Registration of MRI-CT Image for Improving Abdominal Tumor Localization using Particle Swarm Optimization

Abstract

Image registration is primary step in many real time image processing and play an increasingly important role in many clinical application. Image registration of overlaying imagesof the same scene taken at different time, viewpoints and sensor. The registration geometrically align two images the reference and sensed images. The abdominal tumor was increasing and it was very difficult to diagnosis of the tumor by using magnetic resonance image (MRI) alone orcomputerized tomography (CT) alone. The aim of this study apply the mutual information in abdominal images that was obtained using multimodalities CT and MRI to obtain high quality image and improve the radiotherapy planning. Intensity based registration algorithm was developed in the present study for integrating the two images based on the information theory, normalize mutual information (NMI)and entropy correlation coefficient (ECC) in detecting to correspondent between it and use the affine transformation algorithm to transform one of these image because of the deformation that occurs during the acquisition of images in order to align these images. To adjust the values of the transformation parameters to yield high values for similarity measures (NMI and ECC) particle swarm optimization was used as optimization method. The wiener filter was used in filtration step after that CT and MRI images became very sharp and clear. After registration process the images are became very clear and it can produces high quality image when used the PSO and find optimal similarity measures NMI and ECC. The research is recommend that using PSO algorithm in other modalities like ultrasound images or positron emission tomography (PET) and use other optimization method to enhance the quality of similarity measure in order to gain best transformation parameters value in minimum time.