1.6 Biomedical research  (Page 3/5)

Grid technology for distributed medical data management

Providing patients with “google-like” secure access to their medical records requires the information to be available for querying and retrieval. Google is able to query and search for any data published on the Internet. However, it will be absolutely necessary to ensure the security of this Internet environment before storing any medical data on it. An alternative is provided by grid technology which allows distributed data to be queried securely according to personal access rights. Some platforms in medical data management (Erberich et al. 2007) of paediatric data (Freud et al. 2007) or medical radiography data (Warren et al. 2007) already benefit from grid technologies to manage medical data securely thanks to dedicated grid middleware services such as MDM8 or Globus Medicus. The use of grids overcomes the difficulties inherent in a centralized storage system, especially high cost and complexity. Grids also make it possible to store data where or very close to where they are produced. Through grid authentication, authorization and accounting, only duly authorized persons can gain access to data which are encrypted and made anonymous when they are transmitted (Mohammed et al. 2007).

Early attempts at epidemiological applications of grids (Blanquer and Hernández 2005) have demonstrated their relevance for patient customized research. Users ought to be able to take it for granted that the security mechanisms are sufficient to protect their data; that the results of their research will be private and available to third parties only if designated; that the system will meet the concerns of the ethical and legal committees of their research institutions; that the services are reliable, efficient and permanent; that they do not have to change significantly their current procedures; protocols or workflow, and finally that the data is somehow automatically organised and gathered, and thus available for further exploitation. In the next chapter, we will present an epidemiological application of grids for cancer surveillance which is currently being used in France. Another attractive field of application for grid technology is computer-intensive analysis of distributed medical images. The impact of grid technology comes from the secure management of distributed images together with the capacity to gain access to large computing resources on demand to analyze them. In the field of oncology, the use of Computer-Aided Detection (CAD) for the analysis of mammograms was addressed by the MammoGrid project as early as 20059. Other efforts focus on using grid computing resources to plan radiotherapy treatment (Benkner et al. 2001) a case of the use of this technology currently exploited in collaboration with a French Cancer Treatment Centre will be further documented in the case study 2.

Case study 1 - cancer surveillance network

Cancer screening programs aim at the early detection of the malignant tumors in order to improve the prognosis. Most EU countries have launched a national program for breast cancer screening (von Karsa et al. 2007). In France, when a woman is positively diagnosed with a risk of tumour, cancer associations are responsible for providing a second diagnosis on the mammograms and have to follow-up the pathology data about the tumour, which are stored by the laboratories. At present, the patient’s data are faxed on request or carried physically by the patient to the associations where they are recorded again. This process is costly and error prone as data has to be typed and reinterpreted twice. The cytopathology data are also relevant for epidemiological analysis. The INVS (Sanitary Surveillance Institute), the French equivalent of the (E)CDC in the USA, is in charge of publishing indicators about global health and particularly about cancer. To produce its indicators, the INVS relies on regional cancer registries (CRISAPs) set up to collect relevant information to support statistical and epidemiological studies about cancer incidence, mortality, prevalence or screening. CRISAPs (Centre de Regroupement Informatique et Statistique en Anatomie et cytologie Pathologiques) are like regional data warehouses collecting anonymous data from pathology laboratories or from healthcare establishments involved in cancer treatment.