Multi-modal medical image fusion can obtain more comprehensive and high quality image by integrating the complementary information of medical images, which can provide more accurate data for clinical diagnosis and treatment. To preserve the detailed information and structure information of the source image, in this paper, a novel medical image fusion method exploiting multi-scale edge-preserving decomposition and sparse representation is proposed. In our method, medical source images are decomposed into low-frequency (LF) layers and high-frequency (HF) layers by the weighted least squares filter. The rule which combined by Laplacian pyramid and sparse representation is employed to fuse LF layers. The HF layers are merged using max-absolute fusion rule. Finally, the fused LF and HF layers are combined to obtain the fused image. Experimental results prove that our method outperforms many other methods in terms of both visual and quantitative evaluations. © 2018 Elsevier Ltd