Teeth Segmentation in Cone-Beam Computed Tomoghraphy Images using Watershed and Morphology Operators

Abstract

In dentistry, computer-based techniques such as pre-operative planning, implant implantation, and surgery are being developed day to day. One of the radiographic methods used to study the anatomy of the tooth is the use of Cone Beam Computed Tomography, which reduces the twodimensional radiography limitations and has a higher resolution than computed tomography. Dental segmentation in digital display plays an important role in computer algorithms for extraction of features, orthodontic simulation for rearrangement teeth. In the last decade, Achieving and implementing the mentioned processes is one of the most important issues of research. One of the topics of segmentation algorithm is the increase in accuracy for the detection of teeth and pulp borders, the area related to enamel and also the reduction of artifacts. For this, various methods have been presented in previous articles. In this study, a completely new approach is proposed. In this study, in addition to segmentation of the teeth, the pulp and enamel of the tooth are also segmented. The proposed procedure for the segmentation of the teeth includes increasing contrast, reducing artifacts, morphological operators, detecting the edge, and the marker controlled watershed segmentation. The proposed method for the pulp segmentation, includes increasing contrast, decreasing artifacts, detecting the edge, and dividing the marker controlled watershed segmentation. The proposed method for segmentation of the enamel includes increasing contrast, reducing artifacts, and local threshold. Using a median filter on CBCT images shows a noticeable noise reduction. Our method is more accurate and more sensitive for teeth segmentation. We overcome the over-segmentation phenomenon. The results of the enamel and pulp segmentation algorithm show that the proposed algorithm has high accuracy and specificity.