Evaluation of a new method for metal artifact reducing in computed tomographic images

Background: The common streak artifacts in computed tomographic (CT) images result from metal implants in patients. Metal artifact suppresses and obstructs diagnosis or misdiagnoses as it occurred in ten percent of the patients’ tomographic images. Objectives: To develop the method for metal artifact reduction in CT images using MATLAB software and implement it in phantoms with the metal artifact as well as in patients with the metal artifact in the head and neck region. Materials and methods: The new method of metal artifact reduction in CT images using MATLAB software. The homogeneous polymethylmethacrylate (PMMA) phantom, the Alderson Rando phantom, and patients with a metal implant in the head and neck region were scanned by the Philips Brilliance Big Bore CT system. Commercial orthopedic metal artifact reduction (OMAR) software and a new method software were applied to the CT images of phantoms and patients. The quantitative analysis of image quality on a metal artifact of the head and neck region was evaluated in the percent noise. The qualitative analysis in clinical imaging was evaluated in scoring by two radiologists with the same experience. Results: In the Alderson Rando phantom, the new algorithm indicated higher efficiency in metal artifact reduction than OMAR software. In contrast, for the patient at head and neck CT images with metal artifact reduction, OMAR, and the new method showed comparable results. The new method suppressed the artifact in homogeneous PMMA, Alderson Rando phantoms, and patients with a metal implant in the head and neck region with approximately 40%, 40%, and 60% percentage of noise reduction, respectively. The qualitative analysis by two radiologists showed comparable results of OMAR and the new method. Conclusion: The efficiency of metal artifact reduction of the new method is better than no correction and OMAR in homogeneous PMMA phantom and Alderson Rando phantom. However, the efficiency of OMAR is better than the new method, and no correction regarding the percent noise.