EGEE 2007

After the first twenty months of intense work, the Health-e-Child development team demonstrated its achievements at the world’s leading conference on Grid computing, Enabling the Grid for E-sciencE (EGEE) 2007, which took place from October 1st to the 5th, 2007 in Budapest, Hungary. The visual demonstration of the HeC platform and gLite-based grid infrastructure simulating a clinician logging in to the system and analysing patient data was one of 12 out of 50 submitted proposals (covering a variety of domains) selected to demonstrate their work on-site. Following evaluation by the EGEE External Advisory Committee, five presentations were singled out as “best live demos”.

The demonstration illustrated a clinician/user connecting to the HeC virtual organisation using a normal USB key containing his/her credentials together with basic portable applications and is thus able to securely browse and add patient data, regardless of location and connectivity. Immediately after authentication, the so-called Peer-to-Peer Patient Privacy network (P2P3) is established within the LAN,

which enables users to store and share private patient information outside of the grid and map it to deidentified records inside the grid. Thanks to the joint use of the grid and P2P technologies, client applications are user-friendlier and can, e.g. resolve universal unique pseudonyms (patient identifiers) against real patient names (provided they are allowed to) while still complying with ethical requirements. Such a feature is undeniably helpful for physicians to efficiently retrieve the patient records they are in charge of from the large distributed case database.

The live demo aptly depicted how the clinician is able to access patients’ medical records from different clinical centres, which are made seamlessly available via query in the HeC Gateway. This latter constitutes a secure portal of on-line medical services (developed as a Services-Oriented Architecture – SOA) for storing and manipulating multi-modal and multi-centric patient data. Thus, the clinician can select a patient and preview attached information ranging from medical records to images. The clinician can also enhance the patient data by anonymously attaching new records. The process of uploading, downloading and more generally invoking Gateway functionality is graphically illustrated through progress bars “a la FireFox”, which give real-time indications on operation(s) status(es). Various operations are available to the clinician, from upload, to preview, to query, to download of patient data, as well as access of more complex tools such as the CaseReasoner developed by HeC partner Siemens.

During the demonstration, innovative techniques for patient similarity visualisation like combined distance/heatmaps, treemaps and neighbourhood graphs were illustrated. Similar cases are detectable both in the whole integrated case database, and in some subsets of interest (e.g., high-grade tumours, males, etc), defined in the form of a simple filter. For each patient in the case database, it is possible to visualise and compare related images from the patient history. In combination with basic statistics, class distribution histograms and scatter plots for the case base under study, this will be a universal tool for gridbased decision support for the diseases studied in HeC.

Additional features worthy of mention were demonstrated, including the 3D Registration Tool for brain MR scans developed by HeC partner INRIA. This application enables the clinician/user to visualise the evolution of brain tumours: by selecting two MR scans of a given patient in the case database, the clinician retrieves a series of graphical tools to register, fade in/out and ultimately compare the two data sets to understand how the disease evolves. Various graphical approaches are available from fading to grid-mapping which highlight differences found between images.

Finally, from a platform standpoint, the SoAutomator application was introduced. SoAutomator enables end-users to discover the high-level functionality of the Gateway (e.g., query, run algorithm, invoke external tools, etc.) and to further combine these under the form of graphical workflows which can then be submitted to the system for execution. As part of this SOA demonstration, an accompanying Shell-like application was presented, which allows more experienced users to navigate through the Gateway’s lower-level functionality and to delve more deeply in the system, ultimately reaching application programming interfaces (API) and their basic testing using command lines.