DEVELOPMENT OF AN ARTIFICIAL ATRIUM FOR THE

ADVANCEMENT OF CATHETERIZATION SURGERIES

C. Coyne, A. Kurz, C. Munger, S. Shrago

ABSTRACT

BACKGROUND

Atrial fibrillation (AF) is a heart arrhythmia in

which aberrant electrical signals are conducted
through cardiac tissue, causing random
contractions of the atria and improper heart rates.
Catheter ablation, a corrective surgery for this
arrhythmia, involves the transmission of electrical
pulses by a special catheter to the conductive
segments of the cardiac tissue, thus modifying the
electrical properties of these areas and preventing
improper electrical activity. Currently, the tracking
methods of ablation surgeries require X-ray and
CT-based monitoring, exposing the patient and
surgeon to dangerous radiation. Additionally, the
physician must manually manipulate the catheter
tip throughout the long surgery, as opposed to an
automated or digital-based method of catheter
control.

A proper heart beat is initiated by an electrical

signal originating in the sinoatrial (SA) node in the
right atrium, depicted in Figure 1. This electrical
signal forces a contraction of the right atrium. At
the same time, the left atrium is stimulated, in
addition to the atrioventricular (AV) node, which
dampens the speed of the signal from the atria to the
ventricles. In turn, the ventricles are given time to
fill as the atria then contract. The ventricles then
contract, delivering blood from the heart to the rest
of the body.

Dr. Grant H. Kruger of the SM Wu Manufacturing

Center at the University of Michigan has developed
a feed-forward robotic catheterization system that
eliminates the need for high-radiation monitoring
and allows the physician more fidelity with
automation controls. Currently, the only test
environment for Dr. Krugers robot has been a
stationary box of fluid. However, the system must
be further advanced before it can be deployed in
clinical settings. Specifically, its feed-forward
inverse kinematic model must be tested and
developed in settings that mimic forces and
disturbances normally present in arrhythmic heart
environments.

Figure 1. Diagram of human heart with AF catheter positions.1

AF is caused by the generation of aberrant electrical

signals from trigger sites originating in the atrial
walls. These atypical signals force high-frequency,
random cyclic contractions of the atria.
In ablation surgeries, diagnostic catheters are
inserted to identify the AF trigger sites, while an
ablation catheter is deployed to transmit the
corrective signal to these locations. In current
settings, the positioning of these catheters are
manually controlled through bending the distal end
or rotating the handle. This causes exhaustive, nonergonomic strain on the operator, while the use of

A dynamic environment that simulates flow

patterns present in the human atrium as well as
the disturbances triggered by atrial fibrillation
was thereby developed by this team for purposes of
advancing Dr. Krugers catheterization system.

1

G. Bian, M. Lipowicz, and G. H. Kruger. SelfLearning of Inverse Kinematics for Feedforward

Control of Intracardiac Robotic Ablation Catheters,

pp. 1-6.


X-ray and CT monitoring systems expose patients
and medical personnel to high amounts of radiation.
Studies indicate that catheter ablation surgeries
have higher efficacy than drug-based therapies.

accuracy and proper operation of the system. In the

case of atrial fibrillation, forces resulting from
cardiac contractions will certainly deform the distal
shaft of the catheter and thus result in positioning
error. In his paper Self-Learning of Inverse
Kinematics for Feedforward Control of
Intracardiac Robotic Ablation Catheters, Dr.
Kruger holds that the three sources responsible for
positioning error are 1) disturbances due to blood
pressure and flow, 2) mechanical contact with the
endocardium, and 3) friction and viscoelastic
effects due to the catheter design. Other sources of
positioning error are the contact between the
catheter tip and the endocardium as well as material
properties of the catheters elasticity responsible for
causing some deflection.

Dr. Kruger has developed a robotic system that

controls the location of catheters using a feedforward control algorithm, eliminating the need for
high-radiation monitoring systems and relieving the
operator of manually positioning the catheter for
long, sustained periods of time. The results of using
a feed-forward loop for catheter control are
depicted in the below images generated from a
magnetic tracking system.

