Expiration CT Gas Trapping Measures with Texture-Based Radiomics Improves Association with Lung Function and Lung...

Area of Science:

• Pulmonary imaging and diagnostics
• Radiomics and computational pathology
• Chronic Obstructive Pulmonary Disease (COPD) research

Background:

• Current computed tomography (CT) methods for quantifying gas-trapping in the lungs lack spatial voxel analysis.
• There is a need for improved quantitative methods to assess lung function and disease progression in COPD.

Purpose of the Study:

• To evaluate if adding expiration CT texture-based radiomics features enhances existing gas-trapping measurements.
• To determine the impact on models for lung function, lung function decline, COPD classification, and visual gas-trapping.

Main Methods:

• Analysis of CT chest imaging from the CanCOLD cohort (n=1111) including full-inspiration/expiration scans.
• Quantitative CT measurements (LAA≤-856HU, E/I MLA, RVC_856-950) and texture-based radiomics (RadScore) were performed.
• Multivariable regression models assessed associations with lung function parameters, lung function decline, COPD classification, and visual gas-trapping.

Main Results:

• Both conventional CT gas-trapping measurements and CT RadScore were independently significant predictors (p<0.05) for all assessed outcomes.
• Combining CT gas-trapping and CT RadScore in the same model led to significant improvements in all model performance metrics (p<0.05).

Conclusions:

• Expiration CT texture-based radiomics provides crucial information on the spatial distribution of gas-trapping, beyond just its extent.
• Integrating radiomics with conventional CT measures significantly enhances the predictive power for lung function, its decline, and COPD classification.