Comparison of Filtering Techniques in Ultrasound Color Flow Imaging

Abstract

The article considers filtering techniques used to suppress clutter signals from moving tissues and to improve reliability of blood flow estimation. It compares polynomial and adaptive bases such as the result of empirical mode decomposition and singular vectors obtained through Karhunen−Loève transform. Filtering techniques are examined using a computer-simulated model, Doppler flow phantom and in vivo data. Filters are compared in terms of computational complexity, ability to retrieve flow profile without errors and through ROC curve analysis. Polynomial regression filters with tissue phase shift compensation were found to be the best fit for clutter suppression in terms of computational demands and accuracy of velocity estimation.