Quantification of Airway Structures by Persistent Homology.

Area of Science:

• Medical imaging analysis
• Pulmonary medicine
• Computational geometry

Background:

• Chronic obstructive pulmonary disease (COPD) assessment relies on metrics like lung parenchyma density and airway wall area.
• Existing methods may not fully capture the complex geometric changes in the airway tree.

Purpose of the Study:

• To introduce novel morphological metrics for quantifying tubular structures in CT images.
• To apply these metrics for identifying airway irregularities in patients with COPD.
• To evaluate the complementary information provided by these metrics compared to conventional COPD assessments.

Main Methods:

• Automatic extraction of the three-dimensional airway shape from lung CT volumes, represented as a rooted tree.
• Computation of two novel metrics, treeH0 and radialH0, using persistent homology.
• treeH0 quantifies airway branch length distribution for structural complexity.
• radialH0 assesses irregularities in the airway's luminal radius.

Main Results:

• The proposed metrics quantify airway geometry and identify irregularities in patients with COPD.
• The novel metrics provide information complementary to conventional COPD assessment tools.
• The developed metrics demonstrate association with clinical outcomes in COPD patients.

Conclusions:

• Novel morphological metrics derived from persistent homology offer a new approach to airway analysis in CT imaging.
• These metrics enhance the understanding of airway geometry in COPD, complementing existing assessment methods.
• The findings suggest potential for improved clinical outcome prediction in COPD through advanced geometric analysis.