超声语音识别系统的研发及临床应用

The Research and Clinical Application of Ultrasound Speech Recognition System

目的研究语音识别技术在超声诊断中的应用价值。方法应用计算机领域智能语音识别技术研发超声语音识别录入系统,三名医生使用该系统对15名患者进行了检查,分析系统使用的语音识别率。结果三名医生平均语音识别率分别为83.27%、78.70%和54.36%,总的平均语音识别准确率为71.93%,系统在使用过程中出现过一次语音识别失灵现象。结论将语音识别技术引入超声检查,具有潜在应用价值:优化超声检查工作流程、提高工作效率(语音自动录入和生成报告),但目前系统还在研发中,识别率暂未达到95%的预期效果,还需在模版、降噪、可靠性方面继续进行研究。

Objective To research the clinical value of speech recognition technology in ultrasound diagnosis. Methods Ultrasound speech recognition input system was developed using intelligent speech recognition technology in computer field. The system was used in 15 patients by 3 doctors, and the speech recognition rate was analyzed. Results The average speech recognition rates of three doctors were 83.27%, 78.70% and 54.36% respectively, with the overall average speech recognition rate of 71.93%. There had been a speech recognition avoidance during the process. Conclusions Speech recognition technology used in ultrasound diagnosis has potential value. But its speech recognition rate does not reach 95% of the desired effect, and the system needs to be improved in template, noise-canceling and reliability.