Abstract. Bio-medical imaging (BMI) is currently confronted to similar issues than those of manufacturing industries twenty years ago : the growing amount of data, the heterogeneity and complexity of information coming from diverse disciplines, have to be handled by various actors belonging to different organizations. The researchers of the GIN (Neuroimaging Functional Group) laboratory study brain maps of anatomical and functional cognitive activation of hundred-subject cohorts, acquired with Magnetic Resonance Imaging (MRI). Therefore they want to manage the whole process of their research studies, from raw data to analysis results. Even if some data management systems have been developed to meet the requirements of BMI large-scale research studies, there are still many efforts to do in the integration of all the data and processes along a research study, from raw to refined data. So, the use of the Product Lifecycle Management (PLM) concepts to handle the complexity and characteristics of BMI data is proposed. A PLM neuroimaging datamodel that has been designed in collaboration between the GIN laboratory, Roberval laboratory and Cadesis company to meet the needs of the GIN, is described.

The data management needs of the neuroimaging community are currently addressed by several specialized software platforms, which automate repetitive data import, archiving and processing tasks. The BIOMedical Imaging SemanTic data management (BIOMIST) project aims at creating such a framework, yet with a radically different approach: the key insight behind it is the realization that the data management needs of the neuroimaging community-organizing the secure and convenient storage of large amounts of large files, bringing together data from different scientific domains, managing workflows and access policies, ensuring traceability and sharing data across different labs-are actually strikingly similar to those already expressed by the manufacturing industry. The BIOMIST neuroimaging data management framework is built around the same systems as those that were designed in order to meet the requirements of the industry. Product Lifecycle Management (PLM) systems rely on an object-oriented data model and allow the traceability of data and workflows throughout the life of a product, from its design to its manufacturing, maintenance, and end of life, while guaranteeing data consistency and security. The BioMedical Imaging-Lifecycle Management data model was designed to handle the specificities of neuroimaging data in PLM systems, throughout the lifecycle of a scientific study. This data model is both flexible and scalable, thanks to the combination of generic objects and domain-specific classes sourced from publicly available ontologies. The data integrated management and processing method was then designed to handle workflows of processing chains in PLM. Following these principles, workflows are parameterized and launched from the PLM platform onto a computer cluster, and the results automatically return to the PLM where they are archived along with their provenance information. Third, to transform the PLM into a full-fledged neuroimaging framework, we developed a series of external modules: DICOM import, XML form data import web services, flexible graphical querying interface, and SQL export to spreadsheets. Overall, the BIOMIST platform is well suited for the management of neuroimaging cohorts, and it is currently used for the management of the BIL&GIN dataset (300 participants) and the ongoing magnetic resonance imaging-Share cohort acquisition of 2,000 participants.

Abstract. Product Lifecycle Management (PLM) domain is at a key point in its development: its concepts and technologies are mature. PLM systems not only manage documents but information associated to a product along its lifecycle, such as Bills-Of-Material (BOM) or requirements at different levels of granularity. All the dependencies between concepts lead to complex relationships from which it is not easy to get a coherent overview. The purpose of the paper is to know whether PLM systems are able to deal relationships complexity. Two case studies -one form manufacturing industry and the other one from a new application domain of PLM, Bio-Medical Imaging -are developed in the paper. They show that hierarchical browsing of existing PLM systems is not suitable to manage relationships complexity and must evolve to graph browsing.