Skip to Main content Skip to Navigation

Développement méthodologique pour l’optimisation de l’imagerie des propriétés électromagnétiques en IRM

Abstract : It is now well accepted that electromagnetic (EM) properties of biological tissues are characteristic features related to their contents or their structure. It has been an old issue to seek for ways to estimate these properties in vivo, and has thus led the scientific community to develop numerous specific tools, from raw measurement technologies to imaging methods more recently. On a fundamental level, these works make it progressively possible to reveal some specific physiological mechanisms and are contributing to improve our understanding of the living. Recently, Magnetic Resonance Imaging (MRI) has become a privileged tool in this framework, allowing among other things the reconstruction of electrical and magnetic properties in different frequency ranges, with its distinctive resolution power. Recent progress suggest it could be possible to provide a tissue-specific electromagnetic MR contrast, that we consider as promising new biomarker from a clinical perspective. In a quest to give new insights for a better understanding of in vivo electromagnetic phenomena, as well as contributing to a comprehensive approach of EM modelling, we have endeavoured to develop an innovative electromagnetic mapping method with an MR scanner. In the MR community, the study of EM properties led to the development of two main research fields: the so-called “low-frequency imaging”, under 1 MHz, and the “high-frequency imaging”, above 50 MHz, depending on the device used for stimulation. Given both these approaches, we have considered strategies that avoid additional hardware, and that could provide qualitative as well as quantitative results in the context of a classical clinical examination then. Low-frequency methods have been evaluated with simulation tools and have been progressively dismissed for practical and theoretical reasons: in that frequency range, information is polluted by the noise. Conversely, we have developed a new mapping method for electrical properties in the high frequency range, built from existing methodologies. Importantly, we wanted to provide a method that could easily be translated to clinical applications at a reasonable computational cost. To that end, we first performed simulation studies, and then MR acquisitions with specific dedicated EM phantoms. We finally used volunteers’ in vivo data to assess the performance or our algorithm in a realistic context. Our reconstruction method fits particularly well with acquisition schemes based on gradient-recalled echo with ultrashort echo-times (UTE), or more dramatically with zero echo-time (ZTE). By isolating the local EM signature in the MR signal, we use them to provide quantitative maps of electromagnetic properties, and we are able to estimate the sensitivity of these reconstructed maps to our model parameters. Our simulation results first and foremost show that our method improves the overall theoretical reconstruction quality as compared to existing related mapping techniques. Qualitative results confirm the possibility of a direct distinction, in terms of contrast, between media with variables electromagnetic properties. Quantitative results are encouraging, we observe satisfactory absolute values for reconstructed EM properties in the given frequency range. Our framework contributes to the development of EM imaging in MRI, and gives new insights for reconstruction model optimization. Efforts are still needed to achieve better use of UTE/ZTE sequences and to improve the overall quality of our reconstructions. After final numerical optimization, the reproducibility of the method will be evaluated in several test organs before its integration to a standard clinical protocol.
Document type :
File URL :
Complete list of metadatas
Contributor : Thèses Ul <>
Submitted on : Tuesday, February 2, 2021 - 12:01:08 PM
Last modification on : Friday, February 5, 2021 - 4:48:22 PM


  • HAL Id : tel-03128553, version 1


Paul Soullié. Développement méthodologique pour l’optimisation de l’imagerie des propriétés électromagnétiques en IRM. Imagerie. Université de Lorraine, 2020. Français. ⟨NNT : 2020LORR0152⟩. ⟨tel-03128553⟩



Record views