State research institute develops more accurate AI-based chest X-ray COVID-19 diagnosis algorithm

By Park Sae-jin Posted : May 25, 2020, 14:28 Updated : May 25, 2020, 14:28

[Courtesy of Korea Advanced Institute of Science & Technology]

SEOUL -- A South Korean research institute has developed an artificial intelligence-based chest X-ray diagnosis technology that is more accurate than medical specialists. This technology could help health officials fighting COVID-19 at the frontline with limited resources.

Currently, reverse transcription-polymerase chain reaction (RT-PCR) and computerized tomography (CT) scans that require expensive machines are widely used for COVID-19 screening. The Korea Advanced Institute of Science & Technology (KAIST) said on Monday that an X-ray diagnosis algorithm developed by its research team used artificial intelligence (AI) and deep learning technologies.

Research organizations around the world have tried to create AI-based diagnosis methods using X-ray images that require a huge amount of data for AI to study and learn.

KAIST said that its pre-processing and patch-based convolutional neural network has allowed AI to increase the accuracy of diagnosis by studying a relatively small number of trainable parameters for COVID-19 diagnosis. The technology also provides highlighted images to show where COVID-19 lesion patterns are found.

KAIST predicts that its technology would enable the efficient use of limited medical resources. The research was published through an online version of IEEE transactions on medical imaging, an international scientific journal.

 
기사 이미지 확대 보기
닫기