Skip to main content

INtegrating Magnetic Resonance SPectroscopy and Multimodal Imaging for Research and Education in MEDicine

Periodic Reporting for period 1 - INSPiRE-MED (INtegrating Magnetic Resonance SPectroscopy and Multimodal Imaging for Research and Education in MEDicine)

Reporting period: 2019-01-01 to 2020-12-31

Medical Imaging plays a central role in modern healthcare systems by the development of the most advanced techniques such as Magnetic Resonance Imaging (MRI) / Magnetic Resonance Spectroscopy (MRS) and Positron Emission Tomography (PET).

Furthermore, development of multi-parametric MR and hybrid imaging (MR/PET) by combining non-invasive metabolic information of MRS (MRSI) with high sensitivity given by radiotracers of PET is crucial for the improvement of diagnosis and treatment monitoring of diseases.

Therefore, the clinical translation of these powerful hybrid techniques is necessary to focus on different diseases such as cancer and neurodegenerative brain diseases and requires high level of expertise in several methodological domains: 1) Acquisition of multi-parametric and multimodal MRS(I)/PET data, 2) Image and spectra processing, and 3) Data analysis using machine learning approaches.
The major objective of the INSPiRE-MED project is to train 15 Early-Stage Researchers (ESR) in a complex and challenging branch of medical imaging, namely MRS(I) combined with PET, which requires a solid background in physics for acquisition developments and/or in computer science for data analysis using latest machine-learning methods, to develop (pre-)clinical applications.

These goals will be achieved by:

1. Providing core scientific knowledge, essential to break down barriers between basic research and clinical use based on high experience of INSPiRE-MED partners:
• New acquisition methods of pre- and clinical imaging,
• New version of data processing software (jMRUI),
• New ML and DL methods for clinical diagnosis and prognosis.

2. Guiding the ESRs into this integrated scientific culture by training events:
• “hands-on” research training and workshops,
• Complementary “soft” skills courses from local institutions and SME industrial partners,
• Exchange with medical doctors.

3. Creating new relationships between academic research, health practice and health industry:
• Training in successful academic or industrial environments through secondments,
• Organizing joint-programs between software developments and clinical applications.
The network is organized in 6 WPs including 3 scientific research WPs, a WP for Dissemination and Exploitation, a WP for Training and a last WP for management. The preliminary results of the 5 WPs are the following:

• WP1 “Next generation methodology for enhanced MRS(I) in (pre-)clinical research and use” has developed 3 tasks including 1) a multi-echo MRS in single excitation implemented as basis for fingerprinting acquisition, spectra simulation, improved data fitting by incorporating multi-parameter approach into jMRUI through plugin for FiTAID and machine learning implementation and tests have been initiated for fingerprinting, 2) high-speed MRSI methods implemented in a prostate clinical application at 3T and on a heteronuclear (Deuterium) 7T MR system to increase spatial and temporal resolution, 3) diffusion-weighted MRS STE-LASER techniques implemented on a preclinical 14.1T system in a C hepatic encephalopathy rodent model. These works have led to 5 conference communications and 1 publication submission.

• WP2 “Multi-parametric and multimodal metabolic imaging approaches for (pre)-clinical application” has developed 3 tasks including 1) development of methods for simultaneous MR/PET in brain tumor of rodent model, 2) 3D semi-LASER MRSI of the prostate w/o water suppression in human volunteers and 3) multiparametric MRI/MRS (QSM, T2*, T1, MRS) combined with dopamine receptor-sensitive PET scanning in Tourette syndrome. These works have led to 4 conference communications and 2 publication submissions.

• WP3 “Boosting processing techniques for quantitative clinical MRS(I) and multimodal analysis by machine learning methods” has developed 3 tasks including 1) pre-processing and fusing longitudinal multimodal data into jMRUI, 2) integrate automatic pipeline into a user-friendly visualization tool, 3) design unsupervised, semi-supervised and supervised classifiers (deep learning) for diagnosis and prediction of disease progression in brain tumors and multiple sclerosis. These works have led to 6 conference communications and 3 publication submissions.

• WP4 « Communication, Dissemination & Exploitation »: A website has been developed for presentation of the project and the activities performed. Despite the current “Covid”, the ESR are encouraged in participating to virtual conferences, and submitting conference abstracts and journal publications to largely disseminate the project results.

• WP5 « Training programme »: the 15 ESRs have been enrolled in doctoral studies hosted the university organizations and are fully integrated in their respective research laboratories. Most of the secondments have been postponed to 2021 due to Covid-19 and virtual exchanges have been intensified. Meanwhile three Workshops (MRS, jMRUI, machine learning) could be organized virtually due to the Covid-19.
The 12 academic partners with an established collaborative track record in R&D and 9 industrial partners from the broad and competitive imaging sector give INSPiRE-MED fellows access to a broad ranging, practically-focused education, leading them to successfully participate in developing breakthrough tools for clinicians. MRI/MRS(I) and PET are unique techniques providing information non-invasively on anatomy, function and metabolism that are extremely innovative for diagnosis and therapy follow up in numerous disease models and patients. Despite such potential, clinical uptake of MRS(I) is lagging behind that of MRI and PET, mainly because of limited availability of efficient, robust and automatic software tools. Thus, INSPiRE-MED aims at establishing MRS(I) as an additional tool integrated into clinical routine imaging, but also extending its benefits with next-generation research methods. For this purpose, INSPiRE-MED is creating novel MRS(I) methodology, integrate MRS(I) into multimodal MR/PET clinical metabolic imaging protocols, develop latest machine learning methods for data analysis and provide a clinical version of jMRUI ( a worldwide unique data processing tool for MRS(I).