Articles

Neuro-Imaging life-cycle management

Given that the number and nature of neurological research images are rising extremely rapidly, the BIOMIST Project is designed to adapt a data processing system already used in manufacturing industrial sectors to facilitate neurological image data management. After all, the life cycle follow-up process for an industrial product may not prove so different from a follow-up in neurological research protocols … when it comes to data management!

Neuro-Imaging life-cycle management

What is the connection between the life-cycle of a product used in the automobile or aeronautical sectors and medical imagining used in neurological research? They could all in fact use a common data processing tool. The software package here is one commonly used for product lifecycle management (PLM) by manufacturing sectors to preserve throughout a complete lifecycle, the data appertaining to a given product.

It is a protocol that enables the manufacturer to follow all the data from design stage to decommissioning and scrapping. Naturally, there can be no question of just applying a manufacturing PLM software to neurological imaging, but rather to draw from the underlying principles and build on these to create a tailor-made tool.

In the framework of the second thematic axis of the call to project 'Contint 2013' launched by the French Government's HE agency ANR, the BIOMIST project (an acronym in French for Semantic management of biomedical imaging data used for research, ANR-13-CORD-0007) corresponds to the specification, proposing a module focused on biomedical imaging integrated in a Siemens PLM product.

Complex and heterogeneous data

"The problems facing engineers in the automobile or aeronautical worlds are identical to those for neurological research scientists", underlines Alexandre Durupt, who works at the UTC Roberval Laboratory, a partner to the project. The mass, the diversity, the heterogeneity and technicity of the data collected are really considerable, while the needs in terms of sharing them with numerous users, located in different sites, remains identical. The objective therefore consists of adapting an existing PLM tool to the specific case of neuro-functional imaging.

In addition to UTC, the project partners are the Institute Charles Delaunay, the Neuro-functional Group (GIN, CNRS, CEA, and the University of Bordeaux) and an SME, Cadesis, specialists in integrated data handling systems for industrial customers. "Whereas a PLM tool is used to monitor the lifecycle of an industrial product, we substitute the concept of a scientific investigation", explains Philippe Boutinaud, Head of R&D at Cadesis and executive manager for the BIOMIST Project.

Analysing and defining data dependencies

The BIOMIST Project, now in mid-stream, has already succeeded in integrating a data processing system used in industry, feeding it with a data base containing neurological information.The data in question covers brain images and also information related to the groups and categories of patients that have been scanned: their behavioural, genetic or demographic data and characteristics.

"It is now possible", adds Alexandre Durupt, "to submit queries about this information and to share the data and the answers". The system seems well adapted to the field of neuro-sciences, where the data is more voluminous and where the correlations and dependencies among data are more difficult to identify. In this context, BIOMIST focuses on the analysing and defining of the dependencies among data, using a graph display system to visualise the connexions.

"The objective of the research scientists is to seek and validate various hypotheses related to how the brain functions. When you have a tool that shows the correlations that exist between different areas of the brain, this proves interesting for all the brain specialists" underscores Philippe Boutinaud. The design and development of the tool started in 2013 with a programmed schedule running 348 months so BIOMIST is essentially coming into the last lap and final stretch.

Very soon, it will be produced as part of i-Share, a research programme about students' health status spanning several years. In the long term, the display graphs will be finalized and the product will then interest also the pharmaceutical world. "During the clinical and pre-clinical phases, numerous images have been added to the data levels collected as research scientists investigate the properties of new molecules", explains Philippe Boutinaud, for whom such a tool will enable the research communities to store, preserve and organise data and results.

Preserving the data and monitoring research protocols is no longer viewed as a luxury occupation, inasmuch as the possibility of replicating a results is the only way open to guaranteeing a reliable result. This validation requirement is not anecdotic since a publication dated back to 2003 showed that almost one quarter of all research results published has never been successfully replicated.