These sources of error in the positioning of

catheters in Dr. Krugers system are far from
negligible. To address them, the systems inverse
kinematic algorithm must be further developed by
deploying a test environment that mimics the flow
patterns, disturbances and other physical
parameters present in a human heart with atrial
fibrillation.
While an artificial heart for the specific purpose of
addressing surgery for AF hasnt yet been
pioneered, several products exist that provide
suitable means of comparison and benchmarking.
These products include an artificial ventricle
employing soft robotics, the SynCardia Total
Artificial Heart (TAH), and a hybrid biomechanical
device developed by a team of engineers at a
European defense and aerospace agency.

While the application of feed-forward control and

magnetic-based validation have significant
potential for advancing ablation surgeries, there are
sources of error to be addressed.

Among all benchmark products, the artificial

ventricle developed by Dr. Ellen T. Roche of
Harvard University has proven the most useful in
understanding various soft robotic technologies for
replicating complicated motions. In addition, Dr.
Roches work has undergone substantial validation
and has proven to be a low-cost method for a
product that is notoriously expensive. However, it
was not developed for application with intracardiac
catheters, and therefore is not suitable for this
application.

In feed-forward control, one difficulty is the

presence of multiple disturbances within the
system, which must be predicted a priori to ensure

This team was tasked with developing a beating

heart model to provide Dr. Kruger with a natural
heart environment that would precisely simulate

Figure 2. Catheter tip position data gathered by magnetic tracking

system


both the anatomy and physiology of the heart as
well as the sources of disturbance present in a heart
with atrial fibrillation. This model would then be
used in conjunction with Dr. Krugers robotic
system in an effort to advance the mathematical
control model.

event that his tracking system may not be used with

any version of this groups model.
Dimensional Requirements
Accurate inner dimensions (10/10): In order for the
product to simulate the environment of a human
heart, the physical dimensions of the atria and
ventricles must lie within the normal range of a
human heart.

METHODS
Stakeholder Needs and Specifications
The primary user of the beating heart model will be
Dr. Kruger'. The model must accurately simulate
the environment of a human heart with AF so that
the users robotic system can be operated and tested
within our product. In a meeting with Dr. Kruger, it
was confirmed that the following requirements
form the basis of his needs.

Accurate motion (8/10): The products output

motion due to the pumping mechanisms must
accurately simulate the beating motion of a heart.
Accurate flow properties (10/10): The blood-like
fluid within our heart model must behave exactly
like the blood in a human heart. If it doesnt, the
model will not accurately simulate the environment
of a human heart and thus cannot be used as a
proper environment for the users equipment. The
specific fluid properties are discussed in the
Engineering Specifications section.

Material Requirements
Compatibility with catheter (10/10): The devices
inlet at the right atrium must permit the insertion
of a catheter. Inside the model, the septum must
allow the penetration of the catheter to reach the
left atrium.

Blood-like viscosity (5/10): While viscosity might

be addressed by the previous paragraph, the user
has made a separate point for this team to use a fluid
with a viscosity very similar to that of blood. If the
model uses a fluid without this property, the
catheters will be subject to unintended motion and
forces.

Conductive targets (8/10): In catheter ablation,

conductive targets that transfer aberrant electrical
signals within the heart must be detected by a
catheter (and later, treated with an electrical signal
to alter the tissue). This teams product must
contain conductive targets that will simulate these
areas in a human heart.

Miscellaneous Requirements

Transparency / translucency (5/10): Dr. Kruger has

confirmed that the beating heart model should
allow the audience to view activity inside of the
heart. Thus, the user of this groups product has
made it clear that a transparent or translucent
material would be preferable over an opaque one.

Variable settings (9/10): The heart models ability

to be switched through a variety of beat
frequencies and amplitudes is important because
these conditions differ among patients.
Easy replacement of atria (3/10): In the event that
an atrium wall is punctured by a catheter, a
method to rapidly reproduce and reinstall an
atrium is preferred by the user.

