Accepted Article This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/mp.12719 This article is protected by copyright. All rights reserved. Article Type: Research Article Dual-energy-based Metal Segmentation for Metal Artifact Reduction in Dental Computed Tomography Mohamed A. A. Hegazy, Mohamed Elsayed Eldib, Daniel Hernandez, Myung Hye Cho, Min Hyoung Cho, Soo Yeol Lee a) Department of Biomedical Engineering, Kyung Hee University 1732 Deogyeong-daero, Yongin-si, Gyeonggi-do 446-701, Korea a) Correspondence: sylee01@khu.ac.kr Purpose: In a dental CT scan, the presence of dental fillings or dental implants generates severe metal artifacts that often compromise readability of the CT images. Many metal artifact reduction (MAR) techniques have been introduced, but dental CT scans still suffer from severe metal artifacts particularly when multiple dental fillings or implants exist around the region of interest. The high attenuation coefficient of teeth often causes erroneous metal segmentation, compromising the MAR performance. We propose a metal segmentation method for a dental CT that is based on dual-energy imaging with a narrow energy gap. Methods: Unlike a conventional dual-energy CT, we acquire two projection data sets at two close tube voltages (80kV p and 90kV p ), and then, we compute the difference image between the two projection images with an optimized weighting factor so as to maximize the contrast of the metal regions. We reconstruct CT images from the weighted difference image to identify the metal region with global thresholding. We forward project the identified metal region to designate metal trace on the projection image. We substitute the pixel values on the metal trace with the ones computed by the region filling method. The region filling in the metal trace removes high-intensity data made by the metallic objects from the projection image. We reconstruct final CT images from the region-filled