My research focus is the development of ultrasonic techniques for non-invasive imaging and flow visualization, primarily to elucidate the connections between vascular tissue changes and local hemodynamics. The development and progression of vascular disease involves an inherent feedback loop whereby local hemodynamics (e.g. oscillating shear) affect tissue changes (e.g. plaque development), which in turn affect the hemodynamics (e.g. turbulence). Hence, to investigate the role of hemodynamics in vascular disease, it is important to be able to map out and characterize the local hemodynamics relative to the tissue geometry and composition (available through various imaging modalities, incl. ultrasound), as well as follow any correlated tissue and hemodynamic changes.
During my PhD work, I designed and implemented a unique in vitro vascular test facility (Poepping et al. 2002) incorporating anthropomorphic flow phantoms (Poepping et al. 2004) to investigate the hemodynamic characteristics associated with the carotid artery bifurcation and elucidate their relationship with various Doppler ultrasound (DUS) spectral characteristics. I demonstrated the potential of DUS for revealing regions of clinically relevant hemodynamic factors, such as turbulence and recirculation, with high temporal and spatial resolution, through the 4-D (time-varying 3-D) visualization of unique DUS spectral parameters, such as turbulence intensity (Fig. 1) and spectral broadening index (Poepping et al. 2001). Images obtained using digital particle imaging (DPI) within the same anatomical models presented an independent confirmation of the flow patterns observed using the Doppler ultrasound data, and were compared with the 3-D velocity patterns obtained using computational fluid dynamics (CFD) (Steinman et al. 2000), thus also providing an indirect validation of the CFD simulations. Direct comparisons of flow velocities from CFD and DUS are difficult because of factors associated with the DUS technique. Namely, a finite sample volume and transducer aperture will result in a distribution of velocities corresponding to each given measurement site, unlike for CFD. A recent collaborative paper addresses these issues, by combining DUS modeling with CFD velocity fields, towards the development of a novel virtual DUS tool (Koshniat et al. 2004). Such work provides insight into the complex interaction between the ultrasound acoustics, signal processing, local hemodynamics, and possible clinical interpretation, under controlled conditions.
Collectively, these results have demonstrated the different flow patterns that arise from arterial stenoses (lumen constrictions) of different severity and symmetry (eccentricity), including distinction of post-stenotic recirculation zones, helical flow, and regions of elevated wall shear stress. While clinical evaluation of atherosclerotic stenoses focuses on maximum stenosis severity, and corresponding peak flow velocities, the above results indicated that other factors merit serious evaluation, such as the affect of stenosis eccentricity on turbulence and shear stress. DUS offers a means to obtain such information.
For the development of DUS techniques with clinical potential, anatomically realistic models are essential test objects in order to produce realistic spatial and temporal flow variations, mimicking complex in vivo flow, within a single test object. To enable DUS in vitro studies in more realistic anthropomorphic geometries, I have developed a new gel-elastomer tissue-mimicking material (TMM), during my post-doctoral work. The material is both acoustically well matched to tissue and sufficiently strong to withstand physiologically realistic stresses. Additionally, this material can be produced with varying elasticity, which will enable simulation of progressing arterial stiffness due to vascular disease (Poepping et al., Tissue Elasticity Conference, 2004; British patent pending).
For in vivo measurements, distortion of the DUS beam, due to acoustic speed inhomogeneities of overlying lesions, is a concern. In recent collaborative efforts towards investigating the cause of an US imaging artefact, known as the refractile or edge shadowing artefact, ensemble-averaged US B-mode images of lesion-mimicking phantoms were compared with numerical modelling of ultrasound image formation (Steel et al. 2004). This work demonstrated the distortion of the beam at a boundary with an acoustic speed contrast between a fluid-filled cavity and the surrounding TMM, adding insight into the relative contributions of refraction, critical-angle effects, and propagation along the cavity-medium interface. While this study was directed at an imaging artefact, overlying lesions or plaques will distort the DUS beam in similar ways, which we are now able to investigate using numerical modelling.
Fig. 1: Central plane slice from a colour-encoded map of the turbulence intensity at peak systole in models with varying stenosis severity and eccentricity. Turbulence intensity is the variation in the mean velocity, derived here from the same time point within 10 consecutive cardiac cycles. Increased turbulence intensity reflects a greater fluctuation about the mean velocity and hence is an indication of disturbed or turbulent flow.