SALT LAKE CITY - Installing a voice recognition system can offer time savings and workflow benefits over traditional radiology report dictation and transcription. But radiology departments need to evaluate potential systems carefully in order to choose the one that best fits the way they operate.
That’s the advice of Mary Freeh of the University of Utah Medical Center in Salt Lake City. Freeh shared her department’s experiences in acquiring voice recognition technology in a SCAR presentation on Thursday. Freeh’s facility is currently rolling out voice recognition in a phased approach.
The department’s primary goal in adopting voice recognition was to minimize its impact on users. This led the VR acquisition team to develop an evaluation system for rating products from two voice recognition vendors.
There were five major criteria on the evaluation checklist:
- Ease of use and functionality of the system, such as support for ICD-9 coding and the ability to edit and approve reports off-site.
- The system’s ability to interface with the university’s existing RIS and PACS networks, through features such as the ability to create a worklist on the PACS workstation from completed exams in the RIS.
- The functionality of the product’s speech engine, such as its ability to recognize a large number of radiology-specific terms and to recognize words from speakers with foreign accents.
- The quality of vendor support, including radiologist training and on-site installation and service support.
- The stability and customer history of the vendor providing the technology.
Of particular importance is RIS integration, Freeh said. "If you are looking at voice recognition, make sure that this piece works," she said. "Your voice recognition isn’t going to work if you don’t connect to your RIS."
The voice recognition system should have the ability to update the RIS with addenda to reports, she said. The team also wanted the ability to send a preliminary report to the RIS, because the department often sends preliminary reports to referring physicians.
While PACS integration is important, it may take more time to accomplish, Freeh said. The department’s voice recognition system isn’t currently integrated with its PACS, but they’re expecting to implement this feature some time this summer. The department probably won’t roll out voice recognition to the entire department until PACS integration is available and has been fully tested.
Support for ICD-9 coding is also crucial. All of the department’s radiologists must pick an ICD-9 code when dictating a report, and the evaluation team wanted to ensure that the voice recognition system would support searches by ICD-9 code and that its codes were accurate.
The team found that one of the products under evaluation had an ICD-9 code database, but it was out of date. The hospital, which ended up purchasing the system, is working with the vendor in updating the database.
"Our immediate statement to them was, ‘You have to get this coding system updated,’ because it doesn’t do us any good to try to add a code and it’s not in the list," Freeh said. "That’s an active project."
Another important feature is error reports. The system should generate an error report for problems that have occurred in either the voice recognition system or during the interface with the RIS. This enables the department to catch mistakes such as when a scheduled exam or dictated report doesn’t get transferred from one system to the other.
Ultimately, the team’s final criteria -- company stability and history -- ended up being the most important, Freeh said. This was particularly important because the hospital was burned once in the past when it purchased a product from a firm that was acquired by another company that subsequently discontinued support of the product.
By Brian CaseyAuntMinnie.com staff writer
May 4, 2001
Related Reading
Speech recognition turns in 90% accuracy rate, April 2, 2001
Voice recognition provides speed, cost-benefit advantages, November 29, 2000
Click here to post your comments about this story in our PACS Digital Community. Please include the headline of the article in your message.
Copyright © 2001 AuntMinnie.com
