Ultrasound elastography is a promising imaging modality for the differentiation between benign and malignant tissue, such as the detection of stiff tumors in the (female) breast. In order to deduce the elastic tissue properties and reconstruct the spatial distribution of Young's modulus (E-modulus), the inverse problem governed by the equilibrium equations of linear elastostatics must be solved using internal tissue displacement estimates which are, in practice, subject to spatially non-stationary measurement errors.
In this paper we investigate the novel E-modulus reconstruction approach of taking into account the spatially non-stationary errors of ultrasonic displacement estimates within the field-of-view. The application of spatially adaptive weight factors derived from the mean-square displacement estimation errors by means of an appropriate confidence measure leads to an improved reconstruction quality that strikes the best balance between the two opposing reconstruction goals of "achieving high image homogeneity" and "keeping high-frequency spatial information", which both are diagnostically important. We demonstrate that over- and under-regularization within the field-of-view can be significantly reduced leading to an improved image quality. The results presented here are derived from extensive simulations and phantom experiments. The simulation results will be compared to those of an earlier study.