Denoising DWI based on regularized filter

The Rician noise introduced into the diffusion weighted images (DWI) can bring serious impacts on tensor calculation and fiber tracking. To decrease the effects of the Rician noise, many anisotropic diffusion denoising methods have been presented. Among all these methods, the P&M filtering method is most popular. Although efficient in alleviating the effects of the noise, the classical P&M filter has shortcomings such as instability and ill-posedness. In this paper, a regularized anisotropic diffusion filter was presented and applied to restore the DWI. The presented filtering strategy displayed well posedness and good preservation of edges. To evaluate its efficiency in accounting for the Rician noise, the PSNR and MSSIM metrics were used for the first time. The results acquired from the synthetic and real data proved the better performance of the presented filters.