7th Dutch Bio-Medical Engineering Conference
January 24th & 25th 2019, Egmond aan Zee, the Netherlands
10:30   Arteries
Chair: Alina van der Giessen
15 mins
Adaptive Velocity Compouding for Blood Vector Velocity Imaging in Carotid Arteries
Anne Saris, Rik Hansen, Stein Fekkes, Jan Menssen, Maartje Nillesen, Chris de Korte
Abstract: Atherosclerosis is the primary cause of ischemic heart failure and stroke in westernized societies. It starts with the formation of lipid-rich accumulations in the vessel wall, which can grow and mature into plaques. These plaques can become unstable and rupture which can result in a stroke or myocardial infraction. It is known that certain arterial segments are more susceptible to plaque development [1, 2]. For example, at inner curves of vessels, downstream of a stenosis and at the outer walls of bifurcations, disturbed flow occurs. These regions are characterized by low wall shear stress (WSS) and they are more prone to atherosclerosis [1, 2]. For early detection of atherosclerosis, visualization and quantification of complex flow and the resulting WSS seems important. Conventional B-mode and Doppler ultrasound imaging allow visualization of plaque geometry and estimation of blood velocities. However, complex flow patterns cannot be visualized and quantified using the conventional techniques. This limitation originates from the angle dependency of these techniques, where only the velocity component along the US beam direction is estimated. With the aim of capturing the complex flow patterns that are associated with plaque development and progression, we developed a technique for the quantification of 2D blood velocities in the carotid artery. We performed 8000 angled plane wave ultrasound acquisitions (-20° and 20°) per second using a 7.8 MHz linear array transducer connected to a Verasonics Vantage 256 research ultrasound system. Adaptive clutter filtering was performed, where the velocity cut-off point was set dynamically based on the tracked vessel wall velocity. Inter-frame displacements were estimated using a 2-step cross-correlation-based algorithm. Subsequently, 2D blood velocities were obtained by either compounding axial displacement estimates obtained at both angles, or by projecting the angled estimates obtained at only one angle. This is decided adaptively, were RF signal power and velocity variance thresholding are used to determine the quality is the angled displacement estimates. The performance of the method was evaluated experimentally using a straight tube flow setup. Thereafter, initial in vivo evaluation was performed in healthy carotid arteries (n = 2), early-stage stenosed arteries (n = 1) and arteries which underwent endarterectomy with stent placement (n = 2). Straight vessel experiments demonstrated the technique performed with a maximum velocity magnitude bias of -3.7% (2.8% standard deviation) and an angle bias of -0.16° (0.41° standard deviation). In vivo, complex flow patterns were visualized in both healthy and diseased carotid arteries and quantified using a vector complexity measure that increased with increasing wall irregularity. This measure is a first example of new potentially clinical relevant parameters which might aid in early detection of atherosclerosis. REFERENCES [1] SS Dhawan, et al., Expert Rev Cardiovasc Ther, 8:545-56, 2010 [2] CJ Slager, et al., Nat Clin Pract Cardiovasc Med, 2: 456-64, 2005
15 mins
Novel Approach for Material Properties Estimation of Atherosclerotic Plaque Tissue
Ronald van den Berg, Stéphane Avril, Frank Gijsen, Ali Akyildiz
Abstract: Fatal cardiovascular events are majorly caused by atherosclerotic plaque rupture, a biomechanical event that occurs when local plaque stresses exceed strength. Plaque stresses can be assessed by computational (finite element, FE) models and accurate plaque material parameters are a must for accurate plaque stress computations[1]. Current techniques for material parameter estimation, such as inverse FE modelling[2], have large computational costs (hours or days). This study aims to develop and validate an approach to much faster obtain plaque properties by employing the virtual fields method (VFM)[3]. VFM employs the virtual work principle and utilizes full-field deformation measurements of the investigated structure. Energy balance equations, obtained by prescribing kinematically admissible virtual fields, are solved for unknown material stiffness parameters. In this study, firstly full-field displacement maps of atherosclerotic plaque for physiological intraluminal pressure were computed with three FE plaque models, based on histological slides of human carotids. Computed full-field deformation maps were then used as input for the developed VFM approach. VFM-estimated plaque stiffness values were finally compared against the ground truth values prescribed in FE models to validate the approach. In the single-component cases of an idealized and a realistic plaque; VFM-estimated plaque stiffness was <1% different from the ground-truth value(200 kPa). For more complex, three component plaque case; soft and calcified region stiffness values were 13% and 3% off from ground-truth(10&1000 kPa, respectively), yet plaque stiffness was still <1% off. Even when a mean noise of 15% was added to the deformation input, the developed VFM approach satisfactorily estimated the plaque stiffness, where the difference was only 5%. VFM estimation procedures took <30 seconds. In this study, VFM was used for atherosclerotic plaques for the first time. Study results demonstrate the potential of VFM for plaque stiffness estimation if full-field deformation data is available. The short computational time required and the possibility to use it with clinically measured deformation data make the developed method a great potential method of assessing plaque tissue stiffness in-vivo. Near-future plan is employing the developed VFM approach on existing ultrasound full-field deformation data from ex-vivo inflation experiments with atherosclerotic porcine iliac arteries and human coronaries[2].
