Mr Ferran Borreguero (Universitat Politècnica de València)
Medical Imaging Biomarkers - a quantitative analysis of specific physical properties by applying computational methods to medical images - is revealing as a highly effective and harmless procedure for supporting clinical diagnosis and therapy follow-up. The EuroBioImaging Node of Valencia Region (BIM-CV) is developing a use case on top of the INDIGO DataCloud Platform for the analysis of Osteoporosis. BIM-CV requires: 1) Persistent Storage in the form of POSIX volumes that can be mounted remotely; 2) Privacy and security. Data volumes must be available only to users provided of an access credential; 3) User customised software. The capability of defining their own software configuration has been traditionally one of the main hurdles in scientific clouds; and 4) Efficient execution. Projects may require resources or have external resources available, and the applications should work seamlessly. Automatic elasticity is taken for granted. We have implemented a prototype based on the next procedure: 1) Specification of the dependencies as Ansible Yaml and application topologies as TOSCA documents. 2) Upload of images and IPR-protected software on a OneData volume. 3) Deployment of an elastic cluster where VMs mount the ONEData volume. 4) Execution of jobs in the SLURM queue embedded on containers that mount locally the shared volume and run the biomarker job. INDIGO-DC middleware and services provides BIMCV with: 1) The availability of an efficient batch queue; 2) The convenient & secure access to remote volumes accessible through POSIX; 3) The Convenient integration of user-specific code, including IPR restrictions; 4) Single sign-on, integrated deployment, QoS.
Mr Ferran Borreguero (Universitat Politècnica de València) Dr Ignacio Blanquer (UPVLC) Dr Luis Martí-Bonmatí (Hospital Universitario y Politécnico la Fe) Dr Ángel Alberich-Bayarri (QUIBIM)