Ultrasound imaging systems have been an essential tool for medical diagnosis over last two decades, due its noninvasive nature, low cost, portability, real-time image formation, etc. However, one of the fundamental problems in ultrasound image is the poor image quality caused by speckle noise. The presence of speckle noise affects human interpretation of images or medical diagnosis.
This paper presents a novel approach for 2-dimensional B-mode ultrasound image enhancement in the sense of subjective quality. Our assumption is that although the frequency band of speckle covers a wide range, most annoying artifacts arise from the components in a certain frequency band. Hence, by determining the corresponding speckle frequency band and removing it with a notch filter, we can alleviate the speckle noise and get a fine and smooth speckle pattern. Through the wavelet domain process, which can handle both spatial and frequency domain information, we reduce the speckle noise efficiently while enhancing edge features. The final processing results are proved to show great improvement on subjective image quality.
