Sites Inria

Version française


Jean-Michel Prima - 27/05/2014

A Neuroimaging Platform for Multiple Sclerosis Research

C.Barilllot et  J.Guillaumont équipe  Visages Christian Barillot et Justine Guillaumont, de l'équipe Visages

Shanoir is an open source neuroimaging platform meant to facilitate the exploitation of medical data. Created by Visages, an Inria research team in Rennes, France, this server now hosts the data collected by hospitals affiliated with OFSEP, the French Observatory of Multiple Sclerosis, a body funded by the Government with mission of assembling a nationwide MS study cohort.

Multiple Sclerosis (MS) affects approximately 70,000 persons in France, with 2,000 new cases reported each year. The disease causes of variety of symptoms, some of them severe. It results in the thinning of the myelin sheath that normally insulates the nerve fibers, in particular those located in the brain. That is why neuroimaging offers valuable contribution to diagnostic, patient monitoring as well as research studies.
In January 2011, the French Government green-lighted the OFSEP project for the creation of a  nationwide MS data base in which hospitals are invited to pool together clinical and biological data, as well as biomedical images. The French Research Agency (ANR) will grant €10M to this endeavor over 10 years. At the moment, the cohort includes 36,000 persons whose cases are being monitored by hospitals. It is the Shanoir platform that has been selected for storing the image data.

Harmonizing Heterogeneous Data

This neuroimaging platform was developed by Visages, a neuroinformatics research team (2) at Inria center, in Rennes. Shanoir relies on OntoNeuroLOG, an ontology that was specifically designed to represent knowledge in the neuroimaging field. This vocabulary base enables the harmonization —for mutualization purpose— of distributed and heterogeneous data in neuroimaging.
Research Engineer Justine Guillaumont  is in charge of this tool deployment within the hospitals network for the account of OFSEP. “In any given medical study, the statistical aspect is a weighty stumbling block. In order to reinforce the findings, it is necessary to gather a large number of patients meeting the preset criteria. In practice, a hospital doesn't always have enough cases to undertake a significant study.  Thus multicentric research studies came up in which  several centers pool together their respective patients. Until now, assembling a neurimaging database for such purpose required the collection, duplication and transportation of a massive number of CD-Roms on which each center used to store its own files. It was both tedious and expensive. Through a web server, Shanoir will facilitate the mutualization and uniformization of all this data, thus improving their magagement, archiving, indexing and sharing.
In multicentric research studies, one of the problems comes from the heterogeneity of MRI equipment used by the hospitals. Different constructors and models. Different levels of definition. Also, at times, different naming habits from one establishment to another. To cope with this diversity, in Shanoir, for any given study, a specific profile (study card) is attributed to each center. The study card corresponds to the implementation of the protocol on each machine.

‘De-identified’ Data

Then, through one single query, the neurologist, radiologist or  researcher will instantly access the data mutualized by different centers. ” Interestingly, such information will be ‘de-identified’ in order to protect patients' anonymity. Security devices also ensure the confidentiality of medical research conducted for instance by pharmaceutical companies.
As Team Leader Christian Barillot explains, “users can access their data through the web interface. But they might as well chose to work through their favorite tool for visualizing or processing those data. It could be for instance Edmus, a specific application dedicated to MS clinical monitoring (visualization), or MedInria, ” Inria's toolbox for medical image processing.
Fourteen hospitals have joined the platform so far. Dozens of others are about to do the same. “First feedback is very good, Guillaumont points out. We wish to keep collecting users' observations, including MRI operators. They play an important role in the production chain. ” The experience-gathering  might also help Shanoir to cross French borders. “Some entities abroad have expressed  interest, Barillot remarks. Among them the Harvard Medical School. ’’

The project is supported by the French Government through the Investissement d'Avenir funding plan as a result of the Cohorts 2010 call for proposals. It gathers about 40 MS centers. It is supported by a consortium composed of the  Eugène Edmus Foundation, Claude Bernard Lyon 1 University (UCBL1) and Lyon hospitals (HCL). It was initiated by the Neurologist Christian Confavreux who died last September.

Keywords: VISAGES Christian barillot INRIA Rennes - Bretagne Atlantique Justine Guillaumont Sclérose en plaque Shanoir MedInria