In this paper, a new fusion algorithm for combining MRI and CT images of a brain based on gradient domain guided image filter and the improved sum-modified laplacian is presented. Gradient domain guided filter smoothens the source images to provide base images and subtraction of base images from source images gives detail images. Next, the obtained detail and base layers of two source images are fused with the help of fusion rules such as maximum and improved sum-modified laplacian. Finally the fused image is obtained by taking the linear combination of final base and detail images. Our method not only preserves the source information along edges but also reduces halo artifacts, pixel distortion. An experimental result reveals our method gives the better results compared with existing methods both qualitatively and visually.