Chest computed tomography in severe bronchopulmonary dysplasia: Comparing quantitative scoring methods.

Area of Science:

• Pulmonary Medicine
• Radiology
• Neonatology

Background:

• Bronchopulmonary dysplasia (BPD) is a common complication in extremely preterm infants, often associated with structural lung abnormalities.
• Accurate quantification of lung damage in BPD is crucial for patient management and research.

Purpose of the Study:

• To evaluate three new Hounsfield unit (HU) based chest-CT scoring methods for quantifying lung damage in BPD.
• To assess the intra- and inter-observer variability of these methods.
• To compare the HU methods with the established Perth-Rotterdam-Annotated-Grid-Morphometric-Analysis (PRAGMA)-BPD score and clinical data.

Main Methods:

• Chest CT scans from infants with severe BPD were analyzed at a median of 7 months corrected age.
• Quantification of hyper- and hypo-attenuated regions was performed using PRAGMA-BPD and three HU-based methods (automated, semi-automated, manual).
• Reproducibility was assessed using intraclass correlation coefficients (ICC) and Bland-Altman plots; correlations with clinical data were evaluated using Spearman rank correlation.

Main Results:

• The manual HU method and PRAGMA-BPD demonstrated excellent intra- and inter-observer reproducibility (ICCs 0.80-0.97).
• The manual HU method showed high correlation with PRAGMA-BPD for both hyper- (ρ=0.92) and hypo-attenuation (ρ=0.79).
• Automated and semi-automated HU methods exhibited poorer correlation, especially for hypo-attenuation.

Conclusions:

• The manual Hounsfield unit (HU) method and PRAGMA-BPD score offer the best reproducibility for quantifying CT abnormalities in BPD.
• These findings support the use of the manual HU method as a reliable tool for assessing lung damage in BPD.