Signal processing in diagnostic ultrasound: Algorithms for real-time estimation and visualization of blood flow velocity

Ultrasound Color Flow Imaging (CFI) has become a valuable tool in a wide rangeof medical applications where information about blood flow can be related to thediagnosis of disease, as for instance in the cardiovascular system. The modalityprovides a map of blood velocity and direction in a twoor three-dimensional regionof interest, where abnormal blood flow patterns can be detected and investigated.The work presented in this thesis is devoted to the development of CFI signalprocessing for improved estimation and visualization of blood velocity. The thesisconsists of three technical contributions, and one chapter describing preliminaryclinical and experimental results of using one of the methods described. The differentcontributions are written in article form and can be read individually. A thoroughbackground chapter is also included to introduce the unfamiliar reader to conceptsand challenges present in ultrasound imaging and CFI specifically.Two thesis chapters address the problem of separating the weak signal from bloodin CFI, typically dominated by signal from surrounding tissue. In chapter three,an adaptive filter approach for the removal of this clutter signal prior to velocityestimation is described. An adaptive filter algorithm suitable for real-time performanceis developed, and shown to provide satisfactory results even in excessive tissue clutterconditions. In chapter four, a different approach for dealing with the clutter signalis described. Assuming the statistical properties of the clutter signal known, weanalyze the properties of maximum likelihood estimation (MLE) of blood velocity,compared to conventional methods. We further address issues related to the practicalimplementation of this model-based estimation scheme.A technique for the visualization of the two-dimensional blood velocity vector haspreviously been introduced. In chapter five, the real-time implementation of thistechnique, called Blood Flow Imaging (BFI), is described and evaluated. The methodis not limited by angle-dependency or velocity aliasing as conventional CFI, and isshown to have potential within different imaging contexts. Finally in chapter six,clinical and experimental pilot studies are described where the potential of the BFImodality has been investigated. It is shown that BFI can provide a more detailedand intuitive image of flow conditions, that can be beneficial in both vascular andcardiac applications when the blood flow direction plays a major role. Throughthe investigations, practical restrictions and potential improvements of the currentimplementation have also been mapped.