VIT students' AI-based app can help detect COVID-19 by studying a person's Chest X-Rays

An app developed by two PhD students of Vellore Institute of Technology (VIT) can apparently look at X-Rays and detect whether the person has COVID-19, using Artificial Intelligence. The app developed by Techraion SystemWorks a start-up by two students Ankush Rai and Abhishek Rai, is currently undergoing clinical trials.

The app which the duo plan to deliver to hospitals and doctors free of cost, upon approval, can apparently detect the virus in a few seconds, looking at X-Rays. The developers claim that if the person is infected by the novel Coronavirus, the app can also tell what stage they are in, making it easier for the doctors. "The blood testing to check for Coronavirus usually takes around four to five days. There is often a problem in identifying the stage in which the infected individual is," says Ankush. "Coronavirus causes lung infection which in turn causes pneumonia. Now normally, pneumonia can be bacteria or Corona induced. It is difficult to differentiate between the two. But this app will tell distinctively if the patient's pneumonia is induced by Coronavirus," he says.

Ankush goes on to explain the working of the app. "A doctor can take a picture of the patient's X-Ray and pass it to the app and identify if its corona or bacteria-induced pneumonia. It gives the result in 1 or 2 seconds and also identifies the severity of the situation," he says. Usually, COVID-19 infected patients develop pneumonia in 7 days and the condition worsens in 14 days.

The app can identify these stages accurately. "It also shows the areas of the lungs that are infected by the virus. We can automate the diagnosis here and arrange for the person to be put in ventilation easily, using the app," says Ankush. "The main AI application is running on the server and app server as the client hence with it we will be able to process and caters to every doctor across the country," he adds.

The duo is currently in talks with BiRac and BHEL and a few medical practitioners about the practicality of the app. "We hope to get it validated soon, remove the drawbacks and get it running in 2-3 months," he says.