Compatibility with tracking (10/10): Tracking the

position of the catheter in real time is a critical
requirement for the user. The tracking device is a
magnetic system, so the user requires this group to
not include any metallic elements that may interfere
with the tracking device.

Ease of operation (1/10): The user prefers a

system that is simple and easy to use.
Engineering Specifications
The team formulated a range of engineering
specifications to attribute quantifiable values to the
users requirements. These specifications are
divided
into
five
categories:
material,

Compatibility with ultrasound (5/10): The ability to

monitor the system using ultrasound technology is
preferred. The user requires this capability in the


manufacturing, fluids, electrical, and budget. In a
quality functional deployment (QFD) matrix, their
values were weighted (and ultimately summated
and analyzed) according to the user requirements.
From this process, each engineering specification
was designated an importance rating from 1-9. In
the list below, a +, -, or N is indicated alongside
every engineering specification to indicate a
positive, negative, or N/A (not applicable)
desirability attribute of that property as it relates to
our design.

any powders or additives in the product) is an

appropriate unit.
Measurement unit: Conductivity (S/m)
Pressure, volume & flow detection (+) (8/9): Dr.
Kruger has emphasized a comprehensively
validated and simple model over an unsubstantiated
yet complex model. The material selected must
therefore allow for the incorporation of small
measurement sensors to corroborate accurate
pressure profiles, chamber dimensions, heartbeat
waveforms, etc.
Measurement unit: Pressure (Pa), Volume (mL),
Flow rate (mL/min)

Material Specifications
Sensitivity to touch (+) (1/9): Since the material
must react to various stimuli (i.e. physical contact,
electrical impulses), the materials strain is used to
gauge its sensitivity.
Measurement unit: Strain: ()

Manufacturing Specifications
Manufacturability (+) (9/9): In accordance with the
users request for an easily replaceable atrium or
part, this team believes the products
manufacturability (its simplicity and ease of
manufacturing) is an appropriate specification.
Measurement unit: Subjective scale: 1-5

Flexibility (+) (3/9): Since the material must be

resilient to withstand probing and other forms of
physical contact, the Youngs Modulus (stiffness)
of the material is used to gauge flexibility.
Measurement unit: Youngs Modulus: E (Pa)

Inner volume (-) (5/9): In accordance with the

users request for exact dimensions of the human
heart, this group looks to conserve as much inner
volume as possible. The amount of volume
conserved (as a difference in volume in comparison
to that of a human heart) is the appropriate metric.
Measurement unit: Volume (m3)

Porosity (-) (3/9): To maintain the proper blood

flow characteristics, the heart models walls cannot
become filled with blood. Therefore, a proper
metric of the material is its porosity (ratio of
volume of materials voids to its actual volume).
Measurement unit: Percentage: % ()

Number of molds in composite (-) (9/9): In order to

provide a system in which an atrium or part may be
replaced, a system with variable parts (in this case,
molds) is preferred. However, it is preferred that the
number of molds is not excessively high, which
would
overcomplicate
the
design
and
manufacturing.
Measurement unit: Quantity: #

Bonding ability (+) (9/9): In order for the heart

model to be mounted to a surface (and to prevent
against any fluid leakage), its material must be able
to be sealed. For the seal to stay bonded, the
materials bonding property must be specified on a
subjective scale.
Measurement unit: Subjective scale: 1-5
Durability (+) (4/9): As previously stated, the
product must withstand various stimuli and probes.
It must also survive long periods of use and
catheterization. The materials durability is an
appropriate metric for this.
Measurement unit: Yield strength: Y (Pa)

Fluids Specifications
Volumetric flow rate (N/A) (2/9): Keeping the
fluids behavior in line with that of a human heart
must be tracked by several fluid properties. One
which stays constant throughout a system with
equal mass transfer is the volumetric flow rate.
Measurement unit: m3/s

