An Integrated De-Speckling Approach for Medical Ultrasound Images Based on Wavelet and Trilateral Filter

Abstract

Speckle noise is a granular artifact in medical ultrasound images. It causes serious problems and hinders the development of automatic diagnosis technology. A novel and integrated approach is proposed to address the de-noising problem by using wavelet transformation and trilateral filter. Firstly, a dynamic additive model is developed to account for the medical ultrasound signal with speckle noise.Secondly, in accordance with the statistical property of the additive model, an adaptive wavelet shrinkage algorithm is applied to the noisy medical signal. Particularly, the algorithm is significant to the high-frequency component of the speckle noise in the wavelet domain. Thirdly, but most importantly, the low-frequency component of the speckle noise is suppressed by a trilateral filter. It simultaneously reduces the speckle and impulse noise in real set data. Finally, a lot of experiments are conducted on both synthetic images and real clinical ultrasound images for authenticity. Experimental results show the proposed method not only guarantees noise reduction, but also performswell on edge sharpening and offers greater flexibility. Compared with other existingmethods, experimental results show that the proposed algorithm demonstrates an excellent de-noising performance, offers great flexibility and substantially sharpens the desirable edge.