynamic blood flow and oxygenation simulations) and build digital signal dictionaries. 
  • Develop "deep reinforcement learning" type AI tools that drive virtual MRI scanners and automatically optimize acquisition patterns (sequencing of radiofrequency pulses, relaxation time, phase, etc.).
  • Acquire in vivo brain data (human + small animal) and test the tools when the level of breathed oxygen is increased (hyperoxia) or lowered (hypoxia), then possibly in pathological situation (stroke).
MRF principle

"Ma thèse en 180 secondes" contest
The video below shows my participation at the francophone contest MT180. The objective : vulgarize my thesis topic, in french, in less than 3 minutes (180 seconds) in front of a multidisciplinary jury. It resumes in a funny way my thesis work.
MT180 IRM "fingerprinting" et intelligence artificielle pour la prise en charge des patients AVC

 

Fonctions et responsabilités

PhD student in Medical Imaging and Artificial Intelligence

Discipline(s) enseignée(s)

Biophysique et imagerie médicale, Informatique, INFORMATIQUE SCIENTIFIQUE, Neurosciences
PYTHON Introduction
Molecular Biology

Curriculum vitae

  • 2021-2024: PhD Thesis Grenoble Institut Neurosciences
    • MR Fingerprinting
  • 2021: M2 Internship Grenoble Institut Neurosciences - Deep Learning and Medical Imaging
    • Participation in a research project for the segmentation of tumors in brain MRI in patients with glioblastoma
    • R&D team Pixyl
  • 2018-2021: Master Grenoble-INP PHELMA
    • Biomedical Engineering - Medical Imaging cursus
  • 2016-2028: CPP-La Prépa des INP
    • Two years of intensive scientific courses to prepare at Engineering School
Membres
Associés renforcés
Associés simples