NURBS-based detection of age-related variability of the human brain surface

A technique based on statistical and curvature analyse of non-uniform rational b-splines (NURBS) deformations to detect and visualize the profound variations in surface patterns during the evolution of the human brain from childhood to adolescent is presented. Initially, a deformable NURBS surface is used to segment the brains from normalized magnetic resonance (MR) images. The NURBS surface of a template brain is then deformed until it conforms to a representative subject image. Variability in significant deformations between the subjects is captured by regression analysis of the deformed vectors while the curvature Laplacian is computed to detect the variations in sulci and gyri patterns. We also examine the variability between different age ranges by separating the subjects into 4 groups based on their ages. The technique is demonstrated using the surface patterns of an ensemble of healthy male subjects between ages 4-18.