Tamaki Ichikawa MD, Presenter: Nothing to Disclose

Taro Takahara MD, PhD, Abstract Co-Author: Nothing to Disclose

Yutaka Imai MD, Abstract Co-Author: Nothing to Disclose

Yasuki Kishi, Abstract Co-Author: Nothing to Disclose

Accracy rate of recognition on continuous speech recognition(CSR) systems is about 96%, it takes longer time for correction of errors in reports than time of transcription.The purpose is to evaluate usefullness of radiological reporting with a comnbination method of CSR system and error correction by transcriptionists.

Four transcriptionists were classified 2 groups; A group of 2 members had more than 10 years of experience of transcription, B group of 2 members had less than 3 months of experience of transcription. All transcriptionists heard same 100 dictation speech files and created radiological reports using two methods. One was to utilize the conventional human transcription and the other utilizes a combination method using with CSR conjoined with manual error correction by thranscriptionists. On the speech files under spontaneous dictation using CRS system, 301 words of false recognition occurred out of 7349 words, average recognition accuracy rate 96.0% and total time of speech recognition was 2116 seconds. We compared the accuracy rate and report creating time in conventional transcription with a combination method in both groups. Statistical difference was tested by Mann-Whitney U-test. And we evaluated individual dependence of transcriptinists on accuracy rate and report creating time in both methods by using Post Hoc test. We used a CSR system AmiVoice (Advanced Media, Inc., Tokyo, Japan).

Accuracy rate was more than 99.7% in all transcripnists and in both methods. There was no significant difference in accuracy rate between the conventional transcription and a combination method. In conventional transcription, the members of B group can took two times as late as the members of A group to transcript. There was no significant difference in the error correction speed depend in combination method on the years in an experience.

Combination method of CSR system and error correction by transcriptionists is useful and accurate for creating of many radiological reports speedily.

Ichikawa, T, Takahara, T, Imai, Y, Kishi, Y, Radiological Reporting with Combination Method of Continuous Speech Recognition System and Error Correction by Transcriptionists.  Radiological Society of North America 2005 Scientific Assembly and Annual Meeting, November 27 - December 2, 2005 ,Chicago IL. http://archive.rsna.org/2005/4408103.html