15 mins
Carotid Plaque Composition Differentiation in Asymptomatic Individuals Using Compound Ultrasound Strain Imaging Validated by Magnetic Resonance Imaging
Hendrik Hansen, Inge van den Munckhof, Eveline van der Kolk, Anton Meijer, Marinette van der Graaf, Joost Rutten, Chris de Korte
Abstract: The rupture of vulnerable plaques in the carotid artery is a main cause of transient ischemic attacks and stroke. To detect atherosclerotic plaques with a rupture prone composition, dedicated ultrasound strain imaging techniques have been developed to accurately and noninvasively measure deformations/strains of the carotid artery wall induced by the pulsating blood flow. Multiple studies have confirmed the relation between strain and plaque composition. However, most studies applied strain imaging methods that only allow imaging in longitudinal cross-sections, implying that they only allow imaging of plaques in the top or bottom part of the wall. This study for the first time investigates if strains obtained with a strain imaging technique designed for transverse planes correlate with MRI–based plaque composition in an asymptomatic population. Ultrasound radiofrequency (RF) data of 51 carotid plaques from 31 asymptomatic participants with an increased cardiovascular risk profile were recorded by an experienced sonographer in a dedicated three-angle (0°, +20° and -20°) focused acquisition mode using a Samsung Medison Accuvix V10 ultrasound system equipped with an L5-13IS transducer. Additionally, MR angiography and high resolution MR vessel wall data were obtained using a Siemens 3.0 T MAGNETOM Skyra MRI scanner (Erlangen, Germany) with a dedicated 4-channel Machnet Phased Array Carotid Coil (Eelde, the Netherlands). Principal radially-oriented strains were derived from the ultrasound RF data using our previously published compound strain imaging approach and cumulated over the diastolic phase. The resulting strain map was normalized with respect to pulse pressure measured with a sphygmomanometer. Two strain parameters were calculated from the strain map values in those parts of the circumference where the wall was at least 2 mm thick: 1) the median value, and 2) the percentage of strain values exceeding 0.5% strain. We studied the performance of both strain parameters for differentiating plaques classified as fatty, haemorrhagic, calcified, or aspecific/fibrous by an experienced neuroradiologist based on the MRI data and blinded to the ultrasound data. Strain parameters of the plaques with a composition associated with vulnerability, i.e. fatty (n=9) and haemorrhagic plaques (n=5), were significantly higher than the strain parameters of plaques with stable composition, i.e. calcified (n=14) and aspecific/fibrous plaques (n=23) (Mann-Whitney test, P<0.05). Overall, both strain parameters were highest for fatty plaques compared to the other plaques, although no significant differences between the individual groups were detected (Kruskall-Wallis test P<0.05). Median strain and percentage strain >0.5% for the vulnerable plaques were 0.95% and 53.4% versus -0.02% and 47.1% for the stable plaques, respectively. Thus, even in this small asymptomatic population, compound strain imaging discriminated between plaques with vulnerable and stable features.
