- Email: [email protected]
Vector Flow Visualization of Urinary Flow Dynamics in a Bladder Outlet Obstruction Model
Ultrasound in Med. & Biol., Vol. -, No. -, pp. 1–10, 2017 Ó 2017 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/$ - see front matter
http://dx.doi.org/10.1016/j.ultrasmedbio.2017.07.006
d
Original Contribution VECTOR FLOW VISUALIZATION OF URINARY FLOW DYNAMICS IN A BLADDER OUTLET OBSTRUCTION MODEL TAKURO ISHII, BILLY Y. S. YIU, and ALFRED C. H. YU Schlegel Research Institute for Aging and Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON, Canada (Received 24 December 2016; revised 9 June 2017; in final form 8 July 2017)
Abstract—Voiding dysfunction that results from bladder outlet (BO) obstruction is known to alter significantly the dynamics of urine passage through the urinary tract. To non-invasively image this phenomenon on a time-resolved basis, we pursued the first application of a recently developed flow visualization technique called vector projectile imaging (VPI) that can track the spatiotemporal dynamics of flow vector fields at a frame rate of 10,000 fps (based on plane wave excitation and least-squares Doppler vector estimation principles). For this investigation, we designed a new anthropomorphic urethral tract phantom to reconstruct urinary flow dynamics under controlled conditions (300 mm H2O inlet pressure and atmospheric outlet pressure). Both a normal model and a diseased model with BO obstruction were developed for experimentation. VPI cine loops were derived from these urinary flow phantoms. Results show that VPI is capable of depicting differences in the flow dynamics of normal and diseased urinary tracts. In the case with BO obstruction, VPI depicted the presence of BO flow jet and vortices in the prostatic urethra. The corresponding spatial-maximum flow velocity magnitude was estimated to be 2.43 m/s, and it is significantly faster than that for the normal model (1.52 m/s) and is in line with values derived from computational fluid dynamics simulations. Overall, this investigation demonstrates the feasibility of using vector flow visualization techniques to non-invasively examine internal flow characteristics related to voiding dysfunction in the urethral tract. (E-mail: [email protected]) Ó 2017 World Federation for Ultrasound in Medicine & Biology. Key Words: Urinary flow, Voiding dysfunction, Urinary tract phantom, Vector flow visualization, Doppler ultrasound, High frame rate imaging.
INTRODUCTION
help urologists determine whether such treatment is necessary, it is essential to first assess the severity of voiding dysfunction. Yet, it is not trivial to perform this diagnosis non-invasively. Techniques such as uroflowmetry (Schafer et al. 2002) and ultrasound-based bladder volume measurement (Stevens 2005) can merely provide overall indications of voiding dysfunction, and they lack the provision of insight into the anatomic and urodynamic factors that deter the urine passage process. To extend beyond these solutions, it is necessary to devise strategies that can monitor urinary flow dynamics in LUTS patients before and after an intervention (in addition to evaluating the shape of urinary tract). Medical imaging is undoubtedly a potential solution to such a need. Among existing imaging modalities, some are capable of visualizing the flow of biofluids inside the human body, such as magnetic resonance imaging (MRI) (Stalder et al. 2008; Sughimoto et al. 2016) and optical imaging (Kurata et al. 2015; White et al. 2003). However, most of these imaging techniques have rarely been applied to urinary flow visualization. In the case
Urethral voiding dysfunction that leads to weak urinary flow, urine dribbling and urine splitting is well considered as a major category of lower urinary tract symptoms (LUTS). This problem is mainly attributed to mechanical and physiological aging of the vesico-urethral system. Despite being benign in nature, urethral voiding dysfunction would hamper a patient’s quality of life on a chronic basis, and thus it is increasingly regarded as a healthcare concern in today’s aging society. In recent years, we have seen a growing interest in the potential to devise appropriate strategies for addressing urethral voiding dysfunction (and LUTS in general), such as pharmacological and surgical interventions that aim to alter the internal urethral shape during micturition (Tubaro et al. 2015). To
Address correspondence to: Alfred C.H. Yu, Schlegel Research Institute for Aging and Department of Electrical and Computer Engineering, University of Waterloo, 200 University Avenue West, Waterloo, ON N2L 3G1, Canada. E-mail: [email protected] 1
2
Ultrasound in Medicine and Biology
of MRI, its temporal resolution is inadequate to track fast urinary flow (can be .2 m/s velocity), and logistically it is challenging to manage a patient’s micturition process inside the MRI scanner bore; in the case of optical imaging, the penetration depth is insufficient (Weiss et al. 1989) for the purpose of non-invasively imaging the urinary tract. Alternatively, it is possible to leverage computational fluid dynamics (CFD) to render flow profiles based on anatomic information obtained from MRI or computed tomography (Torii et al. 2007). Alas, proper development of a CFD model for urinary flow dynamics is after all not straightforward because of the need to account for the predominance of autonomic nerve activity, as well as dynamic changes in the size of the urinary tract, which can expand to 10 mm in diameter during peak micturition. In contrast, ultrasound imaging is perhaps better suited for urinary flow visualization. Previous work by others have indeed shown the feasibility of using ultrasound to observe the urinary tract and its related flow dynamics (Arif et al. 2014, 2015; Ding et al. 2000; Gratzke et al. 2015; Ozawa et al. 2010). Nevertheless, these published investigations were all conducted using conventional ultrasound technology (on the basis of scanline–based imaging principles) that typically renders flow information in two formats: (i) full-view color flow images at video-range frame rates (20–30 fps) (Ding et al. 2000; Ozawa et al. 2010), or (ii) M-mode or radiofrequency data plots over a single scanline (Arif et al. 2014, 2015). They fell short of providing timeresolved rendering of urinary flow dynamics across the entire imaging view, nor can they offer information about the multi-directional nature of urinary flow patterns. To overcome the temporal resolution limitation, one can possibly develop high-frame-rate ultrasound imaging methods that are based on plane wave excitation (Lu et al. 2006; Montaldo et al. 2009) and synthetic aperture principles (Jensen et al. 2006; Karaman et al. 1995). As for the need to visualize multi-directional urinary flow, vector flow estimation techniques can be devised (Jensen et al. 2016a, 2016b). Nevertheless, neither of these concepts has been applied to urinary flow diagnostics. In this paper, we report the first application of a highframe-rate ultrasound vector flow visualization technique called vector projectile imaging (VPI) to facilitate analysis of urinary flow dynamics. It is our intent to investigate the feasibility of leveraging VPI to derive diagnostically relevant insight into the fluid mechanics within the urinary tract, particularly in LUTS cases with urethral voiding dysfunction. To perform this investigation, we have designed a novel anatomically realistic urinary flow phantom that can resemble urine passage under controlled conditions. Note that VPI is a novel tech-
Volume -, Number -, 2017
nique developed by our team earlier (Yiu et al. 2014), and it is based on multi-angle plane wave excitation and least-squares Doppler vector estimation principles (Yiu and Yu 2016). We have initially applied VPI to achieve time-resolved visualization of carotid hemodynamics. Here, we seek to expand the clinical application domain of VPI by translating this technique toward use in urinary flow diagnostics.
METHODS AND MATERIALS Male lower urinary tract phantom design Background description. Bladder outlet (BO) obstruction was chosen as the LUTS disease model in this study. Well regarded as a common lesion in the posterior urethra causing voiding symptoms, BO obstruction is estimated to affect 21.5% of the worldwide population (Irwin et al. 2011). It refers to the condition where sclerosis and narrowing of the BO are both observed during micturition. Its emergence is often attributed to (i) lateral-lobe prostate hypertrophy or (ii) urethral mucosa hyperplasia as a consequence of local inflammation. To our knowledge, the flow dynamics in urinary tracts with BO obstruction have not been visualized using noninvasive medical imaging techniques. From an anatomic standpoint, BO obstruction occurs in the upstream part of the male lower urinary tract, which is illustrated in Figure 1a. Note that the urinary tract length can be as long as 20 cm (Kohler et al. 2008). Its prostatic and membranous segments are located inside the body, while the bulbous segment is located within the penis. In the presence of BO obstruction, urethral turbulence is expected to emerge. Such turbulence would lead to excessive loss of fluid energy and in turn cause voiding dysfunction symptoms (Ishii et al. 2014a; Yamanishi et al. 2000). Model features and design justifications. Urinary tract models with and without BO obstruction were designed to facilitate comparative analysis of VPI’s urinary flow visualization performance in healthy and pathological urinary flow scenarios. The overall geometry of the urinary tract is illustrated in Figure 1c. It was drafted using SolidWorks (Dassault Systemes, Waltham, MA, USA) in the form of a spline surface, as generated by stacking together multiple cross sections that were defined from five key anatomic features of the urethra: (i) bladder outlet (BO); (ii) proximal prostatic urethras (PU); (iii) verumontanum (V); (iv) urethral sphincter (S); and (v) bulbous urethra (BU). For each crosssectional plane, its shape and dimensions were set in accordance with a recent study on CFD-based urinary flow visualization (Ishii et al. 2014b).
Vector flow visualization of urine passage d T. ISHII et al.
3
Fig. 1. Design of the urinary tract. (a) Anatomic features of the lower urinary tract. (b) Endoscopic views of BO. Median bar is shown in the BO obstruction image. (c) Outline of urinary tract of the disease model. (d) Shape of the cross sections; contour colors correspond to labels in (c). BO 5 bladder outlet; BU 5 bulbous urethra; PU 5 proximal prostatic urethras; S 5 urethral sphincter; V 5 verumontanum.
