Tooth Level Set Segmentation Based On CBCT Image

As more and more people attach importance to oral health,there is a large amount of clinical demand for improved tooth function and aesthetic treatment,and doctors'requirements for the application of tooth segmentation technology are gradually increasing.The emergence of CBCT provides important data information for computer-aided orthodontic diagnosis and treatment,but the problems of the contact boundary between the tooth and the adjacent tooth,periodontal tissue,and jaw bone is blurred,and the topology of the tooth structure changes make it more difficult to segment,However,the level set method performs well in the topology change of medical image segmentation.Therefore,this work aims at the practical application problems of clinical orthodontic in the efficiency,accuracy and stability requirements of tooth segmentation,combined with the features of CBCT image data and knowledge of tooth anatomy structure,researches the tooth level set segmentation algorithm,and completes the realization of three-dimensional tooth structure.The specific research content is as follows:Data pre-processing.According to the CBCT image data information,piecewise linear transformation and median filtering are used for image enhancement and noise reduction,and maximum intensity projection is used to select the tooth boundary area for easy segmentation processing.Formulate the realization plan of tooth segmentation.Combined with the development of tooth segmentation and based on the level set theory and tooth anatomy structure,the implementation scheme of tooth segmentation is mainly introduced.For the initial layer of molars,a fast level set segmentation based on edge detection factors is proposed,and experiments are performed with specific teeth.The results verify that the running time is improved.Improved single tooth hybrid level set segmentation.For the situation that the topology change of the segmented tooth root is not timely and the curve is easily affected by the adjacent teeth and periodontal tissue,resulting in the wrong segmentation boundary,this work makes full use of the inter-layer information of the CBCT image sequence to optimize the definition of prior information items and regular terms,and introduces a constrained shape item,build an improved hybrid level set framework.An iterative segmentation experiment of interlayer sequences in the root direction of the improved hybrid level set for dental data.The average values of the four indicators Dice,VD,MSSD,and ASSD are 89.20%,35.31mm~2,1.43mm,and 0.20mm,respectively.The performance is relatively good,which slightly improves the timeliness of the topology change of the root branch,and better guarantees the stability during the model evolution.Three-dimensional realization of total oral tooth segmentation.Aiming at the existence of ambiguous areas at the boundary of the adjacent tooth crowns,a hybrid double level set mutually exclusive model is proposed for total oral tooth segmentation.The single teeth of the initial layer are separated and then combined with rapid segmentation processing.Alternate parity is used as the initialization of full oral segmentation.this work completes iterative segmentation of the interlayer sequence and three-dimensional optimization processing,complete the realization of the three-dimensional structure of the total oral teeth and analyze the results.The above work has carried out targeted experiments through multiple teeth in the clinic.The results show that the improved segmentation model proposed in this work has relatively good performance in the application efficiency,accuracy and stability of tooth level set segmentation.