Osseous and digital subtraction angiography image fusion via various enhancement schemes and Laplacian pyramid transformations

Abstract

Image fusion is a significant medical imaging tool which integrates the complimentary information from various sources into a single frame for enhanced visual perception. The fusion of osseous and vascular information is used for the localization of various medical abnormalities. The emergence of gradient reversal artefacts and halo effects in the fused image are of major concern for the researchers. In this paper we propose an image fusion technique to mitigate the artefact issues in case of bone and vessel image fusion. An ideal image fusion rule transfers maximum information from source images to fused image with least amount of distortion or loss. In this regard we have transformed the source images with the help of KL transformations and Ripplet transform. The artefacts are controlled via anisotropic diffusion filtering. The Laplacian pyramidal based fusion is employed to fuse the mask and DSA images. For the validation of our proposed methodology conventional as well gradient based metrics along with human visual perception are employed. The proposed methodology outperforms eight other state-of-the-art image fusion techniques with far better visual results. The entire algorithm is implemented in MATLAB 2012 with core i5 processor.