In our urinary tract model, the BO cross section was drafted as a triangle-like closed curve. This design choice was made on the basis of cystourethroscopic imagery and anatomic knowledge of the bladder trigon (Ishii et al. 2014b). In the case with BO obstruction, the bladder outlet shape included a median bar: the characteristic anatomic feature for this pathology (see Figure 1b). The vertical diameter of the bladder outlet was set to 2 mm and 3.5 mm, respectively in the BO obstruction geometry and the healthy geometry, in line with a previous report (Tojo et al. 1994). Other anatomic features of our urinary tract model were drafted as ellipsoids of different size. Specifically, vertical diameters of 5.6, 6.0, 4.0 and 6.0 mm were respectively defined for proximal prostatic urethra, verumontanum, urethral sphincter and bulbous urethra, as shown in Figure 1d. The same diameter values were used for both the healthy and diseased urethra geometry in order to focus on investigating the impact of BO obstruction on urinary flow dynamics. Consistent with published data (Humphrey 2014; McNeal 1981), the length of prostatic, membranous and bulbous urethra was respectively set to 30, 20 and 30 mm. Also, the prostatic urethra was angled at 35 . The angle between inlet and outlet was assumed to be 90 in this study. For each cross-sectional curvature, four points were marked at 0 (dorsal side), 130 , 180 (ventral side) and 260 angles. This process was done so that the
four guide spline curves (associated with the marked points of same degree in each cross section) can collectively define the final shape of the urinary tract model. It is worth noting that the urinary tract was assumed to be an expanded rigid object. Such an assumption is valid for our purpose, because our experimental aim is to test the efficacy of using VPI to track urinary flow dynamics during peak micturition when the flow profile is generally constant. Phantom fabrication protocol. A core-filled replicate of the drafted urinary tract geometries was built with the use of 3-D printing technologies to facilitate phantom construction using a wall-less investment casting protocol that was reported earlier (Ho et al. 2017). For this process, the casting box used to mount the urinary tract replicate was developed such that it incorporated a designated dorsal side located at 20 mm away from the posterior urethra. Note that these distances are aligned with typical values in the human male body (Humphrey 2014; McNeal 1981; Prada et al. 2007). After the urinary tract replicate was mounted onto the casting box, a tissue-mimicking solution was slowly poured into the box cavity. This solution was based on the same polyvinyl-alcohol–containing formula published elsewhere (Ho et al. 2017). Also kept the same were the freeze-thaw process required to crystalize the tissue mimic and the chloroform immersion procedure needed to remove the urinary tract replicate.
4
Ultrasound in Medicine and Biology
Flow circuit setup. The flow circuit that was used to drive the urinary tract phantom was set up as shown in Figure 2. The inlet of the fabricated phantom was connected to a cylinder (50 mm in diameter) that mimicked a pressure-driven boundary condition. The phantom outlet, which corresponded to the excretion of urine mimicking fluid, was connected to an unpressurized buffer tank through a 20-cm long, 8-mm diameter flexible tube. In each instance of fluid passage through the urinary tract, the excreted fluid would be relayed from the buffer tank back to the inlet cylinder by using a gear pump that operated at constant flow mode (PQ12; Greylor Co., Cape Coral, FL, USA). The pump’s flow rate was tuned to keep the fluid level in the inlet cylinder to exert 300-mm,H2O head pressure at the phantom inlet. The urinary mimicking fluid used in this work was composed of distilled water and 50 mg/L of silicon carbide scatterers that had a mean size of 37 mm (357391; Sigma-Aldrich, St. Louis, MO, USA). Following previous reports by others (Arif et al. 2015), the viscosity, density and scatterer concentration of this fluid mixture were aligned with that for actual urine. In each experimental run, the rate at which this fluid flowed through the urinary phantom was found via dividing the measured outflow volume by the time elapsed.
Fig. 2. Schematic of the experimental setting and probe locations. The urinary tract was fixed as posture in the body. Instead of bladder, a cylinder was connected to the phantom. To generate static pressure boundary condition at bladder outlet, fluid was continuously supplied from outlet tank to keep the water level of the cylinder. Dimensions in the figure are in millimeters.
Volume -, Number -, 2017
Ultrasound visualization of urinary flow dynamics Overview of imaging framework. Time-resolved visualization of urinary flow dynamics was achieved with the use of the VPI technique that we established previously (Yiu et al. 2014). In brief, this imaging innovation is based on the use of plane-wave excitation principles to achieve high imaging frame rates that are well beyond the video display range. It derives local flow vector information (with axial and lateral velocity components) at every pixel position in the entire imaging view through a multiangle, least-squares vector Doppler estimation methodology (Yiu and Yu 2016), and these flow vectors were visualized through a dynamic rendering algorithm that depicts the trajectory of individual vector projectiles. Note that, to combat against slow-time aliasing artifacts, a phase-unwrapping algorithm has been incorporated in the VPI signal processing chain (Yiu et al. 2014). Transducer placement. During imaging data acquisition, an L14-5 linear array transducer (Analogic Ultrasound, Peabody, MA, USA) was placed on the dorsal side of the urinary flow phantom such that the transducer aperture was parallel to the urethral tract and the imaging view encompassed the BO. The corresponding field of view was similar to that obtainable from a trans-rectal scan (Mihmanli et al. 2006). This scan configuration was adopted so that the urinary tract was not too deep from the probe surface (2030 mm). Also, the insonation angles used for multi-angle Doppler flow vector estimation would be within an 80 –100 range, where the detected frequency shifts would not be aliased. Imaging system and parameters. The L14-5 transducer was connected to a research-purpose ultrasound imaging platform (SonixTouch; Analogic Corporation, Peabody, MA, USA), and it was driven according to the imaging parameters specified in Table 1. As illustrated in Figure 3, in a typical VPI imaging experiment, two data acquisition sequences were executed. First, to obtain a background B-mode image of adequate quality, unfocused plane wave pulses were sequentially fired from 24 steering angles (in equal increments over the angle span 210 to 10 ). Second, for high-frame-rate vector estimation purposes, non-steered plane wave pulses (0 ) were repeatedly fired at 10-kHz pulse repetition frequency (PRF) for a 2-s duration. Effectively, the raw acquisition frame rate of VPI cine loops is 10,000 fps. For each transmit firing event, raw pre-beamformed data were acquired from all array channels through a channel data acquisition system (Cheung et al. 2012). As explained in our earlier work (Yiu et al. 2014), the acquired data sets were processed offline via a high-throughput beamforming and flow estimation platform that was based on graphics processing unit (GPU) technology (So et al. 2011; Yiu et al.
Vector flow visualization of urine passage d T. ISHII et al.
Table 1. Hardware specification and imaging parameters Parameter
VPI
B-mode
a. Imaging platform Number of Tx/RX channels 128 Array pitch (mm) 0.3048 Pre-beamformed data sampling 40 rate (MHz) Pre-beamformed data resolution 12 (bit) b. Data acquisition Imaging frequency (MHz) 5 Tx pulse duration (number of 3 2 pulses) Number of Tx steering angle 1 24 0 212 21 & 112 Tx steering angle ( ) Pulse repetition frequency (kHz) 10 Effective data acquisition rate 10 0.416 (kHz) Maximum imaging depth (mm) 60 Data acquisition duration (s) 2 0.1 c. Beamforming Pre-beamformed data filter 5 passband (MHz) Filter design method Equiripple Lateral and depth pixel dimension 0.1 (mm/pixel) d. Slow-time data processing Normalized clutter filter cut-off 0.05 Filter design method Equiripple Sliding window for flow estimation 64 Sliding window step size 4 e. VPI visualization Nominal frame rate (fps) 2476 Launch point density (%) 3 Mean projectile lifetime (frames) 30 VPI 5 vector projectile imaging.
2011) and a Matlab interface (v. 2016a, Mathworks Inc., Natick, MA, USA). Using this platform, beamforming was performed from three receive steering angles (210 , 0 , 10 ), so as to realize a one-transmit, three-receive multi-angle configuration for least-squares Doppler vector estimation (Yiu and Yu 2016). Accordingly, B-mode images, VPI cine loops, and representative still frames were obtained. For reference, Doppler spectrograms were also derived at pixel positions of interest. These supplementary data were derived by performing short-time Fourier transforms on the slow-time ensemble of the identified pixel position, similar to our previous work (Yiu et al. 2014). Computational fluid dynamics simulations To facilitate comparative analysis, CFD simulations based on the Navier-Stokes equation were also carried out in this investigation using SolidWorks Flow Simulation software (v. 2015, Dassault Systemes). Briefly, based on the same urinary tract geometries as those used for phantom construction, the CFD solver computed the Navier-Stokes equation for incompressible fluid with the k-ε turbulence model by the finite volume method (Versteeg and Malalasekera 2007). As boundary
5
conditions, we respectively specified a constant inlet (BO cross-section) pressure of 104,267 Pa (300 mm H2O) and a constant outlet pressure of 101,325 Pa. Fluid properties were set based on the viscosity and density of water at 37 C temperature, which are respectively 0.685 3 1023 Pa$s and 992.48 kg/m3. For computation stability, the outlet length was extended by ten times of the outlet diameter, so that the urinary tract geometry used for the CFD simulations was essentially equivalent to the experimental phantoms that we fabricated. Other computational routines, such as spatial discretization, were the same as that described in our previous study (Ishii et al. 2014b). RESULTS VPI visualization of urinary flow in normal and diseased models VPI was found to be capable of depicting differences in the flow dynamics of normal and diseased urinary tracts. This observation is summarized in Video 1 (playback rate: 50 fps), which renders the urinary flow dynamics for a diseased model with BO obstruction (left panel) and a normal model (right panel). For comparative analysis, the bottom portion of Video 1 shows the Doppler spectrogram at the pixel position that corresponds to the intersection between the vertical line and horizontal line drawn in the VPI video frame (i.e., at the distal end of prostatic urethra). A key result to be noted is that, in the diseased model, flow jetting can be observed at the BO because of its anatomic narrowing. The flow velocity is consequently faster than that at the BO of the normal model. This trend is well summarized in the still-frame VPI snapshots shown in Figure 4. Another important observation is that vortices are present in the proximal and distal parts of the prostatic urethra as urine flowed past the obstructed BO in the diseased model. Because of the sustained presence of these vortices, jittering of flow trajectories can also be observed in the vicinity of the proximal prostatic urethra. These vortices and flow path jittering are not present in the VPI cine loop for the normal model. Such a trend is consistent with the Doppler spectrogram reference data, which showed that at the distal end of the prostatic urethra, the detected Doppler frequencies only spanned a limited range as indicative of constant flow speed. It is in contrast with the diseased model’s corresponding Doppler spectrogram that showed the presence of a broad range of positive and negative Doppler frequencies as indicative of chaotic flow patterns. Comparative analysis of flow patterns in the urinary flow models The flow vectors rendered by VPI were found to be consistent with CFD computation results. This finding is
6
Ultrasound in Medicine and Biology
Volume -, Number -, 2017
Fig. 3. Schematic of duplex VPI algorithm. Two sequences of pulse waves were insonated and stored separately through a DAQ tool. Steered pulse wave data were processed to generate a sequence of high-resolution B-mode images, while VPI was processed with second sequence RF data. L1-AC and L-S denote lag-one autocorrelation and least squares algorithm, respectively. DAQ 5 data acquisition; RF 5 radiofrequency; VPI 5 vector projectile imaging.
