New Artificial Intelligence Platform Detects COVID-19

by Karishma Abhishek on Nov 24 2020 11:15 PM

New Artificial Intelligence Platform Detects COVID-19
New artificial intelligence (AI) platform that detects COVID-19 by analyzing X-ray images of the lungs, was developed by researchers at North-western University, Evanston, Illinois.The study remains to be published in the journal Radiology.
The team developed a machine-learning algorithm called DeepCOVID-XR, which was tested in the largest COVID-era dataset of 17,002 X-ray images, among which 5,445 were obtained from COVID-19-positive patients from sites across the North-western Memorial Healthcare System.

An algorithm is a basic set of rules designed for a computer to perform/train to a particular task. The DeepCOVID-XR algorithm could analyze 300 random test images, almost 10 times faster and 1-6% more accurately (82% accuracy) than a team of specialized thoracic radiologists with 76-81% accuracy.

"We are not aiming to replace actual testing. X-rays are routine, safe and inexpensive. It would take seconds for our system to screen a patient and determine if that patient needs to be isolated", said Northwestern's Aggelos Katsaggelos, an AI expert and senior author of the study.

AI and COVID-19:

Though the presence of the virus is not confirmed by this new AI, the time required for the conventional COVID-19 test – hours or days, to receive the results is eventually overcome with the use of the technique. This can help triage COVID-19 patients for isolation at a much faster rate and lower cost.

The technology has its limitations in the case of asymptomatic COVID-19 patients or when they don’t exhibit any lung signs on their chest X-rays. This limits the diagnosis from a radiological point of view for even the clinicians, suggesting that the DeepCOVID-XR algorithm is not a replacement for conventional COVID-19 testing.

The algorithm is still under the research phase for its further validation. Hence this study offers the AI platform to be publicly available for designing new data sets in various aspects of radiology.