Automated quantification and analysis of mandibular asymmetry

We present an automated method of spatially detailed 3D asymmetry quantification in mandibles extracted from CT and apply it to a population of infants with unilateral coronal synostosis (UCS). An atlas-based method employing nonrigid registration of surfaces is used for determining deformation fields, thereby establishing detailed anatomical point correspondence between subjects as well as between points on the left and right side of the mid-sagittal plane (MSP). Asymmetry is defined in terms of the vector between a point and the corresponding anatomical point on the opposite side of the MSP after mirroring the mandible across the MSP. A principal components analysis of asymmetry characterizes the major types of asymmetry in the population, and successfully separates the asymmetric UCS mandibles from a number of less asymmetric mandibles from a control population.