An improved method for image registration in digital subtraction angiography based on mutual information and edge overlap degree

This paper studies the mutual information method in the image registration of digital subtraction angiography (DSA). It also studied the principle of edge overlap between two images and bought it into the similarity measurement. This article improved the goal function of registration by integrating the overlap degree of edge with mutual information. Besides, the author analyzed the deformable model in DSA registration, proposed a simple and efficient transform model. The new similarity measure method in this paper not only uses the gray-scale information of the original images, but also uses the pixel's relationship on two-dimensional level of edge image which compensates for the drawback of mutual information and improves the accuracy of registration. The Matlab simulation experiments prove that this is an effective and reliable image registration method.