Abstract
A new method for quantitative evaluation for high resolution computed tomography (HRCT) of the lungs was developed by assessment of the distribution of radiological densities within the lung slices. To enable effective reduction of data and improve the sensitivity of detection of abnormalities, the density distributions were analysed by curve fitting through the gamma variate model. The output of two variables proved most representative: the most frequent density (Hoansfield units; HU) and width of distribution (HU). The method was applied to seven patients with early asbestosis (positive histological finding and International Labour Office (ILO) profusion score up to 0/1), 15 patients with advanced stage of asbestosis (positive histological finding and ILO score above 1/2), and 13 normal controls. All patients with early asbestosis had isolated reduction of diffusing lung capacity to carbon monoxide (DLCO), whereas all patients with advanced asbestosis had reduced DLCO and restrictive disease; two of them also had an obstruction pattern. The most frequent densities were significantly greater in the advanced asbestosis group (-567 HU) when compared with both the early asbestosis group (-719 HU; p = 2 x 10(-6)), and controls (-799 HU; p = 0), and they also discriminated significantly between the early asbestosis group and controls (p = 0.0002). Significantly stronger linear correlations were established between DLCO and the most frequent densities (r = 0.86) than between DLCO and HRCT score (r = 0.57) or ILO score (r = 0.34). It is concluded that fitting the curve of the density distribution enables a more objective assessment of HRCT pulmonary scans, especially in the early stage of asbestosis.