Purpose: To compare sub-solid nodules detectability (SSND) between ultra-low-dose chest digital tomosynthesis (ULD-CDT) with/without iterative reconstruction (IR) and chest radiography (CR) by using low-dose computed tomography (LDCT) as the standard of reference (SOR).
Materials and methods: Institutional Review Board approved this study and written informed consent was obtained. In a single visit, 79 subjects underwent ULD-CDT at 120 kV and 10 mA, CR and LDCT (effective dose: 0.171, 0.117 and 3.52 mSv, respectively). Sixty-three coronal images were reconstructed using CDT with/without IR. SOR as to SSN presence was determined based on LDCT images. Seven radiologists recorded SSN presence and locations by continuously-distributed rating. Receiver-operating characteristic (ROC) analysis was used to compare SSND of ULD-CDT with/without IR and CR, in total and subgroups classified by nodular longest diameter (LD) (> or < 9 mm) and mean CT attenuation value (CTAV) (> or < -600 Hounsfield of Unit (HU)). Detection sensitivity (DS) was compared among 4 groups classified by combination of the identical thresholds: nodular LD (9 mm) and mean CTAV (-600 HU) in each of ULD-CDT with/without IR and CR with Friedman and Wilcoxon signed rank test.
Results: SSND for total 105 SSNs as well as larger SSNs with nodular LD of 9 mm or more at ULD-CDT with IR was higher than either that at ULD-CDT without IR or CR, as the areas under the ROC curve were 0.66 ± 0.02, 0.59 ± 0.01 and 0.52 ± 0.01, respectively (p < 0.05). DS at ULD-CDT with IR was 69.5 ± 10.8% in groups with larger (LD > 9 mm) and more-attenuated (>-600 HU) SSNs, and higher than in the other 3 groups (p < 0.05).
Conclusion: ULD-CDT with IR demonstrated better SSND than that without IR or CR, with increased DS for larger and more-attenuated SSNs compared with the remaining ones.
Keywords: Chest X-ray; Chest digital tomosynthesis; Iterative reconstruction; Low dose computed tomography; Sub-solid nodule.
Copyright © 2018 Elsevier B.V. All rights reserved.