1.6 Biomedical research  (Page 4/5)

Healthcare professionals in laboratories are reluctant to release data because of cost and also because they lose some control over the data they have produced. An alternative is for clients to query databases of the pathology laboratories. A grid, federating the laboratories, would provide a secure framework enabling the screening associations to query databases and fill their local patient files (De Vlieger et al. 2009). No action is required by physicians to put their data on the network. Thanks to the grid security architecture, the cytopathologists are able to define and modify the access rights of the users querying their data.

Several projects in Europe have studied or are currently exploring the advantages of grid technology with regard to breast cancer, particularly computer-aided diagnosis of mammograms, most notably the e-Diamond (Brady et al. 2003 ) and MammoGrid (Warren et al. 2007) projects. If a sentinel network is able to federate pathology databases, it can be used by the epidemiological services of the National Institute for Health Surveillance (Institut National de Veille Sanitaire) and the regional epidemiological observatory. In the present case, it means that women could consult their own data in the pathology laboratories as well as see mammographic images stored in the radiology services through the proposed network. A cancer surveillance network is presently being implemented in the Auvergne region in France within the framework of the AuverGrid regional grid initiative (http://www.auvergrid.fr). It uses grid technology developed by EGEE, such as the AMGA metadata catalogue (Koblitz, Santos and Pose 2008) and the MDM Medical Data Manager (Montagnat et al. 2006), as well as by the Health-e-Child project, for example, the Pandora Gateway (http://www.health-e-child.org).

Case study 2 - application in radiotherapy

Radiotherapy is one of the three major treatments for cancer. It has demonstrated its efficacy in curing cancer and is also the most cost effective strategy. From a technology point of view, radiotherapy is a highly complex procedure, involving many computational operations for data gathering, processing and control. The treatment process requires large amounts of data from different sources that vary in nature (physics, mathematics, biostatistics, biology and medicine), which makes it an ideal candidate for healthgrid applications. Nowadays, in radiotherapy and brachytherapy, commercial treatment planning systems (TPS) use an analytical calculation to determine dose distributions near the tumor and organs at risk. Such codes are very fast (execution time below one minute to give the dose distribution of a treatment), which makes them suitable for use in medical centres.

For some specific treatments using very thin pencil beams (IMRT) and/or in the presence of heterogeneous tissues such as the air-tissue, lung-tissue and bonetissue interfaces, it appears that Monte Carlo simulations are the best way to compute complex cancer treatment by keeping errors in the dose calculation below 2%. The accuracy of Monte Carlo (MC) dose computation is excellent, provided that the computing power is sufficient to allow for extreme reduction of statistical noise. In order to finish MC computations within an acceptable time period for interactive use, parallel computing over very many CPUs has to be available. In this way, MC dose computations could become standard for radiotherapy quality assurance, planning and plan optimisation years before individual departments could afford local investment that is able to support MC. With the objective of making Monte Carlo dose computations the standard method for radiotherapy quality assurance, planning and plan optimisation, we are participating in the development of a Monte Carlo platform dedicated to SPECT, TEP, radiotherapy and brachytherapy simulations together with 21 other research laboratories which are involved in the international collaboration OpenGATE (http://www. opengatecollaboration.org, Jan et al. 2004). This GATE software with its accuracy and flexibility was made available to the public in 2004 and now has a community of over 1000 users worldwide.