A new healthcare tool that applies artificial intelligence technology to improve the accuracy of COVID-19 detection in chest x-rays has been developed and shared by Birmingham City University researchers.

DeTraC, created by computer vision and data scientists Professor Mohamed Gaber and Dr Mohammed Abdelsamea from the School of Computing and Digital Technology, uses machine learning to assess and diagnose using large datasets of images from several hospitals across the world.

The technology is now publicly available for the World Health Organisation and the global medical community as an open-source program. 

The announcement arrives on World Health Day, following a period of 10 days where the two academics – in collaboration with researcher Asmaa Abbas from Assiut University in Egypt - worked to adapt and deploy their diagnostic tool in response to updates from WHO tracking the spread of the virus. {module INSIDE STORY}

DeTraC, which stands for Decompose, Transfer and Compose, is a convolution neural network that can be trained using a limited number of medical images.

"We believe that our work will open the door for a number of other researchers to help medical professionals to improve their diagnosis by providing unbiased solutions directly from the images. It will boost artificial intelligence research in medical image processing and analysis – which could ultimately lead to a faster diagnosis of COVID-19," said Dr Mohammed Abdelsamea.

The Birmingham City University scientists’ work is the latest in a series of announcements made by the UK institution utilising the expertise, knowledge, resource and capacity of staff and students in order to contribute to the global fight against Coronavirus.