A Liver Phantom Study: CT Radiation Dose Reduction and Different Image Reconstruction Algorithms Affect Diagnostic Quality.

Abstract

The aim of this study was to evaluate whether iterative reconstruction techniques for different dose levels and/or reduction of tube potential can increase liver lesion detectability. An anthropomorphic liver phantom was scanned at different dose levels (CTDIvol 15 mGy, 7.5 mGy, 5 mGy, and 2.6 mGy) and tube potential levels (120 kV, 100 kV, and 80 kV). Images were reconstructed with the following algorithms: filtered back projection (FBP), adaptive statistical iterative reconstruction (ASiR) 40%, and a model-based iterative reconstruction (Veo). The presence or absence of lesions was assessed independently on a 4-point scale by 4 readers. The areas under the receiver operating characteristic curve were calculated. Veo improved detectability of hyperdense liver lesions compared with both FBP and ASiR 40% at most dose levels (15 mGy, 7.5 mGy, and 5 mGy with P < 0.05). Veo also improved detectability at reduced tube potential compared with FBP (120 kV, 100 kV, and 80 kV at 5 mGy with P < 0.05) and ASiR 40% (120 kV and 100 kV at 5 mGy with P < 0.05). For ASiR 40%, the area under the receiver operating characteristic curve was significantly larger compared with FBP only at dose levels 7.5 mGy and 2.6 mGy at 120 kV. In general, the reduction of tube potential reduced the lesion detectability. This study shows that iterative reconstruction algorithms, in particular Veo, improve lesion detectability in a liver phantom. However, a too aggressive dose reduction may result in poorer image quality. Results considering different tube potentials diverged, thus careful consideration is necessary upon tube potential reduction.