Radiology is the fastest-growing sector of health care in the United States by far. According to some estimates, it is growing at a rate of $100 billion annually, with a yearly growth rate of 20%. Health care facilities are seeking solutions that allow them to ensure appropriate high-cost exam processing, deliver results more efficiently, improve communication between physicians, and analyze years of data to improve the practice of radiology. Moreover, they are looking for ways to harness the power of existing IT infrastructure.

In spite of its birth pangs, as digital technology has advanced over the years, voice recognition technology (VR), by all accounts, has now surpassed traditional transcription methods for saving time and increasing accuracy in radiological reporting. For some, it has now become a “must.”

How effective is the tool? John Floyd, MD, who led a sizeable study comparing voice recognition with traditional dictation, said radiologists in one office increased their 1-hour completion rate to 82% from just 6%.

According to Dictaphone, a leading supplier of speech recognition (SR) equipment, and other VR suppliers, interpreting physicians can now turn around reports within seconds, instead of days. Since radiologists can sign off on a completed report while viewing the original image, the need to double-check reports returning from transcription is eliminated. As a result, speech recognition not only boosts radiologist efficiency, it also improves referring physician satisfaction, enhancing patient care and yielding significant cost savings.

With this closed-loop technology, health care providers are experiencing a savings of well over $200 million annually. Dictaphone claims that its PowerScribe for Radiology reduces transcription costs by up to 85% and can pay for itself in 6 months or less. Indeed, one in three radiologists in the United States now uses PowerScribe for Radiology, including 61% of the nation’s top 100 hospitals (National Benchmark for Success, Solucient, 2005) and 100% of Honor Roll Hospitals (US News & World Report, 2006).

PowerScribe fits today’s “on demand” browser-based requirements, providing remote connectivity from multiple locations or home for dictation, review, and sign-off. Radiologists open a browser and connect to a web server to access their unique user profiles and speech files. In most cases, the user interface is clear and easy to use, presenting all essential information in understandable windows and tabs.

Typically, VR systems go through a learning curve phase so the software can recognize normal speech and understand it in full context. The physician does not need to be a skilled typist or proficient at data entry. Nor is there a change required to dictation workflow, recording method, or location. This continuous learning improves performance and accuracy.

MedQuist’s senior product manager for speech recognition, Chris Spring, agrees that VR has Star Trek-like understanding of speech. “The punctuation is inserted automatically,” he said. “The ‘uhs’ and ‘ums’ are ignored, and the other idiosyncrasies of speech are understood in context. This saves an enormous amount of time and promotes consistency. I think it makes doctors like me 30% more productive. Because of VR, I now get to go home at night. It’s just that simple.”

Lynn Kosegi, director of professional services at M*Modal (Multimodal Technologies Inc), another leading SR provider, noted that their system “works with deep integration, which allows it to check a report and alert the physician for missing data and relevant information. It also integrates all data and can combine data, such as including a PACs image with the text documentation. It focuses on context rather than individual words and then structures the information for use in health record systems and medical reports.”

According to Floyd, of Radiologist Consultants of Iowa, VR “turnaround time” is the key. “Time is critical,” he said. “If the system can’t translate accurately within a limited amount of time, then everything else is a work-around. VR has an extreme impact. With VR, we are reducing turnaround time from as much as 3 days down to minutes.”

Research Confirms Results

Floyd is the author of an extensive survey of SR technology in contrast to traditional transcription methods. In his study, presented at the RSNA conference of 2006, his evidence is convincing. Comparing hundreds of reports among several rural hospitals using VR technology versus traditional transcription methods, he said, “The error rates for SR in this specific practice are not statistically different from, and may be better than, the error rate for traditional transcription.”

He suggests that a significant factor behind the improvement resides in giving the radiologists financial ownership in the VR system, which provides income incentive for good performance. “Psychological ownership of the final product may lead to improved effort through personal pride in the process,” he said.

“Referring doctors were happy, outpatient and inpatient as well as ER clinicians; calls to the department for ‘wet’ readings markedly declined; ‘second’ reads of previous dictations were markedly reduced; and our test group of rural hospitals now have similar TAT?(turnaround time) as our large urban hospitals,” Floyd said.

He points out that a further advantage is VR’s ability to integrate with PACS.

