A Study of Quantifying Facial Symmetry using Hyper-dimensional Computing Technique
This study combines the characteristics of 3D facial contour images that can represent the 3D features of the overall face, and hyper-dimensional (HD) computing that can be used to calculate the similarity of the 3D contour images of the composite image of left and right faces (that is, the overall facial symmetry). We propose an assessment system of facial symmetry using HD computing technique for measuring the similarity of 3D facial contour composite lines.200images were recruited in a retrospective cohort study for building the assessment model of facial symmetry. 3D facial photos were captured by the Cone Beam Computed Tomography (CBCT) and 3D contour images were extracted from 3D facial images used for the feature images. We also developed a web-based system for clinicians to automatically assess facial symmetry based on our proposed HD computing technique. Additionally, or prospective study in clinical application, the proposed assessment model of facial symmetry using HD computing was applied for the automatic assessment of facial symmetry in 258 patients before and after or thognathic surgery (OGS). The experimental results revealed that the average accuracy of quantitative facical symmetry is 86% using HD computing model. For prospective study in clinical application, the average degree of improvement in facial symmetry after surgery was 26.7%.