Download Comparison Among Different High Porosity Stent Configurations

Breigh N. Roszelle1
Daniel Felix Ritchie School of Engineering
and Computer Science,
Department of Mechanical
and Materials Engineering,
University of Denver,
Clarence M. Knudson Hall,
2390 South York Street #200,
Denver, CO 80208
e-mail: [email protected]
Priya Nair
School of Biological
and Health Systems Engineering,
Arizona State University,
Tempe, AZ 85287
L. Fernando Gonzalez
Department of Neurological Surgery,
Jefferson Medical College,
Philadelphia, PA 19107
M. Haithem Babiker
School of Biological
and Health Systems Engineering,
Arizona State University,
Tempe, AZ 85287
Justin Ryan
School of Biological
and Health Systems Engineering,
Arizona State University,
Tempe, AZ 85287
David Frakes
School of Biological
and Health Systems Engineering,
Arizona State University,
Tempe, AZ 85287;
School of Electrical,
Computer, and Energy Engineering,
Arizona State University,
Tempe, AZ 85287
Comparison Among Different
High Porosity Stent
Configurations: Hemodynamic
Effects of Treatment
in a Large Cerebral Aneurysm
Whether treated surgically or with endovascular techniques, large and giant cerebral
aneurysms are particularly difficult to treat. Nevertheless, high porosity stents can be
used to accomplish stent-assisted coiling and even standalone stent-based treatments that
have been shown to improve the occlusion of such aneurysms. Further, stent assisted coiling can reduce the incidence of complications that sometimes result from embolic coiling
(e.g., neck remnants and thromboembolism). However, in treating cerebral aneurysms at
bifurcation termini, it remains unclear which configuration of high porosity stents will
result in the most advantageous hemodynamic environment. The goal of this study was to
compare how three different stent configurations affected fluid dynamics in a large
patient-specific aneurysm model. Three common stent configurations were deployed into
the model: a half-Y, a full-Y, and a crossbar configuration. Particle image velocimetry
was used to examine post-treatment flow patterns and quantify root-mean-squared velocity magnitude (V RMS ) within the aneurysmal sac. While each configuration did reduce
V RMS within the aneurysm, the full-Y configuration resulted in the greatest reduction
across all flow conditions (an average of 56% with respect to the untreated case). The experimental results agreed well with clinical follow up after treatment with the full-Y configuration; there was evidence of thrombosis within the sac from the stents alone before
coil embolization was performed. A computational simulation of the full-Y configuration
aligned well with the experimental and in vivo findings, indicating potential for clinically
useful prediction of post-treatment hemodynamics. This study found that applying different stent configurations resulted in considerably different fluid dynamics in an anatomically accurate aneurysm model and that the full-Y configuration performed best. The
study indicates that knowledge of how stent configurations will affect post-treatment
hemodynamics could be important in interventional planning and demonstrates the capability for such planning based on novel computational tools. [DOI: 10.1115/1.4026257]
Introduction
Treatments for cerebral aneurysms have improved dramatically
over the past few decades; however, large and giant aneurysms
remain notoriously difficult to treat [1]. Large aneurysms are those
with a sac diameter of >10 mm while giant aneurysms are often
defined as having a sac diameter of >25 mm [2,3]. Because of
increased complications during and after treatment, such as hemorrhage, cerebral compression, and thromboembolism, giant
aneurysms account for the highest mortality rates among cerebral
aneurysms [4]. An analysis of mortality and morbidity of patients
who underwent clipping surgery on unruptured cerebral aneurysms by Raaymakers et al. showed that the mortality rate of
patients with posterior circulation nongiant aneurysms was 3.0%,
while the mortality rate of patients with posterior circulation giant
aneurysms was 9.6% [5]. Because surgical clipping has achieved
such limited success in treating giant aneurysms, other techniques
1
Corresponding author.
Contributed by the Bioengineering Division of ASME for publication in the
JOURNAL OF BIOMECHANICAL ENGINEERING. Manuscript received September 2, 2013;
final manuscript received December 11, 2013; accepted manuscript posted
December 16, 2013; published online February 5, 2014. Editor: Victor H. Barocas.
Journal of Biomechanical Engineering
including parent artery occlusion (PAO) have emerged. The goal
of PAO is to reduce blood flow into the aneurysmal sac, and the
technique can be executed either surgically or with endovascular
devices such as detachable balloons and/or coils [6].
Over the past decade, embolic coiling has become the gold
standard for treating cerebral aneurysms. Coiling is less invasive
than traditional surgical methods and also more effective according to the International Subarachnoid Aneurysm Trial [7].
Unfortunately, large and giant aneurysms, especially those with
neck sizes of over 5 mm, are difficult to treat even with the most
advanced coiling techniques [8]. Larger aneurysms treated with
coiling demonstrate increased incidence of poor initial occlusion,
high rates of neck remnants, poor coil stability, and recanalization
rates of over 50% [9]. However, flexible high porosity stents can
be used to aid in the coil embolization of large and giant aneurysms. The stents are deployed directly into the lumen of the
diseased parent vessel, across the aneurysmal neck. This provides
a bridging scaffold to both support coils and serve as a nidus for
neointimal growth over the aneurysmal neck [10]. Because
treating giant aneurysms with coiling alone often results in neck
remnants, the use of a stent or stents to support the coils and
maintain the patency of the parent vessel can be highly
C 2014 by ASME
Copyright V
FEBRUARY 2014, Vol. 136 / 021013-1
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
beneficial [11]. However, it remains unclear which configuration
of stents is most beneficial in addressing aneurysms at bifurcation
termini.
