What dictation software do radiologists use?

The radiology departments we read about dictate their reports into a reporting system rather than a general dictation app. The best-known product in this group is PowerScribe, a radiology reporting system that connects to the department's Radiology Information System (RIS) and Picture Archiving and Communication System (PACS). We could not find an independent survey of which products radiologists use, so this answer names the systems we could document and says what they do. Your department's reporting system decides the answer for you.

Radiologists dictate into a reporting system, not a general app

A reporting system does more than turn speech into text. It opens the study, shows the images beside the report, and sends the finished report back to the record. A general dictation app can write text into a window, but it does not connect to the imaging system. Specialist products are built around that connection, as the professional dictation guide explains.

PowerScribe shows that pattern. A 2007 announcement described Yale-New Haven Hospital's radiology department adopting PowerScribe, and said the system would integrate with the hospital's RIS from GE Medical and its PACS software from FUJIFILM Medical Systems ¹. The same announcement said the hospital processed more than 365,000 reports a year and planned to reduce turnaround for them ². That is a vendor's plan from 2007, not a measured result, and the integration depends on the exact RIS and PACS versions in use.

Speech recognition cut turnaround, and one audit found more errors

The strongest result we found is on speed. After speech recognition was introduced at a 150-bed community hospital, the median report turnaround fell from 24 hours to 1 hour ³. That is one hospital with a before and after design, so other changes may have contributed.

The accuracy result runs the other way. In a 2010 audit of 50 MRI reports per method from two radiologists, 42 and 30 finalized voice-recognition reports contained errors, against 6 and 8 transcriptionist reports ⁴. The audit is small, and its system is older than the ones sold today. It still shows why a faster report is not automatically a better one.

It depends on your department and your reports

• Your reporting system. If your department already uses a reporting system with an integration, a general app may not be worth testing for final reports.
• Your RIS and PACS versions. An integration works on the versions it was tested with, so check them against the vendor's list.
• Your report types. Structured templates for common studies need different vocabulary from a free-text consult letter.
• Your error audit. A tool that is fast on a demo can still produce more errors on real studies, so audit a sample of finalized reports.

A general dictation app is enough for letters, admin notes and drafts that a radiologist reads and signs in a document. It is not a substitute for the reporting system if the report has to be finalized inside the department's record.

What to test before you buy or change systems

1. Ask which reporting system and version your department uses, and which RIS and PACS it connects to.
2. Dictate a sample of real reports, including a complex one, and count the corrections needed before signing.
3. Audit finalized reports for errors, comparing speech-recognition reports with transcriptionist reports, as the 2010 audit did.
4. Ask for turnaround and error data from sites that use the same version, and check the date of any figure.

For the product name, see the PowerScribe glossary entry. For the Epic route, see whether Epic has built-in dictation. For the veterinary equivalent, see the best dictation app for veterinarians. The dictation software hub lists every guide on this topic.