• Tsutomu Gomi, Miwako Nozaki, Tohoru Takeda, Tokuo Umeda, Kazutoshi Takahashi, and Masahiro Nakajima.
• School of Allied Health Sciences, Kitasato University, Kitasato 1-15-1, Minami-ku, Sagamihara, Kanagawa 252-0373, Japan. Electronic address: gomi@kitasato-u.ac.jp.
• Acad Radiol. 2013 Nov 1; 20 (11): 1357-63.

Rationale And ObjectivesTo compare initial evaluations of chest dual-energy subtraction digital tomosynthesis (DES-DT) and dual-energy subtraction radiography (DES-R) for detection of pulmonary nodules.Materials And MethodsDES-DT and DES-R systems with pulsed x-rays and rapid kV switching were used to evaluate pulmonary nodules (>4-6 mm, 2 nodules; >6-8 mm, 2 nodules; >8 mm, 32 nodules). Multidetector computed tomography was used as a reference. A filtered back-projection algorithm was used to reconstruct low-voltage (60 kVp), high-voltage (120 kVp), and soft-tissue or bone-subtracted tomograms of the desired layer thicknesses from the image data acquired during a single tomographic scan. DES-R images were processed from the low- and high-voltage images. To detect the pulmonary nodules, we used both systems to examine 36 patients with and 36 patients without pulmonary nodules. Two radiologists and three doctors of pulmonary medicine (average experience, 18 years) performed receiver operating characteristic (ROC) curve analysis to evaluate the results.ResultsThe ROC analysis results suggested that the detection ability was significantly better for DES-DT than for DES-R (P < .0001; 95% confidence interval: DES-DT, 0.94 [0.83-0.99]; DES-R, 0.76 [0.68-0.85]; sensitivity: DES-DT, 87.7 ± 2.9%; DES-R, 53.8 ± 3.5%; specificity: DES-DT, 78.3 ± 5.6%; DES-R, 78.4 ± 3.4%; accuracy: DES-DT, 83.1 ± 3.8%, DES-R, 66.1 ± 2.0%). When the nodules were no longer superimposed over the normal structures, their characteristics and distribution could be observed much more clearly.ConclusionCompared with DES-R, DES-DT provided greater sensitivity for detection of pulmonary nodules, particularly for the larger ones.Copyright © 2013 AUR. Published by Elsevier Inc. All rights reserved.

22 keywords...

hide…