Multimodal medical image fusion using L0 gradient smoothing with sparse representation

Multimodal medical image fusion technique is an important and essential coalescing technique for the medical images with different modalities. The fused medical image carries more useful information than two or more relevant individual medical images of modality. To enhance, preserve edge and feature information and to remove noise of the source images a novel medical image fusion method has been developed with multiscale edge preserving decomposition L-0 smoothing with sparse representation (SR) in nonsubsampled contourlet transform (NSCT) domain. The NSCT-based image fusion method provides richer information in the spatial and spectral domains simultaneously. In this method, initially, L-0 gradient smoothing filter is applied in two different modal clinical images separately to decompose the source images into two layers such as low frequency layer (LFL) and high frequency layer (HFL) which preserves the information and improves quality. To maintain the curve edges and the energy of the source medical images, the LFLs of different modal images are combined by using the NSCT - SR fusion rule also to protect the detailed information of input medical images and reduce redundant information, the HFLs are combined by max-absolute combination rule. By combining the reconstructed LFL and HFL, the resultant fused image is obtained. The experimental results show that the proposed work can provide better results than current methodologies in terms of both visual consistency and quantitative analysis.

Automated summary

The proposed method combines low frequency layers of different modal images to preserve detail and improve quality, while combining high frequency layers to maintain curve edges and image energy. Experimental results show that this method outperforms current methodologies in terms of visual consistency and quantitative analysis.