Fig. 4. Urine flow behavior visualized by VPI. Still-frame snapshots are provided for diseased model with BO obstruction (left) and normal urinary tract phantom (right). For both models, a Doppler spectrogram is shown as reference for a pixel position marked by the intersection between vertical and horizontal lines in the VPI snapshot image. BO 5 bladder outlet; VPI 5 vector projectile imaging.
Vector flow visualization of urine passage d T. ISHII et al.
7
Fig. 5. Comparison of magnitude flow velocity between VPI and CFD. Diseased model with BO obstruction (left); normal urinary tract phantom (right). Results are shown in a duplex format that depicts flow velocity magnitude as color codes and flow trajectories as streamlines. CFD 5 computational fluid dynamics; VPI 5 vector projectile imaging.
illustrated in Figure 5. The time-averaged flow maps obtained from VPI are shown for both the normal and diseased models (Fig. 5, left), and the corresponding CFD maps are depicted (Fig. 5, right). Flow vector information is rendered in the form of a duplex plot that shows the local velocity magnitude as color codes and the flow direction in the form of streamlines. For the normal and diseased models, the spatial-maximum flow velocity magnitude was estimated to be 1.52 m/s and 2.43 m/s, respectively. The higher peak flow for the diseased model is seemingly attributed to the BO obstruction. These measured values are generally aligned with CFDderived values, which yielded 1.83 m/s and 2.49 m/s, respectively for the spatial-maximum flow velocity. DISCUSSION Summative interpretation of findings Our laboratory investigation has generally demonstrated that VPI can visualize the internal flow characteristics of the lower urinary tract during micturition. Both laminar flow passage and complex flow characteristics, such as jetting and vortexing, can be tracked using VPI with a fine temporal resolution of 0.1 ms (Video 1). The maximum flow velocity magnitude, as derived from the flow vectors used to generate VPI cine loops, was found to range 2–3 m/s (Figs. 4 and 5). These numbers are in line with clinical observations in young boys
(Nurnberger 1985) and experimental results of a previous investigation on urinary flow turbulence (Arif et al. 2014). The urinary tract can be technically regarded as a pipeline system in which, in the absence of structural defects within the tract, urine can be efficiently passed through to the outlet (Schafer et al. 2002). If geometric defects like BO obstruction are present, flow energy loss would inevitably take place, leading to symptoms such as weak or intermittent urination (Ishii et al. 2014a; Yamanishi et al. 2000). We have observed such a phenomenon experimentally using the ultrasoundbased VPI technique that depicted a narrowed flow channel in the diseased phantom geometry (Video 1 and Fig. 4; left panel). In the urinary tract phantom with BO obstruction, its corresponding VPI cine loop has identified the emergence of a flow vortex at multiple places and has rendered their spatiotemporal behavior. An upstream vortex was found to be located between the BO and the proximal prostate urethra, and a secondary vortex was observed at the curving point in the prostatic urethra. Note that, while the upstream vortex is essentially attributed to obstruction-induced urethral narrowing, the secondary vortex is likely caused by the high fluidic shear at the flanks of the flow jet. From a diagnostic standpoint, it is after all of significance to be able to detect vortex locations via medical imaging techniques such as VPI because of two biomechanical reasons. First, these vortex locations often
8
Ultrasound in Medicine and Biology
represent the physical origin of undesired fluid pressure loss in a diseased urethra (Hou et al. 2016; Ishii et al. 2015; Ku et al. 2010; Tojo et al. 1994). Second, these vortices may alter the physical stress exerted against the urethral wall, thereby remodeling the urinary tract geometry in an undesirable way that favors the trapping of residual urine (Gustafson et al. 2004, Major et al. 2002, Ostergard 1979). Non-invasive visualization of these intra-urethral vortices, together with knowledge of the urinary tract anatomy, can potentially help clinicians to more informedly plan for surgical resection of the urethra, thereby realizing a less invasive, focal therapeutic approach to the BO obstruction problem in urethral pathology. Remarks on scanning configuration For the VPI technique to operate properly, probe location and insonation angle to the flow direction both need to be chosen properly so that the fast flow through the urinary tract can be consistently tracked without spurious errors as a result of Doppler aliasing. In our experiments, the transducer aperture was placed to resemble a trans-rectal scan configuration. The corresponding range of insonation angles was 80 –100 , and it effectively facilitated the imaging of urodynamics with up to 4.0-m/s flow velocity magnitude. Nevertheless, from an application standpoint, one caveat of performing transrectal imaging is that it may bring physical and psychological stress during the patient’s micturition process. To make the imaging process more patient friendly, a trans-perineum scanning configuration can possibly be considered in which the transducer is placed extracorporeally. Yet, it is not technically straightforward to properly acquire VPI cine loops with a trans-perineum field of view. The reason is because, in a trans-perineum field of view where the urinary flow tract is positioned at 50– 80 mm depth (Baxter and Firoozi 2013), the pulse repetition frequency must necessarily be lowered to 7 kHz or lower to maintain depth resolvability. Doing so would make the multi-angle Doppler flow vector estimator naturally more prone to slow-time aliasing artifacts. Of particular concern are the cases where slow-time aliasing extends beyond a full-cycle phase wraparound as a result of extremely high flow speeds, which are prone to arise in urinary flow as the maximum BO flow velocity can reach 2.0 m/s even with mild (,15%) obstruction (Ishii et al. 2014b). These challenging aliasing cases cannot be resolved with simple phase unwrapping algorithms such as the one included in the initial implementation of VPI (Yiu et al. 2014). Under a similar rationale, it would be challenging to make use of a trans-abdominal scanning configuration to acquire VPI cine loops of urinary flow. Also, it should be emphasized that the optimal imaging parameters may
Volume -, Number -, 2017
differ between subjects because the maximum urine flow velocity is inherently influenced by multiple factors such as the detrusor pressure, the structural and mechanical properties of urethra and sexual anatomic differences. As such, clinical usage guidelines will later need to be established as similar to those for B-mode imaging. Perspectives for further development A few directions of practical interest may be pursued to extend beyond the proof-of-principle demonstration of VPI-based urinary flow visualization reported here. First, from a technical development standpoint, it would be worthwhile to address the Doppler aliasing problem currently faced by VPI’s vector flow estimation so that the velocity dynamic range can be extended. One solution would be to incorporate the use of aliasing correction techniques that are based on staggered data acquisition principles (Posada et al. 2016). If this technical challenge can be overcome, it will effectively facilitate the use of VPI in a trans-perineal scanning configuration that is more patient friendly than trans-rectal imaging. Another technical topic of interest would be to develop new imaging techniques to monitor the deformation of the urinary tract at the start of micturition. Doing so would help couple the flow dynamics observations of VPI with the structural dynamics of the bladder outlet. A third direction to be pursued is to perform in vivo trials of VPI in urology clinics to establish the clinical feasibility of this new technique. The focus of these trials will be to refine operational procedures during in vivo scanning, especially patient posture and transducer alignment to ensure that patients can perform micturition normally during the imaging process. CONCLUSION High-frame-rate vector flow visualization (in the form of VPI) has been applied to non-invasively examine internal flow characteristics related to voiding dysfunction in urinary flow phantoms. This new application of VPI showed promise in locating positions within the urinary tract where disturbed urine passage is evident. Such diagnostic information opens new possibilities in developing non-invasive functional indices that describe voiding dysfunction, thereby assisting the clinical workflow in managing the care of patients with LUTS. Acknowledgments—This work has been supported in part by Natural Sciences and Engineering Council of Canada (RGPIN-2016-04042), Canada Foundation for Innovation (36138), Hong Kong Innovation and Technology Fund (GHP/025/13 SZ) and Research Grants Council of Hong Kong (GRF 785113 M). Parts of this investigation were conducted at the University of Hong Kong.
Vector flow visualization of urine passage d T. ISHII et al.
SUPPLEMENTARY DATA Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.ultrasmedbio.2017.07.006.
