A grid enabled PSE for medical imaging: experiences on MedIGrid

A key issue in designing applications codes which are able to effectively take advantage of grid computing environments is to deal and manage all the software components complexity, this is even more true for medical imaging applications. We are concerned with development of a medical imaging PSE for advanced and grid computing environments. Specifically, MedIGrid is a distributed application for the management, processing and visualization of biomedical images that integrates a set of software and hardware components, or, more specifically, a set of grid collaborative applications useful to nuclear doctors. It results from the interaction of computing scientists devoted to the design and deployment of new tomographic reconstruction algorithms techniques, computer scientists researchers in the field of distributed and parallel architectures and physicians involved in nuclear medicine. We'll describe MedIGrid general architecture, as well as improvements and enhancements, essentially regarding the semantic power of the GUI (the Portal) (its language and its realization), and the entire phase of monitoring and migration of the parallel computational kernel for images reconstruction, mainly based on a performance contract inked taking into account historical performances and an ad hoc developed performance model.