Postoperative facial prediction for mandibular defect based on surface mesh deformation.

Area of Science:

• Biomedical Engineering
• Computer Science
• Medical Imaging

Background:

• Current methods for predicting post-operative facial contours after mandibular defect reconstruction often fail to preserve crucial geometric features and lack interpretability.
• Accurate pre-operative planning is essential for optimizing aesthetic and functional outcomes in oral and maxillofacial surgery.

Purpose of the Study:

• To develop and introduce a novel predictive model for post-operative facial contours in patients with mandibular defects.
• To address the limitations of existing methodologies in geometric feature preservation and model interpretability.

Main Methods:

• The study utilizes surface mesh theory and deep learning, employing surface triangular mesh grids instead of traditional point cloud approaches.
• A Mesh Convolutional Restricted Boltzmann Machines (MCRBM) model is used to extract latent variables and generate a three-dimensional deformation field.
• The approach aims to enhance the preservation of geometric information and improve model interpretability.

Main Results:

• Experimental evaluations show the proposed model achieves a prediction accuracy of 91.2%.
• This accuracy represents a significant improvement compared to traditional machine learning-based methods.
• The model demonstrates enhanced geometric information preservation and interpretability.

Conclusions:

• The developed model presents a promising advancement for pre-operative planning in oral and maxillofacial surgery.
• It significantly enhances the accuracy of post-operative facial contour predictions for mandibular defect reconstructions.
• The proposed approach offers substantial improvements over previous methods for facial contour prediction.