Asan Medical Center and Samsung Medical Center Joint Research Team Develop AI Based Quantitative CT Analysis Technology to Precisely Track Pulmonary Fibrosis Progression

Increase of More Than 4 Percent in Pulmonary Fibrosis Score Over One Year Linked to Higher Mortality Risk, Providing an Objective Disease Assessment Marker

Findings Published in the American Journal of Respiratory and Critical Care Medicine (Impact Factor: 21.7)

▲ (From left)  Professors Jooae Choe of the Department of Radiology and Ho Cheol Kim of the Division of Pulmonology and Critical Care Medicine at Asan Medical Center, together with Professor Ho Yun Lee of the Department of Radiology at Samsung Medical Center.

Idiopathic pulmonary fibrosis (IPF) is a chronic, progressive lung disease of unknown cause in which the lungs become increasingly scarred and stiff, leading to a gradual decline in respiratory function. Once fibrosis develops, lung function is difficult to restore, and the prognosis is poor, with a five-year survival rate after diagnosis that is lower than that of many major cancers.

Because the rate of disease progression varies considerably among patients, accurate assessment and longitudinal monitoring are essential. A Korean research team has now proposed a new AI based approach that quantifies the progression of pulmonary fibrosis and predicts patient prognosis, providing an objective marker for disease assessment.

A joint research team led by Professors Jooae Choe of the Department of Radiology and Ho Cheol Kim of the Division of Pulmonology and Critical Care Medicine at Asan Medical Center, together with Professor Ho Yun Lee of the Department of Radiology at Samsung Medical Center, recently developed an objective benchmark for assessing disease progression and predicting prognosis in patients with idiopathic pulmonary fibrosis by quantifying fibrotic lung lesions on computed tomography (CT) images using AI based software.

Using the automated quantitative CT analysis technology, the researchers compared baseline CT scans with follow up scans obtained one year later in patients with idiopathic pulmonary fibrosis. They found that patients whose pulmonary fibrosis score increased by more than 4 percent over one year were at significantly higher risk of requiring lung transplantation or dying.

The findings are expected to help clinicians identify patients with idiopathic pulmonary fibrosis who are at high risk of disease progression early and determine personalized treatment strategies by providing a standardized, objective assessment marker instead of relying solely on visual interpretation of CT images.

The study was recently published in the American Journal of Respiratory and Critical Care Medicine (AJRCCM) (Impact Factor: 21.7), one of the world's leading journals in respiratory and critical care medicine.

Until now, disease progression in idiopathic pulmonary fibrosis has primarily been assessed using pulmonary function tests. The most commonly used measures are forced vital capacity (FVC), which evaluates the maximum amount of air a patient can exhale, and diffusing capacity of the lungs for carbon monoxide (DLCO), which measures the lungs' ability to transfer gas.

While pulmonary function tests provide an overall assessment of changes in lung function, they cannot accurately determine where fibrosis has progressed within the lungs or quantify its extent. As a result, gradual progression of fibrosis may not be clearly reflected in pulmonary function measurements alone, potentially leading clinicians to overlook meaningful disease progression.

Although computed tomography (CT) allows direct visualization of the lungs, image interpretation can vary among radiologists, and accurately measuring subtle changes and progression of fibrotic patterns over time has remained challenging. More importantly, there has been no standardized tool capable of consistently quantifying the subtle changes detected on CT images.

To overcome these limitations, the research team applied an AI based quantitative CT analysis technology that objectively measures fibrotic lung lesions using AI powered software. The technology automatically quantifies reticular opacities and honeycombing on chest CT images, enabling repeated measurements based on standardized criteria and providing an objective assessment of changes in the extent of pulmonary fibrosis over time.

The researchers analyzed clinical data from 524 patients diagnosed with idiopathic pulmonary fibrosis at Asan Medical Center between January 2011 and April 2020 who underwent both chest CT and pulmonary function tests at baseline and again one year later.

To externally validate the findings, the team also evaluated an independent cohort of 224 patients diagnosed with idiopathic pulmonary fibrosis at Samsung Medical Center between July 2008 and December 2022 who met the same inclusion criteria.

Using the automated quantitative CT analysis technology, the research team calculated pulmonary fibrosis scores at baseline and one year after follow up and assessed changes over the one-year period. The researchers then compared these changes with one-year changes in forced vital capacity (FVC) and diffusing capacity of the lungs for carbon monoxide (DLCO) to determine the threshold at which an increase in the pulmonary fibrosis score could accurately reflect actual disease progression.

In addition, the team evaluated whether the fibrosis score could predict patient outcomes by using transplant free survival, defined as the length of time a patient remained alive without undergoing lung transplantation, as the primary outcome measure.

The analysis showed that an increase in pulmonary fibrosis score was associated with disease progression assessed by conventional pulmonary function tests. A 5 percent decline in forced vital capacity (FVC) was associated with a 2.72 percent increase in pulmonary fibrosis score, while a 10 percent decline in diffusing capacity of the lungs for carbon monoxide (DLCO) was associated with a 4.52 percent increase in the score.

The study also found that patients whose pulmonary fibrosis score increased by more than 4.05 percent over one year had a higher risk of requiring lung transplantation or death. This finding was consistently observed in the external validation cohort. In particular, patients in the external validation cohort who exceeded this threshold had an approximately 2.8 fold higher risk of lung transplantation or death compared with those who did not. The three-year prognostic analysis further confirmed that the increase in pulmonary fibrosis score was a significant predictor of patient outcomes.

When baseline pulmonary fibrosis scores and changes in fibrosis scores at one year follow up were added to an existing prognostic model based on sex, age, and pulmonary function test results, the model's predictive performance improved. These findings suggest that quantitative markers obtained through CT based analysis can serve as complementary tools for assessing risk and predicting prognosis in patients with idiopathic pulmonary fibrosis.

Professor Jooae Choe of the Department of Radiology at Asan Medical Center said, “Previously, visual assessment of CT images could vary depending on the radiologist, and there were limitations in objectively measuring and reporting whether pulmonary fibrosis had progressed and to what extent over time. This study established a clinically meaningful threshold for changes in fibrosis scores, demonstrating that quantitative CT analysis can be used as an adjunctive marker for regular follow up and treatment planning in patients with idiopathic pulmonary fibrosis.”

Professor Ho Cheol Kim of the Division of Pulmonology and Critical Care Medicine at Asan Medical Center said, “Once lung damage occurs due to idiopathic pulmonary fibrosis, it is difficult to reverse, making early diagnosis and treatment essential. This study is significant because it established an objective standard for assessing disease progression in idiopathic pulmonary fibrosis and confirmed the potential of this approach as an imaging biomarker for evaluating treatment efficacy in future clinical trials.”