Conductivity (+) (6/9): The heart model must

contain conductive targets to simulate aberrant
patches of conductive tissue in patients with AF.
The materials conductivity (or the conductivity of

Intracardiac pressure (N/A) (2/9): Intracardiac

pressure is a measure of the pressure within a heart
chamber (right/left atria, ventricles) at any point in
4


the heartbeat cycle. The heartbeat cycle is complex,
with many fluctuations valve openings/closures
within a small interval of time. Thus, it is important
to keep track of these pressures and design our
pump system around them.
Measurement unit: mmHg

our budget, a calculation of the total cost is

employed.
Measurement unit: Price (USD)
Discussion of User Requirements, Engineering
Specifications
After carefully gathering user requirements,
assigning them engineering specifications and
determining the importance level of each, our team
had a solid preliminary concept of the beating heart
model.

Viscosity (N/A) (3/9): In accordance with the users

request for a fluidic viscosity equal (or extremely
similar) to that of blood, the team found viscosity
to be an appropriate specification for this aspect.
Viscosity is also important, since it can give the
team an idea of how the fluid will behave against
the inner heart wall material.
Measurement unit: (m2/s)

According to Dr. Kruger, the most crucial elements

of the model would be its compatibility with
catheters (weight: 10/10), accurate inner
dimensions (weight: 10/10), accurate flow
properties (weight: 10/10) and variable settings
(weight: 9/10). Indeed, the model would have to
accommodate the insertion of catheters; otherwise,
it could not be used in conjunction with the robotic
system. The accurate simulation of flow patterns
present in atrial fibrillation was also essential, since
the feed-forward model could not be advanced
otherwise.

Deviation of flow speed from human system (-)

(3/9): In order to determine the degree of accuracy
of our system, the team could take blood flow
speeds ascertained from a human system and
compare values against those to produce a
percentage deviation.
Measurement unit: Percentage: % ()
Electrical Specifications
Number of circuits (-) (6/9): The amount of variable
settings and degree to which the user can operate
the system with ease is, in part, determined by the
amount of circuitry in the system. This group
quantifies this as the number of total circuits in the
system.
Measurement unit: Quantity: # ()

Concept Generation and Selection

Paramount to the concept generation process of this
project was the formulation of a functional
decomposition. The many processes at hand were
categorized and from the functional decomposition,
the team found the three primary functions of the
model to be: 1) beat heart, 2) pump fluid and 3)
sense touch.

Lines of code (+) (7/9): To achieve the same user

requirement listed in the previous point, the
computer code that drives the actuators and reads
user inputs must be versatile and detailed. It is for
this reason that a plus sign (+) is designated for this
quality, but this team is mindful of the risks of an
overcomplicated code.
Measurement unit: Quantity: # ()

Before discovering the research of Harvard

Universitys Whitesides Group, our team
deliberated the use of very rudimentary objects to
achieve the pulsations and contractions present in
atrial fibrillation. Such objects included
combinations of balloons, springs, fluid nozzles
and motors.

Budget Specifications
Total cost of materials (-) (8/9): The user has
specified a target cost of 400 USD. For this reason,
this group takes a conservative approach on
spending and will make every effort to perform a
cost/benefit analysis to determine whether a
particular material is worth its price. To measure

The work of Dr. Ellen T. Roche of Harvard

University was discovered and studied extensively
because it thoroughly investigated the use of
affordable soft robotic actuators for purposes that
were very similar to those in this project. Dr. Roche
referred us to the Soft Robotics Toolkit, a small
firm established by Harvard students that made
available the CAD drawings and assembly
instructions for various soft robotic actuators,
5


including pneumatic air muscles (PAMs) and
pneumatic networks (PneuNets).

further stressed the importance of transparent

materials, and advised against using silicone since
the material would attenuate ultrasound signals
during the verification phase. Indeed, Dr. Chen
urged the team to use PVC, but it was later
discovered that the curing behavior of PVC
presented unresolvable challenges in the
manufacturing process. Thus, silicone was
ultimately selected. One measure taken to prevent
the formation of air bubbles (and consequent air
pockets that would weaken the material, disturb the
transmission of ultrasound signals, and reduce
translucency) included the use of a vacuum
chamber after mixing of silicone constituents.

