Purpose To evaluate image quality and diagnostic confidence of a raw-data-based iterative reconstruction technique (SAFIRE) in reduced-dose CCTA images in comparison with standard-dose filtered back projection (FBP) images.

Materials and methods 107 consecutive patients (72 males; 35 females), referred for a CCTA were prospectively included using a dual-source CT system in a high pitch (n = 51) or a sequential mode (n = 56) according to heart rate (mean DLP = 204.6 mGy.cm). From each acquisition, three series of images were reconstructed: standard-dose images reconstructed with FBP and considered as the reference standard (Group 1); and two series of reduced-dose images obtained with prototype software simulating a 30% dose reduction, and reconstructed with FBP (Group 2) or SAFIRE (Group 3). Two readers blindly evaluated each series for (a) objective noise and CNR; (b) coronary border sharpness, lesion severity; and (c) diagnostic confidence level using a 5-point scale.

Results In Group 2, there was a significant increase in noise compared to Group 1 (36.8 HU ±6.73 vs 30.4 HU ±5.20; p < 0.0001) and a CNR impairment (15.6 ± 4.3 vs 18.7 ± 4.5; p < 0.0001). In Group 3, despite the 30% dose reduction, SAFIRE restored the objective image quality: mean noise = 31.1 HU ±5.4 (p = 0.8) and CNR = 18.5 ± 5.0 (p = 0.7). However the diagnostic confidence was altered when compared with Group 1 (p < 0.0001), mainly rated as moderate with a blurred aspect of the coronary borders (81/107 [75.7%], p < 0.0001) and a significant number of artefactual non-flow-limiting soft plaques described in vessels considered as normal in Group 1 (105/428 [24.5%], p < 0.0001).

Conclusion Raw-data-based iterative reconstruction allowed significant noise reduction but may be associated with blurring of the coronary luminal borders, which can decrease diagnostic confidence. When reporting reduced-dose CCTA with iterative reconstruction, false smooth plaque artefacts must be considered in diagnostic assessment and subsequent patient management.