To iMAR or not to iMAR: Quantitative impact of MAR algorithms on image quality in a phantom study

Introduction

The purpose of this study was to describe the impact of the latest metal artefact reduction algorithm (MAR) on a photon-counting CT (PCCT) in a phantom study simulating metal hip prostheses.

Methods

Two phantoms were scanned on a clinical PCCT with different configurations of laterally placed metal inserts. Acquisitions were performed at 140 kVp and reconstructed as virtual monoenergetic images (VMI) at 60–190 keV in 10-keV steps using a standard abdominal kernel, with and without iterative MAR (iMAR). Artefacts were quantified with the Root Mean Square of the image Difference (RMSD). Image quality was assessed by means of noise properties (noise, NPS), frequency response (TTF) and detectability index (d') and compared with the reference (70 keV VMI, no implants).

Results

increasing VMI energy did not always result in a reduction of artefacts measured by RMSD; the minimum values were always obtain with iMAR, which provided a further 32 % and 60 % best RMSD reduction with two steel and two titanium rods, respectively. Combining low-energy VMI and iMAR resulted in the closest NPS texture to the metal-free reference. The detectability of the high-contrast task with two steel rods decreased by 27–66 % with VMI alone compared to the reference, and by only 8.5 % with iMAR at 70 keV. The detectability with iMAR was never smaller than with VMI alone (within one standard deviation).

Conclusions

iMAR showed better and more consistent noise and artefact reductions compared to VMI technique alone for all configurations. In 10 cases out of 12 the best detectability was observed with a combination of iMAR and VMI at 60 or 70 keV.

Implications for practice

iMAR should be activated in the presence of metal hip implants to effectively reduce metal artefacts, particularly in combination with low VMI energies.