An Improved Anisotropic Diffusion Filtering Approach for noise reduction in MRI
Nikita Joshi, Sarika Jain · 2021
A confined, non-invasive and accurate diagnosis procedure is preferred for investigation of numerous human body disorders. Magnetic Resonance (MR) Imaging is one such crucial and popular diagnostic procedure and therefore these MR images need to be noise free to reduce the chances of error in diagnosis. An improved and novel approach has been suggested to denoise MR images by utilizing unbiased non local means (UNLM) method and anisotropic diffusion method. The proposed method's performance has been compared with other denoising techniques and the results were evaluated on metrices like Peak Signal to Noise Ratio (PSNR), Mean Square Error (MSE) and Structure Similarity Index (SSIM) values. The proposed method produced successful results and the process of anisotropic diffusion was hence improved. The MR images were well denoised and fine structures were also preserved along with the edges.