Steven M. Defossez, MD, medical director of North Shore Magnetic Imaging Center in Peabody, Mass, and director of MRI at Beverly Hospital, also maintains the importance of the increased accuracy and time of VR technology. “The results are dramatically impressive,” he said. “The increase in quality and accuracy is very important. Prior to our acquisition of voice recognition, our report turnaround at the hospital was typically one-half day to 2 days. Now our report turnaround time is a matter of minutes. One of our local competing hospitals, which does not have voice recognition, had report turnarounds as long as 5 days.”

Yet there are other advantages as well. Kosegi noted the M*Modal system helps ensure compliance with the Physician Quality Reporting Initiative by encoding clinical facts such as medications, dosages, allergies, and measurements. “We support a clinical vocabulary for various disciplines.”

And Defossez adds another key advantage. Critical results management reporting software is now available from Dictaphone. This software program is called Vocata. “Utilizing Vocata with the few keystrokes, in addition to the report being instantly faxed to the referring physician’s office, and subsequently mailed in paper form to their office, critical results are simultaneously paged to the referring physician’s digital beeper or cell phone,” he said. “If they have not retrieved these results within a certain period of time, the software alerts the radiology department that a person-to-person phone call must be made to ensure that the referring physician has received these critical results. In this way, all critical findings must be viewed by the referring physician, and the software documents that this retrieval of critical findings has been performed. This fulfills the radiologist’s legal and moral obligation to communicate the critical findings with the referring physician.”

Another advantage of VR technology is versatility. “Physicians can enjoy full control of their own clinical documentation, with no capital expense or change in workflow,” said Joel Fontaine, M*Modal business development director. “Physicians can choose multiple options: immediate ‘edit in place’ and sign-off; batch processing for multiple report review and sign-off; or transfer for Medical Language Specialist (MLS) editing and completion with return to the physician for sign-off. Physicians can also select their own templates to begin the transcription process, cutting down on actual dictation time.”

According to Fontaine, M*Modal technology captures and comprehends any physician’s dictation without training by factoring into the process the way people really speak, regardless of dialect, grammar, mispronunciation, or speaking rate. Second, it automatically creates a unique combination of narrative text, and structured, encoded data converts dictation into narrative text with proper grammar and punctuation. Third, it simultaneously and efficiently validates the accuracy of the text and data for physician sign-off, which shows the narrative and data in a way that becomes intuitive for editing. And last, it easily integrates actionable data into existing PACs, RIS, and EHR by creating a Meaningful Clinical Document, which allows for advanced functions not currently possible with conventional transcribed reports or scanned content.

But the versatility doesn’t stop there, said Defossez. “Dictaphone has a database that includes only words routinely utilized in a radiology report. Thus, the word ‘contrast’ is not mistyped as ‘contras’ because the radiologist-specific word database does not contain extraneous words. This constrained vocabulary improves VR accuracy. This also allows a radiologist report to be completely comprehensive. An additional feature that increases the radiologist’s efficiency is the color of the typing, which appears on the screen as the radiologist dictates. The program responds with normal black type when it interprets the canned report section as okay without the need for editing. On the other hand, blue typing is displayed when the section is problematical and must be reviewed. This saves the radiologist time, as they need to proof only the blue type. The black typing has been proofread the first time and no longer needs to be proofread each subsequent time it is used.

“I especially like this editing feature,” said Defossez. “The Dictaphone microphone has a rewind button. If I dictate something in error, I can hit the rewind button and hear ‘chatter,’ as the virtual digital tape rewinds. I can then press play and listen to the point where I want to redictate. At that point, I simply push record and start talking. This simultaneously erases everything I had previously dictated after that point, and re-records and transcribes what I am now dictating. Being able to rewind and talk over a portion of my dictation saves time,” he said. “Otherwise, I would have to directly edit the report, using mouse and keyboard, or delete one word at a time using a button on another microphone.”

It can be said that VR providers offer a wide range of alternatives of speech-driven clinical documentation solutions that significantly help health care organizations speed report turnaround time, reduce costs, and improve quality of care. Also, VR technology enables radiologists to have complete control of reporting. It provides access to relevant clinical content and delivers, manages, and audits critical patient findings while calculating performance metrics, business indicators, and forecasting trends.

All vendors have different and exclusive approaches, and various packages can be added to increase performance, but most feature a web-based, front-end speech recognition solution that can help radiology departments significantly reduce report turnaround time and lower transcription costs by as much as 75% to 100% a year.

James Markland is a contributing writer for Axis Imaging News. For more information, contact .