Impact of Midline Rotational Standardization on Estimates of Facial Asymmetry

Objective: : Inherent to the study of landmark-based facial asymmetry are the determination of a midline and a method of superimposition to align the reference and opposing portions of the face.  Choice of the method of standardization can have a substantial impact on the results of facial asymmetry studies.  The goals of this study were to develop a methodology for superimposition using translation and rotational standardization and to determine the impact of rotational standardization on landmark-based measures of asymmetry.       

Method: Three-dimensional stereophotogrammetric images of 182 unrelated adults (118F/64M) were soft-tissue landmarked using 3dMD Software and configurations were translated to the origin via the landmark subnasale.  The left configuration was reflected over the X-Y plane and the Euclidean distance between each pair of 9 bilateral landmarks was calculated to measure landmark asymmetry without rotational standardization.  To compute the standardized asymmetry scores, the landmarks nasion, subnasale, and sublabiale were chosen to define the mid facial plane of the translated configurations. Standardization of the configurations to the X-Y plane at Z=0 and the X-Z plane at Y=0 utilized iterative methods of rotation to determine optimal vertical position of the mid-facial plane. Similar reflections and calculations of the Euclidean distance were performed for the standardized data.  Standardized and non-standardized landmark asymmetry scores were examined for differences using the Wilcoxon Signed Rank test to determine the impact of standardization on the estimate of asymmetry.  

Result: Significant differences existed between the unstandardized and the standardized asymmetry scores for endocanthion (Wilcoxon p<0.0001), exocanthion (p=0.0286), subalare (p=0.0383), crista philtri (p<0.0001), and chelion (p=0.0407).  For these landmarks, the estimate of asymmetry was significantly larger in the unstandardized data compared to the standardized data.

Conclusion: Insufficient rotational standardization of facial landmark data may result in an inaccurate estimate of bilateral landmark asymmetry, particularly in landmarks close to the mid-facial plane.