IBM builds computer grid for breast cancer diagnosis and screening

24

Author: JT Smith

IBM and the University of Pennsylvania
today announced a powerful computing Grid that aims to bring advanced
methods of breast cancer diagnosis and screening to patients across the
nation, while reducing costs.

Built with open standards, the University of Pennsylvania Grid is a massive
distributed computer that delivers computing resources as a utility-like
service over the Internet. Enabling up to thousands of hospitals to store
mammograms in digital form, the Grid will provide analytical tools that
help physicians diagnose individual cases and identify cancer “clusters” in
the population. It will also give authorized medical personnel
near-instantaneous access to patient records and reduce the need for
expensive film X-rays.

“Once a patient’s mammograms are loaded into the system, they can be
evaluated with powerful tools that isolate abnormalities very quickly by
comparing current X-rays with those from previous years,” said Dr. Robert
Hollebeek, director of the university’s National Scalable Cluster Lab.
“Traditional film X-rays of individual patients are often scattered among
various medical facilities, making them hard to find when needed. This Grid
will help ensure that all of a patient’s vital data is provided to
authorized physicians very quickly, efficiently and securely.”

Hospitals are connected to the Grid via secure Internet portals that allow
authorized physicians to upload, download and analyze digitized X-ray data.
Advantages of the Grid include:

  • Fast data retrieval — authorized physicians have immediate access to
    a patient’s complete history of mammograms, no matter where or when the
    X-rays were taken.

  • Computer-assisted diagnosis — X-ray data can be scanned with powerful
    software that identifies potential tumors and other problems, helping
    physicians diagnose patient illnesses.

  • Pattern identification — Sophisticated algorithms can uncover
    patterns that appear in the population, such as cancer “clusters,” or
    abnormal concentrations of the disease in a particular community.

  • Cost savings — Each year, the average hospital spends $4 million to
    develop X-ray films. Estimates indicate that participation in the Grid will
    result in an average yearly cost savings in the millions of dollars.

  • Training — A suite of educational tools will be deployed on the Grid
    to help doctors, medical students and interns learn more about breast
    cancer and related diseases.

Now in its early stages of deployment, the University of Pennsylvania Grid,
in collaboration with a group from Oak Ridge National Laboratory (ACT at
BWXT), connects hospitals at the University of Pennsylvania, University of
Chicago, University of North Carolina, and the Sunnybrook and Women’s
College Hospital in Toronto. It is funded by the National Library of
Medicine.

In the future, the University of Pennsylvania will work to extend the Grid
to additional medical institutions. The design of the system is capable of
serving thousands of hospitals.

The Power of IBM eServer

The University of Pennsylvania Grid is built with a three-tier architecture
that leverages the strengths of IBM eServer and open protocols from Globus.
At the user level, each participating hospital is equipped with a portal
consisting of two IBM eServer xSeries systems. One xSeries machine serves
as a temporary repository for the digital data, and the other is a link to
the next generation of the Internet, called Internet2.

Once the data is loaded into the portal, it is transmitted to a
metropolitan hub — an IBM eServer Cluster 1600 UNIX system. When the Grid
is fully deployed, data from several metropolitan hubs will be funneled to
a high-capacity regional hub, which is now being prototyped with an IBM
eServer 1300 Linux system.

The three-tiered system — running AIX, Linux and Windows — illustrates
the inherent heterogeneity of computing Grids.

Teams from the University of Pennsylvania and IBM are partnering to develop
an ultra-high capacity DB2 database to serve as the secure repository for
the digitized X-ray data.

IBM’s Grid Leadership

The University of Pennsylvania Grid is the latest in a series of Grid
projects that illustrates IBM’s leadership in this space.

Earlier this month, IBM was selected to build the North Carolina
Bioinformatics Grid, which will be developed in collaboration
GlaxoSmithKline Inc., Biogen, the University of North Carolina System, Duke
University and other organizations.

In August, IBM was selected by a consortium of four U.S. research centers
to build the world’s most powerful computing Grid, an interconnected series
of Linux clusters capable of processing 13.6 trillion calculations per
second. The Grid system — known as the Distributed Terascale Facility
(DTF) — will enable thousands of scientists around the country to share
computing resources over the world’s fastest research network in search of
breakthroughs in life sciences, climate modeling and other critical
disciplines.

IBM is also partnering with several centers in the UK National Grid to
provide key technologies and infrastructure for the project, which is
linking a massive network of computers throughout the United Kingdom.

In addition, IBM is building a powerful computing Grid for universities in
the Netherlands.

Open Standards

Just as electricity is delivered to homes over an electrical grid,
Computing Grids allow geographically distributed organizations to share
applications, data and computing resources. A new model of computing,
Grids are clusters of servers joined together over the Internet, using
protocols provided by the Globus open source community (Globus.org) and
other open technologies, including Linux (R).