After careful consideration, the team decided to

proceed with the use of pneumatic networks, since
these actuators could be downsized, easily adhered
to the silicone wall of a heart model, and rapidly
reproduced, if need be.
For methods used to pump fluid, our team sought to
purchase a diaphragm-style or peristaltic fluid
pump. Yet, the cost of such a pump was
prohibitively high, so instead, a bilge pump was
purchased, and proved to be a very economical
alternative.

Before silicone was selected as the concept, another

method under consideration was the use of
conductive powder within the heart wall, which
would provide the user with the capability to
monitor the position of the catheter tip via
transmission of electrical signals. This concept was
quickly discarded and would only be considered if
the use of transparent materials proved to be
impossible or ineffective.

Pump choice aside, the precise location of the

pump(s) was a source of intense deliberation for
this project. Before this team decided to pursue the
development of only one atrium (due to budget and
time constraints), it intended to construct a
complete four-chamber heart, and thus questioned
what the precise arrangement of pumps within the
system would be. The two main arrangements in
question were 1) a singular pump that begins in the
right atrium and achieves fluid propulsion in a
single loop throughout the heart, and 2) the use of
two pumps with the right and left sides of the heart
split into two independent loops for ease of
construction and modeling.

Results of Concept Selection

Ultimately, the above selected concepts were
successfully incorporated into the final atrium
model. The atrium was made of silicone, and was
evacuated of air in a vacuum chamber to ensure
minimal formation of air bubbles (which helped to
achieve a clear color and mitigate attenuation of
ultrasound signals). To accomplish the beating and
randomized contractions of AF, the pneumatic
networks were successfully downsized and
modified for use on the atrium wall. Finally, the
bilge pump also proved successful and error-free as
a source of propulsion for fluid.

As stated previously, the team ultimately produced

a singular left atrium instead of an entire heart, due
to budget and time constraints. Dr. Kruger
acknowledged that these restrictions would limit
the ability to produce an entire heart. In the end, one
bilge pump was used to drive the fluid through the
atrium.
The ability of the model to sense touch and allow
the user to locate the catheter tip was essential. In
comparison to the concepts generated for the
previous two functions, the ones for this objective
were straightforward.

Retrospective Changes
In retrospect, very little should have been changed
with respect to the components used in the atrium.
Yet, difficulty was encountered in using an
ultrasound phantom to analyze the fluid profiles
within the atrium. This team believes a higher
emphasis could have been placed on the use of
proper blood-mimicking fluid to generate proper
images in the ultrasound machine. While a bloodmimicking fluid was still concocted from water,
glycerol and finely crushed glass beads, it did not

The use of transparent silicone, PVC or other

plastic substances for the heart model was believed
to be the simplest and most effective way of
satisfying the users ability to view the position of
the catheter tip. Dr. Roland Chen of the SM Wu
Manufacturing Lab at the University of Michigan
6


allow the ultrasound to produce satisfactory
images.
Furthermore, in a meeting with University of
Michigan cardiologist Dr. Hakan Oral, the team
learned that a vital function of the AF simulation
should involve the motion resulting from a patients
respiratory system (respiratory gating). While this
was brought to the attention of the team far too late
in the semester, the simulation of this function is
certainly an area for future work on the model.

Figure 4. Tetrahedral volume mesh used for computational fluid

dynamics (CFD) simulations.

Engineering Analysis
The four design drivers behind the atrium model
were fluid dynamics, durability, anatomical
accuracy, and intracardiac pressure.

