The aims of this work were to measure the accuracy of one continuous speech recognition product and dependence on the speaker's gender and status as a native or nonnative English speaker, and evaluate the product's potential for routine use in transcribing radiology reports. IBM MedSpeak/Radiology software, version 1.1 was evaluated by 6 speakers. Two were nonnative English speakers, and 3 were men. Each speaker dictated a set of 12 reports. The reports included neurologic and body imaging examinations performed with 6 different modalities. The dictated and original report texts were compared, and error rates for overall, significant, and subtle significant errors were computed. Error rate dependence on modality, native English speaker status, and gender were evaluated by performing ttests. The overall error rate was 10.3 +/- 3.3%. No difference in accuracy between men and women was found; however, significant differences were seen for overall and significant errors when comparing native and nonnative English speakers (P = .009 and P = .008, respectively). The speech recognition software is approximately 90% accurate, and while practical implementation issues (rather than accuracy) currently limit routine use of this product throughout a radiology practice, application in niche areas such as the emergency room currently is being pursued. This methodology provides a convenient way to compare the initial accuracy of different speech recognition products, and changes in accuracy over time, in a detailed and sensitive manner.
Osteophytes associated with spondylosis have been implicated as a cause of multiple extraspinal manifestations. Symptoms are more likely to occur with the large osteophytes associated with diffuse idiopathic skeletal hyperostosis. In the thoracic region, osteophytes have been reported infrequently as a cause of extraspinal complications. We report a case in which an anterior thoracic vertebral osteophyte was responsible for chronic obstructive pneumonia due to obstruction of the right main stem bronchus. The patient's condition improved considerably after surgical resection of the compressing thoracic osteophyte.
R ADIOLOGY REPORTS in most medical settings are generally dictated by the radiologists and then transc¡ by a human transcriptionist, resulting in a text report. The radiologist then finalizes the transcribed report after reviewing it and assuring the accuracy of the text. Time delays between the various stages of this process usually mean that the final reports are available only after several hours or more have passed following interpretation of the examination.The emergence of automatic speech recognition software has suggested that all reading rooms operate in the direct dictation mode without involving the human transc¡ When used in conjunction with electronic systems for managing the text information (radiology information system IRIS]) and image information (picture archiving and communication system [PACS]), speech recognition software may allow all finalized radiology examinations to be delivered to clinicians within minutes of interpretation by the radiologist.Early speech recognition software products required the user to speak in a discontinuous manner, so that each individual word could be identified and transcribed. 1-70verall accuracy, as determined in one study of a discrete speech recognition system, was reported to be 97.6%. 7 The requirement for discontinuous speech made these products impractical for routine use in a high-volume radiology reading room. Newer products allow the user to speak in a more natural, continuous manner) Our aims in the current work include measurement of the accuracy of one continuous speech recognition product, investigation of the impact on accuracy of the gender of the speaker and status of the speaker asa native or non-native English speaker, and evaluation of the potential for routine clinical use of the system for radiology report transcription. METHODSIBM MedSpeak/Radiology software, version 1.1 (IBM Corporate Offices, Annonk, NY) was evaluated. This software allows continuous speech to be transcribed to text as it is spoken. Six speakers, three males and three females, familiar with medical and radiological terrninology participated in the study. Two of the speakers were non-native English speakers. Each speaker performed the minimum enrollment (training) procedure, and dictated a set of 12 preselected reports. The reports included neurologic and body imaging examinations performed with six different imaging modalities.Once the o¡ and dictated reports were compared, each discrepancy was classified as one of four different error types. Class 0 errors involved no change in meaning with respect to the original repon text, and the transcribed text was grammatically correct. Class 1 errors also involved no change in meaning, but the transcribed text was grammatically incorrect. Class 2 errors were those in which the meaning of the transcribed report text was different than that of the original report text, but the error was judged to be obvious. Class 3 errors also involved a change in meaning as compared with the original report text, but the error was judged not t...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.