Published Paper Details:

In this manuscript, for the removal of Rician noises from MR images a fourth order PDE (partial differential equation) based nonlinear filter is proposed. The filter implemented contains two terms firstly, the data fidelity term and secondly, prior function term. By minimizing the negative log likelihood of Rician probability density function (pdf) we get the fidelity term. A prior function is introduced because the solution of the first term has ill-posed nature. This prior function is a nonlinear anisotropic diffusion based filter. To maintain the trade off balance between these two terms a regularization parameter is introduced. In terms of PSNR at various noise levels the comparative study and performance analysis of the suggested model with other standard methods is presented for Brain Web dataset. From the results obtained, it is clear that the method implemented is performing better noise removal as compared to other methods.