REFERENCES Arif M, Idzenga T, de Korte CL, van Mastrigt R. Development of a noninvasive method to diagnose bladder outlet obstruction based on decorrelation of sequential ultrasound images. Urology 2015; 85:648–652. Arif M, Idzenga T, van Mastrigt R, de Korte CL. Estimation of urinary flow velocity in models of obstructed and unobstructed urethras by decorrelation of ultrasound radiofrequency signals. Ultrasound Med Biol 2014;40:938–946. Baxter C, Firoozi F. Pelvic floor ultrasound. In: Fulgham FP, Gilbert RB, (eds). Practical urological ultrasound. New York, NY: Springer New York; 2013. p. 143–153. Cheung CCP, Yu ACH, Salimi N, Yiu BYS, Tsang IKH, Kerby B, Azar RZ, Dickie K. Multi-channel pre-beamformed data acquisition system for research on advanced ultrasound imaging methods. IEEE Trans Ultrason Ferroelectr Freq Control 2012;59:243–253. Ding YY, Ozawa H, Yokoyama T, Nasu Y, Chancellor MB, Kumon H. Reliability of color Doppler ultrasound urodynamics in the evaluation of bladder outlet obstruction. Urology 2000;56:967–971. Gratzke C, Bachmann A, Descazeaud A, Drake MJ, Madersbacher S, Mamoulakis C, Oelke M, Tikkinen KA, Gravas S. EAU Guidelines on the assessment of non-neurogenic male lower urinary tract symptoms including benign prostatic obstruction. Eur Urol 2015;67: 1099–1109. Gustafson KJ, Creasey GH, Grill WM. A urethral afferent mediated excitatory bladder reflex exists in humans. Neurosci Lett 2004; 360:9–12. Ho CK, Chee AJY, Yiu BYS, Tsang ACO, Chow KW, Yu ACH. Wallless flow phantoms with tortuous vascular geometries: Design principles and a patient-specific model fabrication example. IEEE Trans Ultrason Ferroelec Freq Contr 2017;64:25–38. Hou C-P, Lin Y-H, Chen C-L, Tsai Y-L, Chang P-L, Tsui K-H. Impact of the static prostatic urethral angle on men with lower urinary tract symptoms. Urol Sci 2016;27:47–50. Humphrey PA. Male Urethra and external genitalia anatomy. In: Brandes SB, Morey AF, (eds). Advanced male urethral and genital reconstructive surgery. Valley Stream, NY: Humana Press; 2014. p. 17–23. Irwin DE, Kopp ZS, Agatep B, Milsom I, Abrams P. Worldwide prevalence estimates of lower urinary tract symptoms, overactive bladder, urinary incontinence and bladder outlet obstruction. BJU Int 2011; 108:1132–1138. Ishii T, Kambara Y, Yamanishi T, Naya Y, Igarashi T. Urine flow dynamics through prostatic urethra with tubular organ modeling using endoscopic imagery. IEEE J Transl Eng Health Med 2014a;2: 1800709. Ishii T, Nakamura K, Naya Y, Igarashi T. Therapeutic designing for urethral obstruction by virtual urethra and flow dynamics simulation. Minim Invasive Ther Allied Technol 2015;24:141–147. Ishii T, Naya Y, Yamanishi T, Igarashi T. Urine flow dynamics through the urethra in patients with bladder outlet obstruction. J Mech Med Biol 2014b;14:1450052. Jensen JA, Nikolov SI, Gammelmark KL, Pedersen MH. Synthetic aperture ultrasound imaging. Ultrasonics 2006;44(Suppl 1):e5–e15. Jensen JA, Nikolov SI, Yu ACH, Garcia D. Ultrasound vector flow imaging: Part I. Sequential systems. IEEE Trans Ultrason Ferroelec Freq Contr 2016a;63:1704–1721. Jensen JA, Nikolov SI, Yu ACH, Garcia D. Ultrasound vector flow imaging: Part II. Parallel systems. IEEE Trans Ultrason Ferroelec Freq Contr 2016b;63:1722–1732. Karaman M, Li PC, Odonnell M. Synthetic-aperture imaging for smallscale systems. IEEE Trans Ultrason Ferroelectr Freq Control 1995; 42:429–442. Kohler TS, Yadven M, Manvar A, Liu N, Monga M. The length of the male urethra. Int Braz J Urol 2008;34:451–454.
9
Ku JH, Ko DW, Cho JY, Oh SJ. Correlation between prostatic urethral angle and bladder outlet obstruction index in patients with lower urinary tract symptoms. Urology 2010;75:1467–1471. Kurata T, Li Z, Oda S, Kawahira H, Haneishi H. Impact of vessel diameter and bandwidth of illumination in sidestream dark-field oximetry. Biomed Opt Express 2015;6:1616–1631. Lu JY, Cheng J, Wang J. High frame rate imaging system for limited diffraction array beam imaging with square-wave aperture weightings. IEEE Trans Ultrason Ferroelectr Freq Control 2006;53: 1796–1812. Major H, Culligan P, Heit M. Urethral sphincter morphology in women with detrusor instability. Obstet Gynecol 2002;99:63–68. McNeal JE. The zonal anatomy of the prostate. Prostate 1981;2:35–49. Mihmanli I, Kantarci F, Gulsen F, Kadioglu A. Transrectal voiding sonourethrography for diagnosis of a prostatic urethral calculus. J Ultrasound Med 2006;25:1455–1457. Montaldo G, Tanter M, Bercoff J, Benech N, Fink M. Coherent planewave compounding for very high frame rate ultrasonography and transient elastography. IEEE Trans Ultrason Ferroelectr Freq Control 2009;56:489–506. Nurnberger N. Normal values of urine stream velocity in boys. Z Urol Nephrol 1985;78:539–544. Ostergard DR. The neurological control of micturition and integral voiding reflexes. Obstet Gynecol Surv 1979;34:417–423. Ozawa H, Igarashi T, Uematsu K, Watanabe T, Kumon H. The future of urodynamics: Non-invasive ultrasound videourodynamics. Int J Urol 2010;17:241–249. Posada D, Poree J, Pellissier A, Chayer B, Tournoux F, Cloutier G, Garcia D. Staggered multiple-PRF ultrafast color Doppler. IEEE Trans Med Imag 2016;35:1510–1521. Prada PJ, Fernandez J, Martinez AA, Rua ADL, Gonzalez JM, Fernandez JM, Juan G. Transperineal injection of hyaluronic acid in anterior perirectal fat to decrease rectal toxicity from radiation delivered with intensity modulated brachytherapy or EBRT for prostate cancer patients. Int J Radiat Oncol Biol Phys 2007;69:95–102. Schafer W, Abrams P, Liao L, Mattiasson A, Pesce F, Spangberg A, Sterling AM, Zinner NR, van Kerrebroeck P, International Continence Society. Good urodynamic practices: Uroflowmetry, filling cystometry, and pressure-flow studies. Neurourol Urodyn 2002;21: 261–274. So HKH, Chen J, Yiu BYS, Yu ACH. Medical ultrasound imaging: To GPU or not to GPU? IEEE Micro 2011;31:54–65. Stalder AF, Russe MF, Frydrychowicz A, Bock J, Hennig J, Markl M. Quantitative 2 D and 3 D phase contrast MRI: Optimized analysis of blood flow and vessel wall parameters. Magn Reson Med 2008; 60:1218–1231. Stevens E. Bladder ultrasound: Avoiding unnecessary catheterizations. Medurg Nurs 2005;14:249–253. Sughimoto K, Shimamura Y, Tezuka C, Tsubota K, Liu H, Okumura K, Masuda Y, Haneishi H. Effects of arterial blood flow on walls of the abdominal aorta: Distributions of wall shear stress and oscillatory shear index determined by phase-contrast magnetic resonance imaging. Heart Vessels 2016;31:1168–1175. Tojo M, Yasuda K, Yamanishi T, Hattori T, Nagashima K, Shimazaki J. Relationship between bladder neck diameter and hydraulic energy at maximum flow. J Urol 1994;152:144–149. Torii R, Oshima M, Kobayashi T, Takagi K, Tezduyar TE. Numerical investigation of the effect of hypertensive blood pressure on cerebral aneurysm—Dependence of the effect on the aneurysm shape. Int J Numer Meth Fluids 2007;54:995–1009. Tubaro A, De Nunzio C, Puccini F, Presicce F. The evolving picture of lower urinary tract symptom management. Eur Urol 2015;67: 271–272. Versteeg HK, Malalasekera W. An Introduction to computational fluid dynamics: The finite volume method. Harlow, Essex: Pearson Education Limited; 2007. Weiss GH, Nossal R, Bonner RG. Statistics of penetration depth of photons re-emitted from irradiated tissue. J Mod Opt 1989;36:249–259. White BR, Pierce MC, Nassif N, Cense B, Park BH, Tearney GJ, Bouma BE, Chen TC, de Boer JF. In vivo dynamic human retinal blood flow imaging using ultra-high-speed spectral domain optical Doppler tomography. Opt Express 2003;11:3490–3497.
10
Ultrasound in Medicine and Biology
Yamanishi T, Yasuda K, Sakakibara R, Hattori T, Tojo M. The effectiveness of terazosin, an alpha1-blocker, on bladder neck obstruction as assessed by urodynamic hydraulic energy. BJU Int 2000;85: 249–253. Yiu BYS, Lai SSM, Yu ACH. Vector projectile imaging: Time-resolved dynamic visualization of complex flow patterns. Ultrasound Med Biol 2014;40:2295–2309.
