Finding 3D Parametric Representations of the Deep Cortical Folds

Marc Vaillant, Christos Davatzikos and R. Nick Bryan Neuroimaging Laboratory Department of Radiology Johns Hopkins School of Medicine 600 N. Wolfe street, Baltimore MD 21287 WWW: http://ditzel.rad.jhu.edu Abstract Parametric representations of anatomical structures provide useful mathematical descriptions for many medical imaging applications, including morphological analysis of the brain. In this paper, we develop a methodology for obtaining a parametric representation of the deep cortical folds of the brain utilizing characteristics of the cortical shape. We rst nd a mathematical representation of the outer cortical surface using a deformable surface algorithm. Using the principal curvatures of the resulting surface, we then identify the edges on the sulci on it, and we initialize active contours along them. An external force eld guides an active contour to the deep edge of a sulcus, along the medial surface of a cortical fold. A parametric description of a sulcal surface is obtained as the active contour traverses the sulcus, sweeping a surface resembling a convoluted ribbon embedded in 3D. In this paper we present results using magnetic resonance images.