During stent-assisted coiling, stents are deployed first in order
to keep coils in place as they are deployed. In some cases coiling
is performed immediately after stenting, but in other cases the two
steps may be staged over days or even months as separate interventions [12]. During the time between staged stenting and coiling, there is often some thrombosis within the aneurysmal sac,
indicating that stents alone can effect hemodynamic changes that
facilitate aneurysmal exclusion [13]. It is also noteworthy that the
later coiling stage of stent-assisted coiling can in some cases fail,
making the earlier stenting stage the effective conclusion of the
treatment. Accordingly, the configuration of deployed stents can
have considerable impact on the overall outcome of treatment.
Determining the best stent configuration to promote beneficial
hemodynamics is, thus, an important consideration in treatment
design.
Currently, there is no consensus as to which stent configuration
is best for treating aneurysms at bifurcation termini [14]. When
dealing with basilar tip aneurysms, the characteristic vascular geometry allows for several different configurations to be deployed
within the branching arteries. These include the half-Y, full-Y,
and crossbar configurations, all of which are shown in Fig. 1. A
previous study by our group investigated different stent configurations in an idealized wide-neck aneurysm model both experimentally and computationally, and it was clear that each configuration
led to different flow patterns [15]. Because the study examined
only a single idealized model, characterizing flows in additional
anatomical models may help improve understanding of the effects
that stent configuration has under more clinically realistic conditions. Therefore, this study considered a model of a large patientspecific aneurysm to further explore the flow characteristics
effected by three different stent configurations. Specifically, we
used particle image velocimetry (PIV) to assess fluid dynamic outcomes after treatment with the different stent configurations and
then compared the experimental findings to clinical follow up
with favorable agreement. We also applied novel finite element
(FE) models and computational fluid dynamics to simulate the
course of treatment pursued in vivo, thereby, demonstrating
potential for interventional planning.
Methodology
In Vitro Experiments
Modeling. The basilar tip aneurysm examined in this study was
selected because of its large sac and neck sizes, which made it a
very poor candidate for coil embolization alone. Specifically, the
aneurysm had a maximum sac diameter of 14.3 mm and a minimum neck diameter of 5.4 mm at the parent vessel. It was also
selected because post-treatment clinical follow up data (imagebased and otherwise) were available. A three-dimensional model
of the aneurysm is shown in Fig. 2.
In order to reconstruct the aneurysmal geometry, computed tomography image data were first enhanced and segmented using custom MATLAB code (Mathworks, Natuck, MA) [16,17]. Thresholding
and region growing segmentation operations were then performed
within Mimics software (Materialize, Ann Arbor, MI) to form a 3D
computational model of the aneurysm and associated vessels. The
computational model was translated into a physical core model
made of wax using an R66þ Solidscape 3D printer (Solidscape,
Merrimack, NH). The wax core was then recast to form a refined
metallic core so as to achieve regulated smoothness along the vessel
walls. A final in vitro flow model was then made by casting the
refined metallic core in urethane and melting the core out, leaving a
lost-core model suitable for optical experiments.
Device Deployment. Three different configurations of
Neuroform stents (Stryker, Freemont, CA) were investigated: the
021013-2 / Vol. 136, FEBRUARY 2014
Fig. 1 Illustrations of three different stent configurations: (a)
half-Y, (b) crossbar, and (c) full-Y
half-Y, full-Y, and crossbar configurations. Each configuration is
shown in Fig. 1. Experiments were also run on the untreated case
for comparison.
Flow Measurement. In order to mimic in vivo flow conditions,
a neurovascular flow loop apparatus was developed. The loop was
filled with a blood analog comprised of water, aqueous sodium
iodide, and glycerol designed to have a viscosity of 3.16 cP at a
25 C operating temperature. The refractive index of the fluid was
matched to that of urethane to facilitate PIV measurements, and
8 lm fluorescent particles (Thermo Scientific, Waltham, MA)
were seeded into the fluid.
Flow within the loop was driven by a peristaltic pump (Harvard
Apparatus, Holliston, MA) through flexible polyvinyl chloride
Transactions of the ASME
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Computational Simulations
Fig. 2 Computational model of the patient-specific aneurysm
tubing. Both steady and pulsatile conditions were investigated at
flow rates of 3, 4, and 5 ml/s. During steady flow experiments, the
fluid was directed through a compliance chamber to remove pulsatile effects. During pulsatile flow experiments, the chamber was
removed and the pump was configured to reproduce a vertebral
flow waveform from literature [18]. In both cases, the flow rate
through the loop was continuously monitored with a flow meter
(Omega Engineering, Inc., Stamford, CT). Inflow to the aneurysm
model was also monitored using a fluid-filled pressure transducer
(Harvard Apparatus, Holliston, MA) attached to the tubing
directly upstream of the model. The pressure transducer was connected to an amplifier unit (Harvard Apparatus, Holliston, MA)
and waveform measurements were monitored using an
oscilloscope.
