ABSTRACT: Iterative reconstruction techniques (IRTs) may improve image quality for low-dose imaging compared with filtered back projection (FBP) reconstruction.
We compared the results of an IRT for low-dose thoracic aortic computed tomography (CT) imaging with those from FBP reconstruction.
Data from 50 patients who underwent 256-slice CT for evaluation of the thoracic aorta were reconstructed with FBP and an IRT (iDose(4)) at 3 noise-reduction strengths (levels 2, 4, and 6). A blinded reader graded image quality (scale, 1-5; 5 = high diagnostic confidence) and the extent of shoulder artifact (scale, 1-5; 5 = no artifact) on all reconstructions. A second reader evaluated a subset of 20 cases to determine interreader and intrareader reproducibility. The mean and SD of attenuation were measured at 5 locations along the thoracic aorta and both subclavian arteries.
Image noise (SD of attenuation) improved with IRT relative to FBP (aorta: FBP, 31.4 ± 8.6 HU; IRT level 2, 25.1 ± 6.9 HU; level 4, 21.7 ± 6.2 HU; level 6, 17.2 ± 5.4 HU; P < 0.0001; subclavian arteries: FBP, 92.7 ± 34.6 HU; IRT level 2, 50.1 ± 17.1 HU; level 4, 48.9 ± 18.6 HU; level 6, 45.2 ± 19.2 HU; P < 0.0001), whereas mean attenuation was unchanged. Increasing image quality was observed in the aorta and through the shoulders as the contribution from IRT to the final images increased (P < 0.0001). Significant differences were noted between readers in image quality assessment of the aorta but not through the shoulders.
IRT is associated with reduced noise and shoulder artifact and allows for low-dose aortic CT imaging.
Journal of cardiovascular computed tomography 04/2012; 6(3):205-13.