InterSpeech 2021

Automatic Radiology Report Editing through Voice
(3 minutes introduction)

Manh Hung Nguyen (VinBrain, Vietnam), Vu Hoang (VinBrain, Vietnam), Tu Anh Nguyen (VinBrain, Vietnam), Trung H. Bui (Independent Researcher, USA)
We present a system that allows radiologists to edit the radiology report through their voices. This is a function in our bigger system at VinBrain LLC that uses AI algorithms to assist radiologists with chest x-ray diagnosis, the system can suggest the abnormalities, then bases on the radiologist’s confirmations or conclusions to automatically generate the report using predefined templates. We then allow the radiologist to freely edit the report using voice. The system combines two components, the first is the Speech Recognition System (SRS), and the second is the Natural Language Understanding System (NLUS) that executes the user’s command. The user can delete, modify or add an arbitrary whole sentence. In addition, we successfully developed an SRS for such a non-mainstream language as Vietnamese and adapted it for the radiology domain.