3D brain surface matching based on geodesics and local geometry.pdfVIP

Computer Vision and Image Understanding 89 (2003) 252–271 /locate/cviu 3D Brain surface matching based on geodesics and local geometry Yongmei Wang,a,* Bradley S. Peterson,b and Lawrence H. Staiba a Departments of Diagnostic Radiology and Electrical Engineering, Yale University School of Medicine, P.O. Box 208042, 333 Cedar Street, New Haven, CT 06520-804 2USA b Department of Psychiatry, Columbia University, New York, NY 10032, USA Received 4 February 2002; accepted 15 October 2002 Abstract The non-rigid registration of surfaces is a complex and difficult task for which there are many important applications, such as comparing shape between deformable objects and com- paring associated function. This paper presents a new approach for brain surface matching by determining the correspondence of 3D point sets between pairs of surfaces. The algorithm is based on shape using a combination of geodesic distance and surface curvature. There are two major procedures involved. An initial sparse set of corresponding points is first generated by matching local geometrical features. Geodesic distance interpolation is then employed hierar- chically in order to capture the complex surface. By this scheme, surface correspondence and triangulation are computed simultaneously. Experiments applied to human cerebral cortical surfaces are shown to evaluate the approach. It is shown that the proposed method performs well for both surface matching and surface shape recovery. 2003 Published by Elsevier Science (USA). Keywords: Surface matching; Correspondence; Corresponding points; Geodesics; Geometrical features; Shortest paths; Shape; Triangulation 1. Introduction The shape analysis of the human cerebral cortex is of tremendous int