Chest X-ray for Clinical Evaluation, Idiopathic Pulmonary Fibrosis (IPF), Incidental Findings, Interstitial Lung Disease (ILD), Lung Disease, Interstitial
Conditions
Brief summary
The goal of this observational study is to learn how well an artificial intelligence (AI)-based chest X-ray analysis software can incidentally detect interstitial lung disease (ILD), which appears as interstitial opacity, on chest X-rays taken for other reasons, and whether these AI-flagged findings represent true interstitial opacity. The main question it aims to answer is: How often does an AI-flagged interstitial opacity correspond to true ILD? This retrospective study uses existing records: researchers review each participant's follow-up computed tomography(CT), CT report, and final diagnosis to confirm true ILD and reticular opacity.
Interventions
DEVICEVUNO Med®-Chest X-ray™
VUNO Med®-Chest X-ray™ is artificial intelligence (AI)-based software that supports the detection and diagnosis of abnormal findings on chest radiographs. It automatically identifies abnormal findings and provides information on their type and location to aid clinical decision-making.
Sponsors
Chung-Ang University Hospital
VUNO Inc.
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
19 Years to No maximum
Healthy volunteers
Yes
Inclusion criteria
* Adults aged 19 years or older * Visited the pulmonology and allergy clinic (outpatient or inpatient) at Chung-Ang University Hospital (Seoul or Gwangmyeong) and underwent chest radiography from January 2022 to December 2024 * A follow-up CT performed after the index chest radiograph * Reticular/interstitial opacity detected on the index radiograph by VUNO Med®-Chest X-ray™
Exclusion criteria
* Prior history of ILD or ILD-related disease before the index chest radiograph, or a CT report containing terms related to interstitial opacity * Non-frontal (non-posteroanterior/anteroposterior \[PA/AP\]) chest radiograph view position * Missing CT report or final clinical diagnosis
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Positive predictive value (PPV) of AI-detected interstitial opacity | From the index chest radiograph to the reference standard confirmation (the first follow-up CT after the index chest radiograph and/or final clinical diagnosis), up to 3.5 years | Positive predictive value (PPV) of the AI flag for interstitial opacity is the proportion of AI interstitial-opacity-positive index radiographs confirmed as true positives by the radiologist reference standard (consensus review of the paired follow-up CT, CT report, follow-up diagnoses, and the index radiograph). PPV = true positives / all AI interstitial-opacity-positive cases. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Comparison of AI finding scores between true-positive and false-positive cases | From the index chest radiograph to the reference standard confirmation (the first follow-up CT after the index chest radiograph and/or final clinical diagnosis), up to 3.5 years | The AI finding scores (e.g., interstitial opacity, consolidation, nodule, pleural effusion) were compared between interstitial-opacity true-positive and false-positive cases using the Mann-Whitney U test. Scores are summarized as the median (first-third quartile, Q1-Q3). |
Countries
South Korea
Outcome results
None listed