Despeckling Ultrasound Thyroid Images

A Hybrid Approach Utilizing Directional Sticks and Partial Differential Equations

Abstract

Speckle is a type of noise that occurs in ultrasonic imaging that negatively impacts the clarity of the image. This noise can make it challenging for physicians to accurately diagnose conditions. Speckle filtering is an essential step in the preprocessing stage to ensure that valuable information is not lost. So the search for despeckling filters is a difficult task based on their performances and statistics. In this paper, we evaluated different denoising algorithms based on their statistical parameters, like mean, MSE, PSNR, ENL, SSI, SMPI and elapsed time. Based on the above parameters, the filter that uses local statistical parameters as a design performs better than the filters discussed.