Abstract
Medical image fusion is important in biomedical applications for non-invasive diagnosis. Multimodal image fusion reduces defects associated with images created from single modalities, which often result in less informative and therefore less useful images. Fused images created from two or more images exhibit increased accuracy of the information content and better visual properties. However, existing fusion methods have not properly addressed some of the poor visual properties and insufficient accuracy of the fused images. A novel method for improving the visual properties and information content of the fused image is described in this paper. The proposed algorithm combines Gabor filtering, maximum pixel selection and Pulse-Coupled Neural Network (PCNN) implementation aimed at creating an improved fused image. In order to quantify this improvement, two fused images are generated and evaluated under standard performance criteria, and then compared with results from existing image fusion methods.