Impact of iterative reconstruction on the diagnosis of acute pulmonary embolism (PE) on reduced-dose chest CT angiograms.

Abstract

To evaluate the impact of iterative reconstruction on the detectability of clots. Fifty-three patients were enrolled in a study comparing reduced-dose and full-dose images, available from the same dual-source data set. From each acquisition, three series of images were generated: (1) full-dose images (from both tubes), reconstructed with filtered back projection (FBP) (group 1; standard of reference), (2) reduced-dose images (from tube A only; 60 % dose reduction) reconstructed with FBP (group 2) and iterative reconstruction (SAFIRE) (group 3). In group 1 (mean DLP: 264.6 mGy.cm), (1) PE was diagnosed in 8 patients (15 %) with 82 clots in the central (n = 5), segmental (n = 39) and subsegmental (n = 38) arteries and (2) mean level of noise was 30.56 ± 5.07. In group 2 (mean DLP: 105.8 mGy.cm), a significant increase in noise (44.56 ± 6.24; p < 0.0001) (1) hampered detection of PE in one patient and (2) altered detection of peripheral clots (12 false-negative and 2 false-positive results). In group 3, image noise was not significantly different from that in group 1 (p = 0.1525; effect size: 0.2683), with a similar detection of PE compared to group 1 (p = 1). Reconstruction of reduced-dose images (60 % dose reduction) with SAFIRE provided image quality and diagnostic value comparable to those of full-dose FBP images. • Iterative reconstruction does not alter the detection of endoluminal clots. • Iterative reconstruction allows dose reduction in the context of acute PE. • Iterative reconstruction allows radiologists to approach the prospects of submilliSievert CT.