Affine Transform Assisted Firefly Algorithm in Image Registry to MRI and CT Brain Images

Abstract

Medical multimodality images create an essential need of Image registration. In this work, we focused on one of the components of the methods of which is the measure of similarity used to match the images. We are particularly interested in the iconic approach, which takes into account only the information carried by the intensities of the pixels of the images to be recalibrated. This approach has the advantage of being fully automatic, since no prior segmentation is necessary. The main contribution of this research work is to propose new measures of similarity based on cumulants and the development of edge worth and for some of them they approximate the Mutual Information, which is a measure of similarity of reference in registration. Tests on these new measures show their effectiveness for registration medical images. In addition, the generosity of the proposed approach allows the use of these measures in various situations. Further we proposed and analysed a novel multimodal image registration method for medical imaging using firefly algorithm (FF). We optimized the registration parameters of affine transform with firefly algorithm to register computed tomography (CT) brain image over magnetic resonance image (MRI) to maximize mutual information. We tested our method over images of different sizes and modality.