Automatic Tooth Crown Segmentation Of CBCT Based On Level Set

Nowadays,health is frequently listed on the hot search list of various search engines,which shows that the attention of contemporary people to health is unprecedented,and the most important thing to be ignored is oral problems.Whether it is an oral disease or beautification of the oral cavity,it is inseparable from the study of the teeth in the oral cavity.In order to reduce the burden of doctors,the current research on tooth segmentation is becoming the research focus and hotspot of various scholars.At this time,the birth of CBCT imaging technology has been widely concerned,and it provides important data support for oral and maxillofacial diagnosis and treatment.At present,there are many methods for segmentation of CBCT tooth image,but due to the complexity of the structure of the tooth,CBCT tooth image is essentially a multi-object segmentation.The boundary between adjacent teeth is blurred,the root division of the tooth cannot be tracked in time,and the process of segmenting the tooth will be affected by the periodontal tissue,which makes the segmentation of the tooth extremely difficult.The smoothness of the level set and other characteristics make it perform well in the field of medical images,and it can overcome the problem of needling on boundary caused by deep learning to segment teeth.Therefore,this paper studies CBCT tooth images on the basis of the level set model based on edges-the level set of distance regularization.The main contents of the study are as follows:CBCT data preprocessing.Considering the relevant characteristics of CBCT imaging,it is necessary to preprocess CBCT images,using bilateral filtering to filter and denoise CBCT images,and propose a method to eliminate external implants in CBCT images.And introduce segmented grayscale the linear change further preprocesses the dataset.Fully automatic and fast level set segmentation of single tooth crown.First,how to select the initial layer of dataset is introduced in detail,and The gray information processing of the boundary is introduced to make the tooth boundary more clearly recognized in the process of segmentation.The level set algorithm is improved,the MICO(Multiplicative Intrinsic Component Optimization)model is introduced to grade the tooth data grey level,and the result of smoothing the image is used as the initialization of the level set to realize automation.Fully automatic segmentation of the full oral crown.A priori probability term and shape constraint term are introduced to improve the situation that the root segmentation is not timely and the tooth boundary segmentation is not accurate.The regularization term is optimized to make the segmentation more stable,and the iteration stop condition is optimized to enhance the adaptability of the algorithm,and the introduction of MICO model to achieve a single-tooth crown fully automatic division on the basis of the introduction of mutually exclusive horizontal set model to solve the problem of adjacent interdental boundary blur.Then,the segmentation results of single tooth and full teeth were reconstructed and displayed.