Retrospective image registration with genetic algorithms

Abstract

A description of image registration is provided. Several most popular brain im-age registration techniques are explored and their characteristics are examined. The drawbacks of most existing registration techniques invoke the need for new image registration techniques. Detailed research on the distance functions and optimization algorithms is carried out. Based on the research, two distance func-tions, namely the Euclidean distance and the chamfer distance and the genetic algorithm based optimization technique are adopted. Two novel intramodality image registration techniques based on voxel intensity and object boundary and one novel intermodality image registration technique based on object boundary are developed by using the selected distance functions and genetic algorithms. Their applications to brain image registration of functional MR images and struc-tural- MR images are studied. The comparison of the two intramodality image registration techniques is provided as well as the comparison between the adopted optimization technique and some other optimization techniques.