PinkTech Design develops AI algorithm to classify X-Rays of COVID-19 patients

27 July 2020 | News

X-Rays and CT (computed tomography) scans can be an equally effective method to detect COVID-19 as well as present greater detail and accuracy in the results

Image credit- shutterstock.com

Image credit- shutterstock.com

New Delhi based Pink Tech Design, an IT innovation company, has recently developed an AI algorithm to classify X-Rays of patients as being COVID19 positive. With the development being successful, it can pre-screen patients who need COVID19 confirmatory tests.

The number of coronavirus cases in India and around the world has been growing exponentially. While researchers are attempting to learn more about SARS-CoV-2 and develop a cure or vaccine, equal efforts are being made by governments and various organisations to improve the testing process, making it more precise and accurate, faster and affordable so that tests can be performed on the majority of the population to control the spread in an effective way.

Once the company got access to a dataset of COVID-19 positive X Rays and COVID-19 negative X Rays released by Kaggle, they processed the data and trained their COVID-19 detector with Keras and TensorFlow. After many optimisations, the model gave 90% accuracy with a sensitivity of 0.8 and specificity of 1.0. Though these are initial results, the development has made it clear that X-Rays and CT (computed tomography) scans can be an equally effective method to detect COVID-19 as well as present greater detail and accuracy in the results.

Speaking about the development, Dr. Kanav Kahol, CEO, PinkTech Design said, “The project has led us to some extremely promising results and we are keen to build on this success rapidly to help in the fight against Coronavirus. The AI model employed in the study is able to predict results with great accuracy. However, our research team continues to strive for even more robust and reliable results. I am incredibly proud of my team for developing this. These are times of quick action and we are happy to provide this model free of cost to any healthcare institute.”

Comments

× Your session has been expired. Please click here to Sign-in or Sign-up
   New User? Create Account