Medical image registration by maximizing mutual information based on combination of intensity and gradient information

In recent years, mutual information has developed as a popular image registration measure especially in multimodality image registration. However, based on Shannon entropy, it focuses on the relationships between corresponding individual pixels and not those neighboring pixels. It ignores the spatial information contained in the images such as edges and corners that might be useful in the image registration. Thus we propose the adaptation of mutual information measure which incorporates the spatial information by combining intensity and gradient information. Mutual information value now is calculated from the gradient value and intensity value of the images. Salient pixels in the regions with high gradient value contribute more in the estimation of mutual information of image pairs being registered. Results of normalized mutual information, gradient-based mutual information and new proposed method are presented for rigid registration of medical images. We show that the new method yield better registration accuracy and it is more robust to noise than normalized mutual information.