Variability in densitometric assessment of pulmonary emphysema with computed tomography

The objectives of this study were to investigate whether computed tomography (CT) densitometry can be applied consistently in different centers; and to evaluate the reproducibility of densitometric quantification of emphysema by assessment of different sources of variation, ie, intersite, interscan and inter- and intraobserver variability, in comparison with intersubject variability.

In 5 different hospitals, 119 patients with emphysema were scanned using standardized protocols. In each site, an observer performed a quantitative densitometric analysis (including blood recalibration) on the corresponding patient group (n=23-25) and one observer analyzed the entire group of 119 patients. After several months, the latter observer analyzed all data for a second time. Subsequently, different sources of variation were assessed by variance component analysis with and without volume correction of the data.

Inter- and intraobserver variability marginally contributes to the total variability (<0.001%). The interscan variability was 0.02% of the total variation after application of volume correction. The intersite variability was 48% as a result of one deviating CT scanner. Air recalibration normalized deviating air densities in CT scanners. Within sites, the intersubject variability ranged between 93% and 99% based on the analysis of 2 subsequent CT scans of the patients.

Almost all variability in the density measurement of emphysema originates from differences between scanners and from differences in severity of emphysema between patients. Lung densitometry with multislice CT scanners is a highly reproducible measurement, especially if corrected for lung volume, because this reduces interscan variability.