A new metal artifact reduction algorithm based on a deteriorated CT image

BACKGROUND: Computed tomography (CT) is an established imaging technology primarily used as a non-invasive diagnostic tool that reconstructs axial images. However, significant problems with metal artifacts remain. A metal artifact is a strong radial noise in an image, which makes it difficult to diagnose patients and inspect products containing metal implants. Historically, studies related to metal artifact reduction used projection data, though the data is not typically saved after processing. OBJECTIVE: This study proposes a new metal artifact reduction algorithm that does not require projection data, for new applications and for accurate diagnostic techniques. METHODS: The algorithm utilizes reconstructed images and is based on iterative reconstruction. By reproducing an accurate forward projection on simulation and combining it with an iterative calculation, discrepancies causing metal artifacts are eliminated. RESULTS: Validation was completed with numerical phantom models. Our results indicate that the proposed algorithm effectively reduces metal artifacts, even if numerous complex-shaped metal pieces were embedded in the cross-section. CONCLUSIONS: We developed a novel reconstruction algorithm for metal artifact reduction in CT imaging that only requires reconstructed images and projection conditions. Any historical CT data containing metal artifacts could be improved with this method.