Mutual information based image registration for MRI and CT SCAN brain images

摘要

Image registration is the process of transforming the different sets of data, which is acquired by sampling of the same scene or object at different times or from different perspectives, into one coordinate system. This paper presents an algorithm for recovering translation parameter from two images that differ by Rotation, Scaling, Transformation and Rotation-scale-Translation (RST) also known as similarity transformation. The proposed algorithm utilized information-theoretic ideas for image registration. Registration is assumed to correspond to maximizing mutual information. Mutual information compares the statistical dependence between two images. Here experiments are carried out on Brain images. For CT scan and MRI scan, patients needs to lie still. There may be some discomfort from having to remain still for several minutes for some patients having chronic pain or specially for children. So resultant CT scan or MRI scan may be rotated, scaled or translated. So it must be registered correctly as reference image for diagnosis purpose. Simulations are shown which shows the performance of the method presented.