In the process of image fusion,there is often a serious loss of edge points.In order to solve this problem and improve the effect of fused image,a new fusion algorithm using non-subsampled shearlet transform (NSST) is proposed in this paper. First of all,the medical images that havebeen already registered are decomposed into a series of high-frequency sub-band coefficients and a low-frequency su b-band coefficient by using NSST.Then regional maximum difference comparison method is used to fuse low-frequency coefficients,and the high-freq uency coefficients are fused by using improved edge energy method and edge strength method.Finally,th e fusion coefficients are reconstructed by using the inverse NSST.Four parameters,such as information entropy (IE),edge strength (ES),space frequency (SF) and standard deviatio n (SD),are selected to evaluate the fused images.Co mpared with the other four algorithms,the algorithm has better fusion effect,esp ecially in the space frequency and edge strength index,and the other indexes are in the middle and upper level.Compared with the inovative image fusion method b ased on improved NSST by Liu jian,SF values of four groups are increased by 6.4%,9.2%, 1.4% and 2.5%,SD values are increased by 1.8%,－1.1%,6．7％ and －0．8％,IE v alues are increased by －0．8％,1．4％,－0．8％ and 0.5%,and ES values are increased by 2.8%,0.6%,1.5% and 3.4%.The experimen tal results show that the fusion image obtained by this algorithm is excellent i n clarity,contrast and edge information richness,and this algorithm has a good a pplication prospect.