Three-dimensional assessment of facial asymmetry using dense correspondence, symmetry, and midline analysis

It is considered normal for facial structures to exhibit mild asymmetry between left and right sides. An automated, landmark-independent method was developed to accurately assess and quantify facial asymmetry in 3 planes of space and describe a midline deviation of each subject and ultimately establish thresholds of significance.

The subjects were 279 healthy young Western Australian white adults (134 females and 145 males) with a mean age 22.17 years ± 0.63, (minimum 20.58 years-maximum 24.42 years) without craniofacial anomalies. They were randomly selected from participants in the Raine Study-Generation 2. Surface facial images were obtained using a 3dMDface scanning system (3dMD Inc, Atlanta, Ga). Images were standardized using the dense correspondence technique. An automated landmark detection method was applied, and measurements performed on color deviation maps to quantitatively assess facial asymmetry.

Based on asymmetrical projections over the total facial surface area, the proportion of female and males with moderate asymmetry (2-5 mm) was 52.3% and 58.4%, respectively, and with severe asymmetry (>5 mm) was 7.1% and 7.7%, respectively. Most asymmetry occurred in the coronal plane (x-axis), followed by the transverse plane (z-axis) and the least asymmetry in the sagittal plane (y-axis). Males were statistically more asymmetrical (P <0.05) in the coronal and transverse planes (males: coronal 36.5%, transverse 15.2%; females: coronal 31.8%, transverse 12.3%). The midline was deviated to the right in all females and in all but 1 male subject.

This study presents an automated, rapid and accurate method of assessing 3-dimensional facial asymmetry (using symmetry and midline analyses). Analyses revealed that >50% of the faces of young adults are >2 mm asymmetrical, based on total facial surface area.