PIV was performed using a Flowmaster 3D stereo PIV system
(LaVision, Ypsilanti, MI) comprising a 532 nm wavelength Solo
PIV III dual cavity pulsed YAG laser (New Wave Research, Fremont, CA) and two Imager Intense cross correlation CCD cameras. A thickness of 0.5 mm was maintained for the laser light
sheet. The fluorescent particles illuminated by the laser in the
working fluid experience peak excitation and emission at 542 and
618 nm, respectively. Because stents can cause intense laser
reflections, low-pass optical filters with a 572 nm cutoff (Omega
Optical, Brattle Bro, VT) were installed on the cameras. During
data collection, a total of seven parallel planes were measured,
each 1 mm apart. The test section of the aneurysm selected for
measurement allowed for complete coverage of the aneurysmal
neck.
Experimental Flow Data Processing and Analysis. Two hundred image pairs were acquired (at a rate of 5 Hz) at each imaging
plane and for each flow condition. Flow velocity vectors were calculated using a cross correlation algorithm within DaVis software
(Lavision, Ypsilanti, MI). Interrogation windows with an initial
size of 32 by 32 pixels and a final size of 16 by 16 pixels were
used. An overlap of 50% was specified between neighboring windows. A single velocity flow field was then averaged within each
measurement plane for each flow condition.
Root-mean-squared velocity magnitude (V RMS ) was used to
describe flow conditions inside the aneurysm. Since the metric
considers all data points measured within the aneurysmal sac, a
reduction in V RMS indicates a reduction in overall fluid dynamic
activity within the aneurysm. V RMS is calculated as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n
1X
V rms ¼
jV i j2
n i¼1
where n is the number of data points within the aneurysmal sac
and |Vi| indicates the flow velocity magnitude at point i.
Journal of Biomechanical Engineering
Finite Element Device Deployments. In order to demonstrate
and assess interventional planning capabilities, FE modeling was
used to simulate deployment of the full-Y stent configuration
(which was the course of treatment pursued in vivo). A computational model of the Neuroform stent was first designed in Pyformex2. A description of the computational stent geometry can be
found in Babiker et al. [15]. The computational stent was imported
into Abaqus (Simulia, Providence, RI) where it was meshed using
28,503 four-node linear quadrilateral shell elements of Abaqus element type S4R. Shell elements were chosen because they are computationally less expensive to simulate than conventional 3D solid
elements. Shell elements also perform similarly to 3D solid elements, as demonstrated by Hall and Kasper who showed good
agreement between stents modeled with shell and 3D solid elements during stent expansion simulations [19]. An artificial shell
thickness of 70 lm was applied to the shell elements in Abaqus.
Superelastic Nitinol material behavior was also imposed using a
user defined subroutine in Abaqus, which is described by Gong and
Pelton [20] and Auricchio et al. [21]. A material density of 6.8 g/
cm3 was specified in order to match Nitinol.
Stent deployment in the computational aneurysm model was
simulated in Abaqus/Explicit by (1) crimping the stent into a microcatheter, (2) advancing the microcatheter to the site of the aneurysm, and (3) unsheathing the stent in a step-by-step process. This
technique is similar to the one reported in a previous study by our
group [22]. The aneurysm was modeled as a rigid body with 18,557
quadrilateral elements. The crimper was modeled as a cylindrical
shell with 10,653 quadrilateral elements and the material properties
of steel. A radial displacement boundary condition was imposed on
the crimper to uniformly compress the stent into a 0.54 mm microcatheter. The microcatheter was also modeled as a cylindrical
sheath with 27,189 shell elements, a 30 GPa elastic modulus, a 0.3
Poisson ratio, and a 7 g/cm3 density. After crimping, the microcatheter was advanced along the vessel centerline to the site of the aneurysm, as shown in Fig. 3 (panes 1 and 2 and 6 and 7).
Microcatheter advancement was performed through the kinematic
coupling of a control point with circular nodes lining the microcatheter tip. Displacements were imposed onto the reference point to
guide the microcatheter through the vessel centerline, which is similar to the technique used by Ma et al. [23]. This approach simulates
microcatheter navigation in the clinic, where a guidewire is used to
direct the microcatheter to the site of the aneurysm.
After microcatheter advancement, the stent was divided into
eight subsets and radial constraints were applied to each in order
to constrain the stent radius to that of the microcatheter. Step-bystep, the radial constraints on each subset were relaxed and the
stent subset was allowed to expand, as shown in Fig. 3 (panes
3–5). This approach simulates stent unsheathing in the clinic
where the stent is fixed in place using a pusher and the microcatheter is slowly pulled back to unsheath the stent.
Surface-to-surface contacts were modeled using the “general
contact” algorithm in Abaqus/Explicit. Stent-to-microcatheter,
microcatheter-to-vessel, stent-to-vessel, and stent-to-stent contact
interactions were modeled using frictional coefficients of 0.07,
0.1, 0.07, and 0.07, respectively. The frictional coefficients were
estimated from the literature [24–26].
Computational Fluid Dynamic Simulations. After simulating
stent deployment in Abaqus, deployed stent geometries were
shelled in Geomagic studio (Raindrop Geomagic, Research Triangle Park, NC) with a 70 lm thickness and merged. The shelled
stent geometries and the original computational aneurysm model
were then imported into ANSYS ICEM (ANSYS, Canonsburg,
PA), where the inlet and outlets of the aneurysm model were
extruded. A mesh density function was applied near the stents and
the Octree algorithm was used to discretize the stent and blood
2
Please see pyformex.berlios.de.
