An automated method to assess the distribution of low attenuation areas on chest CT scans in chronic pulmonary emphysema patients

Abstract

We developed an automated method to recognize the lung in a computed tomographic (CT) image. With computer-assisted analysis, we were able to describe the continuous low attenuation (less than -960 Hounsfield units) areas (CLA) on chest CT scans. The size (CLAs) and number (CLAn) of the CLA and the percentage of total lung area occupied by low attenuation area (LAA%) were measured using CT scans obtained from 24 patients with chronic pulmonary emphysema (CPE) and 13 control patients. The automated algorithm recognized the lung areas successfully in all patients. The CLAs and LAA% were significantly higher, and CLAn was significantly lower in patients with CPE than in controls. There was a significant correlation between CT parameters and pulmonary function test results. The histograms of the size of CLA could be represented as a power function in each patient. This automated method should be useful in objectively defining the affected areas in the lungs of patients with CPE.