Research On Full Crown Prosthesis Design Based On Convolution Neural Network

The application and promotion of CAD/CAM technology in the field of oral prosthesis have gradually replaced the traditional manual tooth restorations.The digital design of full crown restoration can not only improve the production efficiency,but also improve the accuracy of restoration.However,manual interactive operations are required in the design,and the quality of restorations depends on the proficiency and subjective ideas of doctors.With the successful application of deep learning in the field of oral prosthesis,dental restoration will be promoted to become more intelligent,personalized and automatic.In this paper,the full crown restoration is taken as the research goal,and the key technologies in the design are studied.The main research contents are as follows:(1)Based on 3D convolutional neural network,an extraction method of the tooth preparation line is proposed.Firstly,the tooth preparation dataset is established.Secondly,based on convolutional neural network,the preparation segmentation network model is established.Thirdly,the network model is trained and tuned to realize the automatic segmentation of the preparation point cloud.Finally,point cloud boundary extraction and spline curve fitting and interpolation are used to achieve the tooth preparation line.(2)Based on radial basis function,an internal surface generation method is proposed.According to different thickness of the adhesives in different regions of the preparation model,the radial basis function is used to calculate the implicit surface by defining the control points,so as to generate the internal surface.The method realizes the unequal offset of the preparation model and has higher stability.(3)Based on generative adversarial network,a method for designing personalized full crown restoration is proposed.Firstly,on the basis of the depth image dataset of crown,the network model is established and trained to repair the depth image.Secondly,the methods of pixel-distance mapping and delaunay are used to design the personalized occlusal surface.Finally,the transition surface is designed between the internal surface and the occlusal surface.And the Laplacian deformation method is adopted to adjust the occlusal surface to achieve the personalized full crown restoration.On the basis of the above research,the digital design modules of the full crown restoration are developed based on VS2015,VTK graphics library and QT interface,which are conformed stability and suitable by actual design.