A durability analysis was conducted to analyze the

deformation limits of our product. To better
understand the deformation limits of the heart wall,
a finite element simulation in SolidWorks was
performed. An upper limit of deformation was set
to the tear strain limit (19.60 mm) of the material
based on literature values. 2 As anticipated, the
deformation of the heart wall (0 mm to 8.30 mm)
was well within the 19.60 mm limit, but the analysis
can be used to preliminarily validate that the
operating pressure is within the anatomical pressure
inside the atrium. Below is a contour map from the
deformation analysis. The model was deformed
from its original shape in the simulation. The
expanded model was used to visually check that the
operating deformation is close to the simulated
deformation for anatomical pressure.

As stated, precise replication of fluid dynamics of

an arrhythmic heart environment was perhaps the
most important aspect of this model, as Dr. Kruger
explained. As a means of theoretical modeling,
ANSYS Fluent was used with the intention of
identifying the flows of AF characterized by large
vortex patterns. Below are ANSYS-generated flow
profiles. In systole, there are proper flow dynamics,
wherein flow initiates at each of the four pulmonary
veins, forms small vortex-like patterns near each
orifice, and leaves under laminar conditions
through the mitral valve. In diastole, there is
increased turbulence throughout. These conditions
closely resemble the flow patterns in a heart with
atrial fibrillation.

Figure 5. Deformation analysis reveals that the upper limit of

deformation is safely within material deformation limits.

Figure 3. ANSYS Fluent flow simulations of systole (left) and

diastole (right) are consistent with flow patterns in AF conditions.

The metrics under consideration for anatomical

accuracy were wall thickness (which ranges from
1.5 mm to 6 mm), inner diameter (30-40 mm for
men, 27-38 mm for women), and internal relaxed
volume (18-58 mL for men, 22-52 mL for women).

To produce such accurate and detailed ANSYS

simulations, a tetrahedral volume mesh used for
computational fluid dynamics (CFD) was
produced, pictured in Figure 4.

2

Sil-Poxy Data Sheet, Smooth-On, Inc.


Anatomical accuracy was achieved by obtaining a
CAD model of the heart based on MRI scans of
human patients (and taking into account the above
ranges), pictured below.

Figure 7. The Wigger diagram visually illustrates the relationship

between a variety of different variables throughout the cardiac cycle,
including intracardiac pressures, ventricular volume, ECG signal, and
heart sounds.3

RESULTS

Figure 6. The model used to generate the wax core has an inner
diameter of 30.37 mm, which falls within the normal range of both
male and female atria.

Final Design and Prototype

Pictured below and on the next page are an up-close
photograph of the atrium model, two images
(frontal view and rear view) of the atriums CAD
model, and a photograph of the overall system.

To achieve proper replication of intracardiac

pressure, the waveforms of an electrocardiogram
(ECG) were studied. The P-wave of the WCG
indicates the depolarization in the sinoatrial node,
which in turn causes muscular contraction of both
the left and right atria. This contraction is marked
by an intracardiac pressure of 6-12 mm Hg in the
left atrium during normal heart rhythm. Yet, the
smooth P-wave is absent in patients with AF, and is
instead replaced by a fibrillation wave (F-wave).

Pulmonary inlets (2 of 4)

Outlet to mitral valve

Pneumatic networks (2 of 4)

Results from the ECG could then be mapped into a

pressure profile via a Wigger diagram (pictured to
the right). Drawing from the relationship between
atrial pressure and electrocardiogram waveform for
ordinary heart rhythm, a similar approach could be
used to replicate intracardiac pressures found in AF
settings.
The final piece of this analysis is the incorporation
of small pressure sensors into the atrium model
wall. This approach (ECG to pressure profile)
would allow the derivation of pulse patterns, and
therefore internal pressures, directly from literature
research and actual patient data.

Figure 8. Photograph of final model with affixed

pneumatic networks and fluid tubing.

A. C. Guyton and J. E. Hall, 2006, Textbook of

Medical Physiology, Elsevier Inc., Philadelphia, PA,
Chap. 9.


