Semi-automatic Segmentation Of Tooth And Application In CT Image

In clinical orthodontics and prosthodontics, it is necessary for doctors to have enough anatomical information on the root and crown of an individual tooth. Tooth segmentation in CT images which provides doctor the above information aims to help doctors on clinical analysis and formulate reasonable treatment.The similar gray scale between teeth blurs the boundaries between them in CT images. The same problem exists between teeth and alveolar bone. It is difficult to realize individual tooth segmentation with the common way. At present, some scholars use level set algorithm to realize the segmentation of dentition in CT images, but this method failed to separate single tooth from others.In view of the above clinical background and research situation, this paper tried to seek an interactive method to extract individual tooth in CT images. This method can avoid the problem of the automatic segmentation method, and on the other hand, it can make full use of doctors’ clinical experience and knowledge of anatomy.This paper mainly focuses on the following aspects of work:(1) Tooth segmentation based on imitated oral panorama: with the help of imitated oral panorama technology and its advantage that can display the whole dentitions in one picture, this paper proposed a semi-automatic method to realize tooth segmentation. In order to facilitate the interactive operation, this paper made some further improvement in the imitated oral panorama.In this method, we firstly isolate the target tooth from others and map it to 3D CT data, then creat a bounding box to realize the segmentation. This method effectively decrease the interactive manipulation compared to direct interactive segmentation in 3D model reconstructed from CT data.(2) Interactive segmentation method: for tight closure of the upper and lower dentitions or the case of overlapping teeth in CT images, there is no effective segmentation method. Therefore, we designed an interactive teeth segmentation method based on contour lines. In this method, we extract the tooth contours along the main axis direction of it, and then reconstruct the dental 3D model using the improved shortest diagonal method. Experiments show that, it is effective to separate the connected teeth.(3) The application of teeth segmentation: according to the clinical significance of dental segmentation in CT images, this paper completed the application on two aspects: the use of principal component analysis to calculate direction of the teeth, and a dentist has verified the effectiveness of this method; based on the symmetry of molar shape, we use mirror symmetry methods to reconstruct a three-dimensional model of the missing tooth.