Korean experts have created an SI system that distinguishes between Covid-19 and bacterial pneumonia based on analysis of computed tomography images of the lungs. The system is more than 98% accurate and can be used in other diseases.

Effective assistance to patients requires, first of all, an adequate diagnosis. In the case of Covid-19, diagnosis is mainly based on inaccurate and time-consuming genetic testing. Therefore, it is difficult to quickly and accurately identify patients infected with SARS-CoV2. At the same time, patients can spread the virus while waiting for the test result.

A CT-assisted lung examination allows faster diagnosis, but requires the eye of a radiologist, and scans often resemble images of lungs with bacterial infection.

In a new study published in Medical Imaging Science, scientists from Korea’s Daegu Kyungpook Institute of Science (DGIST) describe an AI-based algorithm that works great when replacing a radiologist.

“As scientists, we have also been affected by the epidemic, and we have passionately used our knowledge in medical imaging to create a system that will speed up diagnosis and improve work in hospitals,” the researchers say.

The system works with over 98 percent accuracy. It can also learn from limited data, including incomplete data.

The researchers say it can also be used to analyze other types of medical images, and the innovative architecture of the algorithms created makes it easier for clinicians to understand how they work.

According to the researchers, such programs may be especially useful in poor countries with limited access to specialists.

See also  Why is exposure to space dust an inevitable aspect of space travel?

More information on the source publishing page (https://www.sciencedirect.com/science/article/abs/pii/S1361841521001511#)

Marek Matakz

mat / beech /