IBM has joined forces with U.S. researchers to design a new computer network that may one day help improve breast cancer diagnosis and detection.
Working in conjunction with University of Pennsylvania, IBM is overseeing the creation of a complex computing grid linking together several hospitals.
Computing grids enable geographically distributed hardware systems to share applications, data and computing resources.
Unlike simple Internet connectivity, grids can more readily leverage all of the assets tied into it, in some cases creating virtual supercomputers. For example, IBM was selected in August by a consortium of four U.S. research centers to help create the worlds most powerful grid, which when completed in 2003 will be capable of processing 13.6 trillion calculations per second.
In this instance, the grid will enable participating hospitals to quickly access medical data at distant locations and share analytical programs that can aid in diagnosis, such as specialized software used to review X-rays and identify potential problems areas.
The grid will also utilize sophisticated algorithms to uncover possible cancer patterns that appear in the population. By culling through data input at various hospitals, the system will be able to spot an abnormal concentration of the disease in a particular community or ethnic group.
Another advantage of grid computing is that it allows administrators to more closely manage data spread across a variety of computers running various operating systems. In this instance, the grid will use hardware running operating systems from Microsoft Corp., IBMs proprietary version of Unix called AIX, as well as Linux.
While the grid will initially join only four hospitals, IBM said its designing the system to enable potentially thousands of medical centers to connect into it.
By demonstrating the effectiveness of the grid model, IBM and researchers hope to persuade other medical institutions to join in the effort that could have far-reaching impact in breast cancer research.
“The idea here is that hospitals are creators of mammography data. If you can aggregate this data, you would have much better insight into the population at large,” said David Turek, vice president of Linux clustering for IBM in Poughkeepsie, N.Y.
For example, the computing grid will have the ability to cross-reference breast cancer data against certain environmental or ethnic factors. Such information could prove vital in identifying individuals at risk and enable earlier detection and treatment of people with breast cancer.
IBM is building the grid using a three-tier architecture that includes IBMs Intel-based xSeries servers, its eServer Cluster 1600 UNIX and eServer Cluster 1300 Linux system.
The software that will be used to manage the grid will be based on open protocols developed in conjunction with the Globus Project, a collaborative research and development effort supported by many of the worlds leading computer companies focused on fostering the development of grid computing.
Each participating hospital will be equipped with two IBM xSeries servers, one of which will serve as a temporary depository for digital data, and the other as a link to the grid.
Once data is loaded into the portal, it will be transmitted to a metropolitan hub consisting of an IBM eServer Cluster 1600 UNIX system. Data from the metropolitan hubs will then be funneled to an IBM eServer Cluster 1300 Linux system, which will serve as a high-capacity regional hub.
In addition, IBM is working with researchers to develop an ultra-high capacity DB2 database to serve as the secure repository for the digitized X-ray data.
The first medical centers being connected to the grid include hospitals at the universities of Pennsylvania, Chicago and North Carolina, as well as the Sunnybrook and Womens College Hospital, in Toronto.
Funding for the project is being provided by the National Library of Medicine in Bethesda, Md.