RSNA programs exploit power of info networks

January 1, 2007
John C. Hayes

The 2006 RSNA meeting made clear that the ability to reach outside the radiology department via digital information networks offers the opportunity to enhance business and educational opportunities and improve the quality of patient care.

The 2006 RSNA meeting made clear that the ability to reach outside the radiology department via digital information networks offers the opportunity to enhance business and educational opportunities and improve the quality of patient care.

The Integrating the Healthcare Enterprise project demonstrated the ability of its XDS-I integration protocol to allow PACS from different vendors at separate facilities to connect and share images via the Internet. And research showed that a system in Philadelphia that is using the XDS-I protocols could improve efficiency and quality of care by sharing images for patients who visited multiple facilities in the network.

RadLex, the RSNA's project to develop a common lexicon for radiology reporting and image classification, released its first full version and made it available on the Internet. RadLex will allow more efficient and targeted searching of educational and clinical materials than has been possible with standard Internet search mechanisms such as Google, said Dr. Curtis P. Langlotz, chair of the RSNA's RadLex Steering Committee and associate chair for informatics at the University of Pennsylvania.

RadLex's terminology for anatomy and pathology offers a method of intelligent searching, according to Langlotz.

"You can search on larger concepts, as opposed to Google terms," he said. "If I'm interested in inflammatory bowel disease of the small bowel, and I search Google, I get only exact matches on those words."

RadLex, on the other hand, knows that Crohn's disease is an inflammatory bowel disease and that the ileum is part of the small intestine, he said. If cases are coded with RadLex terminology and structure, it becomes possible to search for more general terms, such as inflammatory bowel disease and small bowel, and find all the cases for all the different parts of that larger structure and all the specific subtypes of that disease category. So, unlike a Google search, a RadLex-enabled search would find cases indexed by "Crohn's disease."

"That's one of the primary values of common terminology. You'll be more likely to find cases relevant to your interest," Langlotz said.

RadLex started out as a way to characterize teaching files, and, accordingly, it has been adopted by the RSNA for indexing teaching files under the MIRC. But development has implications in other areas as well.

"We've had contacts from vendors of other teaching file systems interested in incorporating RadLex," Langlotz said. "We've heard from developers who are building radiology order entry and radiology reporting systems who have expressed interest in RadLex. If you have a standard lexicon available, that gives you the power to create innovative reporting technologies."

Next up for Radlex are vendor-independent terms to describe devices, procedures, and imaging sequences, Langlotz said. RadLex is available on the Web at RSNA.org/RadLex and at RadLex.org.

Individual presentations at the scientific sessions and posters reinforced the network themes and added new twists. Presentations showed how the XDS-I profile was put to work to share images in Verona, Italy, and Shanghai. Other presentations showed how networks were linked to a single PACS in Greenland and 10 facilities in southern Maine.

It's been assumed that sharing images among facilities could reduce duplication and overlap, leading to new efficiencies, but this meeting was perhaps the first to present data actually suggesting such improvements could happen. Dr. Elliot Menschik, chairman and CEO of Hx Technologies, presented the initial results from the XDS-I-enabled Philadelphia Health Information Exchange during a scientific session at the RSNA meeting. Hx Technologies has been the driving force behind the PHIE.

Researchers quantified the overlap in patient population between two major rivals in the Philadelphia market: Thomas Jefferson University and the University of Pennsylvania Health System. During a 19-month time period, they found that there were nearly 10,100 relevant priors that could be shared between the two facilities. They also estimated that within a 120-day time period, as many as 2280 duplicate exams could be performed at the two facilities, at a cost of nearly $900,000.

"Reducing duplication has the potential to improve quality of care and reduce cost when regional imaging exchanges enable radiologist access to relevant historical images and reports," Menschik said.

A demonstration at the RSNA meeting consisted of a central facility at the Lakeside Learning Center, where visitors could watch images move from one system to the next, and at remote stations on the exhibit floor, where vendors joined the demonstration with image and data collection points at their exhibition booths.

"The significance is that this fits radiology into the evolving concept of the nationwide health information networks and efforts that are taking place in other countries," said Christopher Carr, director of informatics at the RSNA.

More than 900 attendees visited the centralized and remote sites to view it, he said.

The demonstration was coordinated by Dr. David S. Mendelson, an associate professor of radiology at the Mount Sinai School of Medicine and a member of the RSNA Radiology Informatics Committee.

"We encouraged the development of a solution and demonstrated it in an open environment," Mendelson said. "We were pleased to have 15 vendors demonstrate that they could share images. We also demonstrated to our membership that this is real, has value, and can be done, so they will begin to look for this capability in products. We are hopeful that we will see more vendors adopting the profile."

Vendors who participated in the XDS-I demonstration were Acuo, Agfa, Aspyra, Fujifilm, GE Healthcare, HX Technologies, IBM, Initiate Systems, Kodak, McKesson, MIE, Siemens, SoftMedical, and Toshiba.