Research On Methods Of Three-Dimensional Segmentation Tooth From CBCT Images

Cone beam computer tomography (CBCT), as a method to acquire three-dimensional medical images, is more and more popular in the clinical dental practices. Compared with the conventional CT, CBCT has the capability of producing sub-millimeter resolution, with short scanning time and radiation dose reduction, which makes the 3-dimensional images of maxillofacial skeleton applied to dental practices possible. On the basis of these CBCT images with 3D image segmentation methods, dental clinicians can obtain the anatomical information of the tooth to make the accurate diagnosis and take a personalized treatment. Furthermore, when the information is combined with computer aided manufacturing it can improve the precision and productivity of dental repairing materials. So researching the 3D dental images segmentation methods may be significant and important for clinical dental practices.Initially, the basic principles and features of some typical segmentation methods were analyzed in this thesis. Then considering the dental CBCT image characteristics, the method of deformable surface of 3D triangle mesh model was proposed. The method uses a deformable model which is initialized from an icosahedron and then evolves to fit the teeth's surface by the application of the locally adaptive external forces computed from the image data and internal forces coming from the model itself. The experimental results indicate that the proposed method is robust and accurate.The main contribution of this thesis is that we have improved a deformable model segmentation algorithm applicable to dental images segmentation. On the basis of segmentation results, we implemented the functionality of mesh clipping, multi-planar reconstruction to provide the teeth structure information from different angles for dentists.