Multimodal Image Fusion of Magnetic Resonance and Computed Tomography Brain Images – A New Approach
Leena Chandrashekar*
and Sreedevi AElectrical and Electronics Engineering Department, VTU, Bangalore, India, 560098.
Corresponding Author E-mail: leenamaheshnikam10@gmail.com
Abstract: Multimodal images are studied to confirm the presence or classify the brain tumors like Glioblastoma. Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) are the commonly used multimodal images to confirm the tumors. Each of the images represents a unique and crucial attribute of the tumor. Experts perform an elaborate manual analysis for each of these images and in some cases the conclusions on the tumors can be indecisive. The efficiency of analysis is mostly based on the expertise of the experts. Moreover, the multimodal images succumb to uncertainty, since they represent diverse information of the tumor. We propose a technique to overcome these issues by fusing multimodal images using Non-sub Sampled Contourlet Transform. The quality of the fused images can be further improved by performing a contrast and edge enhancement technique based on Contrast Limited Adaptive Histogram technique and Particle Swarm Optimization. The paper proposes a novel approach that can promote the visual quality of the fused images. The proposed technique provides improvement in standard deviation, entropy, structural similarity index and unique image quality index of the fused images.
Keywords: Contrast Limited Adaptive Histogram Equalization; Multimodal Image Fusion; Non-Local Means Filter; Non-sub Sampled Contourlet Transform; Particle Swarm Optimization. Back to TOC