Fusion of multimodal medical images using nonsubsampled shearlet transform and particle swarm optimization

Medical imaging has been an indispensable tool in modern medicine in last decades. Various types of imaging systems provide structural and functional information about tissues. But most of the time both kinds of information are necessary to make proper decision. Image fusion aims at gathering complementary information of different sources in one image to be more informative. This paper proposes a new method for this purpose. In proposed method, source images are first decomposed using nonsubsampled shearlet transform. Extracting most of relevant information and merging them to achieve the best weights for fusion task is done by principal component analysis and particle swarm optimization. Fused image is provided by merging source images according to weights achieved from previous steps. Quantitative and qualitative analysis prove outperformance of our methods compared to well-known fusion methods. The experimental results show improvement compared to subsequent best method, in terms of peak-signal-to-noise-ratio (+ 8.85%), entropy (+ 3.48%), standard deviation (+ 16.3%), and quality index (+ 14.84%).