Research On The Key Technologies In Modeling Of Facial Form Prediction Based On Feature-Driven

In recent years,with the development and the cross application of computer technologies and digital technologies,precision medicine becomes the development trend of healthcare industry based on digital medicine.In this paper,to aim at the difficulties and requirements of the facial form prediction in the digital oral health care,the prediction method of the facial form based on characteristics-driven is studied for three common situations that are in urgent need of the facial form prediction.The new tooth arrangement algorithm and facial form prediction methods driven by tooth and gingiva are proposed.Digital tooth arrangement in orthodontics and facial form prediction in different situations can be achieved.The research results are able to provide the doctors with new technologies and methods in the digital oral health care.The main research contents and the achievements of this paper are as following:(1)To aim at the features that consume a great deal of time,materials and human resources in the traditional teeth arrangement,and the neglect of the position relations between roots of teeth and alveolar bone in the existing digital teeth arrangement,the digital teeth arrangement of complete teeth models with double feature constraints space based on the information of crowns and roots of teeth is proposed.The automatically teeth arrangement can be achieved under the balance of double feature constraints space,and quickly and accurately provide the digital teeth arrangement results to doctors.(2)The servo system consisted with teeth,gingiva and face is constructed,and the mixed deformation method based on the volume mesh Laplace transformation technique and the volume mesh mass-spring model is researched.The models registration based on curvature Hausdorff distances,and the extraction method of the action scope based on ray sampling and surface projection are proposed to preprocess gingival soft tissue model and facial soft tissue model.The volume mesh generation of gingival soft tissue model and facial soft tissue model based on constraint Delaunay tetrahedron generation algorithm is designed.The deformation method of gingival soft tissue model based on the volume mesh Laplace transformation technique is proposed.And the deformation method of facial soft tissue model based on the volume mesh mass-spring model is raised.The face deformation prediction after the location change of teeth can be achieved.That is helpful for doctors to predict how the tooth movement affects the face deformation.(3)The prediction method of facial deformation based on statistical models is explored combining with sample data.This method designed the feature template based on the curvature characteristic to replace the facial deformed region.To aim at the treatment of habit of specific doctor and the treatment need of plenty of doctors respectively,the prediction methods conforming to the individual habit and the public demand are constructed.BP neural network model and principal component analysis are designed to learn the changes between preoperative and postoperative models respectively.The elastic model of skin tissue with complexity mechanical properties can be achieved.The feature template after deformation can be outputted based on the prediction.The preoperative model can be deformed with Laplace transformation technique,and the predicted postoperative model can be obtained.The iterative adjustments of the denture to fit and observe the facial forms of the patients in the traditional complete denture design process can be not necessary.(4)The separate area deformation prediction of the facial model based on ARAP cluster algorithm and elastic characteristics is carried out combining with elastic characteristic of personalized skin.The feature template based on linear regression model is constructed according to the facial muscle anatomical structure.The ARAP method is optimized by cluster.The preoperative model can be simulated with ARAP cluster algorithm.The weight space is constructed,and the classification method of elastic degree is designed based on Gabor filter.The deformation results are more closed to skin elasticity of the patient.The subarea and the adjusted deformation of the facial model based on anatomical structure in the oral treatment to meet the demands of the doctors is managed.(5)On the basis of the above-mentioned research,the teeth arrangement and facial form prediction software system(OralTreatSystem)is developed with function library in VC++ and OpenGL.The system can predict the deformation of facial and gingival soft tissue model in oral treatment,and simulate teeth arrangement in orthodontic treatment to assist in teeth arrangement plan and design the brackets.The part function of the system has been put into clinical test in Peking University School of Stomatology.