Registration And Fusion Techniques For Multimodal Brain Images

Loading...
Thumbnail Image
Date
2010-08
Authors
Al-Azzawi, Nemir Ahmed Fuaad
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
The research presented in this thesis is concerned with the problem of multimodal medical image registration and fusion for brain images. The work on image registration conducted in this thesis was based on feature and intensity. The feature-based approach used nonsubsampled contourlet transfrm (NSCT) to extract salient edges and control points. Then, mutual information was adopted to the register feature points where translation parameters were calculated by using particle swarm optimization. As result, the registration performance was perfect. The intensity-based nonrigid registration was developed and evaluated based on free-form deformation (FFD). Kullback-Leibler distance (KLD) and particle swarm optimization (PSO). The procedure yielded quality images with average target registration error (TRE) below 4 mm. Furthermore, the proposed method could run 2-3 times faster than the traditional nonrigid registration mutual imformation method.
Description
Keywords
Registration and fusion techniques , for multimodal brain images
Citation