Pulmonary inlets

Catheter insertion sleeve

Figure 9. Frontal view of CAD-modelled left atrium

Downstream mitral
solenoid valve
Heart model

Upstream flutter
solenoid valve

Outlet to mitral valve

Figure 10. Rear view of CAD-modelled left atrium

Air solenoid
manifold
Fluid reservoir
containing pump

Relay switches

LCD
display

Arduino boards

Figure 11. Photograph of final system of heart model, tubing,Validation

fluid reservoir, solenoid valves and control board


While there were many functionalities of the
system, the only ones worth validating were the
fluid parameters. The validation experiment
involving ultrasound imaging proved to be very
difficult in the absence of a commercial-grade
blood mimicking fluid and proper guidance in
using clinical imaging software. Fortunately,
these difficulties can be translated into future
work for the system and are discussed in greater
depth later on.

laboratory, so there was no guidance in using this

technology.

A test was performed to validate that the volume

of the left atrium was within the physiological
range defined in literature. The atrium was
removed from the system and simply filled with
water to determine the volume, which was found
to be 61.4 ml 5.5 ml. Literature values for the
volume of a males left atrium range from 18 mL
to 58 mL. Therefore, with error, our atrium is
within the typical range of standard volume limits
and is thereby valid.

In the fluid system, a direct-acting solenoid valve

was used as the downstream mitral valve. A
direct-acting valve was chosen because it requires
no minimum pressure for operation (and the
pressure downstream of the atrium is only 6-12
mm Hg). This team overlooked the high current
draw of this valve. Indeed, during a presentation,
the valve was left in the on position and fired
continuously for about one hour, during which
time the valve became very hot. Moreover, the
high current draw of the valve generated a
significant back EMF, which in turn severed the
functionality of our button controls on the LCD
screen. The design could have incorporated 1) a
direct-acting solenoid valve rated at a higher
voltage (and thus, one that draws less current) and
2) a snubbing diode to protect circuit elements
from any back EMF generated across the valves.

DISCUSSION
Design Critique
Overall, the design of this system was
exceptional. In particular, it was durable and
contained protective elements that preserved its
functionalities. In retrospect, there are several
areas that could have been slightly improved.

The next validation procedure for the fluid

system was the determination of the relationship
between pump voltage and flow rate. A linear
relationship was found to exist (see Appendix).
According to the calculated relationship, the
proper output of 4.9 L/min was found to exist at
a supplied voltage of 14.86 V. This voltage is
unfortunately above the 13.6 V threshold of the
pump, yet it is important to understand that AF
cuts cardiac output by 20% (down to ~3.92
L/min), which occurs at approximately 13.2 V,
within the voltage limit of the pump.

Another element that could have been protected

more was the pneumatic network actuator placed
on various sites on the atrium. In particular, closer
attention could have been paid to the entrance of
air pressure. An emergency shutoff or diversion
valve could have proven useful in avoiding
unintended air surges to the actuators. At several
points, the Arduino boards were reset and the
relay switches, in turn, were reset, causing the air
valves to switch open for an extended period of
time, delivering too much air to the actuators and
causing material damage that later had to be
resolved with a silicone epoxy.

Identification of flow patterns within the atrium

was a central objective of the validation
experiments, but it proved to be unfeasible. First,
the cost of a commercial-grade blood-mimicking
fluid (whose viscosity is precisely that of blood,
and which contains microscopic particles to
facilitate monitoring with an ultrasound
phantom) was prohibitively high. A fluid
comprising water, glycerol and crushed glass
beads was concocted in the lab, but the mixture
was not sufficient enough for the ultrasound to
capture satisfactory images. Second, the
ultrasound phantom was connected to a computer
running clinical imaging software, which was
foreign to the team. The graduate student familiar
with the software had been away from the

Future Work
This system has deliver phenomenal results, but
there are several improvements that should be
made in the future.

10


First, the addition of respiratory gating the
resultant cardiac motion of a patients respiratory
motion during surgery is an essential
component of simulating a dynamic environment
for Dr. Krugers system. It is advisable to include
a system that synchronizes this function with the
heartbeat, so that both the intracardiac properties
(i.e. flow profiles) in addition to external motion
(respiratory gating) can be simulated.

