A Study on Image Noise Removal Techniques for Magnetic Resonance Imaging

Abstract

Medical image denoising is very important and challenging area in the field of image processing. MRI (Magnetic resonance imaging) is very popular and most effective imaging technique. During the acquisition, MR images affected by random noise can be modeled as Gaussian or Rician distribution. In last few decades, so many denoising techniques were proposed but there are some limitations with the algorithms, because image edges and fine details are need to preserve. So there is need of compromise between de-noising quality and edge preservation to use images for real time application. Computation time is also very important parameter to implement the algorithms. In this paper, we have done an overview of different denoising algorithm. It observes that NLM (Non-Local Means) filter is much better than other existing state of art methods. Here study is done for enhancement of NLM to improve the performance. Results of different algorithms show that PCA (Principal Component Analysis) based algorithm with NLM performs much better in both quantitative and qualitative manner.