Speckle Noise Reduction in Ultrasound Images Using Various Denoising Techniques

An ultrasound image is generated using ultrasonic waves sent to specific body areas by a transducer. When hitting an object in the body, a large number of reflected ultrasonic waves with random phases interfere constructively and destructively, causing the formation of speckles in the image. Speckle is a granular multiplicative noise that can degrade the quality of the ultrasound image. Several techniques have been applied to remove speckles in ultrasound images, including transform domain filters. In this paper, speckle noise reduction was carried out by applying wavelet, contourlet, NSCT, and NSST techniques. The images used consist of speckled synthetic images and real US images. The performance of the denoising techniques was assessed on the quality of the denoised images and the SNR, NSP, EPI, ENL, and running time. The results reveal that NSST outperforms wavelet, contourlet, and NSCT in terms of reducing speckles in the image, preserving edges, and requiring a relatively short running time.