Titipong KaewlekDiew KoolpiruckSaowapak ThongvigitmaneeManus MongkolsukSastrawut ThammakittiphanSiri On TritrakarnPipat ChiewvitKing Mongkuts University of Technology ThonburiThailand National Electronics and Computer Technology CenterRangsit UniversityMahidol University2018-11-232018-11-232015-12-17Journal of X-Ray Science and Technology. Vol.23, No.6 (2015), 649-666089539962-s2.0-84952942984https://repository.li.mahidol.ac.th/handle/20.500.14594/35901© 2015 - IOS Press and the authors. Metal artifacts often appear in the images of computed tomography (CT) imaging. In the case of lumbar spine CT images, artifacts disturb the images of critical organs. These artifacts can affect the diagnosis, treatment, and follow up care of the patient. One approach to metal artifact reduction is the sinogram completion method. A mixed-variable thresholding (MixVT) technique to identify the suitable metal sinogram is proposed. This technique consists of four steps: 1) identify the metal objects in the image by using k-mean clustering with the soft cluster assignment, 2) transform the image by separating it into two sinograms, one of which is the sinogram of the metal object, with the surrounding tissue shown in the second sinogram. The boundary of the metal sinogram is then found by the MixVT technique, 3) estimate the new value of the missing data in the metal sinogram by linear interpolation from the surrounding tissue sinogram, 4) reconstruct a modified sinogram by using filtered back-projection and complete the image by adding back the image of the metal object into the reconstructed image to form the complete image. The quantitative and clinical image quality evaluation of our proposed technique demonstrated a significant improvement in image clarity and detail, which enhances the effectiveness of diagnosis and treatment.Mahidol UniversityEngineeringMedicineArticleSCOPUS10.3233/XST-150518