Improved framework streamlines Web image exchange

Brazilian computer scientists have designed a system to help medical professionals use the Internet for image exchange more effectively.

The system, called MedISeek, allows researchers to describe medical images, submit them for peer review, and then make them available for search and retrieval through the Web.

This issue of efficient Web retrieval is important because the Web has become such an extensive health information repository that it is increasingly difficult to search for relevant medical information, said Jacob Scharcanski, Ph.D., of the Instituto de Informática at the Universidade Federal do Rio Grande do Sul.

"Most medical information available on the Web is not peer-reviewed, and it is retrieved imprecisely by current keyword search mechanisms, casting doubts on its reliability," he said.

The extensive use of digital imaging and the advent of new imaging modalities have created an explosion of diagnostic imaging techniques. Yet in some areas of medicine, images are generated using digital cameras or even digitization through a scanner, Scharcanski said.

"Techniques for archiving, describing, and communicating the visual information available in such images are needed," he said.

Compared with common keyword-based search engines like Google and Altavista, or even medical image databases currently available on the Web, the MedISeek scheme provides superior precision for indexing and searching medical images, according to Scharcanski.

The conventional means of image digitization does not consider image content or offer a means to describe it.

The MedISeek metadata model is meant to allow users to describe, store, and retrieve medical images and associated diagnostic information, facilitating improved Web information exchange. Under MedISeek, medical images are made available for Web access with an identified source, one or more peer reviews, and a guarantee of authorship, Scharcanski said.

"Our main motivation was to propose a solution for providing access to relevant cases documented by medical images and associated data, which are often confined to resources available only within specific institutions," he said.

The MedISeek prototype was constructed using open source software components. It is intended for a community of Web users willing to share images, information, and knowledge with other members, Scharcanski said.

"It is our intention to make the prototype freely available as soon as a community of individuals, groups, or institutions is identified," he said.

Scharcanski can be contacted at jacobs@inf.ufrgs.br.

For more online information, refer to Diagnostic Imaging's PACSweb section.