Detection of lung cancer using single-exposure dual-energy subtraction chest radiography

Abstract

The aim of this study was to evaluate the detectability of lung cancer by chest radiography with a single-exposure dual-energy subtraction (ES) method. Five radiologists read two sets of chest radiographs from 77 patients (66.5 +/- 9.6 years old) with histologically proven lung cancer measuring <or=3.0 cm and those from 77 normal subjects (65.7 +/- 9.0 years old). The observer tests were performed in two sessions: standard computed radiography (CR) images only and a combination of CR and ES images. Receiver-operating characteristic analysis was used for statistical analysis. All tumors were classified into three groups according to the appearance on thin-section CT: (1) nonsolid: tumor shadow disappearance rate (TDR) was 100%; (2) partly solid: TDR was >or=50 but <100%; (3) solid: TDR was <50%. Overall, detectability with the ES method was significantly better than that without ES (mean Az value increased from 0.7673 to 0.8265, P < 0.05). In the subgroup analysis of the nonsolid group and the solid group detectability did not change using the ES method, whereas in the partly solid group detectability with the ES method was significantly better than that without ES (mean Az value increased from 0.7162 to 0.8209, P < 0.005). The ES method improves the detectability of lung cancer by chest radiography, especially of the partly solid group.