FEBRUARY 2014, Vol. 136 / 021013-3
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 3 Image sequence showing the simulated deployment of two Neuroform stents in a full-Y configuration. The final FE simulation result (pane 8) was used in a fluid dynamic simulation.
specify the pressure-velocity coupling. A steady flat 4 ml/s flow
profile was applied at the inlet of the model, and zero pressure
boundary conditions were imposed at the outlets. Although physiologically accurate boundary conditions can be critically important,
we propose that zero pressure conditions are appropriate here based
on the assumptions of equal flow resistance across the outflow
branches and that high porosity stent deployment effects minimal
changes in vascular resistance (the latter assumption is supported
by a previous experimental study from our group [27]). The overall
CFD approach has been described previously by Babiker et al. [15].
Results
In Vitro Results
Effects of Stent Configuration on Aneurysmal Flow. V RMS
reductions within the aneurysmal volume for each treatment (with
respect to the untreated case) are shown in Fig. 4. There was a
reduction in V RMS of at least 25% after any stent treatment regardless of flow conditions. Under steady flow conditions (Fig. 4(a)),
the half-Y configuration was consistently least effective. The
crossbar and full-Y configurations led to similar V RMS reductions
of 50% or above. The full-Y configuration led to the highest V RMS
reductions at 4 and 5 ml/s steady flow rates; however, the crossbar
configuration led to the greatest reduction at 3 ml/s (66.9%).
Under pulsatile flow conditions (Fig. 4(b)), the half-Y and crossbar configurations led to similar V RMS reductions of 25–30%. The
full-Y configuration led to the greatest reductions for all three pulsatile flow rates, with a maximum reduction of 62% at 3 ml/s.
Fig. 4 Reductions in VRMS for all flow conditions explored. The
percentages shown in each column are the reductions in VRMS
with respect to the untreated case.
volumes into approximately 16 106 tetrahedrons. The final mesh
was imported into ANSYS Fluent where the blood volume was
assumed to comprise a Newtonian fluid with the same density and
viscosity as the blood analog solution used in experiments. The
stent volume was modeled as solid. The vessel wall and stent
surfaces were assumed to be rigid, and a no-slip boundary condition was applied at the walls. The SIMPLE algorithm was used to
021013-4 / Vol. 136, FEBRUARY 2014
Aneurysmal Flow Patterns. Flow patterns varied after deployment of each stent configuration as expected. Figure 5 highlights
the flow patterns within a plane near the middle of the measurement volume at 4 ml/s steady flow. Under untreated conditions,
there was a high velocity jet within the aneurysmal sac near the
lower branch vessel, and a region of flow recirculation near the
top branching vessel. Both features are shown in Fig. 5(a). Figure
5(b) shows that the half-Y configuration led to a similar area of
recirculation when compared to the untreated case (near the top
branching vessel corresponding to the unstented side). The half-Y
configuration also led to a high velocity flow region; however,
that region was shifted to the side of the unstented vessel in contrast to the untreated case. The crossbar configuration, as shown in
Fig. 5(c), led to a jet that originated from the parent vessel and
continued on into the center of the aneurysmal sac. Lastly, the
Transactions of the ASME
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 5 Velocity vector flow patterns within the aneurysm for each stent deployment configuration at 4 ml/s steady flow
Fig. 6 Velocity vector flow patterns within the aneurysm for each stent deployment configuration at 4 ml/s pulsatile flow
flow patterns for the full-Y configuration are shown in Fig. 5(d). That
configuration led to a reduction in overall flow velocity and, unlike
the other configurations, did not generate any high velocity jets.
The flow patterns during peak systole of pulsatile flow at 4 ml/s
are shown in Fig. 6. The results are similar to those seen during
steady flow. As shown in Fig. 6(b), there was high velocity flow
into the aneurysm for the half-Y configuration, which is similar to
the untreated case shown in Fig. 6(a). For the crossbar configuration, flow into the aneurysm from the neck plane is reduced in
comparison to the untreated case, as shown in Fig. 6(c). However,
the full-Y configuration as shown in Fig. 6(d) led to greater reductions in flow velocities than either of the other configurations.
Computational Results. Three-dimensional streamtraces for
the untreated and full-Y configurations are presented in Figs. 7(a)
and 7(b), respectively. Figure 7(a) shows a concentrated flow jet
impinging on the aneurysmal fundus that transitions into a recirculating flow pattern near the center of the aneurysm. The impinging
Journal of Biomechanical Engineering
flow jet was noticeably dispersed after the full-Y deployment and
velocity magnitudes within the aneurysm were considerably
reduced, as shown in Fig. 7(b). The reduction in V RMS with
respect to the untreated case was 35.7%.
Wall shear stress (WSS) and spatial wall shear stress gradient
(WSSG) contour plots are presented in Figs. 8 and 9. In the
Figures, WSS and WSSG magnitudes show considerable reductions after treatment. The largest reductions were observed along
the left side of the posterior wall of the aneurysm (oriented as the
right side of the aneurysmal sac in the figures), as shown in panes
in Figs. 8(c) and 8(d) and 9(c) and 9(d).
Clinical Results. The aneurysm we examined was treated clinically with a full-Y configuration. Figure 10(a) shows the aneurysm immediately after treatment; the markers indicating stent
endpoints can be seen in the parent and outlet vessels. The aneurysm was then coiled a month after stenting. During the time
between coiling and stenting, considerable thrombosis
FEBRUARY 2014, Vol. 136 / 021013-5
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 7 Fluid dynamic simulation results showing 3D streamtraces for the untreated case (a) and the full-Y configuration (b) at
4 ml/s steady flow
Fig. 9 Simulated contour plot of the WSSG gradient for the
untreated case (a) and the full-Y configuration (b). Contour
plots are shown for the posterior view of the aneurysm (view
1—(a) and (b)) and anterior view (view 2—(c) and (d)).
