Restoration of MR Images using a hybrid model consisting of two stages is implemented. Application of Fourth Order PDE and the application of relaxed median filter are pre and post stages respectively. The results were calculated using standard parameters like PSNR, SSIM and RMSE. It was found that the hybrid mm gives acceptable results when compared with standard non-linear filters when applied to T1 weighted Thorax Images. For Brain MR Images, the hybrid model filter performs better for a finite range of input noise level (% N ). The algorithm was implemented on synthetic images of T1 weighted Thorax and Brain MRI using MATLAB 2015a. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.