Effective Preprocessing of Medical Images using Denoising Techniques

Abstract

Since the last few decades, image denoising is one of the most widely concentrated areas of research in the domain of image processing. A wide variety of denoising algorithms have been explored to date, but the problem of noise prevention in Magnetic Resonance Images is still a great barrier to the diagnosis and treatment of certain diseases. This paper mainly focuses on the study and analysis of different Denoising algorithms, the type of noise handled, and their efficiency. Preprocessing of medical images is considered one of the important steps that can enhance the accuracy in the prediction of various diseases. The presence of noise and other artifacts are believed to degrade the prediction accuracy which is the important metric that directs physicians to prolong further in providing clinical guidance to the patients. Most of the algorithms perform denoising in the complex domain. Deep learning-based Denoising algorithms are found to produce more promising results than traditional ones. However, the number of training samples and the training time are some limitations worth mentioning. Magnetic Resonance Images are sources of input for medical diagnosis of a variety of diseases. On removal of noise, these images can go a long way in the early diagnosis of numerous fatal diseases and can save lives. The predominant objective of this summary is to direct the researchers to choose prompt denoising techniques appropriate for their applications despite the available limitations in the same. This review is comprehended with the main aim of suggesting effective image denoising approaches that can go a long way in enhancing the quality of biomedical images.