An automated approach to quantitative air trapping measurements in mild cystic fibrosis

Abstract

Purpose: To automatically derive the degree of air trapping in mild cystic fibrosis (CF) disease from high-resolution CT (HRCT) data, and to evaluate the discriminating power of the measurement. Materials and methods: The data consist of six pairs of anatomically matched tomographic slices, obtained during breath-holding in triggered HRCT acquisitions. The pairs consist of an inspiratory slice, at ≥ 95% of slow vital capacity, and an expiratory slice at near residual volume (nRV). The subjects are 25 patients with mild CF and 10 age-matched, normal control subjects. Subjects: Lung segmentation is automatic. The limits defining air trapping in the expiratory slices are determined by the distribution of densities in the expanded lung. They are modulated by density changes between expiration and inspiration. Air trapping defects consist of contiguous low-density voxels. The difference between patients and control subjects was evaluated in comparison to pulmonary function test (PFT) results and lung density distribution descriptors (global density descriptors). Results: In mild CF, air trapping does not correlate with global PFT results, except for the ratio of residual volume (RV) to total lung capacity (TLC); however, the size of air trapping defects was the best discriminator between patients and control subjects (p < 0.005). Of PFT results, only RV/TLC reached significance at p < 0.05. The global density descriptors reached near significance in the nRV images only. Conclusion: Air trapping defined as defect size and measured in an objective automated manner is a powerful discriminator for mild CF.