Fig. 8 Simulated WSS contour plots for the untreated case (a)
and the full-Y configuration (b). Contour plots are shown for the
posterior view of the aneurysm (view 1—(a) and (b)) and the anterior view (view 2—(c) and (d)).
(approximately 50% by clinical standards) occurred on the left
side of the aneurysmal sac as shown in Fig. 10(b). That partial
occlusion facilitated subsequent coil embolization, which led to a
successful treatment outcome. The one-year post-treatment angiogram shown in Fig. 10(c) indicated complete aneurysmal exclusion without any rebleeding.
Discussion
Whether treated surgically or with endovascular techniques,
large and giant cerebral aneurysms are particularly difficult to
treat [4,5]. Fortunately, stent-assisted coiling has shown promise
in recent years as an effective treatment for such aneurysms [28].
However, the best stent configuration for treating aneurysms at
bifurcation termini remains a subject of clinical debate [14]. A
previous study by our group showed that considerable differences
in aneurysmal fluid dynamics resulted after treatment with different stent configurations in an idealized aneurysm model [15]. The
primary goal of this study was to further understanding of the fluid
dynamics that different stent configurations effect under more
clinically realistic experimental conditions. The study also demonstrated potential for interventional planning based on novel
computational tools, which were used here to simulate posttreatment flow conditions that agreed well with clinical outcomes.
Differences in Aneurysmal Fluid Dynamics Among Stent
Configurations. In Vitro experiments revealed considerable differences among aneurysmal fluid dynamics after treatment with
021013-6 / Vol. 136, FEBRUARY 2014
different stent configurations. In that regard, the findings agree
with previous in vitro studies by our group that investigated simpler aneurysm models as well as computational studies by others
involving anatomical models [29,15]. In this study, even in cases
where the reductions in V RMS effected by different stent configurations were similar, the underlying aneurysmal flow patterns
were quite different. For example, under steady flow conditions,
the crossbar and full-Y configurations led to similar reductions in
V RMS ; however, the crossbar configuration led to a concentrated
jet through the aneurysmal neck, which was not observed after
treatment with the full-Y configuration. These two different stent
configurations and the dissimilar flow patterns they effect could,
therefore, lead to important differences in clinical outcomes.
In Vitro results showed that the full-Y configuration led to the
most advantageous fluid dynamic environment among the different stent configurations we considered. Specifically, results
showed that the full-Y led to the largest reductions in V RMS as
compared to the untreated case for all of the flow conditions
examined (with one exception). Further, flow patterns for the fullY showed an overall reduction in high flow velocities, a lack of
concentrated flow jets, and larger regions of low flow velocity
within the aneurysm. Such regions have been correlated with
thrombus formation sites, which can facilitate aneurysmal exclusion, in prior studies [30].
Influence of Aneurysmal Geometry on Post-Treatment
Fluid Dynamics. One critically important factor in determining
the optimal stent configuration is the geometry of the aneurysm
being treated. While the full-Y configuration led to the most advantageous fluid dynamic environment for the geometry we examined, results may differ for other geometries. For example, a case
study by Cross et al. found that the full-Y configuration was a
poor choice for treating one patient’s basilar tip aneurysm because
the intersection of the two stents did not reach the midpoint of the
aneurysmal neck, which reduced stent coverage there and
increased the risk of coil herniation into the parent vessel [14].
The importance of geometry is also highlighted when comparing the patient-specific model from this study to the idealized
model examined by Babiker et al. [15]. Figure 11 compares
in vitro reductions in V RMS with respect to the untreated case, for
each stent configuration, between the patient-specific and idealized aneurysm geometries. Some similarities can be observed
between the two models. For example, the half-Y configuration
was consistently least effective in reducing V RMS within the aneurysm. Further, the crossbar configuration led to a flow jet
Transactions of the ASME
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 10 Digital subtraction angiography images of the large aneurysm at different
stages of treatment: (a) immediately after stent treatment, (b) one month after stent
treatment, before coiling (note the partial occlusion of the sac), and (c) one year after stent assisted coiling
Fig. 11 Comparison of VRMS reductions associated with all
three stent configurations in both the patient-specific anatomical model and a previously examined idealized model
originating from the aneurysmal neck in both cases whereas the
full-Y configuration led to better dissipation of the flow jet. Significant lower recanalization was observed after comparing standalone coiling to half Y and even less need for retreatment after the
Y configuration. Non significant but lower recanalization and
need for retreatment has been shown after a “Y” configuraltion
compared to half Y. However, while similarities do exist, the two
models also effected many fluid dynamic differences. For example, the crossbar configuration was more effective in reducing
V RMS for the idealized case. Further, the half-Y configuration had
a minimal effect on V RMS in the idealized model but reduced
V RMS by more than 24% in the patient-specific case. Therefore,
the geometry of the aneurysm and parent vessel clearly influence
the hemodynamic effects of each stent configuration.
Although they are not a focal point of this study, the considerable effects that experimental inflow conditions had on posttreatment fluid dynamics should also be noted. For example, the
crossbar configuration reduced V RMS by the greatest degree for
one of the steady flow rates, but was far from best under any pulsatile flow conditions. Like the effects that geometry can have on
post-treatment flows, the considerable effects that inflow
conditions can have further underscore the critical importance of
patient-specificity for realistic interventional planning.
