Image database offers virtual second opinion

Physicians making difficult medical diagnoses who seek a second opinion often find the practice complicated by distance.

Research may soon help physicians located anywhere in the world, however, go online for a "virtual" second opinion from a database developed by software engineers at the University of Missouri, Columbia.

WebHIQS is a Web-based database designed by Chi-Ren Shyu, Ph.D., an assistant professor of computer engineering and computer science, that optimizes image retrieval and provides a new resource for physicians.

"The database uses a hybrid approach, combining search methods for images and text to help a doctor find vital information," Shyu said. "This database and others like it will improve healthcare for patients, especially in rural areas where doctors and medical resources are more scarce."

The infant WebHIQS database is already operational for 14 lung diseases.

After uploading a CT study of a patient's lungs to the database's central server, a doctor can search for similar cases in the database, which so far contains a test-bed of some 3000 cases.

What distinguishes WebHIQS is the fact that database searches can be done by image as well as text. The database then provides information on these cases' diagnosis, treatment, and outcome. Patient demographic data are also included to provide additional clues, although this information is encrypted to preserve privacy.

Telemedicine experts were cheered by the potential of WebHIQS.

"This is essentially a large database of images that allows the user to conduct a sophisticated search to find images closely resembling the active patient's image," said Jonathan D. Linkous, executive director of the American Telemedicine Association. "This may prove to be an important supplement for the physician making the diagnosis. Currently, the number of publicly available medical images of various diagnosed conditions is limited."

WebHIQS will provide a virtually unlimited number of prediagnosed cases for differential diagnoses, in contrast to traditional diagnostic resources, according to Shyu.

"Medical systems are disjointed, and there are no search engines for images, only text," he said. "Our system represents a more flexible, instantly updated, richer learning environment that can be used any time any place."

Although not yet released to the general medical community, WebHIQS should be ready for use in about three years, after database accuracy - which averages 83% to 91% - is improved and the number of archived images increased.

Shyu has also begun work on a database for MR images of the brain.