Research On Total Variational Denoising Algorithm For MR Images

Magnetic resonance imaging (MRI) diagnosis is an important medical image diagnosis means like CT, isotope scanning and ultrasound image to facilitate manual or computer-aids analysis. MRI is one of the most active technologies in medical imaging because of its high resolution, non-invasiveness. MRI, in particular, can produce images from human body non-invasively that reveal the structure, metabolism and function of internal organs or tissues. Currently, magnetic resonance imaging technology has become an important method in brain function and anatomy research field. Some important information of MRI images may be submerged due to the presence of noise. Therefore, to filter the interference brought by body thermal noise during imaging procedure, it is necessary to research the de-noise method. It is important for MRI images to be preprocessed before being used for medical analysis and application.Until now, a lot of common de-noise methods have been proposed, including median filter, wiener filter and histogram based filters. The quality of MRI images is improved to some extent, and to satisfy some determinate applications. However, these methods do not solve the defects of the MRI medical images radically. Presently wavelet analysis and the method based on PDE of image de-noising attract attention. But the two de-noising approaches have the tradeoff between de-noising and preserve the edge of the image. After processing edges of images are blurred. Image de-noising based on the anisotropic diffusion filtering can preserve the edge of the image. However, the text??characteristics of MRI images is weakened. As a result, it is difficult for doctor to diagnose different disease using blurred images.In order to solve these problems, a novel texture preserving variational de-noising method for MR images based on the use of adaptive regularization is proposed in this paper .This method combined the advantages of TV algorithm and iterative regularization algorithm based on Bregman distance. The new adaptive regularization method based Total Variational de-noising algorithm uses an adaptive fidelity term which locally controls the extent of de-nosing over image regions according to the gradient information of each pixel. The regularization parameters obtained from the novel aigorithm are variable while traditional algorithms are invariable. So important information, like edge and texture are preserved. The numerical results for de-nosing show improvement in the signal-to-noise ratio (SNR) and execution time over standard model processes, and they are more appealing visually.At first, the present state of research on image de-noising is introduced in this paper. We describe the wide variety of medical image de-noising methods including median filter, wiener filter and histogram based filters and their applications. Then we make a description of theoretical knowledge, characteristics and necessary condition about TV algorithm and iterative regularization algorithm based on Bregman distance in great detail. In chapter 4, we introduce the adaptive regularization method based total variational de-noising algorithm. Chapter 5 briefly summarizes the major contribution of the current work and provides some suggestions for the future research.