The Tooth Segmentattion Algorithm Study Of CBCT Dental Images

Since Rontgen discovered X-ray,human exploration of human visualization technology has never stopped.With the progress and development of technology,more and more visualization tools appear,and the corresponding requirements for medical image processing technology are also getting higher and higher.Medical image segmentation,as an important part of medical image processing technology,plays an important role in human visualization technology.With the improvement and progress of people's living standards and the change of lifestyle,human beings take in a large amount of sugar,fat and so on,which makes the body appear a variety of problems.According to the data,more than 200 million teeth in China are in unhealthy condition,and the number of patients is still increasing every year.Dental health has become an urgent health problem.Cone Beam Computed Tomography(CBCT)and oral X-ray panorama are the most commonly used dental equipment.Compared with traditional multi-slice CT,cone-beam CT has the advantages of low radiation dose,fast imaging speed,high axial resolution and low operation cost.It plays an increasingly important role in the detection of Stomatology and skull.Cone-beam CT images contain the anatomical information of teeth.It is very important to segment a single tooth from the three-dimensional images reconstructed by projection data and to measure quantitatively in orthodontic surgery and denture implantation.The panorama obtained by oral X-ray panorama can reflect the relative position information of teeth and the details of the apex of teeth.Two kinds of pictures have their own advantages.In oral diagnosis and treatment,two kinds of pictures are often needed to cooperate with each other,which requires patients to undergo two X-ray irradiations.In view of the above problems,the following work has been done in this paper.1.This paper introduces the theory of image segmentation in detail,and briefly introduces the most basic image segmentation methods.The segmentation method used in this paper is level set method.Level set can be divided into edge-based level set method,region-based level set method and hybrid level set method.In this paper,three kinds of commonly used level sets are briefly introduced.The advantages and disadvantages of various level set models are analyzed.The development status of tooth image segmentation at home and abroad is analyzed in depth.2.There is little difference in the gray scale of the root surrounded by alveolar bone in tooth segmentation.Molars will split into two or three teeth at the root.It is difficult to satisfy the topological changes in the segmentation.There will be pulp cavity in the teeth,and the internal contour may appear in the segmentation.To solve these problems,a tooth segmentation method based on local Gauss distribution fitting is proposed in this paper.In this method,local Gauss distribution fitting is used as the main algorithm of tooth segmentation.It can segment teeth and alveolar bone accurately,and it can solve the problem of molar splitting well.In addition,this paper proposes a complete segmentation process for tooth segmentation.In view of the small difference of the gray level between the root and alveolar bone,this paper uses the upper segmentation results to construct the narrow band of the current layer,and enhances the contrast of the teeth by calculation.The experiment shows that the gray level difference between the teeth and the alveolar bone can be more obvious.Aiming at the problem of internal contour in tooth segmentation,edge indicators are added to the segmentation model to avoid the problem of internal contour of teeth.To solve the problem of level set regularization,this paper compares the current mainstream regularization methods by experiments.The experimental RD regularization method is the most beneficial to the segmentation method in this paper,which can prevent over-segmentation and tooth edge leakage.Through experiments,this method is compared with CV model and shape-strength prior model segmentation method.The method presented in this paper performs better on molar splitting,root and alveolar bone splitting,etc.3.How to realize the synthesis from cone-beam CT tomography to oral panorama? To solve this problem,a new method of oral panorama reconstruction based on segmentation of dental tomography image is proposed.This method is based on the tooth segmentation mentioned above.At the end of tooth segmentation,the initial layer is selected to get the centroid points according to the segmented teeth,and then the arch line is fitted by cubic spline interpolation.When projecting,it is necessary to remove the interference of non-dental tissues in the oral cavity,so a projection area of 100 pixels width is set in the projection.In this paper,maximum density projection and average density projection are selected for projection.Experiments show that the reconstruction of oral panorama based on tooth segmentation can reduce the workload of doctors,reduce the radiation dose of patients,and reproduce the relationship between the details of teeth and relative position with better energy consumption.