3D medical images denoising

The two-dimensional images are often insufficient to achieve a perfect diagnosis in the medical area; on the other hand the three-dimensional images allow having an interesting deductive vision and there denoising has become a necessity and an essential need. Many methods have been proposed to reduce noise and to preserve edge, are usually used for 2D images and they have been extended to 3D data. The Non-Local Means filter has become the most popular one for denoising medical images, based on a weighted average of voxels inside a search window. In this work, we present a new method in the field of 3D image denoising based on combination between Non-local Means filters and the diffusion tensor. Our proposed algorithm is to normalize the weight average by adding iteratively an anisotropic diffusion stencil. The performance and efficiency of the algorithm are estimated by calculating various quality metrics and compared with other methods.