Iterative correction applied to streak artifact reduction in an X-ray computed tomography image of the dento-alveolar region

Streak artifacts that appear on dental and maxillofacial X-ray computed tomography (CT) images are mainly caused by the presence of metallic prosthetic appliances. Because of the design of common CT hardware, the thin slice thickness method is routinely used in CT examinations of the dento-alveolar region. Thus, within the resulting collection of thin slice images, adjacent CT images will often depict very similar anatomical structures. We took advantage of this aspect and employed iterative correction to reduce the streak artifact caused by metallic materials. The maximum likelihood-expectation maximization (ML-EM) reconstruction algorithm, a type of statistical reconstruction method, was applied to multidetector CT images. The ML-EM is an iterative restoration method that approximates between the processed image and the original projection data. In our study, we processed slices with heavy streak artifacts by using the projection data of an adjacent CT image without any major artifacts. Because the adjacent slices depicted very similar anatomical structures, they became the target of the proposed processing. Thus, the processing is essentially an iterative correction. The iterative correction was carried out 50 times. Processed images at the initial stage were blurred, but some streak artifacts clearly disappeared as the iteration progressed. An ML-EM reconstruction algorithm can be used to modify iterative correction to reduce streak artifacts in dental and maxillofacial CT images. We used an image that contained heavy artifacts in our study, and after 50 iterative correction cycles, only a few weak artifacts remained on the processed image. The final image produced by our iterative correction method depicted clear anatomical structures and developed only marginal deviations from the original image.