Keywords
deep learning, signal processing, MR spectroscopy, HRMAS, parameter estimation
Scientific context
In vivo Magnetic Resonance Spectroscopy (MRS) provides spatially resolved, non invasive and non-ionizing metabolic information on living organs. Quantification of MRS data, i.e. extraction of metabolite concentrations from the MRS signal, is an ill posed problem: robust and accurate quantification remains difficult, which limits the use of MRS in clinical routine. HRMAS MRS (high resolution magic angle spinning) is a method of choice for ex vivo metabolomics. Like with for in vivo MRS, quantitative data must be exacted from spectra using a similar model. Deep learning (DL) has emerged as a practical approach to solve such ill posed problems and our team is the first that have proposed to use deep learning for MRS quantification [1, 2]. The purpose of this proposal is to go further in this work, using strategies dedicated to the MRS and ex vivo HRMAS specificities.
Partners
- M. Sdika, Image and Models Team, CREATIS, (deep learning)
- H. Ratiney, NMR and Optics Team, CREATIS, (MR spectroscopy acquisition and analysis)
- F. Fauvelle, Grenoble Institute of Neuroscience (GIN), (metabolomics and HRMAS acquisition and analysis)
Skills
Strong knowledge in at least one of the following fields is required:
- Machine learning (deep learning)
- Signal processing
- Applied mathematics;
The available code is written in Matlab and Python. The successful candidate is expected to be autonomous and show strong motivation and interest in multidisciplinary research (signal processing, machine learning and MR spectroscopy in a medical context).
Applications
Interested applicants are required to send a cover letter, CV and any other relevant documents (reference letter, recent transcripts of marks,...) to: michael.sdika[at]creatis.insa-lyon.fr and helene.ratiney[at]creatis.insa-lyon.fr.
Gratuity
≈ 580 e/mois.
References
[1] Nima Hatami, Michaël Sdika, and Hélène Ratiney. Magnetic Resonance Spectroscopy Quantifica-
tion using Deep Learning. In MICCAI, Grenada, Spain, October 2018.
[2] Nima Hatami, Michaël Sdika, and Hélène Ratiney. Towards handling artefacts in Convolutional
Neural Networks-based MRS quantification. In ISMRM MRS Workshop, Utrecht, Netherlands,
October 2018.