The Application of Speech Recognition in Radiology Information System

Abstract

The Radiology report is the fundamental means by which radiologists communicate with clinicians and patients. The traditional method of generating reports is time consuming and expensive. Recent advances in computer hardware and software technology have improved Speech Recognition systems used for radiology reporting. In this paper, we design a continuous Mandarin Speech Recognition dictation system for radiology reporting in Radiology Information System based on the CMU Sphinx recognition engine. We compare our system with IBM ViaVoice pro v9.1. The experiment results show that both systems can achieve a high accuracy rate when using recognition for Brain Skull CT report generation in the laboratory environment. Furthermore, our system performs slightly better than ViaVoice when training and testing data are both randomly selected from the same text corpora. We conclude that it is technically feasible to employ Speech Recognition for report generating in Radiology Information System. In summary, the Speech Recognition technology has matured to the point where it is now viable as a mean of providing more timely delivery and better quality of medical services to patients, and this technology will be widely applied in the healthcare industry in the near future.