Enhanced Mutual Information-based Multimodal Brain MR Image Registration Using Phase Congruency

In intensity-based image registration methods, similarity measure plays a vital role. Recently, mutual information and the variations of MI have gained popularity for the registration of multimodal images. As multimodal images have contrast changes, it is difficult to map them properly. To overcome this issue, phase congruency of the images that gives the significant features of illumination changed images. Also, the soft tissues present in the brain images have same intensity value in different regions. Hence, another assumption is that different pixels have unique distribution present in different regions for their proper characterization. For this challenge, utility measure is incorporated into enhanced mutual information as a weighted information to the joint histogram of the images. In this paper, spatial information along with features of phase congruency is combined to enhance the registration accuracy with less computational complexity. The proposed technique is validated with 6 sets of simulated brain images with different sets of transformed parameters. Evaluation parameters show the improvement of the proposed technique as compared to the other existing state of the arts.