Volume -, Number -, 2017 Yiu BYS, Tsang IKH, Yu ACH. GPU-based beamformer: Fast realization of plane wave compounding and synthetic aperture imaging. IEEE Trans Ultrason Ferroelectr Freq Control 2011;58: 1698–1705. Yiu BYS, Yu ACH. Least-squares multi-angle Doppler estimators for plane wave vector flow imaging. IEEE Trans Ultrason Ferroelec Freq Contr 2016;63:1733–1744.
http://dx.doi.org/10.1016/j.ultrasmedbio.2017.07.006
d
Original Contribution VECTOR FLOW VISUALIZATION OF URINARY FLOW DYNAMICS IN A BLADDER OUTLET OBSTRUCTION MODEL TAKURO ISHII, BILLY Y. S. YIU, and ALFRED C. H. YU Schlegel Research Institute for Aging and Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON, Canada (Received 24 December 2016; revised 9 June 2017; in final form 8 July 2017)
Abstract—Voiding dysfunction that results from bladder outlet (BO) obstruction is known to alter significantly the dynamics of urine passage through the urinary tract. To non-invasively image this phenomenon on a time-resolved basis, we pursued the first application of a recently developed flow visualization technique called vector projectile imaging (VPI) that can track the spatiotemporal dynamics of flow vector fields at a frame rate of 10,000 fps (based on plane wave excitation and least-squares Doppler vector estimation principles). For this investigation, we designed a new anthropomorphic urethral tract phantom to reconstruct urinary flow dynamics under controlled conditions (300 mm H2O inlet pressure and atmospheric outlet pressure). Both a normal model and a diseased model with BO obstruction were developed for experimentation. VPI cine loops were derived from these urinary flow phantoms. Results show that VPI is capable of depicting differences in the flow dynamics of normal and diseased urinary tracts. In the case with BO obstruction, VPI depicted the presence of BO flow jet and vortices in the prostatic urethra. The corresponding spatial-maximum flow velocity magnitude was estimated to be 2.43 m/s, and it is significantly faster than that for the normal model (1.52 m/s) and is in line with values derived from computational fluid dynamics simulations. Overall, this investigation demonstrates the feasibility of using vector flow visualization techniques to non-invasively examine internal flow characteristics related to voiding dysfunction in the urethral tract. (E-mail: [email protected]) Ó 2017 World Federation for Ultrasound in Medicine & Biology. Key Words: Urinary flow, Voiding dysfunction, Urinary tract phantom, Vector flow visualization, Doppler ultrasound, High frame rate imaging.
INTRODUCTION
help urologists determine whether such treatment is necessary, it is essential to first assess the severity of voiding dysfunction. Yet, it is not trivial to perform this diagnosis non-invasively. Techniques such as uroflowmetry (Schafer et al. 2002) and ultrasound-based bladder volume measurement (Stevens 2005) can merely provide overall indications of voiding dysfunction, and they lack the provision of insight into the anatomic and urodynamic factors that deter the urine passage process. To extend beyond these solutions, it is necessary to devise strategies that can monitor urinary flow dynamics in LUTS patients before and after an intervention (in addition to evaluating the shape of urinary tract). Medical imaging is undoubtedly a potential solution to such a need. Among existing imaging modalities, some are capable of visualizing the flow of biofluids inside the human body, such as magnetic resonance imaging (MRI) (Stalder et al. 2008; Sughimoto et al. 2016) and optical imaging (Kurata et al. 2015; White et al. 2003). However, most of these imaging techniques have rarely been applied to urinary flow visualization. In the case
Urethral voiding dysfunction that leads to weak urinary flow, urine dribbling and urine splitting is well considered as a major category of lower urinary tract symptoms (LUTS). This problem is mainly attributed to mechanical and physiological aging of the vesico-urethral system. Despite being benign in nature, urethral voiding dysfunction would hamper a patient’s quality of life on a chronic basis, and thus it is increasingly regarded as a healthcare concern in today’s aging society. In recent years, we have seen a growing interest in the potential to devise appropriate strategies for addressing urethral voiding dysfunction (and LUTS in general), such as pharmacological and surgical interventions that aim to alter the internal urethral shape during micturition (Tubaro et al. 2015). To
Address correspondence to: Alfred C.H. Yu, Schlegel Research Institute for Aging and Department of Electrical and Computer Engineering, University of Waterloo, 200 University Avenue West, Waterloo, ON N2L 3G1, Canada. E-mail: [email protected] 1
2
Ultrasound in Medicine and Biology
of MRI, its temporal resolution is inadequate to track fast urinary flow (can be .2 m/s velocity), and logistically it is challenging to manage a patient’s micturition process inside the MRI scanner bore; in the case of optical imaging, the penetration depth is insufficient (Weiss et al. 1989) for the purpose of non-invasively imaging the urinary tract. Alternatively, it is possible to leverage computational fluid dynamics (CFD) to render flow profiles based on anatomic information obtained from MRI or computed tomography (Torii et al. 2007). Alas, proper development of a CFD model for urinary flow dynamics is after all not straightforward because of the need to account for the predominance of autonomic nerve activity, as well as dynamic changes in the size of the urinary tract, which can expand to 10 mm in diameter during peak micturition. In contrast, ultrasound imaging is perhaps better suited for urinary flow visualization. Previous work by others have indeed shown the feasibility of using ultrasound to observe the urinary tract and its related flow dynamics (Arif et al. 2014, 2015; Ding et al. 2000; Gratzke et al. 2015; Ozawa et al. 2010). Nevertheless, these published investigations were all conducted using conventional ultrasound technology (on the basis of scanline–based imaging principles) that typically renders flow information in two formats: (i) full-view color flow images at video-range frame rates (20–30 fps) (Ding et al. 2000; Ozawa et al. 2010), or (ii) M-mode or radiofrequency data plots over a single scanline (Arif et al. 2014, 2015). They fell short of providing timeresolved rendering of urinary flow dynamics across the entire imaging view, nor can they offer information about the multi-directional nature of urinary flow patterns. To overcome the temporal resolution limitation, one can possibly develop high-frame-rate ultrasound imaging methods that are based on plane wave excitation (Lu et al. 2006; Montaldo et al. 2009) and synthetic aperture principles (Jensen et al. 2006; Karaman et al. 1995). As for the need to visualize multi-directional urinary flow, vector flow estimation techniques can be devised (Jensen et al. 2016a, 2016b). Nevertheless, neither of these concepts has been applied to urinary flow diagnostics. In this paper, we report the first application of a highframe-rate ultrasound vector flow visualization technique called vector projectile imaging (VPI) to facilitate analysis of urinary flow dynamics. It is our intent to investigate the feasibility of leveraging VPI to derive diagnostically relevant insight into the fluid mechanics within the urinary tract, particularly in LUTS cases with urethral voiding dysfunction. To perform this investigation, we have designed a novel anatomically realistic urinary flow phantom that can resemble urine passage under controlled conditions. Note that VPI is a novel tech-
Volume -, Number -, 2017
nique developed by our team earlier (Yiu et al. 2014), and it is based on multi-angle plane wave excitation and least-squares Doppler vector estimation principles (Yiu and Yu 2016). We have initially applied VPI to achieve time-resolved visualization of carotid hemodynamics. Here, we seek to expand the clinical application domain of VPI by translating this technique toward use in urinary flow diagnostics.
METHODS AND MATERIALS Male lower urinary tract phantom design Background description. Bladder outlet (BO) obstruction was chosen as the LUTS disease model in this study. Well regarded as a common lesion in the posterior urethra causing voiding symptoms, BO obstruction is estimated to affect 21.5% of the worldwide population (Irwin et al. 2011). It refers to the condition where sclerosis and narrowing of the BO are both observed during micturition. Its emergence is often attributed to (i) lateral-lobe prostate hypertrophy or (ii) urethral mucosa hyperplasia as a consequence of local inflammation. To our knowledge, the flow dynamics in urinary tracts with BO obstruction have not been visualized using noninvasive medical imaging techniques. From an anatomic standpoint, BO obstruction occurs in the upstream part of the male lower urinary tract, which is illustrated in Figure 1a. Note that the urinary tract length can be as long as 20 cm (Kohler et al. 2008). Its prostatic and membranous segments are located inside the body, while the bulbous segment is located within the penis. In the presence of BO obstruction, urethral turbulence is expected to emerge. Such turbulence would lead to excessive loss of fluid energy and in turn cause voiding dysfunction symptoms (Ishii et al. 2014a; Yamanishi et al. 2000). Model features and design justifications. Urinary tract models with and without BO obstruction were designed to facilitate comparative analysis of VPI’s urinary flow visualization performance in healthy and pathological urinary flow scenarios. The overall geometry of the urinary tract is illustrated in Figure 1c. It was drafted using SolidWorks (Dassault Systemes, Waltham, MA, USA) in the form of a spline surface, as generated by stacking together multiple cross sections that were defined from five key anatomic features of the urethra: (i) bladder outlet (BO); (ii) proximal prostatic urethras (PU); (iii) verumontanum (V); (iv) urethral sphincter (S); and (v) bulbous urethra (BU). For each crosssectional plane, its shape and dimensions were set in accordance with a recent study on CFD-based urinary flow visualization (Ishii et al. 2014b).
Vector flow visualization of urine passage d T. ISHII et al.
3
Fig. 1. Design of the urinary tract. (a) Anatomic features of the lower urinary tract. (b) Endoscopic views of BO. Median bar is shown in the BO obstruction image. (c) Outline of urinary tract of the disease model. (d) Shape of the cross sections; contour colors correspond to labels in (c). BO 5 bladder outlet; BU 5 bulbous urethra; PU 5 proximal prostatic urethras; S 5 urethral sphincter; V 5 verumontanum.
