Offre de thèse
ENACT - Optimizing MRI for Energy Systems: Integrating Artificial Intelligence into Imaging Methods and Data Processing
Date limite de candidature
25-04-2025
Date de début de contrat
01-10-2025
Directeur de thèse
PERRIN Jean-Christophe
Encadrement
pas de co-encadrement défini à cette heure
Type de contrat
école doctorale
équipe
Axe Transverse IRMcontexte
The “MRI for Engineering” team at LEMTA is developing research into mass transfer and fluid flows in the fields of energy and energy processes. The methods used are based on nuclear magnetic resonance (NMR), in the time domain (TD-NMR) and imaging (MRI). The subjects studied are highly interdisciplinary, often involving several laboratories and industrial players, with applications such as nuclear safety, membrane processes, mixing phenomena, water transfer in fuel cells, heat storage, etc... Although efforts to develop robust and reliable methods in healthcare have been intense and fruitful, their optimization for energy and process applications still requires further development: in such systems the samples are of variable size and composition, which makes the NMR signal weak and the images noisy. Furthermore, the fluids flow in complex geometries, often at high velocity. The proposed thesis topic is part of a new initiative to implement artificial intelligence (AI) tools at various stages of the experimental workflow, for more efficient data and image processing, and to improve acquisition methods, particularly in magnetic resonance velocimetry (MRV). The thesis will be carried out between the LEMTA laboratory and the startup APREX , which develops AI tools based on adapted algorithms and software suites with the aim of contributing to the digital transition of industry. The new tools and methods will be implemented on the software used to control the new MRI platform at LEMTA.spécialité
Énergie et Mécaniquelaboratoire
LEMTA – Laboratoire Energies & Mécanique Théorique et Appliquée
Mots clés
MRI, AI, images, data processing
Détail de l'offre
The Ph.D. thesis will address the following points:
1. Develop of an MRI image processing methodology specific to the study of energy systems and industrial processes using tools developed by APREX
o Correction of geometric artefacts and aberrations due to magnetic field inhomogeneities
o Correction of aliasing phenomena on velocity maps
o Noise reduction
o Reconstruction of images from incomplete or compressed data
2. Training AI models to improve image quality and reduce acquisition time
o Training of models based on MRI images recorded on samples with controlled geometries produced by additive manufacturing
o Use of AI models to study flow dynamics on these controlled systems, by correlating MRI data with predictive models
o Correction of measurement errors by correlation with computational fluid dynamics simulations
o Develop tools for faster, more robust quantitative analysis
3. Implementation of AI tools in the MRI acquisition software of the Metro'NRJ platform
o Assistance during acquisition by defining optimal parameters
o Improved acquisition frequency
o On-the-fly image processing
Keywords
MRI, AI, images, data processing
Subject details
The Ph.D. thesis will address the following points: 1. Develop of an MRI image processing methodology specific to the study of energy systems and industrial processes using tools developed by APREX o Correction of geometric artefacts and aberrations due to magnetic field inhomogeneities o Correction of aliasing phenomena on velocity maps o Noise reduction o Reconstruction of images from incomplete or compressed data 2. Training AI models to improve image quality and reduce acquisition time o Training of models based on MRI images recorded on samples with controlled geometries produced by additive manufacturing o Use of AI models to study flow dynamics on these controlled systems, by correlating MRI data with predictive models o Correction of measurement errors by correlation with computational fluid dynamics simulations o Develop tools for faster, more robust quantitative analysis 3. Implementation of AI tools in the MRI acquisition software of the Metro'NRJ platform o Assistance during acquisition by defining optimal parameters o Improved acquisition frequency o On-the-fly image processing
Profil du candidat
Candidates should hold a Master's degree or equivalent in one or more of the following disciplines: physics, computer science, engineering, medical sciences. They should also have a strong desire to learn and develop new AI-based physical and software tools. Initial experience in AI, gained through an academic course or internship, is considered a plus.
Candidate profile
Candidates should hold a Master's degree or equivalent in one or more of the following disciplines: physics, computer science, engineering, medical sciences. They should also have a strong desire to learn and develop new AI-based physical and software tools. Initial experience in AI, gained through an academic course or internship, is considered a plus.
Référence biblio
https://hal.univ-lorraine.fr/LEMTA-UL/hal-03953016v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-04137362v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-03227560v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-02736211v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-03798574v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-01693075v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-03227587v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-04311361v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-04222627v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-04452621v1
https://hal.univ-lorraine.fr/LEMTA-UL/hal-03572921v1