Blockwise Robust Singular Value Decomposition for MRI Brain Image Denoising

Abstract

MRI is one of the high-dimensional high-throughput technologies that are playing a major role for diagnosing brain tumor. With the help of MRI images, brain tumor is diagnosed at advanced stages. Correct identification of brain tumor or abnormality, image processing in MRI of brain is highly essential that can reduce the chance of fatal stage. Occasionally, these MR images are introduced with noise during acquisition which reduces the image quality and limits the accuracy in diagnosis. Therefore, preliminary diagnosis of MRI brain images from the hospital may not be always reliable for further analysis because of the presence of noise. Reduction/Elimination of noise in medical images is an important task in preprocessing which is one of the previous crucial parts for further image analysis. Although, several denoising techniques are available in the literature including median filter, box filter, low pass filter, and high pass filter. In this paper, we have measured the influence of the above methods for denoising. We have also proposed a new blockwise robust singular value decomposition technique for denoising image. The results of our analysis showed that blockwise robust singular value decomposition technique gives the better performance compared to the other methods. Therefore, our recommendation is to use the Blockwise robust singular value decomposition technique for brain image denoising.