In our urinary tract model, the BO cross section was drafted as a triangle-like closed curve. This design choice was made on the basis of cystourethroscopic imagery and anatomic knowledge of the bladder trigon (Ishii et al. 2014b). In the case with BO obstruction, the bladder outlet shape included a median bar: the characteristic anatomic feature for this pathology (see Figure 1b). The vertical diameter of the bladder outlet was set to 2 mm and 3.5 mm, respectively in the BO obstruction geometry and the healthy geometry, in line with a previous report (Tojo et al. 1994). Other anatomic features of our urinary tract model were drafted as ellipsoids of different size. Specifically, vertical diameters of 5.6, 6.0, 4.0 and 6.0 mm were respectively defined for proximal prostatic urethra, verumontanum, urethral sphincter and bulbous urethra, as shown in Figure 1d. The same diameter values were used for both the healthy and diseased urethra geometry in order to focus on investigating the impact of BO obstruction on urinary flow dynamics. Consistent with published data (Humphrey 2014; McNeal 1981), the length of prostatic, membranous and bulbous urethra was respectively set to 30, 20 and 30 mm. Also, the prostatic urethra was angled at 35 . The angle between inlet and outlet was assumed to be 90 in this study. For each cross-sectional curvature, four points were marked at 0 (dorsal side), 130 , 180 (ventral side) and 260 angles. This process was done so that the
four guide spline curves (associated with the marked points of same degree in each cross section) can collectively define the final shape of the urinary tract model. It is worth noting that the urinary tract was assumed to be an expanded rigid object. Such an assumption is valid for our purpose, because our experimental aim is to test the efficacy of using VPI to track urinary flow dynamics during peak micturition when the flow profile is generally constant. Phantom fabrication protocol. A core-filled replicate of the drafted urinary tract geometries was built with the use of 3-D printing technologies to facilitate phantom construction using a wall-less investment casting protocol that was reported earlier (Ho et al. 2017). For this process, the casting box used to mount the urinary tract replicate was developed such that it incorporated a designated dorsal side located at 20 mm away from the posterior urethra. Note that these distances are aligned with typical values in the human male body (Humphrey 2014; McNeal 1981; Prada et al. 2007). After the urinary tract replicate was mounted onto the casting box, a tissue-mimicking solution was slowly poured into the box cavity. This solution was based on the same polyvinyl-alcohol–containing formula published elsewhere (Ho et al. 2017). Also kept the same were the freeze-thaw process required to crystalize the tissue mimic and the chloroform immersion procedure needed to remove the urinary tract replicate.
4
Ultrasound in Medicine and Biology
Flow circuit setup. The flow circuit that was used to drive the urinary tract phantom was set up as shown in Figure 2. The inlet of the fabricated phantom was connected to a cylinder (50 mm in diameter) that mimicked a pressure-driven boundary condition. The phantom outlet, which corresponded to the excretion of urine mimicking fluid, was connected to an unpressurized buffer tank through a 20-cm long, 8-mm diameter flexible tube. In each instance of fluid passage through the urinary tract, the excreted fluid would be relayed from the buffer tank back to the inlet cylinder by using a gear pump that operated at constant flow mode (PQ12; Greylor Co., Cape Coral, FL, USA). The pump’s flow rate was tuned to keep the fluid level in the inlet cylinder to exert 300-mm,H2O head pressure at the phantom inlet. The urinary mimicking fluid used in this work was composed of distilled water and 50 mg/L of silicon carbide scatterers that had a mean size of 37 mm (357391; Sigma-Aldrich, St. Louis, MO, USA). Following previous reports by others (Arif et al. 2015), the viscosity, density and scatterer concentration of this fluid mixture were aligned with that for actual urine. In each experimental run, the rate at which this fluid flowed through the urinary phantom was found via dividing the measured outflow volume by the time elapsed.
Fig. 2. Schematic of the experimental setting and probe locations. The urinary tract was fixed as posture in the body. Instead of bladder, a cylinder was connected to the phantom. To generate static pressure boundary condition at bladder outlet, fluid was continuously supplied from outlet tank to keep the water level of the cylinder. Dimensions in the figure are in millimeters.
Volume -, Number -, 2017
Ultrasound visualization of urinary flow dynamics Overview of imaging framework. Time-resolved visualization of urinary flow dynamics was achieved with the use of the VPI technique that we established previously (Yiu et al. 2014). In brief, this imaging innovation is based on the use of plane-wave excitation principles to achieve high imaging frame rates that are well beyond the video display range. It derives local flow vector information (with axial and lateral velocity components) at every pixel position in the entire imaging view through a multiangle, least-squares vector Doppler estimation methodology (Yiu and Yu 2016), and these flow vectors were visualized through a dynamic rendering algorithm that depicts the trajectory of individual vector projectiles. Note that, to combat against slow-time aliasing artifacts, a phase-unwrapping algorithm has been incorporated in the VPI signal processing chain (Yiu et al. 2014). Transducer placement. During imaging data acquisition, an L14-5 linear array transducer (Analogic Ultrasound, Peabody, MA, USA) was placed on the dorsal side of the urinary flow phantom such that the transducer aperture was parallel to the urethral tract and the imaging view encompassed the BO. The corresponding field of view was similar to that obtainable from a trans-rectal scan (Mihmanli et al. 2006). This scan configuration was adopted so that the urinary tract was not too deep from the probe surface (2030 mm). Also, the insonation angles used for multi-angle Doppler flow vector estimation would be within an 80 –100 range, where the detected frequency shifts would not be aliased. Imaging system and parameters. The L14-5 transducer was connected to a research-purpose ultrasound imaging platform (SonixTouch; Analogic Corporation, Peabody, MA, USA), and it was driven according to the imaging parameters specified in Table 1. As illustrated in Figure 3, in a typical VPI imaging experiment, two data acquisition sequences were executed. First, to obtain a background B-mode image of adequate quality, unfocused plane wave pulses were sequentially fired from 24 steering angles (in equal increments over the angle span 210 to 10 ). Second, for high-frame-rate vector estimation purposes, non-steered plane wave pulses (0 ) were repeatedly fired at 10-kHz pulse repetition frequency (PRF) for a 2-s duration. Effectively, the raw acquisition frame rate of VPI cine loops is 10,000 fps. For each transmit firing event, raw pre-beamformed data were acquired from all array channels through a channel data acquisition system (Cheung et al. 2012). As explained in our earlier work (Yiu et al. 2014), the acquired data sets were processed offline via a high-throughput beamforming and flow estimation platform that was based on graphics processing unit (GPU) technology (So et al. 2011; Yiu et al.
Vector flow visualization of urine passage d T. ISHII et al.
Table 1. Hardware specification and imaging parameters Parameter
VPI
B-mode
a. Imaging platform Number of Tx/RX channels 128 Array pitch (mm) 0.3048 Pre-beamformed data sampling 40 rate (MHz) Pre-beamformed data resolution 12 (bit) b. Data acquisition Imaging frequency (MHz) 5 Tx pulse duration (number of 3 2 pulses) Number of Tx steering angle 1 24 0 212 21 & 112 Tx steering angle ( ) Pulse repetition frequency (kHz) 10 Effective data acquisition rate 10 0.416 (kHz) Maximum imaging depth (mm) 60 Data acquisition duration (s) 2 0.1 c. Beamforming Pre-beamformed data filter 5 passband (MHz) Filter design method Equiripple Lateral and depth pixel dimension 0.1 (mm/pixel) d. Slow-time data processing Normalized clutter filter cut-off 0.05 Filter design method Equiripple Sliding window for flow estimation 64 Sliding window step size 4 e. VPI visualization Nominal frame rate (fps) 2476 Launch point density (%) 3 Mean projectile lifetime (frames) 30 VPI 5 vector projectile imaging.
2011) and a Matlab interface (v. 2016a, Mathworks Inc., Natick, MA, USA). Using this platform, beamforming was performed from three receive steering angles (210 , 0 , 10 ), so as to realize a one-transmit, three-receive multi-angle configuration for least-squares Doppler vector estimation (Yiu and Yu 2016). Accordingly, B-mode images, VPI cine loops, and representative still frames were obtained. For reference, Doppler spectrograms were also derived at pixel positions of interest. These supplementary data were derived by performing short-time Fourier transforms on the slow-time ensemble of the identified pixel position, similar to our previous work (Yiu et al. 2014). Computational fluid dynamics simulations To facilitate comparative analysis, CFD simulations based on the Navier-Stokes equation were also carried out in this investigation using SolidWorks Flow Simulation software (v. 2015, Dassault Systemes). Briefly, based on the same urinary tract geometries as those used for phantom construction, the CFD solver computed the Navier-Stokes equation for incompressible fluid with the k-ε turbulence model by the finite volume method (Versteeg and Malalasekera 2007). As boundary
5
conditions, we respectively specified a constant inlet (BO cross-section) pressure of 104,267 Pa (300 mm H2O) and a constant outlet pressure of 101,325 Pa. Fluid properties were set based on the viscosity and density of water at 37 C temperature, which are respectively 0.685 3 1023 Pa$s and 992.48 kg/m3. For computation stability, the outlet length was extended by ten times of the outlet diameter, so that the urinary tract geometry used for the CFD simulations was essentially equivalent to the experimental phantoms that we fabricated. Other computational routines, such as spatial discretization, were the same as that described in our previous study (Ishii et al. 2014b). RESULTS VPI visualization of urinary flow in normal and diseased models VPI was found to be capable of depicting differences in the flow dynamics of normal and diseased urinary tracts. This observation is summarized in Video 1 (playback rate: 50 fps), which renders the urinary flow dynamics for a diseased model with BO obstruction (left panel) and a normal model (right panel). For comparative analysis, the bottom portion of Video 1 shows the Doppler spectrogram at the pixel position that corresponds to the intersection between the vertical line and horizontal line drawn in the VPI video frame (i.e., at the distal end of prostatic urethra). A key result to be noted is that, in the diseased model, flow jetting can be observed at the BO because of its anatomic narrowing. The flow velocity is consequently faster than that at the BO of the normal model. This trend is well summarized in the still-frame VPI snapshots shown in Figure 4. Another important observation is that vortices are present in the proximal and distal parts of the prostatic urethra as urine flowed past the obstructed BO in the diseased model. Because of the sustained presence of these vortices, jittering of flow trajectories can also be observed in the vicinity of the proximal prostatic urethra. These vortices and flow path jittering are not present in the VPI cine loop for the normal model. Such a trend is consistent with the Doppler spectrogram reference data, which showed that at the distal end of the prostatic urethra, the detected Doppler frequencies only spanned a limited range as indicative of constant flow speed. It is in contrast with the diseased model’s corresponding Doppler spectrogram that showed the presence of a broad range of positive and negative Doppler frequencies as indicative of chaotic flow patterns. Comparative analysis of flow patterns in the urinary flow models The flow vectors rendered by VPI were found to be consistent with CFD computation results. This finding is
6
Ultrasound in Medicine and Biology
Volume -, Number -, 2017
Fig. 3. Schematic of duplex VPI algorithm. Two sequences of pulse waves were insonated and stored separately through a DAQ tool. Steered pulse wave data were processed to generate a sequence of high-resolution B-mode images, while VPI was processed with second sequence RF data. L1-AC and L-S denote lag-one autocorrelation and least squares algorithm, respectively. DAQ 5 data acquisition; RF 5 radiofrequency; VPI 5 vector projectile imaging.
