Individual teeth segmentation in CBCT and MSCT dental images using watershed

Teeth segmentation is an important step in human identification and Content Based Image Retrieval (CBIR) systems. This paper proposes a new approach for teeth segmentation using morphological operations and watershed algorithm. In Cone Beam Computer Tomography (CBCT) and Multi Slice Computer Tomography (MSCT) each tooth is an elliptic shape region that cannot be separated only by considering their pixels' intensity values. For segmenting a tooth from the image, some enhancement is necessary. We use morphological operators such as image filling and image opening to enhance the image. In the proposed algorithm, a Maximum Intensity Projection (MIP) mask is used to separate teeth regions from black and bony areas. Then each tooth is separated using the watershed algorithm. Anatomical constraints are used to overcome the over segmentation problem in watershed method. The results show a high accuracy for the proposed algorithm in segmenting teeth. Proposed method decreases time consuming by considering only one image of CBCT and MSCT for segmenting teeth instead of using all slices.