Evaluation of the Reproducibility of Nonverbal Facial Expressions Using a 3D Motion Capture System

Abstract

To evaluate the reproducibility of three nonverbal facial expressions using a three-dimensional motion capture system. Prospective, cross-sectional, controlled study. Glasgow Dental Hospital and School, University of Glasgow, United Kingdom. Thirty-two subjects, 16 males and 16 females. With a three-dimensional video passive stereophotogrammetry imaging system, maximal smile, cheek puff, and lip purse were captured for each subject. Anatomical facial landmarks were digitized on the first frame and then tracked automatically. The same facial expressions were captured 15 minutes later. The magnitude of each expression and speed of landmark displacement were calculated. The landmark motion curves were spatially and temporally aligned to calculate the similarity of the dynamic movements of the same landmarks between the captures. There were no significant differences between individuals for magnitude (P = .892) or for speed (P = .456). There were significant differences in landmark movement similarity (P = .011); similarity was more reproducible for maximal smile. There was no significant gender effect on the difference in magnitude. There was a significant gender effect on speed to reach maximal smile (P = .044) and a pursed-lip expression (P = .038). There was a significant gender effect on landmark movement similarities (P = .031) for cheek puff expression. There were no differences in magnitude and speed for maximal smile, cheek puff, and lip purse between the two captures for all participants. For individual expressions, maximal smile expression had the highest similarity value for individual landmarks.