Related Experiment Video
Updated: Aug 26, 2025
03:39
Unilateral Lung Volume Analysis Using Micro-CT for Enhanced Assessment of Pulmonary Fibrosis in Preclinical Models
Published on: June 20, 2025
263
Deep-learning algorithm to detect fibrosing interstitial lung disease on chest radiographs
Hirotaka Nishikiori1, Koji Kuronuma1, Kenichi Hirota2
1Department of Respiratory Medicine and Allergology, Sapporo Medical University School of Medicine, Sapporo, Japan.
The European Respiratory Journal
|October 6, 2022
Summary
A new deep-learning algorithm can detect chronic fibrosing interstitial lung diseases (CF-ILDs) from chest radiographs. This AI tool demonstrates detection capabilities comparable to expert physicians, aiding early diagnosis.
Area of Science:
- Radiology
- Artificial Intelligence
- Pulmonology
Background:
- Antifibrotic therapies are crucial for managing chronic fibrosing interstitial lung diseases (CF-ILDs), including idiopathic pulmonary fibrosis.
- Early intervention with these therapies is recommended to slow disease progression and prevent acute exacerbations.
- Accurate identification of early-stage CF-ILDs from chest radiographs remains a clinical challenge.
Purpose of the Study:
- To develop and validate a deep-learning algorithm for detecting CF-ILDs using chest radiograph images.
- To assess the algorithm's diagnostic performance against that of medical professionals.
Main Methods:
- A deep-learning model was trained on 921 chest radiographs from patients with and without CF-ILD.
- The algorithm was tested on a separate dataset of 238 chest radiographs.
- Performance was evaluated by comparing the algorithm's accuracy (area under the receiver operating characteristic curve, sensitivity, specificity) with that of pulmonologists and radiologists.
Main Results:
- The deep-learning algorithm achieved an area under the receiver operating characteristic curve of 0.979.
- With a score cut-off of 0.267, the algorithm demonstrated a sensitivity of 0.896 and a specificity of 1.000.
- The algorithm's performance was found to be noninferior to that of expert physicians.
Conclusions:
- A deep-learning algorithm for detecting CF-ILDs from chest radiographs has been successfully developed.
- The algorithm exhibits diagnostic performance comparable to that of experienced clinicians, offering a potential tool for early disease identification.
