Speckle Noise Removal in Ultrasound Images with Stationary Wavelet Transform and Canny Operator

Speckle noise is random multiplicative noise which inhibits the segmentation and recognition of fine details or textures in ultrasound images. In this paper, a method for despeckling of medical ultrasound images is proposed, which is based on stationary wavelet thresholding. Canny edge detection is applied to the low frequency sub-bands to adjust the denoising strategy in the vicinity strong edges. An adaptive NeighShrink algorithm is used for thresholding the wavelet coefficients of the high frequency sub-bands of an image decomposed by stationary wavelet transform. The combination of Canny edge detection and the adaptive NeighShrink effectively preserve edges and removes noise. The method is compared against some popular methods on de-noising the synthetically speckled images and clinical ultrasound images. The experimental results show that our proposed method results in better performance under most conditions over comparator algorithms in terms of Peak Signal to Noise Ratio, Edge Preservation Factor and 2D cross correlation, and is consistently superior by these metrics at higher noise levels.