Pixel-Wise Interstitial Lung Disease Interval Change Analysis: A Quantitative Evaluation Method for Chest Radiographs...

Area of Science:

• Radiology
• Artificial Intelligence
• Pulmonology

Background:

• Interstitial lung disease (ILD) involves progressive pathological changes necessitating accurate diagnosis.
• Early detection and progression assessment are crucial for effective ILD management.
• Conventional visual assessments of ILD can be subjective.

Purpose of the Study:

• To introduce a novel quantitative evaluation method for analyzing pixel-wise changes in ILD using chest radiographs.
• To develop a precise and objective tool for quantifying ILD progression.
• To enhance the monitoring and management of interstitial lung disease.

Main Methods:

• Utilized a weakly supervised learning framework.
• Incorporated the contrastive unpaired translation model.
• Developed a new ILD extent scoring algorithm for quantitative analysis of chest radiographs.

Main Results:

• The ILD extent score achieved 92.98% accuracy in classifying between ILD and normal cases.
• Disease progression was assessed with 85.29% accuracy using an ILD follow-up dataset.
• The quantitative method demonstrated reliability for ILD monitoring.

Conclusions:

• The proposed quantitative method provides a reliable tool for ILD monitoring.
• The ILD extent score offers more precise and objective quantification of disease changes compared to visual assessments.
• This novel approach has the potential to improve the management of interstitial lung disease.