Kernel Conversion Improves the Correlation between the Extent of Emphysema and Clinical Parameters in Chronic Obstructive Pulmonary Disease: A Multicenter Cohort Study

Abstract

Background: Computed tomography (CT) scans are utilized to assess emphysema, aprominent phenotype of chronic obstructive pulmonary disease (COPD). Variability inCT protocols and equipment across hospitals can impact accuracy. This study aims toimplement kernel conversion across different CT settings and evaluate changes in thecorrelation between the emphysema index pre- and post-kernel conversion, along withclinical measures in COPD patients. Methods: Data were extracted from the Korea COPD Subgroup Study database, whichincluded CT scan images from 484 COPD patients. These images underwent kernelconversion. Emphysema extent was quantified using the percentage of low-attenuationareas (%LAA-950) determined by a deep learning-based program. The correlation between%LAA-950 and clinical parameters, including lung function tests, the modifiedMedical Research Council (mMRC), 6-minute walking distance (6MWD), COPD assessmenttest (CAT), and the St. George’s Respiratory Questionnaire for COPD (SGRQ-c),was analyzed. Subsequently, these values were compared across various CT settings. Results: A total of 484 participants were included. Kernel conversion significantlyreduced the variance in %LAA-950 values (before vs. after: 12.6±11.0 vs. 8.8±11.9). Post-kernel conversion, %LAA-950 demonstrated moderate correlations with forcedexpiratory volume in 1 second (r=–0.41), residual volume/total lung capacity (r=0.42),mMRC (r=0.25), CAT score (r=0.12), SGRQ-c (r=0.21), and 6MWD (r=0.15), all of whichwere improved compared to the unconverted dataset (all p<0.01). Conclusion: CT images processed through kernel conversion enhance the correlationbetween the extent of emphysema and clinical parameters in COPD.