Melbourne Health tests Oracle 11g beta

Oracle yesterday heralded Melbourne Health as one of the first Australian customers for its 11g database, which is still a beta test version.

Melbourne Health is the first Oracle 10g user to upgrade to the latest version of its database software. Oracle hopes that over the next year, between of 15 and 30 percent of its 10g customers will follow, said Mark Townsend, Oracle's vice president of server technologies.

Melbourne Health has rolled out the new platform to five member institutions under the Bio21: Molecular Medicine Informatics Model (MMIM) project, which is a AU$16 million federal and Victorian government-funded initiative designed to help researchers and clinicians access and share information between institutions.

According to Naomi Rafael, senior database administrator at Melbourne Health, the updated database will help researchers make better use of the seven million MRI (magnetic resonance imaging) scans, which are currently stored on DAT tapes.

Rafael said the MRI scans are used to create huge 3D images that can help detect brain deformations associated with epilepsy."Someone came along and said, 'We've got 12 years worth of MRI images and we don't know how many gigabytes it is because they're distributed on a thousand DAT tapes. Do you think you could load that up onto a server for us?'"

Chuckling at the severe underestimation, she said the DAT tapes contained two terabytes of scan data.

Rafael's search for a new system covered PAC (picture archiving and communication) systems, file systems as well as outsourced PACs solutions, but it ended when Oracle got wind of the project and, as an existing 10g customer, proposed Melbourne Health join the 11g beta program.

Under the MMIM system, multiple local research repositories are connected via a virtual private network to a "federated data integrator" which contains the MMIM metadata scheme.

For the MRI scans to include Melbourne Health's metadata scheme, all the files have been converted to DICOM (Digital Imaging and Communications in Medicine) files -- an open standard, similar to a ZIP file, which bundles JPEGs, video and other files into a single file and retains metadata used to identify and manage images.

Rafael said that combining the information in both databases was crucial.

"Our clinicians wanted to know if someone had diabetes in one database and the same person had choleric cancer in another institution's database. We wanted a way to find out what happened to that person in those two databases so we devised a way of identifying each person in the database -- patient name, date of birth, Medicare number, sex etc -- which is all stored in 11g."

Once stored in 11g, said Rafael, the system probabilistically matches information in it with other databases. If a match is found, it re-issues the same identification number that occurred in the first instance.

"They can do queries through the FDI (Federated Integration database) and can view data on various repositories," Rafael said.

Over the next 12 months Rafael expects the new system to be rolled out to research and health sites in regional Victoria, New South Wales, Canberra and Tasmania.

Author: Liam Tung