Comparisons Among Experimental, Simulated, and Clinical
Results. Although experiments were the primary source of data
used in this study to evaluate three different stent configurations, a
simulation of the full-Y configuration was also performed and
Journal of Biomechanical Engineering
clinical data describing that configuration’s use in vivo were gathered. Accordingly, three different characterizations of fluid dynamics affected by the full-Y were available. In comparing the
experimental and simulated characterizations of the full-Y, results
agreed well. Both demonstrated reductions in V RMS , similarly
localized regions of low velocity flow, and dissipation of the
impingement jet observed in the untreated model; however, the
experimental model demonstrated greater dissipation and lower
overall flow velocities. Differences between the results can be
attributed to several factors, one of which is that the stent deployments were not matched exactly in terms of rotation and/or absolute endpoints (such matching is extremely if not prohibitively
difficult). Other potential sources of differences between the experimental and simulated results include that the outflow boundary conditions were different (although symmetric in each case),
and that the aneurysmal volumes used to calculate V RMS were
slightly different due to spatial limitations on data collected
in vitro. Nevertheless, the many strong similarities between the
experimental and simulated cases (quantitative and otherwise)
provide confidence that the simulations generated reliable data.
In comparing the experimental and simulated results to in vivo
data, both agreed well with the limited information provided by
clinical follow up. Specifically, digital subtraction angiography
images acquired 3 months after initial treatment with stents
showed partial thrombosis on one side of the aneurysmal sac.
That region is identified in Fig. 8(b). Flow patterns observed in
the experimental and simulated data shown in Figs. 6(d) and 7(b)
show greatly reduced aneurysmal velocities toward both sides of
the aneurysmal sac, including the site where thrombus formed
in vivo (note the presence of low velocity regions below 0.05 m/s
in Fig. 6(d) and the absence of high velocity streamtraces above
0.18 m/s in Fig. 7(b), toward the sides of the aneurysm in each
case). The simulated data also show lower post-treatment WSS
and WSSG magnitudes at the region where thrombus formed.
Areas of low WSS have been proposed as sites where thrombus is
likely to form [31]. While this study was limited to comparing
across experimental, computational, and clinical results for only
one of the stent configurations that was considered, all of the
results matched well for that configuration.
Implications for Interventional Planning. The different fluid
dynamic outcomes of treatment that were observed in this study
underscore the important roles that both experiments and simulations can play in the design of effective interventions by facilitating selection of the best stent configuration among alternatives for
example. Although it would be impractical to perform an experimental study for every patient-specific case prior to treatment,
experiments provide well founded but more general knowledge of
the hemodynamics that different stent configurations will promote. Simulations, on the other hand, can build upon the
fundamental knowledge that experiments contribute and complement such knowledge with patient-specific predictions of posttreatment hemodynamics that can be generated on timescales
FEBRUARY 2014, Vol. 136 / 021013-7
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
compatible with clinical workflows. Together, parallel bench top
experiments and computational simulations can, thus, provide a
well-rounded foundation to aid physicians in patient-specific
interventional planning. For example, understanding that the
full-Y configuration typically promotes low velocity flow regions
toward the lateral boundaries of a bifurcation aneurysm becomes
more meaningful when a patient-specific simulation (considering
realistic geometric and inflow conditions) confirms that result.
More detailed aspects of the deployment plan can then be finetuned with further simulations to optimize the prevalence and
expanse of the low velocity flow regions or other potentially beneficial fluid dynamic features. One aim of future work is to apply
the computational approaches used in this study to a larger patient
population and, thereby, identify the hemodynamic predictors of
clinical outcomes.
Study Limitations. Several limitations of this study are noteworthy. First, only a single anatomical model was considered.
However, results were compared to another model from previous
work to add depth to the study. Second, both the experimental and
computational aneurysm models we investigated were rigid wall
models. Clearly more realistic moving wall boundaries would
result in somewhat different fluid dynamic outcomes; however,
previous studies have shown that aneurysmal fluid dynamics are
dictated primarily by baseline geometry, which we recreated faithfully in this study for a patient-specific anatomical case [32].
Nevertheless, models with compliant vessel walls (both experimental and computational) have been used in our lab and will be
considered in future work (this will include computational models
with fluid structure interaction). Lastly, the comparisons that were
made to clinical follow up data in this study were highly qualitative. That limitation is in large part due to the nature of the clinical
follow up data available, but more quantitative comparisons to
clinical follow up data will be included as available in future
work.
Conclusion
This study characterized the considerable fluid dynamic differences that three different stent configurations affected in an anatomical basilar tip aneurysm model. Understanding of those
differences is important because the hemodynamics that result
from stenting alone may play an important role in the overall hemodynamic outcome of stent assisted coiling. Among the stent
configurations investigated, the full-Y variant performed best. The
aneurysm examined in this study was treated successfully in vivo
with a full-Y configuration, and clinical follow up data agreed
well with experimental results that demonstrated low velocity
flow regions at the in vivo thrombus site. A computational simulation of the full-Y configuration predicted a low velocity flow
region at that site as well, which indicates promise for interventional planning of endovascular treatments. The study also showed
that aneurysmal geometry can have an important impact on the
fluid dynamics effected by a given stent configuration. Future
work will comprise examination of additional anatomical aneurysm models and endovascular devices, both in vitro and in silico,
as well as further development of our computational tools for
interventional planning. As the tools for interventional planning
are developed, the long-term goal is to establish protocols and
identify predictors from the modeling data that can be used to aid
in optimizing the clinical outcomes of cerebral aneurysm
treatments.
