A Study On MRI Denoising Based On NLM

MR image is often degenerated due to some complex noise dependent with the original image in imaging processing. This noise with Rician distribution has a big adverse effect on the medical analysis and diagnosis based on MR images. There may be an inaccurate medical analysis result, wrong diagnosis or conclusion because of the poor image segmentation and reconstruction with the existence of the Rician noise. Removing the medical image noises effectively plays an important role in medical research and drawing a scientific conclusion on MR images.In this paper, we propose a new MRI denoising method base on improved NLM algorithm, which we combine Canny detection with NLM algorithm. This new algorithm makes up for the defect that only consider a single pixel similarity variable in images and reserve much more details in the same way. What’s more, we take the NLM as the pre-filtering in Canny edge detection for the enhancement of the detection accuracy.Through a series of experiments on Brain Web Database data set, we draw a conclusion that the proposed NLM algorithm based on the improved Canny edge detection has a significant effect on denoising the T1-weighted image which contains Rician noise. The experimental results show that the proposed methods are better than the existing NLM method and its improved method in denoising MRI with Rician noise.