Image registration is one of important problems and a fundamental task in computer vision. In the medical image processing fields, image registration and image warping techniques are introduced in order to register the same or different modalities such as CT and MR imaging. Especially, for diagnosing the legions in the head and to make a surgical planning, physicians gain important information from these modalities. In radiotherapy planning, manual registration techniques are performed on MR and CT images of the brain. In general, physicians segment the volumes of interest (VOIs)from each set of slices on the multi modal images manually. However, manual registration of the object area may require several minutes for analysis based on anatomical knowledge. In this paper, we develop a new method for registration of head images with high speed by using an optimization of volume of interest on the local area. The primary objective of this study is to increase accuracy of the registration and for reducing the computational time for the registration from two images of different modalities.
