Research On Application Of 3D Model Segmentation Technology In Digital Dental Medical Field

This paper mainly studies the three-dimensional dental model segmentation and the segmentation of the mandibular canal.In the past few years,digital tooth models are very common for simulation and planning of orthodontic interventions.An important pre-processing of computer-assisted orthodontics is to precisely locate the teeth and separate the teeth from the tooth model with as little as possible human interaction.However,the automatic segmentation of all teeth is not a simple task since the teeth exist in different shapes and their arrangement varies from person to person.This difficulty is exacerbated when there is a problem with severe occlusion and congested teeth,which is a common clinical case.Most of the published methods in this area are either inaccurate or require a lot of human interaction.Based on the previous studies,this paper integrates the existing methods and proposes a fully automated segmentation framework for dental mesh.This method can effectively identify the segmentation boundary between the dental gingival segmentation boundary and the tooth compared with previous research results.However,this fully automated method is still insufficient to deal with complex clinical trial conditions.In order to make the tooth segmentation algorithm industrially feasible,we have also proposed a semi-interactive rapid segmentation framework.A large number of experiments show that our proposed semi-interactive method can fully resolve the complicated clinical problems.Compared with previous dental segmentation studies,the method we propose requires a small amount of human interaction,while is very fast and robust.Another focus of this article is the mandibular neural tube segmentation technique.The mandibular nerve segmentation is mainly used to assist clinical dental implants,and it is combined with three-dimensional dental segmentation technology to make medical aids.Our raw data comes from oral CBCT scan data.Using CBCT data,a 3D point cloud model of the patient’s mouth can be established by volume rendering technology.The value of each point of the 3D point cloud model is the CT value of this point,which reflects the density of the area represented by this point.Based on this value,we can use threshold filtering to easily separate human tissues such as skeletal muscle fat,and then we can observe the different physiological structures of the oral cavity according to the CT value of each point in the dental medical field.However,only the macro descriptions have limited information.We also hope to quantify the specific information of various organizations.For example,in dental implant surgery,we need to obtain the specific position of the mandibular canal or the position of the relative teeth,so that before surgery Some are fully documented and give the best protection to the patient.In summary,the main contributions and contributions of this paper are as follows: 1)Aiming at the task of fully automatic tooth segmentation,an improved scheme based on improved Snake algorithm combined with regional growth algorithm is proposed.This scheme combines the advantages of Snake algorithm and regional growth algorithm in practice.The effect and robustness in the engineering environment are higher than other solutions;2)For industrial-grade tooth segmentation tasks,a segmentation scheme based on the shortest path algorithm of the grid is proposed.This method requires only a small amount of interaction to obtain Ideal segmentation results,and can be real-time segmentation,the speed is much higher than other segmentation schemes;3)Using the clinically collected data,the current various mainstream methods applied to dental segmentation are tested,and the segmentation effect is done Detailed analysis;4)Based on deep learning,a 2D image maturation solution successfully solves the problems in the 3D domain,and a robust mandibular neural tube segmentation scheme is proposed,which provides a new solution for similar problems in this field.