Comparison of speckle reduction techniques in ultrasound images

In this study, filtering techniques which are used to reduce speckle noise are examined which is one of the most important problems of ultrasound images and a systematic approach is developed for performance evaluation. In this scope, Wiener, Lee, Frost and Median filters which are used to filter images are tested in details on simulated and real ultrasound images. The performances of these techniques are compared according to mean square error (MSE), peak signal to noise ratio (PSNR) and contrast to noise ratio (CNR) parameters. Different probe frequencies, active transducer numbers and noise variances which are the parameters that are used to create an ultrasound image are used during the simulation. Field II simulation tool is used to create simulated ultrasound images. Consequently, for image filtering; Lee and Median filters should be used for more experienced physicians, Wiener and Frost filters should be used for beginner physicians.