Computerized identification of airway wall in CT examinations using a 3D active surface evolution approach

Area of Science:

• Medical Imaging
• Pulmonary Medicine
• Computational Anatomy

Background:

• Airway diseases such as asthma and chronic bronchitis are globally prevalent.
• Airway morphological variations can impair lung function and gas exchange.
• Accurate airway wall measurement is crucial for diagnosing and managing airway diseases.

Purpose of the Study:

• To develop and evaluate a novel algorithm for automatic airway wall identification on CT images.
• To assess the accuracy and performance of the proposed algorithm in airway wall estimation.

Main Methods:

• A three-dimensional (3D) surface model is evolved within airway regions using predefined forces.
• The algorithm leverages geometric and density characteristics of airway walls in a distance gradient field.
• Performance was evaluated using a lung phantom and comparison with manual delineations by an experienced radiologist.

Main Results:

• The algorithm demonstrated significantly lower error (0.04–0.36 mm) compared to the full-width at half maximum (FWHM) method (0.16–0.84 mm) in phantom studies.
• Comparison with expert radiologist delineations showed a mean difference of 0.084 mm.
• The algorithm exhibited consistent performance across different CT reconstruction kernels ('lung', 'bone', 'standard').

Conclusions:

• The developed algorithm provides accurate and reliable airway wall measurements from CT images.
• This automated approach has the potential to enhance the diagnosis and monitoring of airway diseases.
• The algorithm achieves accurate results efficiently, taking approximately 4 minutes per clinical CT examination.