6th Dutch Bio-Medical Engineering Conference
26 & 27 January 2017, Egmond aan Zee, The Netherlands
10:30   Ultrasound Imaging I
Chair: Yuan Yang
15 mins
Hendrik Hansen, Mathieu Pernot, Simon Chatelin, Mickael Tanter
Abstract: Rupture of atherosclerotic plaques in carotid arteries is a main cause of stroke. Plaques with a large soft lipid-rich core are more prone to rupture than predominantly stiff fibrous-rich plaques. Detecting the presence of lipids in plaques noninvasively and fast remains challenging. Ultrasound shear wave elastography (SWE) is a fast and noninvasive technique which was developed to differentiate between benign (softer) and malignant (stiffer) breast lesions [1]. In SWE a shear wave is induced in tissue using a focused ultrasonic push. The local propagation speed of this wave is measured by tracking the wave propagation through tissue using a series of unfocussed high frame rate (>10.000 frames/s) ultrasound acquisitions. The speed of the shear wave is directly related to the Young’s modulus of the tissue. Recent studies have shown that the technique can also be applied to detect locally softer (lipid-rich) regions in phantoms [2]. However, these studies only report imaging in a longitudinal imaging plane. In this plane, it is not possible to detect lipids over the full vessel circumference. Therefore, we investigated the feasibility of SWE in transverse imaging planes. A SuperSonic Imagine Aixplorer ultrasound system equipped with a linear array transducer was used to induce shear waves in vessel-mimicking polyvinylalcohol phantoms of varying dimensions and shear moduli using a single focused ultrasonic push (300 µs). The push was followed by plane wave acquisitions (n = 100, PRF ≥ 10 kHz). The raw element data of the separate frames were cross-correlated (1D) to determine wall motion caused by the generated waves in the vessel wall. By determining the moment of maximum wall motion as a function of circumferential position for circular paths with varying radii we were able to derive the velocity of the shear waves over the vessel wall. Wave front propagation at a nearly continuous speed was observed for all phantoms as a function of circumferential position. For homogeneous phantoms, one wave was observed, whereas a slow and a fast wave were observed in two-layered phantoms containing a soft plaque in a harder vessel wall suggesting that soft lipid-rich regions can be detected. Finally, we also tested the technique in vivo in a volunteer using a Verasonics ultrasound system which confirmed that creation of circumferentially propagating waves is also possible in vivo. In conclusion, shear wave elastography in transverse imaging planes is feasible and promising for lipid-core detection in plaques of superficial arteries. REFERENCES [1] J. Bercoff, M. Tanter, and M. Fink, "Supersonic shear imaging: a new technique for soft tissue elasticity mapping," IEEE Trans Ultrason Ferroelectr Freq Control, Vol. 51, pp. 396-409, (2004). [2] E. Widman, et al., "Shear wave elastography plaque characterization with mechanical testing validation: a phantom study," Phys Med Biol, Vol. 60, pp. 3151-74, (2015).