Dr. Xioning Jin, Barry Belmont, Robert Chisena,

Alex Price and Dr. Hakan Oral for their guidance.
AUTHORS
C. Coyne
Christopher Coyne is a fifthyear undergraduate with
interests ranging from
mechatronics and robotics to
artificial heart technologies to
advanced energy solutions.
Hes been in the research lab since freshman year, and
has spent summers working in the semiconductor,
surgical tool, insurance, and automotive industries.

Second, the use of commercial-grade blood

mimicking fluid (instead of an in-lab concoction
of water, glycerol and glass beads) is highly
advised in order to obtain satisfactory results
from the ultrasound trials. Moreover, the
guidance of an ultrasound technician or qualified
researcher is recommended when using the
ultrasound phantom.

A. Kurz
Alex Kurz is an
undergraduate student of
mechanical engineering at
the University of
Michigan, and has interest
in the automotive and
health industries. He has been working as an intern at
the Environmental Protection Agency: National Vehicle
and Fuel Emissions Laboratory for the past three years.

Third, a method by which the atrium can be

produced rapidly say, in one or two steps
instead of a convoluted series of molding would
be preferred. It came to the attention of this team
that such a procedure could be possible.
Specifically, 3D printing a thin shell of the
atrium, which once filled with silicone and left
time to cure, could then be broken and removed,
leaving a wholesome model without any messy,
time-consuming intermediate steps. In fact, the
team developed said mold, but was unable to print
it due to unavailability of resources and a strict
budget.

C. Munger
Chance Munger is an
undergraduate student in the
mechanical engineering program at
the University of Michigan. He has
experience in the aerospace
industry working for Northrop
Grumman on the Lockheed Martin
F-35 Lightning II.

Fourth, the addition of snubbing diodes and other

circuit elements to limit back EMF generated
across the solenoid valves would be preferred, as
the effects of back EMF have presented
challenges in using the LCD displays button
controls. Moreover, the direct-acting solenoid
valve deployed downstream of the atrium draws
high amounts of current (up to 1A), so a valve of
a higher voltage rating (which would in turn draw
less current) is preferable, in addition to a heat
transfer solution to dissipate heat from the valve.

S. Shrago
Samuel Shrago is a fifth-year
undergraduate student of
mechanical engineering with
a minor in German at the
University of Michigan. He
is currently interested in
the biomedical applications of mechanical
engineering. He works a side job as a developer at
the University of Michigan Medical School, where
he works on prototyping low-cost medical sensors
for use in clinical settings.

ACKNOWLEDGEMENTS
We would like to acknowledge our sponsor, Dr.
Grant H. Kruger of the SM Wu Manufacturing
Center at the University of Michigan. In addition,
we would like to thank Professor Jun Ni, Ph.D.,

11


REFERENCES
1. G. Bian, M. Lipowicz, and G. H. Kruger.
Self-Learning of Inverse Kinematics for
Feedforward Control of Intracardiac Robotic
Ablation Catheters, pp. 1-6.
2. Sil-Poxy Data Sheet, Smooth-On, Inc.
3. A. C. Guyton and J. E. Hall, 2006, Textbook
of Medical Physiology, Elsevier Inc.,
Philadelphia, PA, Chap. 9.

12


APPENDIX
Linear relationship between flow rate and bilge pump voltage:
Voltage
(V)

Volume
(mL)

Time
(sec)

Flow Rate
(L/min)

9.0

418
405
461
428
552
536
553
547
679
661
670
670
862
841
838
847

15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15

1.672
1.62
1.844
1.712
2.208
2.144
2.212
2.188
2.716
2.644
2.68
2.68
3.448
3.364
3.352
3.388

AVG
10.0

AVG
11.0

AVG
12.0

AVG

4.9 L/min occurs at: 14.86 V*

*with water as the working fluid

13