Medical Image Denoising Techniques against Hazardous Noises: An IQA Metrics Based Comparative Analysis

Abstract

Medical imaging has become a vital part of the early detection, diagnosis, and treatment of many diseases. That’s why image denoising is considered as a crucial pre-processing step in medical imaging to restore the original image from its noisy circumstance without losing image features, such as edges, corners, and other sharp structures. Ultrasound (US), Computed Tomography (CT), and Magnetic Resonance (MR) are the most widely used medical imaging techniques that are often corrupted by hazardous noises, namely, speckle, salt and pepper, Poisson, and Gaussian. To remove noises from medical images, researchers have proposed several denoising methods. Each method has its assumptions, merits, and demerits. In this paper, a detailed comparative analysis of different denoising filtering techniques, for example, median, Wiener, mean, hybrid median, Gaussian, bilateral, non-local means, and anisotropic diffusion are performed based on four widely-used image quality assessment (IQA) metrics, such as Root Mean Squared Error (RMSE), Peak Signal to Noise Ratio (PSNR), Mean Absolute Error (MAE), and Structural Similarity Index (SSIM). The results obtained in this present work reveal that Gaussian, median, anisotropic diffusion, and nonlocal means filtering methods perform extraordinarily to denoise speckle, salt and pepper, Poisson, and Gaussian noises, respectively, from all US, CT, and MR images.