An image fusion algorithm based on multi-resolution decomposition for functional magnetic resonance images

This paper presents a new functional image fusion algorithm which is the combination of SPM and ICA using multi-resolution decomposition. Firstly, we designed the fMRI experiments and obtained the fMRI image data from different experimental conditions. The brain activated regions were extracted by the SPM and ICA methods respectively. Secondly, by constructing the Laplacian pyramids of the source image, a new fusion rule based on the salience and matching measure is proposed in various resolutions. Finally, the fused functional images are reconstructed by the inverse Laplacian pyramid transformation. The results show that the algorithm can retain the details of the source images and pinpoint exactly the brain functional area associated with the hand action, thus outperforming SPM or ICA for functional regions extraction.