Iterative reconstruction: comparison of techniques for reduced-dose liver computed tomography following transarterial chemoembolization for hepatocellular carcinoma

Abstract

Background: Adaptive statistical iterative reconstruction (ASIR) and model-based iterative reconstruction (MBIR) algorithms have the potential to reduce dose while maintaining image quality. Purpose: To compare computed tomography (CT) image quality and diagnostic performance among three reconstruction techniques - ASIR, MBIR, and filtered back projection (FBP) - after transcatheter arterial chemoembolization (TACE) of hepatocellular carcinomas (HCC). Material and Methods: Of 60 patients that underwent initial TACE for HCCs, half underwent dynamic liver CT with conventional scanning protocol, and the other half with dose reduction to approximately 60% of conventional exposure. All images were reconstructed using three algorithms: FBP, ASIR, and MBIR. For objective analysis, image noise and signal-to-noise ratio (SNR) were compared. For subjective analysis, three radiologists independently assessed image quality. Ability to detect viable HCCs was also evaluated. Results: MBIR and ASIR produced images with less noise and higher SNR compared with FBP regardless of radiation dosage (P<0.017). However, in terms of subjective parameters, such as image blotchiness, artifacts, and overall quality, MBIR was inferior to FBP and ASIR (P<0.001). Regarding diagnostic performance, there were no significant differences among reviewers in the detection of viable HCCs depending on the reconstruction algorithm, regardless of the dose reduction protocol (P>0.017). Conclusion: Although subjective evaluations suggest that MBIR images are of lower quality compared with FBP and ASIR regardless of radiation dosage, there were no significant differences among reconstruction algorithms in diagnosis of viable HCC after TACE.