Acknowledgment
The authors acknowledge funding for this project from the following sources: Brain Aneurysm Foundation Research Grant,
National Science Foundation CAREER Award, American Heart
Association Beginning Grant in Aid, and Women in Philanthropy
Society Category B Grant.
021013-8 / Vol. 136, FEBRUARY 2014
References
[1] Lawton, M. T., and Spetzler, R. F., 1995, “Surgical Management of Giant Intracranial Aneurysms: Experience With 171 Patients,” Clin. Neurosurg., 42, pp.
245–266.
[2] Whittle, I. R., Dorsch, N. W., and Besser, M., 1982, “Spontaneous Thrombosis
in Giant Intracranial Aneurysms,” J. Neurol., Neurosurg. Psychiatry, 45, pp.
1040–1047.
[3] Johnston, S. C., Higashida, R. T., Barrow, D. L., Caplan, L. R., Dion, J. E.,
Hademenos, G., Hopkins, L. N., Molyneux, A., Rosenwasser, R. H., Vinuela, F.,
and Wilson, C. B., 2002, “Recommendations for the Endovascular Treatment of
Intracranial Aneurysms: A Statement for Healthcare Professionals From the
Committee on Cardiovascular Imaging of the American Heart Association
Council on Cardiovascular Radiology,” Stroke, 33, pp. 2536–2544.
[4] Sluzewski, M., Menovsky, T., Jan van Rooij, W., and Wijnalda, D., 2003,
“Coiling of Very Large or Giant Cerebral Aneurysms: Long-Term Clinical and
Serial Angiographic Results,” AJNR Am. J. Neuroradiol., 24, pp. 257–262.
[5] Raaymakers, T. W. M., Rinkel, G. J. E., Linburg, M., and Algra, A., 1998,
“Mortality and Morbidity of Surgery for Unruptured Intracranial Aneurysms: A
Meta-Analysis,” Stroke, 29, pp. 1531–1538.
[6] Arat, A., Islak, C., Saatci, I., Kocer, N., and Cekirge, S., 2002, “Endovascular
Parent Artery Occlusion in Large-Giant or Fusiform Distal Posterior Cerebral
Artery Aneurysms,” Neuroradiology, 44, pp. 700–705.
[7] International Subarachnoid Aneurysm Trial (ISAT) Collaborative Group, 2002,
“International Subarachnoid Aneurysm Trial (ISAT) of Neurosurgical Clipping
Versus Endovascular Coiling in 2143 Patients With Ruptured Intracranial
Aneurysms: A Randomized Trial,” Lancet, 360, pp. 1267–1274.
[8] Vinuela, F., Duckwiler, G., and Mawad, M., 1997, “Guglielmi Detachable Coil
Embolization of Acute Intracranial Aneurysm: Perioperative Anatomical and
Clinical Outcome in 403 Patients,” J. Neurosurg., 86, pp. 475–482.
[9] Murayama, Y., Nien, Y., Duckwiler, G., Gobin, Y. P., Jahan, R., Frazee, J.,
Martin, N., and Vi~
nuela, F., 1998, “Guglielmi Detachable Coil Emolization of
Cerebral Aneurysms: 11 Years’ Experience,” J. Neurosurg., 98, pp. 959–966.
[10] Pumar, J. M., Lete, I., Pardo, M. I., Vazquez-Herrero, F., and Blanco, M., 2008, “LEO
Stent Monotherapy for the Endovascular Reconstruction of Fusiform Aneurysms of
the Middle Cerebral Artery,” AJNR Am. J. Neuroradiol., 29, pp. 1775–1776.
[11] Lylyk, P., Cohen, J. E., Ceratto, R., Ferrario, A., and Miranda, C., 2002,
“Endovascular Reconstruction of Intracranial Arteries by Stent Placement and
Combined Techniques,” J. Neurosurg., 97, pp. 1306–1313.
[12] Akpek, S., Arat, A., Morsi, H., Klucznick, R. P., Strother, C. M., and Mawad, M. E.,
2005, “Self-Expandable Stent-Assisted Coiling of Wide-Necked Intracranial Aneurysms: A Single-Cent Experience,” AJNR Am. J. Neuroradiol., 26, pp. 1223–1231.
[13] Wanke, I., and Forsting, M., 2008, “Stents for Intracranial Wide-Necked Aneurysms: More Than Mechanical Protection,” Neuroradiology, 50, pp. 991–998.
[14] Cross, D. T., Moran, C. J., Derdeyn, C. P., Mazumdar, A., Rivet, D., and Chicoine, M. M., 2005, “Neuroform Stent Deployment for Treatment of a Basilar
Tip Aneurysm via a Posterior Communicating Artery Route,” AJNR Am. J.
Neuroradiol., 26, pp. 2578–2581.
[15] Babiker, M. H., Gonzalez, L. F., Ryan, J., Albuquerque, F., Collins, D., Elvikis,
A., and Frakes, D. H., 2011, “Influence of Stent Configuration on Cerebral Aneurysm Fluid Dynamics,” J. Biomech., 45, pp. 440–447.
