The objectives of the study were to evaluate the image quality and diagnostic performance of sinogram-affirmed iterative reconstruction (SAFIRE) for detecting hepatic metastasis and to estimate the potential radiation dose reduction at abdominal computed tomography (CT).
Materials and Methods
Fifty-nine consecutive patients (mean age, 59.2 years; range, 42-81 years) who had hepatic metastasis and who underwent dual-source abdominal CT were enrolled in this study. The image noise of the liver was compared between full-dose filtered back projection images (FD-FBP) and simulated half-dose images, which used only single-source projection data, and reconstructed with different strengths of SAFIRE (HD-SAFIRE1-5) by 1-way analysis of variance with Bonferroni correction. The diagnostic performance for hepatic metastasis was compared by logistic regression with the weighted least squares method and noninferiority test between the FD-FBP images and the half-dose sinogram-affirmed iterative reconstruction (HD-SAFIRE) images.
As the SAFIRE strength increased, image noise decreased (P < 0.001). The images of HD-SAFIRE2 showed similar image noise for liver parenchyma (13.4 Hounsfield unit [HU]), signal-to-noise ratio for liver parenchyma (9.2), and lesion-to-liver contrast-to-noise ratio (4.5), compared with the FD-FBP images (12.6 HU, P = 0.668; 9.6, P > 0.999; and 4.5, P > 0.999, respectively). The diagnostic accuracy for hepatic metastasis with the HD-SAFIRE2 images (87.5%) was not different from that of the FD-FBP images (87.5%, P > 0.999). The diagnostic performance of the HD-SAFIRE2 images was also noninferior to that of the FD-FBP images (95% confidence interval lower limit difference [−9.16%] excluding the −10% clinical noninferiority limit). The mean (SD) volume CT dose index of FD-FBP was 11.26 (2.66) mGy, and the mean (SD) radiation dose equivalent associated with HD-SAFIRE was estimated to be 5.63 (1.33) mGy.
By applying SAFIRE2, the radiation dose could be reduced by up to 50% compared with the standard-of-care abdominal CT protocol without increasing image noise and without deteriorating diagnostic performance for the detection of hepatic metastasis.