15 mins
High-Risk Coronary Plaque Development Is Related to Multi-Directional Shear Stress in an Atherosclerotic Pig Model
Ayla Hoogendoorn, Annette Kok, Eline Hartman, Suze-Anne Korteland, Lorena Casadonte, Adriaan Coenen, Frank Gijsen, Dirk Jan Duncker, Antonius van der Steen, Jolanda Wentzel
Abstract: Background – Although time averaged shear stress (SS) is established for its involvement in atherosclerotic plaque development, not much information is available on the influence of multi-directionality of shear stress. We investigated the influence of five different multidirectional SS parameters on plaque development and composition. Therefore, we used an adult porcine model of atherosclerosis in which serial imaging was performed and histology was used to assess plaque growth and composition in the coronary arteries. Methods – 10 adult familial hypercholesterolemic swine were put on a high fat diet for 12 months and underwent 3-vessel CT-scan, near-infrared spectroscopy intravascular ultrasound (NIRS-IVUS) imaging and flow measurements at 3 time points. An anatomically optimized 3D geometry of the arterial lumen at baseline, in combination with local flow measurements, was used for calculation of the SS metrics (time-averaged wall shear stress (TAWSS), oscillatory shear index (OSI), relative residence time (RRT), trans-wall shear stress (transWSS) and cross-flow index (CFI)). For the analysis, the arteries were divided into 3mm - 45° sectors (n=3627) which were split into low, mid or high levels of each SS metric per individual artery. IVUS and histology were used to correlate plaque thickness changes and composition to baseline SS levels. Results – Half of the pigs developed large, lumen intruding plaques and in these fast responders, IVUS-based plaque growth was significantly more pronounced in areas of low TAWSS and high OSI/RRT/CFI (p<0.05). In the slow responding animals no relationship between plaque growth and the five SS metrics was observed. In the fast responders, the most advanced plaque type (fibrous cap atheroma (FCA)) was located at higher absolute levels of OSI and CFI compared to less advanced plaques (p<0.05). These results were supported by a more detailed histological analysis, showing that plaques with the highest lipid, macrophage and necrotic core area were exposed to low TAWSS and high OSI and RRT (p<0.05). CFI and transWSS appeared not to influence plaque composition (p=NS). Conclusions – The data derived from this study support the role of multidirectional shear stress in the development of coronary atherosclerosis with regard to plaque size and composition.
15 mins
Towards in Vivo Photoacoustic Imaging of Vulnerable Plaque in Carotid Artery
Min Wu, Richard Lopata
Abstract: The rupture of the vulnerable carotid plaque is the major cause of stroke. Currently in clinics, the diagnosis and treatment of stroke is generally based on the US characterized plaque morphology (lumen stenosis), resulting in overtreatment. It is reported that 9 symptomatic or even 19 asymptomatic patients need to be operated on to prevent a single stroke in a time span of 5 years. This indicates a strong unmet clinical need for a more effective method to improve the assessment of stroke risk for the individual patient. PA/US imaging can provide comprehensive (both morphology and chemical composition) and patient-specific assessment of plaque vulnerability, being a useful tool to better assess the plaque vulnerability and prevent overtreatment. However, in practice, PAUS imaging suffers from low PA SNR mainly due to limited light permissible exposure under some safety regulations. Therefore, it is quite a challenge to translate PAUS imaging into clinics. In this paper, a preclinical study of PAUS imaging of carotid plaque is conducted using a fast PAUS imaging system (about kHz laser pulse repetition rate and a fully integrated, hand hold imaging probe). A motion corrected averaging algorithm for in vivo PA imaging is developed. The high capability of the integrated imaging system and the proposed averaging algorithm is validated by a designed ex vivo PAUS experiment, which is performed to mimic the in vivo intra-operative PAUS imaging of the human carotid plaque. The results may accelerate the clinical translation of PAUS imaging of carotid plaque.
15 mins
Detection of Haemorrhages Using a Single Wavelength and Singular Value Decomposition
Roy van Hees, Frans van de Vosse, Richard Lopata
Abstract: The presence of haemorrhages in the carotid artery is a major risk factor for stroke. The superficial location of the carotid artery makes it very suitable for photoacoustic imaging (PAI). By using multiple wavelengths, it is possible to detect haemorrhages by searching for sites with a high ratio of deoxyhaemoglobin / oxyhaemoglobin. This approach requires at least two wavelengths of light: one wavelength with a high response for both oxy- and deoxyhaemoglobin to estimate the total blood volume and one wavelength with a large difference in response of both constituents. An approach using a single wavelength would lower the requirements of the laser system used and therefore allow for more compact, handheld photoacoustic systems, for the detection of haemorrhages. Aim of this study is to develop a method to detect haemorrhages in the carotid artery using a single wavelength approach combined with novel flow measurement techniques. Conventional flow measurement techniques such as pulsed wave Doppler and colour flow imaging employ a so called “wall filter” to filter out wall motion, as to only measure blood flow. These wall filters employ temporal filtering and although successful at filtering out wall motion, these filters also discard the flow resulting from microvasculature. To properly detect this flow, which is essential for the reliable detection of haemorrhages using a single wavelength approach, a filtering method employing singular value detection is used instead, as described by Charlie Demené et al . Preliminary tests using polyvinyl alcohol gel phantoms, with a lumen with a diameter of 3 mm, a minichannel containing a haemorrhage with a 1 mm diameter and microchannels of a 0.25 mm diameter show that wall filtering using singular value decomposition is a viable approach. Photoacoustic measurements are still ongoing.