[16] Frakes, D., Pekkan, K., Dasi, L., Kitajima, H. D., de Zelicourt, D., Leo, H. L.,
Carberry, J., Sundareswaran, K., Simon, H., and Yoganathan, A. P., 2008,
“Modified Control Grid Interpolation for the Volumetric Reconstruction of
Fluid Flows,” Exp. Fluids, 45, pp. 987–997.
[17] Frakes, D. H., Conrad, C. P., Healy, T. M., Monaco, J. W., Fogel, M., Sharma,
S., Smith, M. J., and Yoganathan, A. P., 2003, “Application of an Adaptive
Control Grid Interpolation Technique to Morphological Vascular
Reconstruction,” IEEE Trans. Biomed. Eng., 50(2), pp. 197–206.
[18] Ford, M. D., Alperin, N., Lee, S. H., Holdsworth, D. W., and Steinman, D. A.,
2005, “Characterization of Volumetric Flow Rate Waveforms in the Normal Internal Carotid and Vertebral Arteries,” Physiol. Meas., 26, pp. 477–488.
[19] Hall, G., and Kasper, E., 2006, “Comparison of Element Technologies for Modeling Stent Expansion,” ASME J. Biomech. Eng., 128(5), pp. 751–756.
[20] Gong, X. Y., and Pelton, A., 2004, “Finite Element Analysis on Nitinol Medical
Applications,” Proceedings of the International Conference on Shape Memory and
Superelastic Technologies, ASM International, Pacific Grove, CA, pp. 443–451.
[21] Auricchio, F., Taylor, R. L., and Lubliner, J., 1997, “Shape-Memory Alloys:
Macromodelling and Numerical Simulations of the Superelastic Behavior,”
Comput. Methods Appl. Mech. Eng., 146, pp. 281–312.
[22] Babiker, M. H., Chong, B. W., Gonzalez, L. F., and Frakes, D. H., 2013,
“Simulating the Effects of Embolic Coils on Cerebral Aneurysm Fluid Dynamics
Using Finite Element Modeling,” Proceedings of the American Society of Mechanical Engineering (ASME) Summer Bioengineering Conference, Sunriver, OR.
[23] Ma, D., Dargush, G., Natarajan, S., Levy, E. I., Siddiqui, A. H., and Meng, H.,
2012, “Computer Modeling of Deployment and Mechanical Expansion of Neurovascular Flow Diverter in Patient-Specific Intracranial Aneurysms,” J. Biomech., 45(13), pp. 2256–2263.
[24] Dunn, A., Zaveri, T., Keselowsky, B., and Sawyer, W. G., 2007, “Macroscopic
Friction Coefficient Measurements on Living Endothelial Cells,” Tribol. Lett.,
27(2), pp. 233–238.
[25] Vad, S., Eskinazi, A., Corbett, T., McGloughlin, T., and Vande Geest, J. P.,
2010, “Determination of Coefficient of Friction for Self-Expanding StentGrafts,” ASME J. Biomed Eng, 132(12), p. 121007.
[26] Takashima, K., Shimomura, R., Kitou, T., Terada, H., Yoshinaka, K., and Ikeuchi, K., 2007, “Contact and Friction Between Catheter and Blood Vessel,” Tribol. Int., 40(2), pp. 319–328.
Transactions of the ASME
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use
[27] Roszelle, B. N., Gonzalez, L. F., Babiker, M. H., Ryan, J., Albuquerque, F. C.,
and Frakes, D. H., 2013, “Flow Diverter Effect on Cerebral Aneurysm Hemodynamics: An in vitro Comparison of Telescoping Stents and the Pipeline,”
Neuroradiology, 55, pp. 751–758.
[28] Biondi, A., Janardhan, V., Katz, J. M., Salvaggio, K., Riina, H. A., and Gobin,
Y. P., 2007, “Neuroform Stent-Assisted Coil Embolization of Wide-Neck Intracranial Aneurysms: Strategies in Stent Deployment and Midterm Follow-Up,”
Neurosurgery, 61, pp. 460–469.
[29] Appanaboyina, S., Mut, F., Lohner, R., Putman, C., and Cebral, J., 2009,
“Simulation of Intracranial Aneurysm Stenting: Techniques and Challenges,”
Comput. Methods Appl. Mech. Eng., 198, pp. 3567–3582.
Journal of Biomechanical Engineering
[30] Kim, M., Taulbee, D. B., Tremmel, M., and Meng, H., 2008, “Comparison of
Two Stents in Modifying Cerebral Aneurysm Hemodynamics,” Ann. Biomed.
Eng., 36(5), pp. 726–741.
[31] Rayz, V. L., Boussel, L., Lawton, M. T., Acevedo-Bolton, G., Ge, L.,
Young, W. L., Higashida, R. T., and Saloner, D., 2008, “Numerical
Modeling of the Flow in Intracranial Aneurysms: Prediction of Regions
Prone to Thrombus Formation,” Ann. Biomed. Eng., 36(11), pp.
1793–1804.
[32] Cebral, J. R., Mut, F., Weir, J., and Putman, C., 2011, “Association of Hemodynamic Characteristics and Cerebral Aneurysm Rupture,” AJNR Am. J. Neuroradiol., 32, pp. 264–270.
FEBRUARY 2014, Vol. 136 / 021013-9
Downloaded From: http://biomechanical.asmedigitalcollection.asme.org/ on 09/16/2015 Terms of Use: http://www.asme.org/about-asme/terms-of-use