Accuracy assessment of Procrustes analysis for computing mid-sagittal plane of three-dimensional facial data

Abstract

Objective: The aim of this study is to assess the accuracy of Procrustes analysis(PA)to compute a mid-sagittal plane(MSP)of three-dimensional(3D) facial data. Methods: Facial surface data from 30 subjects were acquired by a Face Scan optical 3D sensor. Using the data, 30 asymmetrical facial images with a true MSP for control purpose were constructed using the original facial image. The MSPs of the 30 images are then computed using Procrustes analysis. The angle between the true-MSP and the MSP obtained using the PA, and the asymmetry index serve as measures of assessing the accuracy of the Procrustes analysis to compute the MSP of 3D facial data. Results: The mean value and the standard deviation of the angle between the true-MSP and the PA-MSP are 0.62° and 0.49° respectively. The t-test for paired groups is used to assess the differences between the two MSPs in the facial asymmetry index and P values of smaller than 0.05 are considered significant(t=0.783,p=0.440). Conclusions: There are no significant differences between the PA-MSP and the true-MSP in terms of the asymmetry index of the 30 subjects. Thus the Procrustes analysis can be used to compute the MSP of 3D facial data with a significant degree of accuracy.