Teeth segmentation from dental x-ray image by template matching

Abstract

3D digital dental model is used for orthodontic treatment. Teeth segmentation is an important process for a 3D model construction. In this paper, we proposes a method to segment a teeth from a panoramic dental x-ray image. The proposed method consists of three steps: Tooth area identification, template matching, and teeth area segmentation. First, to identify tooth area, the Otsu's threshold and Mahalanobis distance technique are used. Second, teeth template images with different image size are used to match with a teeth in x-ray image. Finally, overlap area from matching multiple templates is used to segment a teeth. To test performance of the proposed method, twenty-five dental images are used. Totally, there are 450 single-rooted tooth, and 250 double-rooted tooth. An accuracy are 42.20 and 49.04 percent for single-root and double-root teeth, respectively. An accuracy should be improved in further. However, this is first fully automatic teeth segment method by using panoramic dental x-ray image. © 2016 IEEE.