iSGTW Feature - Forget me not: brain scans on a grid for Alzheimer's diagnosis

Feature - Forget me not: brain scans on a grid for Alzheimer's diagnosis

The results of a SPM analysis in a patient with dementia: red areas represent areas of impaired brain function.
Images courtesy of Ivan Porro

Early diagnosis of Alzheimer's disease relies on access to a distributed collection of treasured data. Now, a team of Italian scientists is creating a way for hospitals to share these data jewels.

"We had doctors with a real-world problem-they were not able to share data," says Ivan Porro of the University of Genova, Italy. "With distributed data resources, this is intrinsically a grid problem."

A grid solution

Doctors diagnose Alzheimer's disease by comparing the brain scans of "normal" people not suspected of Alzheimer's with a patient's scan, looking for significant differences or signs of neural degeneration.

This test, known as Statistical Parametric Mapping Analysis, is a valuable tool-but one that is difficult to use.

The first challenge is that the comparison requires a large number of "normal" scans: the accuracy of the comparison increases as more data is fed into the algorithm.

However the scanning process is expensive and raises ethical issues since the test involves some risk for the patient.

"You are asking people to accept radioactive fluid in their blood," says Porro. "Most people already at risk for Alzheimer's readily accept this, but healthy people? Normal scans are very precious and very rare. They are the jewels of the hospital data reserves."

Self portraits of an artist descending into Alzheimer's. The drawing on the left is from 1967; on the right is from 1996.
Images courtesy of William Utermohlen and the Galerie Beckel-Odille-Boïcos, Paris

Treasured data still secret but shared

Hospitals, then, while determined keep hold of their own normal scans (if they have any), are very interested in accessing the normal scans of other hospitals. Doctors at their desktop computer will now be able to share these scans through a Web interface, while adhering to strict standards of security and patient privacy.

The first development step-a traditional Web-based application-has been validated by neuroradiologists at San Raffaele Hospital, Milan. The second step-consolidation of a grid-based prototype using gLite middleware-is currently being validated at San Raffaele Hospital and is a candidate for validation by a network of neurological centers in Italy next year.

Once the grid-based version is running successfully, the project would like to offer its services to other hospitals. This application, says Porro, "benefits the doctors directly. They can give earlier and more confident diagnoses. That benefit is conferred to the patient."

While Alzheimer's is still incurable, an early diagnosis allows for treatment to begin as soon as possible. Changes in lifestyle, medication and dietary supplements can try to slow the progress of the disease and give the patient a higher quality of life.

The research groups involved in this project included: BioLab, Department of Communication, Computer and System Sciences, University of Genova, Genoa; Vita-Salute San Raffaele University and Scientific Institute Hospital San Raffaele, Milan; and the Italian National Institute of Nuclear Physics (INFN).

- Danielle Venton, EGEE