Prediction of facial deformation after complete denture prosthesis using BP neural network.

With the accelerated aging of world population, complete denture prosthesis plays an increasingly important role in mouth rehabilitation. In addition to recovering stomatognathic system function, restoring the appearance of a third of the area under the face has become a great challenge in complete denture prosthesis. This study analyzes the interactive relationship between the appearance of a third of the area under the face and complete denture, and proposes a new method to predict facial deformation after complete denture prosthesis. Firstly, to improve computational efficiency, the feature template is constructed to replace the deformed facial region. Secondly, a forecast model of elastic deformation is constructed using BP neural network and predicts elastic deformation amount because of the inhomogeneous, anisotropic and nonlinear material properties of soft tissue. Finally, a new feature template is calculated using deformation amount, and the deformation of preoperative model is simulated using Laplacian deformation technique. The average error rates of different hidden layer nodes in the neural network are analysed. Deformation and postoperative models are superimposed for match analysis. Experimental results show that this method can predict facial soft tissue deformation quickly and accurately.