Korean researchers have developed artificial intelligence that predicts lung function based on computed tomography (CT) test results with an accurate rate of 90 percent or higher.

Two Asan Medical Center (AMC) professors – Lee Se-won of the pulmonary department and Kim Nam-kook of the convergence medicine department – made an AI algorithm that can differentiate patients with lowered lung function based on CT test results.

To do so, they made the algorithm learn low-dose chest CT and lung function test results of 16,148 people who received health checkups at AMC from January 2015 to December 2018

Lung capacity test results are divided into two. One is forced vital capacity (FVC), which is the amount that people breathe out as much as they can after inhaling as much as possible. The other is forced expiratory volume in one second (FEV1), the amount of air people are forced to exhale per second. The AMC team’s AI predicted FVC and FEV1 with respective accuracy of 93 percent and 90 percent.

It also predicted the FEN1/FVC ratio, an indicator used to determine whether there is a high risk of COPD (chronic obstructive pulmonary disease), with about 85 percent accuracy.

“There have been studies on the association between the lung’s anatomical characteristics and lung function as shown in CT images. However, the research that predicts lung function with only CT images using AI algorithm armed with deep learning has just begun,” Professor Kim said. “The study result is quite significant.”

Although various inhalers have been developed and used, they have limitations to curing COPD completely, Kim stressed that it is essential to detect the disease early and delay its aggravation as far as possible by correcting lifestyle.

Therefore, we will continue research to find people with COPD risks without particular symptoms in an early phase and treat them,” he added.

Radiology (IF=29.146), one of the most authoritative journals in radiology published by the Radiology Society of North America, ran the study result in its latest issue.

Low-dose chest CT scans are used to diagnose structural conditions in the lungs, such as lung cancer, lung nodules, and pneumothorax. In general, it is often performed during medical checkups.

To diagnose lung cancer, people with high risks among adults aged 54-74 can get low-dose chest CT scans once every two years with the government’s support.

the other hand, lung function tests that diagnose chronic COPD are influenced by various factors, including examinees and their temporary lung capacity changes depending on test situations. These tests are not included in the state health checkup list. And without specific breathing problems, individuals do not usually receive tests.

COPD has no symptoms until lung function decreases significantly. It is impossible to recover the damaged alveoli after the symptoms appear. Patients can only reduce the damage as much as possible. Therefore, it is crucial to detect and treat it early, according to experts.