Comparison of Recent Least Square Approaches for Fusion of Multimodal Medical Images

Image fusion is a useful technique toward the better analysis of multimodal medical images. Inthis paper some methods are presented for image registration and fusion with focus on CT and MRI images.All proposed methods are Least squares based which are modified by using FFT for faster performance. Todeal with intensity mismatches between multimodal images, some previous methods are reviewed and a newmodified method is also provided. The new method proposed, utilizes the combination of two famousstructural characteristics of images, instead of one, that are gradient and features obtained from phasecongruency model. We applied the proposed method to CT and MRI images of head and thorax. Resultssuggest the priority of the new proposed method over old ones, due to higher correctly aligned percent andsmaller mean alignment error, specially for head images. Visualization of the registration is done viacolor-fusion.

Keywords:

Multimodal image registration , medical image fusion , analysis of CT and MRI images , structural characteristic