15 mins
Ines Beekers, Tom van Rooij, Martin Verweij, Michel Versluis, Nico de Jong, Sebastiaan Trietsch, Klazina Kooiman
Abstract: Ultrasound insonification of gas filled microbubbles can be used to locally enhance drug delivery [1]. To study the underlying mechanisms, the ideal in vitro endothelial cell model includes 3D cell culture and flow. Therefore, we propose to use the elaborate microfluidic channel structure of the OrganoPlate™ to study ultrasound-mediated drug delivery [2]. We assessed the feasibility of acoustic transmission into the OrganoPlate™ by numerical modeling and experimentally studying microbubble behavior. A 3D finite element model of the OrganoPlate™ was created in PZFlex to simulate ultrasound insonification. The pressure amplitude and frequency spectrum transmitted into the microchannels were better preserved for a channel width of 400 μm than for 200 μm. Microbubbles [3] (3 to 8 μm in diameter) were introduced in the OrganoPlate™ and insonified with an ultrasound transducer (20 kPa, 1 to 4 MHz, 8 cycles) while recorded at 17 Mfps using the Brandaris 128 ultra-high-speed camera. The oscillation behavior of the microbubbles was studied and compared to measurements performed in an acoustically transparent OptiCell [4]. The data revealed larger microbubble oscillation amplitudes in the OrganoPlate™ from 1 to 1.6 MHz, resulting from pressure amplification in the microchannels. The measured transmission validates the finite element model, with a RMS deviation of 2.5 dB. The homogeneous acoustic pressure transmitted into the OrganoPlate™ demonstrates its potential to develop an ideal in vitro model to study ultrasound-mediated drug delivery. REFERENCES [1] K. Kooiman, et al., Adv. Drug Deliv. Rev., vol. 72, pp. 28–48, 2014. [2] S. J. Trietsch, et al., Lab Chip, vol. 13, pp. 3548–3554, 2013. [3] K. Kooiman, et al., Eur. J. Lipid Sci. Technol., vol. 116, pp. 1217–1227, 2014. [4] T. van Rooij, et al., Ultrasound Med. Biol., vol. 41, pp. 1432–1445, 2015.
15 mins
Niels Petterson, Emiel van Disseldorp, Frans van de Vosse, Marc van Sambeek, Richard Lopata
Abstract: Combining ultrasound (US) measurements with finite element models (FEMs) is a novel technique to assess mechanical material parameters of vasculature. An example is the estimation of wall stiffness (elastography) in abdominal aortic aneurysms (AAAs). The stiffness of the FEM is calibrated iteratively until the displacements match those measured with US. The abdominal aorta (AA) is located very close to the lower spine and is surrounded by different organs, which have a large influence on the deformation of the wall, leading to possible misinterpretation of the strain data. Most methods do not consider any surrounding tissue when modelling the AA(A), which may lead to inaccurate elastography results. In this study, 4D echography of AA(A)s was used for 3D US speckle tracking. Next, image-based FEMs of increasing complexity were used to quantify the influence of surrounding tissue and the spine on the displacement. Results were compared with speckle tracking. A Philips IU22 system, equipped with an X6-1 transducer, was used to acquire 4D DICOM data of the AA in five healthy volunteers and five AAA patients. The AA(A)s were segmented manually and 3D speckle tracking was performed to acquire wall displacements from diastole to systole. Each geometry was used to generate three FEMs. The first model consisted of the segmented AA(A) (GAA = 100 kPa; GAAA = 900 kPa). The second model included surrounding tissue (G = 20 kPa). Finally, a generic spine (G = 0.9 GPa) was added, modelled as a rod (r = 20 mm), 5 mm beneath the AA(A). After estimation of initial stresses at diastole (p = 80 mmHg), the model was inflated to systole (p = 120 mmHg). Displacements of the US speckle tracking were interpolated on the nodal coordinates of the FEM and compared to the results of each simulation by calculating absolute differences. Both the inclusion of surrounding tissue and the spine improved the accuracy of the displacements predicted by the FEM. Especially the displacement of the lower wall in the depth-direction improved. The displacement error in the upper wall did not improve that much, possibly due to limitations of the surrounding model, e.g. not including the presence of inhomogeneous tissue and probe pressure. Future work should include the comparison of material parameters, and the replacement of the generic spine with a realistic geometry.