Fig. 4. Urine flow behavior visualized by VPI. Still-frame snapshots are provided for diseased model with BO obstruction (left) and normal urinary tract phantom (right). For both models, a Doppler spectrogram is shown as reference for a pixel position marked by the intersection between vertical and horizontal lines in the VPI snapshot image. BO 5 bladder outlet; VPI 5 vector projectile imaging.
Vector flow visualization of urine passage d T. ISHII et al.
7
Fig. 5. Comparison of magnitude flow velocity between VPI and CFD. Diseased model with BO obstruction (left); normal urinary tract phantom (right). Results are shown in a duplex format that depicts flow velocity magnitude as color codes and flow trajectories as streamlines. CFD 5 computational fluid dynamics; VPI 5 vector projectile imaging.
illustrated in Figure 5. The time-averaged flow maps obtained from VPI are shown for both the normal and diseased models (Fig. 5, left), and the corresponding CFD maps are depicted (Fig. 5, right). Flow vector information is rendered in the form of a duplex plot that shows the local velocity magnitude as color codes and the flow direction in the form of streamlines. For the normal and diseased models, the spatial-maximum flow velocity magnitude was estimated to be 1.52 m/s and 2.43 m/s, respectively. The higher peak flow for the diseased model is seemingly attributed to the BO obstruction. These measured values are generally aligned with CFDderived values, which yielded 1.83 m/s and 2.49 m/s, respectively for the spatial-maximum flow velocity. DISCUSSION Summative interpretation of findings Our laboratory investigation has generally demonstrated that VPI can visualize the internal flow characteristics of the lower urinary tract during micturition. Both laminar flow passage and complex flow characteristics, such as jetting and vortexing, can be tracked using VPI with a fine temporal resolution of 0.1 ms (Video 1). The maximum flow velocity magnitude, as derived from the flow vectors used to generate VPI cine loops, was found to range 2–3 m/s (Figs. 4 and 5). These numbers are in line with clinical observations in young boys
(Nurnberger 1985) and experimental results of a previous investigation on urinary flow turbulence (Arif et al. 2014). The urinary tract can be technically regarded as a pipeline system in which, in the absence of structural defects within the tract, urine can be efficiently passed through to the outlet (Schafer et al. 2002). If geometric defects like BO obstruction are present, flow energy loss would inevitably take place, leading to symptoms such as weak or intermittent urination (Ishii et al. 2014a; Yamanishi et al. 2000). We have observed such a phenomenon experimentally using the ultrasoundbased VPI technique that depicted a narrowed flow channel in the diseased phantom geometry (Video 1 and Fig. 4; left panel). In the urinary tract phantom with BO obstruction, its corresponding VPI cine loop has identified the emergence of a flow vortex at multiple places and has rendered their spatiotemporal behavior. An upstream vortex was found to be located between the BO and the proximal prostate urethra, and a secondary vortex was observed at the curving point in the prostatic urethra. Note that, while the upstream vortex is essentially attributed to obstruction-induced urethral narrowing, the secondary vortex is likely caused by the high fluidic shear at the flanks of the flow jet. From a diagnostic standpoint, it is after all of significance to be able to detect vortex locations via medical imaging techniques such as VPI because of two biomechanical reasons. First, these vortex locations often
8
Ultrasound in Medicine and Biology
represent the physical origin of undesired fluid pressure loss in a diseased urethra (Hou et al. 2016; Ishii et al. 2015; Ku et al. 2010; Tojo et al. 1994). Second, these vortices may alter the physical stress exerted against the urethral wall, thereby remodeling the urinary tract geometry in an undesirable way that favors the trapping of residual urine (Gustafson et al. 2004, Major et al. 2002, Ostergard 1979). Non-invasive visualization of these intra-urethral vortices, together with knowledge of the urinary tract anatomy, can potentially help clinicians to more informedly plan for surgical resection of the urethra, thereby realizing a less invasive, focal therapeutic approach to the BO obstruction problem in urethral pathology. Remarks on scanning configuration For the VPI technique to operate properly, probe location and insonation angle to the flow direction both need to be chosen properly so that the fast flow through the urinary tract can be consistently tracked without spurious errors as a result of Doppler aliasing. In our experiments, the transducer aperture was placed to resemble a trans-rectal scan configuration. The corresponding range of insonation angles was 80 –100 , and it effectively facilitated the imaging of urodynamics with up to 4.0-m/s flow velocity magnitude. Nevertheless, from an application standpoint, one caveat of performing transrectal imaging is that it may bring physical and psychological stress during the patient’s micturition process. To make the imaging process more patient friendly, a trans-perineum scanning configuration can possibly be considered in which the transducer is placed extracorporeally. Yet, it is not technically straightforward to properly acquire VPI cine loops with a trans-perineum field of view. The reason is because, in a trans-perineum field of view where the urinary flow tract is positioned at 50– 80 mm depth (Baxter and Firoozi 2013), the pulse repetition frequency must necessarily be lowered to 7 kHz or lower to maintain depth resolvability. Doing so would make the multi-angle Doppler flow vector estimator naturally more prone to slow-time aliasing artifacts. Of particular concern are the cases where slow-time aliasing extends beyond a full-cycle phase wraparound as a result of extremely high flow speeds, which are prone to arise in urinary flow as the maximum BO flow velocity can reach 2.0 m/s even with mild (,15%) obstruction (Ishii et al. 2014b). These challenging aliasing cases cannot be resolved with simple phase unwrapping algorithms such as the one included in the initial implementation of VPI (Yiu et al. 2014). Under a similar rationale, it would be challenging to make use of a trans-abdominal scanning configuration to acquire VPI cine loops of urinary flow. Also, it should be emphasized that the optimal imaging parameters may
Volume -, Number -, 2017
differ between subjects because the maximum urine flow velocity is inherently influenced by multiple factors such as the detrusor pressure, the structural and mechanical properties of urethra and sexual anatomic differences. As such, clinical usage guidelines will later need to be established as similar to those for B-mode imaging. Perspectives for further development A few directions of practical interest may be pursued to extend beyond the proof-of-principle demonstration of VPI-based urinary flow visualization reported here. First, from a technical development standpoint, it would be worthwhile to address the Doppler aliasing problem currently faced by VPI’s vector flow estimation so that the velocity dynamic range can be extended. One solution would be to incorporate the use of aliasing correction techniques that are based on staggered data acquisition principles (Posada et al. 2016). If this technical challenge can be overcome, it will effectively facilitate the use of VPI in a trans-perineal scanning configuration that is more patient friendly than trans-rectal imaging. Another technical topic of interest would be to develop new imaging techniques to monitor the deformation of the urinary tract at the start of micturition. Doing so would help couple the flow dynamics observations of VPI with the structural dynamics of the bladder outlet. A third direction to be pursued is to perform in vivo trials of VPI in urology clinics to establish the clinical feasibility of this new technique. The focus of these trials will be to refine operational procedures during in vivo scanning, especially patient posture and transducer alignment to ensure that patients can perform micturition normally during the imaging process. CONCLUSION High-frame-rate vector flow visualization (in the form of VPI) has been applied to non-invasively examine internal flow characteristics related to voiding dysfunction in urinary flow phantoms. This new application of VPI showed promise in locating positions within the urinary tract where disturbed urine passage is evident. Such diagnostic information opens new possibilities in developing non-invasive functional indices that describe voiding dysfunction, thereby assisting the clinical workflow in managing the care of patients with LUTS. Acknowledgments—This work has been supported in part by Natural Sciences and Engineering Council of Canada (RGPIN-2016-04042), Canada Foundation for Innovation (36138), Hong Kong Innovation and Technology Fund (GHP/025/13 SZ) and Research Grants Council of Hong Kong (GRF 785113 M). Parts of this investigation were conducted at the University of Hong Kong.
Vector flow visualization of urine passage d T. ISHII et al.
SUPPLEMENTARY DATA Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.ultrasmedbio.2017.07.006.
