Wavelet-based ultrasound image denoising: Performance analysis and comparison

Abstract

Ultrasound images are generally affected by multiplicative speckle noise, which is mainly due to the coherent nature of the scattering phenomenon. Speckle noise filtering is thus a critical pre-processing step in medical ultrasound imaging provided that the diagnostic features of interest are not lost. A comparative study of the performance of alternative wavelet based ultrasound image denoising methods is presented in this article. In particular, the contourlet and curvelet techniques with dual tree complex and real and double density wavelet transform denoising methods were applied to real ultrasound images and results were quantitatively compared. The results show that curvelet-based method performs superior as compared to other methods and can effectively reduce most of the speckle noise content of a given image.

Visual Improvement for Hepatic Abscess Sonogram by Segmentation after Curvelet Denoising

“…Noisy -Snr: is the noisy image's signal to noise ratio, some papers add noise values dependent on its variance [6] and [8] or its standard deviation like [12] and [13].…”

“…The Curvelets idea is to represent a curve as a superposition of functions of various lengths and widths obeying the scaling law width ≈ length2 [8]. More than one from the recent researches assures that image denoising in Curvelet Transform domain is the best denoising due to the ability of Curvelet to recover signals in different directions [4], [5], [6], [7], [8], [9], [10], [11], [12], [13].…”

“…Speckle noise filtering of medical ultrasound images (denoising) is important, providing clinicians with enhanced diagnostic ability [6]. More than one from the most recent and modern Researches assured that the best way for ultrasound image denoising is denoising in Curvelet transform domain due to the ability of Curvelet Transform to recover signals in different directions as compared with other methods [4], [5], [6], [7], [8], [9], [10], [11], [12], [13]. The first mention of Curvelet transform was by E. J. Candµes and D. L. Donoho in 1999 [16].…”

A Wavelet-Based Approach for Ultrasound Image Restoration

“…The Curvelets is to represent a curve as a superposition of functions of various lengths and widths [11]. Curvelet transforms gave close and improved delineation to edges [14].…”

“…We used Matlab-R2011b and [33] for calculating PSNR. PSNR and RMSE measures are used in more papers which are about sonograms' denoising [11].…”