NLM based magnetic resonance image denoising – A review

Abstract Denoising Magnetic Resonance (MR) image is a challenging task. These images usually comprise more features and structural details when compared to other types of images. These structural details in MR images provide additional information to physicians for better diagnoses and hence there is a need to preserve these details. Over the past few years, various MR image denoising techniques have been evolved. Among them, the techniques based on Non-Local Means (NLM) have achieved excellent performance by exploiting similarity and/or sparseness among the patches. The evolution of NLM filter has changed the paradigm of research in the area of MR imaging. Many variants of NLM algorithms have been developed till today which in addition to retaining the edge/structural features, improve the signal to noise ratio and computational efficiency. The aim of this paper is to provide an exhaustive review of the published literature on NLM based MR image denoising techniques. A critical review and discussion on the advantages and limitations of these techniques are provided with quantitative result analysis.