Keywords:
Lung, Radioprotection / Radiation dose, Digital radiography, Diagnostic procedure
Authors:
M. Mirzai, C. E. Meltzer, B. Gottfridsson, R. R. Norrlund, J. D. Vikgren, �. A. Johnsson, M. Båth, A. Svalkvist; Gothenburg/SE
DOI:
10.26044/ecr2019/C-3400
Methods and materials
Image collection: Twenty-four chest tomosynthesis examinations of patients without pulmonary lesions wereincluded in this retrospective study.
The Regional Ethical Review Board approved this study and all participants gave written informed consent.
The CTS system used was a modified digital chest radiographic system.
During the image collection the x-ray tube performed a vertical motion around the standard orthogonal posteroanterior (PA) projection.
Sixty low-dose projection images were collected in the angular interval of ±15°using a tube voltage of 120 kV with 3 mm Al + 0.1 mm Cu additional filtration and an acquisition time of approximately 10 seconds.
The tube load (mAs) used for each projection was determined from a PA scout view with automatic exposure control (AEC) and dose ratio 1:10 (the resulting mAs from AEC was multiplied by 10 and distributed evenly over the 60 projections).
These default exposure settings results in an effective dose of 0.13 mSv (scout view included) for an averaged sized patient [6].
Dose simulation: In order to assess the effect of dose reduction on image quality without exposing patients to additional dose due to repeated examinations a previously described method to simulate dose reduction was used [12].
The method is based on the principle of creating a noise image that,
when added to the original image,
simulates that the image has been acquired at a lower dose level.
Each examination was simulated to represent a dose level of 50% of the original dose.
Image analysis: The image data set was assessed using visual grading,
where five radiologists individually rated their confidence of fulfillment of predefined image quality criteria for both the original dose an the dose reduced images (reading order randomized).
In this study 10 quality criteria were used ( Table 1 ).
Criteria 1-9 included structures of interest in the central (criteria 1-6) and peripheral (criteria 7-9) regions of the chest and criterion 10 was defined to grade the presence of artifacts.
The peripheral regions included the lung parenchyma located both superiorly and inferiorly to the highest point of the diaphragmatic dome.
Small sized vessels located superiorly were assessed anteriorly,
posteriorly and laterally in the lungparenchyma (criterion 7),
while vessels located inferiorly were assessed only posteriorly and laterally in the left and right lung,
separately (criteria 8 and 9),
Fig. 2 and Fig. 3.
Statistical analysis: The rating data were analyzed using visual grading characteristics (VGC) [13] and the differences in the reproduction of each structure between the two dose levels was given by the area under the VGC curve (AUCVGC).
An AUCVGC=0.5 represents no difference between the two dose levels.
AUCVGC>0.5 indicate superior reproduction in the dose reduced tomosynthesis images and <0.5 inferior reproduction compared to the original dose images. The statistical analysis of the VGC data was performed using the software VGC Analyzer [14] where a random-reader analysis was used.
The software provides the asymmetric 95% confidence interval of the AUCVGC using a bootstrapping (resampling) technique.
Differences in the reproduction of each structure between the two dose levels was interpreted as statistically significant for confidence intervals not including the value 0.5.