Knee Bone Surface Registration Using the Lipschitz Optimization Algorithm

Abstract

We describe a technique for the registration of three dimensional (3D) knee bone surface points from MR image data sets. This technique is grounded on a mathematical theory – Lipschitz optimization. Based on this theory, we propose a global search algorithm that simultaneously determines the transformation and point correspondences. Compared with the other three registration approaches (ICP, EM-ICP, and genetic algorithms), the new proposed method achieved the highest registration accuracy on animal data.