The performance of metal artifact reduction algorithms in cone beam computed tomography images considering the effects of materials, metal positions, and fields of view

Abstract

The aim of this study was to assess the performance of 2 metal artifact reduction (MAR) algorithms in cone beam computed tomography (CBCT) imaging, considering different materials, metal positions, and fields of view (FOVs). Nine phantoms containing cylinders of amalgam, copper-aluminum (Cu-Al) alloy, and titanium were scanned by using Picasso Trio and ProMax 3D CBCT units with small and medium FOVs. Scans were made with and without MAR algorithms. The standard deviation (SD) of voxel gray values was measured in the neighborhood of the cylinders. Differences in SD were statistically evaluated for effects of MAR and the other parameters, with a significance level of 5%. Significant differences between images with MAR and those without MAR for both devices (P ≤ .0001) were observed. Amalgam showed the largest artifact expression, followed by Cu-Al and titanium. After correction, differences remained only in Picasso Trio images (P = .002). Considering positions, no significant difference in the performance of the MAR algorithm was observed in either device. Considering FOVs, significant differences were observed for ProMax 3D (P = .005), with less artifact expression in the medium FOV after MAR correction. MAR algorithms were effective for artifact reduction despite variation in performance according to device, FOV, and material properties. The position of the metal cylinder within the FOV had no significant effect.