Using a spatial point process framework to characterize lung computed tomography scans.

Area of Science:

• Medical imaging analysis
• Quantitative pathology
• Computational biology

Background:

• Pulmonary emphysema diagnosis relies on visual assessment of lung CT scans, which suffers from poor reliability and time constraints.
• There is a need for objective, quantitative methods to characterize lung diseases like emphysema.

Purpose of the Study:

• To develop spatial point process models for generating quantitative lung biomarkers of emphysema from CT scans.
• To assess the clustering characteristics of emphysematous tissue using these novel biomarkers.

Main Methods:

• Formalized a framework for applying spatial point processes to lung CT imaging data.
• Developed a Shot Noise Cox Process model to quantify emphysematous tissue clustering.
• Employed Bayesian estimation with spatial Birth-Death MCMC (BD-MCMC) for model parameter estimation.

Main Results:

• Validated the accuracy of the BD-MCMC estimation algorithm through simulations.
• Applied the models to the COPDGene cohort's lung CT scans.
• Demonstrated variability in emphysematous tissue clustering across different visual emphysema subtypes.

Conclusions:

• Spatial point process models provide a robust framework for quantitative emphysema biomarker development.
• These quantitative biomarkers reveal heterogeneity in disease presentation not fully captured by visual assessment.
• The developed methods enhance the characterization of pulmonary emphysema for improved diagnosis and research.