Project: Integrating DeepSpeech Voice Dictation for Radiology Reporting

This is a project carried over from last year that was not completed by the student. See this thread : Project: Integrating Voice Dictation for Radiology Reporting

Summary : There is a high burden of data entry tasks when using any medical record. This takes away from patient care as doctors are forced to spend a lot of time entering all the required fields. Voice transcribing/ dictation solutions exist , but are expensive and some do not work for medical jargon

This project will be to integrate into the #toolkit project a trained voice model based on the DeepSpeech project of Mozilla - referenced here

  1. https://github.com/mozilla/DeepSpeech
  2. https://blog.mozilla.org/blog/2017/11/29/announcing-the-initial-release-of-mozillas-open-source-speech-recognition-model-and-voice-dataset/
  3. https://developer.mozilla.org/en-US/docs/Web/API/Web_Speech_API/Using_the_Web_Speech_API#Speech_recognition
  4. https://voice.mozilla.org/en

Mentors @judywawira @sunbiz @namratanehete

Objectives

  • Integrate DeepSpeech into the toolkit application - allowing for uploading of updated training models
  • Train a medical language model for DeepSpeech specific to radiology by creating a simple voice donation tool and training the model
  • Integrating voice dictation for radiology reporting

Technical Requirements Java, MySQL, JavaScript, HTML/CSS, MachineLearning, deep learning

1 Like

This is very similar to this:

This is the one I want to keep and remove the other one from this list

1 Like

Removed it from the list.

Hi, I am really interested in this project as I am a Ph.D. student in the smart healthcare field. I have a question about the training part: are we intending to use existing datasets for reporting about radiology, or we want to create the voice donation tool firts, and then train on the collected data? And thank you,

Planning for voice donation … don’t know any radiology datasets available in the public domain unless you do

1 Like

Hi. I am Avinash Agarwal, a GSoC aspirant and I am really interested in this project as I have always been in favour of using Machine Learning to solve real-life medical problems. I will be very grateful to work in a team which serves this purpose. Could you please help me how to get started? Is there any specific screening task we are supposed to complete first?

Hello,I am Raghuram. I’m really interested in this project but I have a question regarding voice recognition i.e there are cases where the patient’s voice might change like Pharyngitis and Tonsillitis. Are there any measures?? or we should develop a system that detects even with these problems??

The patient wouldn’t be dictating, the Radiologist would be.