Investigation of the effectiveness of eight different metal artifact reduction algorithms in reducing extracorporeal metal artifacts: a phantom study using the Gumbel method

Abstract

Introduction: Metal Artifact Reduction (MAR) processing has been clinically applied to computed tomography (CT) images using various methods. Iterative MAR (iMAR) is an algorithm for reducing metal artifacts from implants and is tailored to the type, shape, and imaging site of a metal object. Various implants have been targeted using iMAR; however, there are some implants and metals that do not have a dedicated iMAR. The potential of iMAR for managing such artifacts has not yet been explored. Utilizing iMAR in unavoidable extracorporeal metal artifact cases could improve diagnosis. We aimed to assess whether the iMAR reduces extracorporeal metal artifacts and enhances image quality. Material and methods: CT was performed on a whole-body phantom with electrocardiogram (ECG) electrodes attached. Images were obtained without the iMAR and with eight different iMAR processings. The CT value profiles were perpendicular to the direction of artifact generation, and the maximum adjacent CT value difference was extracted from each CT value profile as the largest variation. The cumulative probabilities for the largest variations were obtained, and the location and scale parameters were calculated from the cumulative probability plots. Kruskal–Wallis tests and multiple comparisons were performed on nine different images. Results: Regarding the 100 cumulative probability plots of the largest variations obtained from each CT value profile, the coefficients of determination (R2) for all cumulative probability plots were as high as > 0.84, indicating that the features of the extracorporeal metal artifact generated from the ECG electrodes evaluated in this study asymptotically approached a Gumbel distribution. The location parameters showed no significant differences among the nine processed images (p > 0.11), whereas the scale parameters showed significant differences for neuro coil, shoulder implant, extremity implant, and thoracic coil iMAR-processed images compared with controls (p < 0.05). Conclusion: iMAR may improve diagnosis by reducing extracorporeal metal artifacts and enhancing image quality.