A Novel Iterative Prior Fusion-Based Metal Artifact Reduction for CT Imaging: An Experimental Phantom Study

The presence of high-density objects creates significant artifacts in CT images, degrading image quality. Metal artifacts obscure structures and tissues around implants with high density, reducing the diagnostic utility of CT scans. Traditional Metal Artifact Reduction (MAR) algorithms may be able to remove metal-induced streak artifacts, but they may also produce new artifacts (residual artifacts). Furthermore, in extreme situations of metal artifacts, these procedures are unable to successfully decrease the artifact. In this context, we present a unique MAR strategy that combines two widely used approaches: iterative metal artifact reduction and sinogram space correction utilizing the prior fusion-based (FP MAR) method. To lower the weight of the damaged image during the reconstruction phase, the suggested approach uses a fusion-based prior image and an iterative algorithm. The Rando phantom was used to assess the proposed algorithm's accuracy both subjectively and numerically. To evaluate the proposed technique, we compared it with the performance of five commonly used MAR algorithms using quantitative metrics, such as Normalized Root Mean Square Error (NRMSE), Mean Absolute Difference (MAD), and Peak Signal to Noise Ratio (PSNR). In contrast to other MAR techniques, the proposed technique preserved the details of the underlying structures, while other approaches resulted in residual/new image artifacts. According to the quantitative results, the proposed MAR technique demonstrated superior performance over the best performing conventional method with NRMSE=0.16, MAD=31.99, and PSNR=35.37, achieving improvements of 3.7%, 10.9%, and 12.2%, respectively.