Automated Dental Arch Detection Using Computed Tomography Images

The dental arch defection from an x-ray computed tomography (CT) image is an important feature in generating panoramic images as well as in rearranging teeth in orthodontics. This paper introduces an automated approach in dental arch detection. Because teeth have higher intensities than their surrounding area, local entropy thresholding technique is employed to binarize a dental CT image. Next, we use connected component labeling to partially remove metal artifacts. Then, morphological dilation is applied to close the interstices between teeth so the maxilla/mandible region is connected into one piece. After that, morphological thinning operation is used to thin the binary maxilla/mandible region. The thinning result is a rough shape of dental arch. Lastly, we exploit the thinning result in curve fitting method to get a mathematically represented dental arch. We tested our algorithm on the total of 60 dental CT images which are taken from 6 different data sets (ten images per data set). Simulation results demonstrate satisfactory outcomes.