Objective: To evaluate the ability of quantitative parameters derived from low-dose dual-respiratory-phase CT to differentiate the severity of chronic obstructive pulmonary disease (COPD).
Materials and methods: We enrolled 100 COPD patients between October 2023 and October 2024. Inspiratory CT scans were performed using a standard-dose protocol, and expiratory scans were performed using a low-dose protocol. Images were reconstructed with 60% adaptive statistical iterative reconstruction (ASIR). Dual-respiratory-phase image registration was performed on a dedicated "Digital Lung" platform to quantify the following parameters: end-expiratory lung volume (LVEX), end-inspiratory lung volume (LVIN), mean lung density at end-inspiration (MLDIN), mean lung density at end-expiration (MLDEX), and the mean density ratio (MLDEX/IN). The percentages of low-attenuation areas (LAA-910% and LAA-950%), emphysema (Emph%), and functional small airways disease (fSAD%) were also calculated. Correlations between CT parameters and pulmonary function were assessed. Significant variables were entered into univariate and multivariate logistic regression models, and a combined predictive model was constructed. Model performance was evaluated using receiver operating characteristic (ROC) curve analysis with internal validation by bootstrap resampling.
Results: Quantitative parameters, including LVEX, LVIN, LAA-950%, LAA-910%, Emph%, and fSAD%, were significantly correlated with pulmonary function indices (FEV1%pred and FEV1/FVC; all P < 0.01). Among these, LAA-910% and Emph% showed the strongest correlations with FEV1%pred (both r = -0.80). In univariate analysis, LAA-910% demonstrated the highest predictive performance (area under the curve [AUC] = 0.887), followed by Emph% (AUC = 0.884). After adjustment for age and smoking status, multivariate logistic regression identified both Emph% (odds ratio [OR] = 1.298, 95% CI: 1.167-1.488, P < 0.001) and fSAD% (OR = 1.111, 95% CI: 1.040-1.205, P = 0.004) as independent predictors of severe COPD (GOLD 3-4 vs. GOLD 1-2). The combined model incorporating fSAD% and Emph% achieved an AUC of 0.926 (95% CI: 0.883-0.974), with a sensitivity of 80.49% and a specificity of 81.36%. Bootstrap internal validation yielded an optimism-corrected AUC of 0.908, confirming good model generalizability.
Conclusion: Low-dose expiratory CT combined with dual-respiratory-phase quantitative analysis provides reliable imaging biomarkers that correlate strongly with pulmonary function. The predictive model incorporating both fSAD% and Emph% demonstrates excellent diagnostic performance for assessing COPD severity and may serve as a noninvasive adjunct tool for disease stratification.


扫码关注我们
求助内容:
应助结果提醒方式:

