Three-dimensional quantitative assessment of lung parenchyma in cystic fibrosis: preliminary results

Abstract

The aim of this study was to assess the feasibility of three-dimensional (3D) reconstructions and quantitative analysis of the volume of each component of the lung with cystic fibrosis (CF). Twenty-two patients with CF (mean age 17+/-8 years) were included in the study. The patients underwent an unenhanced single-slice spiral computed tomography (CT) chest scan with the following parameters: collimation 3 mm, table feed 6 mm x rot(-1), reconstruction interval 1 mm, soft tissue reconstruction kernel. Four image data sets were obtained: native axial slices, cine-mode display, virtual bronchographic volume-rendered images with algorithm for tissue transition display and virtual endoluminal views. The lungs were segmented manually from the hilum to the visceral pleura on the axial images, and the entire lung volume was calculated. A histogram was generated representing the fractional volume of tissues, the density of which was within a preset range. A curve was then obtained from the histogram. Native axial images and cine-mode display allowed complete evaluation of lung volumes. Virtual bronchography allowed a better assessment of the distribution of bronchiectasis. Virtual bronchoscopy was limited by the fact that it visualised only the surface, without differentiating mucus from the bronchial wall. Manual segmentation and generation of density-volume curves required 41+/-7 min for each lung. Three curve patterns were identified depending on disease severity. Volume-density analysis of lungs with CF is feasible. Its main advantage is that image analysis is not analogical, as the assessment is not performed using scoring systems or similar ordinal scales. This technique cannot differentiate acute from chronic findings, and the predictive value of the curve should be assessed.