"AI CT Analysis Predicts Risk Patients of 'Chronic Obstructive Pulmonary Disease' with 90% Accuracy"

Seowon Lee and Namguk Kim Research Team at Seoul Asan Hospital

An artificial intelligence capable of predicting patients at risk of chronic obstructive pulmonary disease (COPD) with over 90% accuracy using only computed tomography (CT) scan results has been developed.

Professor Se-won Lee from the Department of Pulmonology and Professor Nam-guk Kim from the Department of Convergence Medicine at Seoul Asan Medical Center announced on the 26th that they have developed an AI that can predict COPD risk based on CT scan results by training it with low-dose chest CT scan results and pulmonary function test results.

Professor Se-won Lee (left), Department of Pulmonology, Asan Medical Center, Seoul, and Professor Nam-guk Kim, Department of Convergence Medicine.

COPD is mainly diagnosed through pulmonary function tests that measure lung capacity. It usually shows no significant symptoms until lung function has considerably declined, and once symptoms appear, the damaged alveoli cannot be restored, making early detection and treatment crucial.

The research team created an AI algorithm that distinguishes patients with reduced lung function by training it with low-dose chest CT scan results and pulmonary function test results from 16,148 individuals who underwent health checkups between January 2015 and December 2018.

The lung capacity test results are divided into 'Forced Vital Capacity (FVC),' the amount of air forcibly exhaled after taking a maximum breath, and 'Forced Expiratory Volume in 1 second (FEV1),' the volume of air forcibly exhaled in one second. The AI developed by the team predicted these two values with 93% and 90% accuracy, respectively. Additionally, it successfully predicted the indicator used to determine high-risk COPD groups (FEV1/FVC) with approximately 85% accuracy.

Professor Nam-guk Kim said, "Although previous studies have explored the anatomical features of the lungs visible in CT images and their correlation with lung function, this study is significant as it is still in the early stages of using deep learning-based AI algorithms to predict lung function solely from CT images."

Professor Se-won Lee stated, "Although various inhalers have been developed and are used as treatments for COPD, complete cure through inhaler use alone is difficult. Early detection, lifestyle corrections such as smoking cessation, and delaying disease progression are important. Therefore, we will continue to research diagnostic methods that can detect COPD risk patients without specific early symptoms as early as possible so they can receive prompt treatment."

This study was recently published in 'Radiology (IF=29.146),' one of the most prestigious journals in the field of radiology published by the Radiological Society of North America.

This content was produced with the assistance of AI translation services.