SpeechWorks VR (Voice Recognition)

The SpeechWorks VR (Voice Recognition) product adds a radiologist’s lexicon and learning tools to the Dictation product. It displays the same user screen, but as the radiologist dictates, the speech recognition engine utilizes an underlying radiologist’s lexicon that is tied into what BRIT is calling a Knowledge Tree to translate into text. The lexicon provides the known words and the Knowledge Tree ties into the Metathesaurus*, as well as other Internet-based resources, in real time, to establish the context in which the words are being used. For example, if the radiologist is dictating a chest study, the probability of the word “rib” will be extremely higher than that of the word “wrist”. As the radiologist (and transcriptionist, if one is used) makes corrections, the system learns. All errors are reported back to BRIT for tweaking of the engine. At anytime, the user can revert to an earlier speech model. All new words added can be shared among all SpeechWorks users or just users of the radiologist’s own group. At the end of the dictation, the radiologist can decide to sign the report or send it to a transcriptionist for correction. Once signed off, the report is distributed, the same as the other SpeechWorks products. SpeechWorks VR with its Knowledge Tree will evolve into structured reporting over the next year. Once it is ready, the user will determine when they are ready to make the transition.

*As per the Fact Sheet from the US National Library of Medicine: “The Metathesaurus is a very large, multi-purpose, and multi-lingual vocabulary database that contains information about biomedical and health related concepts, their various names, and the relationships among them.” It is developed and distributed by the National Library of Medicine as part of the Unified Medical Language System ® (UMLS ) project.