Comparative evaluation of emphysema quantification: Standardized %LAV-950 versus DL-based emphysema quantification...

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence in Medicine
• Pulmonary Medicine

Background:

• Emphysema quantification in chronic obstructive pulmonary disease (COPD) is crucial for diagnosis and management.
• Traditional methods like %LAV-950 have limitations in accurately reflecting disease severity.
• Advanced imaging analysis techniques are needed to improve emphysema assessment.

Purpose of the Study:

• To compare the correlation between a deep learning (DL) algorithm and the %LAV-950 threshold method for emphysema quantification.
• To evaluate the association of these quantification methods with pulmonary function test (PFT) parameters (FEV₁ and KCO).
• To assess the performance of DL and %LAV-950 using different CT reconstruction kernels (lung and soft tissue).

Main Methods:

• Retrospective analysis of chest CT scans and PFTs from 101 COPD patients.
• Emphysema quantification using the %LAV-950 method and a DL-based algorithm.
• Analysis of quantitative CT metrics derived from lung and soft tissue reconstruction kernels.
• Pearson correlation and permutation testing to compare CT metrics with FEV₁ and KCO.

Main Results:

• The DL method showed moderate correlations with FEV₁ (lung kernel) and KCO (soft tissue kernel).
• The %LAV-950 method demonstrated only a weak correlation with FEV₁ (soft tissue kernel) and no significant association with KCO.
• Permutation testing confirmed a significantly stronger correlation with KCO for the DL method compared to %LAV-950 using soft tissue kernels.
• DL approach showed more consistent correlations across different kernels and PFT parameters.

Conclusions:

• The deep learning-based algorithm provides a more accurate and consistent quantification of emphysema compared to the %LAV-950 method.
• DL-based emphysema quantification shows stronger associations with key pulmonary function parameters.
• These findings support the integration of AI-driven emphysema quantification into routine CT analysis for improved COPD assessment.