15 mins
Gijs Hendriks, Chuan Chen, Hendrik Hansen, Chris de Korte
Abstract: Background: Ultrasound imaging is used for breast cancer detection especially in women with dense breast in which mammography shows a reduced sensitivity by the higher amount of glandular tissue. Since hand-held ultrasound is operator dependent, the automated breast volume scanner (ABVS) was introduced that consists of a linear array transducer that is translated motor-controlled over the breast while collecting ultrasound data to reconstruct a volumetric breast image. Although clinical studies show high sensitivity, clinicians report high recall-rates due to the detection of many lesions of uncertain malignant potential. Compared to benign lesions, malignant lesions are often stiffer, and more grown into the surrounding tissue (firmly bonded) resulting in decreased strains inside, and shear strain around the lesion respectively. Therefore, we propose to estimate the strain and maximal shear strain induced in the breast from multiple ABVS ultrasound recordings while slightly compressing (1 to 4%) the breast with the transducer. This strain imaging technique can then be used as add-on to increase the specificity of the ABVS. Aim: The aim of this study was to show that it is possible to implement 3-D ultrasound strain imaging on an ABVS-like system and illustrate its feasibility for lesion stiffness and bonding detection. Methods: A breast phantom containing two lesions (three times stiffer compared to surrounding tissue, 10 mm diameter, loosely and firmly bonded) was scanned by an ABVS-mimicking device that consisted of an L12-5 transducer mounted on a translational stage connected to a Verasonics V1 research ultrasound system. First, volumetric ultrasound radiofrequency (RF) data of the phantom were collected with a slight deformation by acquiring 2D ultrasound data while translating the transducer over an ultrasound-permeable membrane of 15x20 cm2 covering the breast phantom (speed 40 mm/s). Next, the acquisition was repeated after lowering the transducer by 0.5 mm, providing volumetric ultrasound data of the phantom in a compressed state. In total the acquisitions took 10 sec, which implies they can be performed within one breath-hold. Strains and maximal shear strains were estimated by 3-D cross-correlation of the collected pre- and post-deformation RF-data followed by least-square strain estimation. Contrast-to-noise (CNR) and signal-to-noise ratios (SNR) were calculated to illustrate the sensitivity of the technique. Results: A 3-D strain volume of the breast phantom was obtained and the lesions showed strain values 3 times lower compared to the environment, which matches perfectly with the phantom’s stiffness characteristics. The lesion stiffness was detected with an SNR and CNR of 15 and 33 dB, respectively. Around the loosely and firmly bonded lesion, the maximal shear strain values were increased by 1% (SNR 8dB; CNR 9dB) and 3% (SNR 12dB; CNR 14dB) respectively compared to the surrounding tissue. Conclusion: It was feasible to implement quasi-static strain imaging in an ABVS-like system. The technique was able to detect lesion stiffness and bonding in a breast phantom with a high SNR and CNR.
15 mins
Anne Saris, Hendrik Hansen, Stein Fekkes, Maartje Nillesen, Marcel Rutten, Chris de Korte
Abstract: Ultrasound imaging is often used in medical practice for blood velocity estimation and visualization in a wide range of clinical settings. Conventional Doppler-based techniques available on commercial systems face difficulties in capturing the true velocity, since they only estimate the velocity component along the ultrasound beam instead of the 3D velocity vector. This induces an angular dependency on the estimates, which is one the major limitations of current blood velocity estimation techniques. High frame rate speckle tracking (ST), using plane waves transmitted at pulse repetition frequencies (PRF) in the order of kHz, has shown potential for true 2D (and 3D) blood velocity estimation. However, two major challenges remain: the reduced image quality, especially in lateral direction, due to the use of unfocused transmissions, and signal dropouts caused by clutter filtering. Suppressing clutter signal is an inevitable step in estimating the blood velocity. For high beam-to-flow angles and low blood velocities however, signal from blood will also be suppressed, causing signal dropouts and consequently less accurate ST results. Methods utilizing multi-angle PW