Improvement of metal artifact reduction in dental CBCT using a CdTe photon-counting detector: Simulation study

Abstract

In dental cone-beam computed tomography (CBCT), the image quality often degrades when patients have metallic objects such as metal implants and/or metal prostheses. Metal artifacts typically appear as streaks and shadows in the reconstructed CT images, limiting their clinical usefulness. Although various metal artifact reduction (MAR) methods are used in dental CBCT, no algorithm that can robustly remove metal artifacts has been universally accepted. Recent research has explored the potential benefits of using of a photon-counting detector (PCD) to advance enhanced MAR. PCD can classify x-ray photons into several different energy bins, allowing for more precise measurements of x-ray attenuation. Therefore, in this study, we propose a new MAR method based on the use of a PCD, named the high-energy trace normalized-MAR (HT-NMAR) algorithm, to ensure improved image quality. The proposed MAR algorithm consists of three main steps: 1) Segmentation of the metal trace using a high-energy binned sinogram, 2) Generation of a residual artifact-reduced prior, 3) Sinogram completion with the segmented metal trace and the prior, followed by CBCT reconstruction. We conducted a simulation on a numerical head phantom with metal inserts using a PCD simulation toolkit to demonstrate its efficacy. Our simulation study indicates that the proposed MAR method considerably reduces metal artifacts in CBCT and shows a better image performance than other existing MAR methods in reducing streak artifacts without any contrast anomaly.