Quantitative phantom analysis of lipiodol artefact mitigation in post-TACE liver imaging using DECT and MAR algorithms

Dynamic computed tomography (CT) is essential for liver tumour detection and transcatheter arterial chemoembolisation (TACE). However, Lipiodol retention post-TACE induces severe metal-like artefacts, compromising diagnostic accuracy. This study quantitatively evaluated the efficacy of combining dual-energy CT (DECT) with a projection-based metal artefact reduction (MAR) algorithm. A cylindrical phantom was designed to simulate small (2 cm) and large (6 cm) tumours, with measurements taken at proximal (1 cm) and distal (5 cm) distances. Virtual monochromatic images (40–140 keV) were reconstructed with and without MAR. Without MAR, the most severe CT number fluctuations and noise occurred at 40 keV and 1 cm proximity. MAR demonstrated the potential to stabilize CT numbers, reducing variations by 75%–90% at low energies and 20%–40% at high energies ( ≥ 90 keV). Furthermore, MAR reduced noise levels (SD) by 50%–90% at 1 cm in the low-energy range (40 keV), where attenuation-induced noise was most severe. Imaging stability was highest at 5 cm, particularly in the 9 o'clock direction. While MAR significantly improves image stability at 40–70 keV, massive embolisations (6 cm) still pose a risk of secondary artefacts. We conclude that a moderate-energy setting ( ∼ 70 keV) at a 5 cm distance may provide a favorable balance between artefact mitigation and contrast preservation under the tested phantom conditions, serving as a foundational reference for future post-TACE clinical protocols.