acquisitions have been proposed to overcome both issues [1, 2]. This study compares the performance of displacement compounding, and zero-degree speckle tracking (0° ST) for 2D blood velocity estimation in simulated and in vivo carotid artery (CA) flow fields. Multi-angle ultrasound RF data (0°, -20° and 20°) of a sophisticated model [3] of the CA (velocity range: 0-1.5 m/s) were simulated at a PRF of 12 kHz to compare the performance of the methods. Data from a healthy CA were acquired with a Verasonics Vantage research ultrasound system to verify the difference in performance in vivo. Simulations showed mean RMSE values for the ensemble averaged axial and lateral velocity estimates at diastole of 10.7 and 22.9 cm/s for 0° ST and 4.0 and 7.7 cm/s for displacement compounding. The compounding method was less affected by clutter filtering. Besides, the use of solely axial information, thereby circumventing the use of the lower resolution lateral information, showed to be beneficial. The performance difference was present over the full cardiac cycle and was supported by the in vivo results. Both simulation and in vivo results show the feasibility of using plane wave imaging for the estimation of the 2D blood velocity field in the CA. The proposed technique also contributes to the assessment of short-lived vortices and complex flow patterns. REFERENCES [1] Fadnes et al., IEEE TUFFC, vol. 62, pp. 1757-67, 2015 [2] Hansen et al., J Biomech, vol. 47, pp. 815-23, 2014 [3] Beulen et al., J of Fluids And Structures, vol. 25, pp. 954-966, 2009
15 mins
Maartje Nillesen, Anne Saris, Hendrik Hansen, Hendrik Hansen, Stein Fekkes, Frebus van Slochteren, Peter Bovendeerd, Chris de Korte
Abstract: Strain imaging techniques can provide quantitative information on the local contractile function of the myocardium. In particular in cardiac phases with high deformation rates, accurate strain assessment remains challenging. Conventional focused ultrasound imaging has limited temporal resolution because of the line-by-line image acquisition. Ultrafast ultrasound imaging can be of great advantage for accurate radio-frequency (RF)-based strain assessment over the entire cardiac cycle. Spherical (diverging) waves are designated for ultrafast cardiac imaging because of the large field of view and imaging depth required in cardiac applications. Multiple unfocused spherical wave transmissions can be combined to improve image quality while preserving high frame rates. A multi-zone RF-based displacement estimation algorithm in combination with a multiple spherical wave transmission scheme is proposed. The technique was evaluated in a cardiac model and in vivo. A transmit sequence of 5 spherical waves as defined to simulate apical and short axis views of a 3D model of a healthy left ventricle [1]. The single transmits (PRF 5 kHz), originating from different positions behind a phased array transducer, were combined using coherent compounding. High frame rate open chest cardiac ultrasound data of a pig were acquired with the same transmission scheme using a Verasonics V-1 system. Axial and lateral displacements were determined using a 2D cross-correlation technique [2]. The grid for beamforming of the RF data was adapted stepwise over imaging depth because of the depth dependent lateral resolution to get more robust displacement estimates. An effective frame rate of 250 Hz was utilized whereas lateral displacement estimation was performed at a 5 times lower frame rate (50 Hz) to obtain more robust estimates. The tissue was tracked over time to determine cumulative displacements and strains. Good agreement was observed between the cumulative displacement estimates and reference displacements derived from the model over the cardiac cycle (axial: RMSE < 1.2 mm, lateral: RMSE < 1.3 mm). In vivo maximum longitudinal strain was around 20% which is in correspondence with clinical studies. These results confirm that cardiac strain imaging at high frame rates using multiple spherical waves is feasible and brings high frame rate full 3D cardiac strain imaging closer. REFERENCES [1] P.H.M. Bovendeerd, W. Kroon, T. Delhaas, “Determinants of left ventricular shear strain”, Am J Physiol Heart Circ Physiol, Vol. 297, pp. 1058-1068 (2009). [2] R.G.P. Lopata, M.M. Nillesen, H.H.G. Hansen, I.H. Gerrits, J.M. Thijssen and C.L. de Korte. “Performance of two dimensional displacement and strain estimation techniques using a phased array transducer”, Ultrasound Med Biol., Vol. 35, pp. 2031-2041, (2009).