REFERENCES Arif M, Idzenga T, de Korte CL, van Mastrigt R. Development of a noninvasive method to diagnose bladder outlet obstruction based on decorrelation of sequential ultrasound images. Urology 2015; 85:648–652. Arif M, Idzenga T, van Mastrigt R, de Korte CL. Estimation of urinary flow velocity in models of obstructed and unobstructed urethras by decorrelation of ultrasound radiofrequency signals. Ultrasound Med Biol 2014;40:938–946. Baxter C, Firoozi F. Pelvic floor ultrasound. In: Fulgham FP, Gilbert RB, (eds). Practical urological ultrasound. New York, NY: Springer New York; 2013. p. 143–153. Cheung CCP, Yu ACH, Salimi N, Yiu BYS, Tsang IKH, Kerby B, Azar RZ, Dickie K. Multi-channel pre-beamformed data acquisition system for research on advanced ultrasound imaging methods. IEEE Trans Ultrason Ferroelectr Freq Control 2012;59:243–253. Ding YY, Ozawa H, Yokoyama T, Nasu Y, Chancellor MB, Kumon H. Reliability of color Doppler ultrasound urodynamics in the evaluation of bladder outlet obstruction. Urology 2000;56:967–971. Gratzke C, Bachmann A, Descazeaud A, Drake MJ, Madersbacher S, Mamoulakis C, Oelke M, Tikkinen KA, Gravas S. EAU Guidelines on the assessment of non-neurogenic male lower urinary tract symptoms including benign prostatic obstruction. Eur Urol 2015;67: 1099–1109. Gustafson KJ, Creasey GH, Grill WM. A urethral afferent mediated excitatory bladder reflex exists in humans. Neurosci Lett 2004; 360:9–12. Ho CK, Chee AJY, Yiu BYS, Tsang ACO, Chow KW, Yu ACH. Wallless flow phantoms with tortuous vascular geometries: Design principles and a patient-specific model fabrication example. IEEE Trans Ultrason Ferroelec Freq Contr 2017;64:25–38. Hou C-P, Lin Y-H, Chen C-L, Tsai Y-L, Chang P-L, Tsui K-H. Impact of the static prostatic urethral angle on men with lower urinary tract symptoms. Urol Sci 2016;27:47–50. Humphrey PA. Male Urethra and external genitalia anatomy. In: Brandes SB, Morey AF, (eds). Advanced male urethral and genital reconstructive surgery. Valley Stream, NY: Humana Press; 2014. p. 17–23. Irwin DE, Kopp ZS, Agatep B, Milsom I, Abrams P. Worldwide prevalence estimates of lower urinary tract symptoms, overactive bladder, urinary incontinence and bladder outlet obstruction. BJU Int 2011; 108:1132–1138. Ishii T, Kambara Y, Yamanishi T, Naya Y, Igarashi T. Urine flow dynamics through prostatic urethra with tubular organ modeling using endoscopic imagery. IEEE J Transl Eng Health Med 2014a;2: 1800709. Ishii T, Nakamura K, Naya Y, Igarashi T. Therapeutic designing for urethral obstruction by virtual urethra and flow dynamics simulation. Minim Invasive Ther Allied Technol 2015;24:141–147. Ishii T, Naya Y, Yamanishi T, Igarashi T. Urine flow dynamics through the urethra in patients with bladder outlet obstruction. J Mech Med Biol 2014b;14:1450052. Jensen JA, Nikolov SI, Gammelmark KL, Pedersen MH. Synthetic aperture ultrasound imaging. Ultrasonics 2006;44(Suppl 1):e5–e15. Jensen JA, Nikolov SI, Yu ACH, Garcia D. Ultrasound vector flow imaging: Part I. Sequential systems. IEEE Trans Ultrason Ferroelec Freq Contr 2016a;63:1704–1721. Jensen JA, Nikolov SI, Yu ACH, Garcia D. Ultrasound vector flow imaging: Part II. Parallel systems. IEEE Trans Ultrason Ferroelec Freq Contr 2016b;63:1722–1732. Karaman M, Li PC, Odonnell M. Synthetic-aperture imaging for smallscale systems. IEEE Trans Ultrason Ferroelectr Freq Control 1995; 42:429–442. Kohler TS, Yadven M, Manvar A, Liu N, Monga M. The length of the male urethra. Int Braz J Urol 2008;34:451–454.
9
Ku JH, Ko DW, Cho JY, Oh SJ. Correlation between prostatic urethral angle and bladder outlet obstruction index in patients with lower urinary tract symptoms. Urology 2010;75:1467–1471. Kurata T, Li Z, Oda S, Kawahira H, Haneishi H. Impact of vessel diameter and bandwidth of illumination in sidestream dark-field oximetry. Biomed Opt Express 2015;6:1616–1631. Lu JY, Cheng J, Wang J. High frame rate imaging system for limited diffraction array beam imaging with square-wave aperture weightings. IEEE Trans Ultrason Ferroelectr Freq Control 2006;53: 1796–1812. Major H, Culligan P, Heit M. Urethral sphincter morphology in women with detrusor instability. Obstet Gynecol 2002;99:63–68. McNeal JE. The zonal anatomy of the prostate. Prostate 1981;2:35–49. Mihmanli I, Kantarci F, Gulsen F, Kadioglu A. Transrectal voiding sonourethrography for diagnosis of a prostatic urethral calculus. J Ultrasound Med 2006;25:1455–1457. Montaldo G, Tanter M, Bercoff J, Benech N, Fink M. Coherent planewave compounding for very high frame rate ultrasonography and transient elastography. IEEE Trans Ultrason Ferroelectr Freq Control 2009;56:489–506. Nurnberger N. Normal values of urine stream velocity in boys. Z Urol Nephrol 1985;78:539–544. Ostergard DR. The neurological control of micturition and integral voiding reflexes. Obstet Gynecol Surv 1979;34:417–423. Ozawa H, Igarashi T, Uematsu K, Watanabe T, Kumon H. The future of urodynamics: Non-invasive ultrasound videourodynamics. Int J Urol 2010;17:241–249. Posada D, Poree J, Pellissier A, Chayer B, Tournoux F, Cloutier G, Garcia D. Staggered multiple-PRF ultrafast color Doppler. IEEE Trans Med Imag 2016;35:1510–1521. Prada PJ, Fernandez J, Martinez AA, Rua ADL, Gonzalez JM, Fernandez JM, Juan G. Transperineal injection of hyaluronic acid in anterior perirectal fat to decrease rectal toxicity from radiation delivered with intensity modulated brachytherapy or EBRT for prostate cancer patients. Int J Radiat Oncol Biol Phys 2007;69:95–102. Schafer W, Abrams P, Liao L, Mattiasson A, Pesce F, Spangberg A, Sterling AM, Zinner NR, van Kerrebroeck P, International Continence Society. Good urodynamic practices: Uroflowmetry, filling cystometry, and pressure-flow studies. Neurourol Urodyn 2002;21: 261–274. So HKH, Chen J, Yiu BYS, Yu ACH. Medical ultrasound imaging: To GPU or not to GPU? IEEE Micro 2011;31:54–65. Stalder AF, Russe MF, Frydrychowicz A, Bock J, Hennig J, Markl M. Quantitative 2 D and 3 D phase contrast MRI: Optimized analysis of blood flow and vessel wall parameters. Magn Reson Med 2008; 60:1218–1231. Stevens E. Bladder ultrasound: Avoiding unnecessary catheterizations. Medurg Nurs 2005;14:249–253. Sughimoto K, Shimamura Y, Tezuka C, Tsubota K, Liu H, Okumura K, Masuda Y, Haneishi H. Effects of arterial blood flow on walls of the abdominal aorta: Distributions of wall shear stress and oscillatory shear index determined by phase-contrast magnetic resonance imaging. Heart Vessels 2016;31:1168–1175. Tojo M, Yasuda K, Yamanishi T, Hattori T, Nagashima K, Shimazaki J. Relationship between bladder neck diameter and hydraulic energy at maximum flow. J Urol 1994;152:144–149. Torii R, Oshima M, Kobayashi T, Takagi K, Tezduyar TE. Numerical investigation of the effect of hypertensive blood pressure on cerebral aneurysm—Dependence of the effect on the aneurysm shape. Int J Numer Meth Fluids 2007;54:995–1009. Tubaro A, De Nunzio C, Puccini F, Presicce F. The evolving picture of lower urinary tract symptom management. Eur Urol 2015;67: 271–272. Versteeg HK, Malalasekera W. An Introduction to computational fluid dynamics: The finite volume method. Harlow, Essex: Pearson Education Limited; 2007. Weiss GH, Nossal R, Bonner RG. Statistics of penetration depth of photons re-emitted from irradiated tissue. J Mod Opt 1989;36:249–259. White BR, Pierce MC, Nassif N, Cense B, Park BH, Tearney GJ, Bouma BE, Chen TC, de Boer JF. In vivo dynamic human retinal blood flow imaging using ultra-high-speed spectral domain optical Doppler tomography. Opt Express 2003;11:3490–3497.
10
Ultrasound in Medicine and Biology
Yamanishi T, Yasuda K, Sakakibara R, Hattori T, Tojo M. The effectiveness of terazosin, an alpha1-blocker, on bladder neck obstruction as assessed by urodynamic hydraulic energy. BJU Int 2000;85: 249–253. Yiu BYS, Lai SSM, Yu ACH. Vector projectile imaging: Time-resolved dynamic visualization of complex flow patterns. Ultrasound Med Biol 2014;40:2295–2309.
Volume -, Number -, 2017 Yiu BYS, Tsang IKH, Yu ACH. GPU-based beamformer: Fast realization of plane wave compounding and synthetic aperture imaging. IEEE Trans Ultrason Ferroelectr Freq Control 2011;58: 1698–1705. Yiu BYS, Yu ACH. Least-squares multi-angle Doppler estimators for plane wave vector flow imaging. IEEE Trans Ultrason Ferroelec Freq Contr 2016;63:1733–1744.