Human hemodynamics further unraveled: ultrafast ultrasound velocimetry for vascular flow imaging
Majorie van Helvert is a PhD student in the Department Multi-Modality Medical Imaging. (Co)Promotors are prof.dr. M. Versluis, prof.dr. M.M.P.J. Reijnen and dr. E. Groot Jebbink from the Faculty of Science & Technology.
The growing body of evidence on the relation between hemodynamics and the onset and progression of vascular disease, highlights the urge for accurate quantification of local blood flow patterns in a real-time, fast and non-invasive manner. Several novel ultrasound techniques, including high-frame-rate contrast-enhanced ultrasound particle image velocimetry (echoPIV), have emerged that overcome important drawbacks of currently available modalities like duplex ultrasound (DUS) and phase-contrast MRI (4-D flow MRI).
Over the course of this thesis the performance of echoPIV in the lab, in healthy subjects and in patients with different types of vascular disease was investigated. The accuracy of echoPIV to quantify femoral bifurcation flow characteristics in the absence and presence of a stenotic lesion was assessed in an experimental setting in Chapter 2. Here, optical PIV was used as the reference technique. An additional comparison to computational fluid dynamics (CFD) models based on echoPIV and DUS-derived flow boundary conditions was made. This validation study showed that echoPIV was able to measure high stenotic velocities up to 330 cm/s and accurately capture steep spatial gradients present in the post-stenotic jet. Furthermore, areas with complex stenotic flow disturbances were well identified. EchoPIV-derived hemodynamic parameters showed good agreement to optical PIV, underlining the potential of echoPIV to identify and quantify areas with disturbed blood flow. DUS-based CFD models presented larger deviations in measured velocities to optical PIV in, which is attributed to inherent limitations of conventional Doppler ultrasound, such as spectral broadening, resulting in increased flow velocity boundary conditions. However, flow velocity patterns and sites with adverse hemodynamic parameters were correctly identified by DUS-based CFD. While conventional Doppler ultrasound is relatively easy to apply in a clinical setting, this study demonstrated that echoPIV can enhance the accuracy of these models by its 2-D aspects, and conversely, echoPIV could be augmented by CFD to achieve insights in 3-D flow velocity patterns and vector-derived hemodynamic parameters.
In Chapter 3 and 4 the performance of echoPIV and native-blood speckle tracking (BST) was investigated in the femoral bifurcation of twenty healthy subjects. In Chapter 3, a comparison between both ultrasound vector flow imaging techniques revealed improved image quality in terms of contrast-to-background ratio in the BST recordings. This effect was not observed in the velocimetry tracking capabilities, where both BST and echoPIV performed similarly well. In addition, a comparison of BST and echoPIV to 4-D flow MRI, as a clinical reference, showed good overall agreement in spatiotemporal velocity profiles and peak systolic velocities. Altogether, this study demonstrated the validity of BST and echoPIV in healthy blood vessels and suggested that the addition of microbubble contrast is not always required in the superficially located femoral bifurcations. This would facilitate clinical translation as the measurements become non-invasive and more widely applicable since microbubble contra-indications do not have to be considered. However, it is difficult to a priori identify who benefits from microbubble contrast administrations and who does not as real-time feedback is currently not available. In Chapter 4, the potential of ultrasound vector flow imaging was further displayed as the previously obtained ultrasound recordings were used to provide detailed descriptions of femoral blood flow characteristics. Several hemodynamic parameters, including vector complexity (VC), vorticity and the number of identified recirculation zones, were obtained. A comparison between different age groups showed largest differences during the late diastolic phase where velocities decreased, and VC increased in the older subjects. This is potentially a result of vascular aging during which blood vessels become stiffer and less compliant. Furthermore, this study confirmed the presence of complex flow phenomena at sites with complex geometry, like inner walls of curvature and outer walls of bifurcations. Although deemed physiological by nature, the presence of these recirculation zones in healthy subjects could, together with other atherosclerotic risk factors, contribute to an atheroprone environment in the long-term.
In Chapter 5 to 7, the feasibility of echoPIV in patients with peripheral arterial disease (PAD) was assessed in three prospective clinical trials. EchoPIV was performed in the femoral artery of patients treated with a stent placement (Chapter 5), in the aorto-iliac artery of patients with a stenotic lesion treated with supervised exercise therapy (Chapter 6), and in the aorto-iliac artery of patients treated with a single stent or stent configuration deployment (Chapter 7). In approximately 90% of the measurements flow quantification with echoPIV was successfully achieved with either optimal feasibility, i.e. flow quantification in the entire imaged vessel and over the full cardiac cycle, or partial feasibility, i.e. flow quantification in parts of the imaged vessel and/or parts of the cardiac cycle. Several limiting issues were identified, including acoustical shadow regions due to arterial calcifications, decorrelation in the PIV algorithm due to high or out-of-plane blood flow velocities, contrast microbubble destruction due to prolonged insonification during diastolic slow or stagnant blood flow, and short vessel segment imaged due to tortuous and complex vessel geometries. Most of these issues were related to the diseased state of the blood vessel and expressed similarly among the different cohorts. Fortunately, the presence of stent material did not adversely affect image quality and PIV tracking performance, indicating the robustness of this modality in patients after endovascular treatment. Moreover, in two patients with optimal blood flow quantification VC and vorticity could be used to discriminate disturbed from undisturbed flow in a within-subject evaluation (Chapter 6).
In Chapter 9, the application of echoPIV in patients with an abdominal aortic aneurysm (AAA) before and after endovascular repair was tested. 2-D flow velocity patterns and temporal velocity profiles corresponded well with 4-D flow MRI datasets in the pre- and postoperative setting. EchoPIV was able to visualize and quantify the multi-scale nature of aneurysm flows including the dominant inflow jet accompanied by regions with recirculation and regions approximating blood flow stasis. The very low flow velocities, presence of bowel gas and increased imaging depth imposes new challenges in these patients. Post-treatment hemodynamics showed increased systolic velocities and decreased VC compared to the preoperative situation. This could be attributed to the change in geometry where the sudden increase in diameter has been replaced by a more or less straight tube which promotes laminar flow.
In conclusion, local hemodynamics play a crucial role in the onset and progression of vascular disease, yet traditional clinical modalities are unable to accurately visualize and quantify these cardiovascular flow features. This thesis investigated the applicability of echoPIV as a potential technique to overcome the existing challenges of conventional approaches. 2-D blood flow quantification with echoPIV was achieved in healthy subjects and a variety of patient cohorts with different types of vascular disease, both before and after treatment. Flow velocity patterns were acquired which provided detailed insights in the existence of complex and disturbed blood flow phenomena. The quantification of echoPIV-derived hemodynamic parameters yielded the differentiation of areas with flow disturbances from areas with laminar flow and highlighted differences before and after endovascular repair. These results underpin the robustness and the additional value of the echoPIV technology. Further development is required to address patient and technique related issues to allow for a wider and improved usability of the modality in the clinic. Many innovations, including artificial intelligence, can aid in clinical translation of echoPIV. If implemented, echoPIV, and cardiovascular flow imaging in general, can add flow-informed decisions to the existing geometry-informed modus operandi. Thus, echoPIV can have major implications for personalized monitoring, treatment and follow-up: a next step to patient-specific healthcare.
