IMPROVED IMAGE REGISTRATION BY USING FOURIER INTERPOLATION

This paper presents a technique for performing two-dimensional rigid-b ody image registration for functional magnetic resonance images (fMRI) . The method provides accurate motion correction without local distort ion. The approach is to perform the translation and rotation in the Fo urier domain. For images sampled on a grid, such as in echo-planar ima ging (EPI), one potential stumbling block to this approach is the comp utational burden of reconstruction, since the rotated image will no lo nger be on the Cartesian grid. A method of approximating rotations via local translations (shearing) is presented, which keeps the data on t he Cartesian grid. This can provide quite accurate approximations with only a moderate amount of computation. A mean squared error (WISE) cr iterion is used for determining the registration parameters. This meth od is tested on several sets of simulated images and shown to have an accuracy ranging from 0.02 to 0.3 pixels for images with SNRs ranging from 100 to 10, respectively. These techniques have been tested on sev eral sets of images. They are shown to work well on real subjects, for both echo-planar and spiral data acquisition schemes. The techniques are used in an activation study in which the subject moved his head du ring image collection. After use of this registration technique, the a ctivation is easily detected.