imageThe merging of radiology information systems (RIS) and voice recognition technology are no longer dreams of the future. Today, they are providing real-life solutions to shorten lengthy turnaround times for imaging reports, and increase the efficiency of radiologists’ time and overall radiology department productivity.

When combined with a picture archiving and communications system (PACS) or another system of digitized image storage, RIS and voice recognition offer a very high percentage of efficiency and accuracy in radiology reporting and medical records. In fact, today’s accuracy rate for voice recognition technology is higher than traditional transcription and/or dictation.

Small community hospitals nationwide, as well as large teaching hospitals, are utilizing voice recognition technology and saving money by reducing transcription costs — among the many other benefits this technology offers.

Recent advances
Software and hardware improvements have made voice technology easier to use and more compatible with certain computer systems.

“We’ve used voice technology here for more than 10 years,” says Jennings Pressly, M.D., director of radiology at Hillcrest Hospital in Simpsonville, S.C. “In the last year, refinements in software and improvements in hardware technology have brought us significant improvement.”

The voice technology system used at Hillcrest is called Rapid Access & Management Systems (RAMS), which was developed by ITS Speech of Greenville, S.C. It also includes image manipulation and image transfer.

Hillcrest Hospital’s voice technology is integrated with RIS, which brings up a patient’s demographic record instantly, and the ICD-9 code. The code is automatically entered into the report so that the radiologist won’t have to look for it. The integration cross-relates the patient’s history with the appropriate code, according to Pressly. The image also is integrated with voice technology and RIS, making a seamless record for the radiologist and referring physician.

The precision of voice technology also has improved, and the integration of several systems provide ease of use.

“The key advancements have been in terms of accuracy and ease of use, and the integration with PACS has been a huge benefit,” says Vishal Wanchoo, global general manager of radiology systems for GE Medical Systems Information Technologies (Milwaukee). “In the past, voice recognition, RIS and PACS were all separate. Now everything is automated. You select the patient, open the case and it’s [dictation] all set to go. In the past, you had to search separately for the patient in voice recognition and in PACS.”

Paul Masotto, administrative director of imaging at St. Mary’s Hospital and Naugatuck Valley Radiology Associates (NVRA) in Waterbury, Conn., agrees.

“The technology continues to improve. The recognition is extremely reliable, vocabularies have expanded and integration systems have become more efficient,” he notes. St. Mary’s and NVRA utilize the Lernout & Hauspie (Stratford, Conn.) PowerScribe, which offers a completely integrated

dictation and transcription solution. The PowerScribe is integrated with RIS and PACS at St. Mary’s and NVRA. St. Mary’s has 10 radiologists on staff. Its voice recognition technology is distributed over a wide area network to five imaging centers.

Accuracy has undergone a significant improvement, especially this past year.

“In 1999, accuracy was in the mid-20 percent range [for voice recognition technology]. Now, most of our customers’ experience a high-90s accuracy range, due to a refinement in the recognition technology,” says Michael Davis, CEO of NCC, LLC in Scottsdale, Ariz., maker of Digital Dictate. “With imaging, you have very constrained structures and a limited range of vocabulary.”

In the last three years, there has been a “big step” in speaking normally and rapidly, according to David Weiss, M.D., radiologist at Chestnut Hill Hospital in Philadelphia. The voice recognition system that he used then was very usable, but more difficult to learn. Today’s systems are much easier to learn and operate.

“Last year, we added another 5 percent in accuracy, which is now between 95 and 99-plus percent,” says Weiss. “It depends on the individual [using it], how his or her voice sounds, how consistent he or she is in what is said and how it is said. If the structure of your chest x-ray reports is about the same [in dictation terminology], then the accuracy is better. The main thrust of vendors in selling now is workflow enhancement, giving the radiologist more tools to do the job more easily.”

Degree of use and penetration
“The penetration of voice recognition is still relatively low. There are about 6,000 institutions total today in the U.S. that could potentially use it,” states Wanchoo. “About 300 radiology facilities use it today, and these are in larger, academic settings. Not necessarily 100 percent of all departments within radiology use voice recognition. Most radiology departments take the phased approach in deploying this technology, one section at a time.”

Hillcrest Hospital’s Pressly agrees than penetration is low, but growing.

“Currently, less than 10 percent of practicing radiologists are using voice recognition technology, because we’re entrenched in hospital systems that have interface problems and pre-existing mainframe computers and can’t tie voice technology into it,” he says. “Newer technology has eclipsed the old problems. Microsoft has voice recognition technology embedded in its software. There were constant interruptions with the initial voice recognition technology. You had to have near silence. Now, with modifications, new noise-cancelling microphones and new hardware, voice recognition technology can make out your words even with background noise.”

Technology has certainly been a factor in the low penetration rate and level of use.

“Up until some of the technology we’ve introduced in the last three to four months, it’s been hard to integrate new technology into an existing environment,” explains NCC’s Davis. “Now it can allow for a more seamless transition, by integrating our technology with the existing field [of traditional transcription practices].” Once the voice recognition technology is integrated alongside traditional transcription, it can then be more fully implemented, while the existing method is phased out.

Please refer to the February 2001 issue for the complete story. For information on article reprints, contact Martin St. Denis