Fast gradient projection algorithm based Rician noise removal in MR images

摘要

MR images are vulnerable to Rician distributed noise due to contrast reducing signal dependent bias. This paper presents a combined gradient projection method with Gabor filter technique for removing the Rician noise present in MR images. The proposed method is based on the function of Gradient projection filter to identify the artefact and denoise the image. For demonstrating the performance of the proposed method, the Rician noise is added to the raw MR images and removed using the proposed method. Further, the proposed algorithm was tested on various brain MR images and the performance of the algorithm was studied by estimating the performance metrics and visual inspection, and compared with the existing denoising methods.