Application of Bilateral Filters for Denoising Rician Noise in MRI Images

Medical imaging reveals the internal structures of the human body. One such imaging modality is Magnetic Resonance imaging, which reveals the soft tissue characteristics of human body. MRI can also detect tumors, hemorrhages and infections in human brain. MRI images are more prone to Rician noise. Thus, preprocessing of MRI brain images must be done to diagnose the brain diseases. The MRI brain images without Rician noise and the images with Rician noise of various noise levels were used in this experiment. The main aim of this work is to eliminate the Rician noise present in MRI brain images and to protect the edges as well as the important details in the brain image. Bilateral filter eliminates the Rician noise by taking weighted average on nearby pixels. In this research work, a comparative study of three such bilateral filters called Standard Bilateral Filter (SBF), Robust Bilateral Filter (RBF) and Weighted Bilateral Filter (WBF) was carried out. The effectiveness of the Bilateral filters was analyzed by means of Peak Signal to Noise Ratio (PSNR) and various other performance metrics. The experimental results show that the Weighted Bilateral Filter excels the other two Bilateral filters for many values of standard deviation. However, for certain other values of standard deviation, the Robust Bilateral Filter also marks a good performance.