Spatio-temporal speckle reduction in ultrasound sequences

Abstract

In this paper we will propose a novel variational framework for speckle removal in ultrasound images. Our method combines efficiently a fidelity to data term adapted to the Rayleigh distribution of the speckle and a novel spatio- temporal smoothness constraint. The regularization relies on a non parametric image model that describes the observed image structure and express inter-dependencies between pixels in space and time. The interaction between pixels is determined through the definition of new measure of similarity between them to better reflect image content. To compute this similarity measure, we take into consideration the spatial aspect as well as the temporal one. Experiments were carried on both synthetic and real data and the results show the potential of our method.