Related Experiment Videos
Report generation using digital speech recognition in radiology
F Vorbeck1, A Ba-Ssalamah, J Kettenbach
1Department of Diagnostic Radiology, University Hospital of Vienna, Austria.
European Radiology
|April 18, 2001
Summary
Continuous speech recognition (CSR) technology can save time in generating radiology reports, with an average 19% faster report generation. While CSR is never slower than manual typing, time savings vary based on speaker, typist, and topic.
Area of Science:
- Radiology
- Medical Informatics
- Speech Technology
Background:
- Radiology report generation is time-consuming.
- Efficient report creation is crucial for clinical workflow.
- Digital tools are being explored to optimize medical documentation.
Purpose of the Study:
- To evaluate the time-saving potential of continuous speech recognition (CSR) in radiology report generation.
- To compare the efficiency and error rates of CSR versus conventional typing for radiological reports.
Main Methods:
- Two radiologists dictated 450 reports across five radiological topics using a German CSR system.
- Reports were generated using both CSR and conventional typing methods.
- Time efficiency and error rates were compared between the two methods.
Main Results:
- CSR systems generated reports 19% faster on average compared to conventional typing.
- The mean error rate for CSR was 5.5%, versus 0.4% for conventional typing.
- Maximum time savings of 28% were observed for sonography reports; CSR was never slower than manual typing.
Conclusions:
- Continuous speech recognition (CSR) is a useful tool for saving time in clinical radiology report generation.
- The degree of time savings and error rates are influenced by factors such as speaker, typist, and the CSR system's vocabulary.
- CSR systems offer a viable alternative to conventional typing, improving efficiency in radiological documentation.