Developing a Canine Model for Inducing Obstructive Sleep
Apnea and Studying Its Effects
Geeta V, Department of Chemistry, Oakland University, MI
Hamza Osto, Department of Biology, University of Michigan, Dearborne, MI
Liliya Goroshko, Department of Biomedical Engineering, University of Akron, OH
Tim Fritzsching, Department of Computer Engineering, University of Akron, OH
Osamah A. Rawashdeh, PhD, Department of Electrical and Computer Engineering, Oakland
University, MI
Robert Hammond, PhD, Beaumont Hospital, Royal Oaks, MI
Abstract: Sleep apnea is breathing disorder characterized by a reduction or cessation of breathing during sleep. Nearly 18
million Americans are estimated to be affected by obstructive sleep apnea (where the airway is occluded) is the most common
form of sleep apnea. The short term to long term consequences of Obstructive Sleep Apnea range from disruptive to life-
threatening. The increasing rate of Obstructive sleep apnea in the past decade made it necessary to develop an animal model to
study this disorder for further understanding of its causes and effects. So, to mimic the OSA, we developed a computer
controlled device to occlude the airway in a trachea tube to different degrees and time intervals in a canine model. We allowed a
user to get feedback of current position of airway occlusion. The wireless transceiver attached to user input and to valve
(attached to trachea tube) made it easier for user to control the occlusion from distant place. Further, miniaturization of this
device with inclusion of fail safe mechanism will allows researchers to induce sleep apnea episodes and monitor short and long
term effects to serve as a data engine for medical researchers to learn more about this serious and spreading problem.
Index Terms: Obstructive sleep apnea, canine model, user interface, sleep phases, sleep state detection, airway occlusion
INTRODUCTION
Sleep apnea is a breathing disorder characterized by a reduction or cessation of breathing during sleep. There are three
types of apnea: central sleep apnea (CSA), obstructive sleep apnea (OSA), and mixed sleep apnea (MSA). OSA is the most
common form of sleep apnea in which the airway is occluded. Nearly 18 million Americans are estimated to be affected by OSA.
Approximately 1 in 5 adults is affected by mild obstructive sleep apnea and 1 in 15 adults has OSA of moderate or worse
severity, as stated in Circulation journal [1]. The consequences of OSA range from disruptive to life-threatening. Disruptive
consequences include daytime fatigue, depression, irritability, sexual dysfunction, and memory loss. Life-threatening
consequences include congestive heart failure, stroke, irregular heart rhythms, cardiovascular disease and fatal car accidents.
OSA is becoming more prevalent due to increase in obesity, diabetes, stress, and alcohol consumption. The lack of an animal
model to study this disorder has hindered the further understanding of its causes and effects. Rather than creating new treatments,
our goal is to create a canine model in order to induce sleep apnea, which could serve as a data engine for medical researchers to
learn more about this serious and spreading problem. We have created a device for occluding the airway by varying degrees,
which induces OSA, and creating a wireless user interface.
OBSTRUCTIVE SLEEP APNEA
Sleep apnea is a disorder characterized by a reduction or cessation of breathing during sleep, deriving from the Greek
word “apnea,” meaning “without breath”. There are three types of sleep apnea, which are Central Sleep Apnea (CSA),
Obstructive Sleep Apnea (OSA), and Mixed Sleep Apnea (MSA), with obstructive being the most common. Only 0.4% of sleep
apnea suffers have CSA, 15% have MSA, but an astonishing 84% have OSA [1].
Central sleep apnea is characterized by breathing that starts and stops throughout the night, since the brain does not send
proper signals to the muscles that control breathing. MSA is a mixture of Central Sleep Apnea and Obstructive Sleep Apnea,
being a transition from central to obstructive features during the events themselves. However, it is the rarest form of sleep apnea.
Lastly, the most common form of sleep apnea is Obstructive Sleep apnea, which affects around 15 million adult
Americans and 3 million children and most of these patients have hypertension and other cardiovascular disorders. OSA occurs
from repetitive interruption of respiration during sleep caused by the collapse of the pharyngeal airway. One obstructive apnea is
a 10 second or more pause in breathing. Obstructive hypopneas (commonly known as snoring) are a decrease in, but not
complete cessation of, ventilation. Diagnosis for OSA begins at an apnea-hypopnea index of more than 5 and excessive daytime
sleepiness [1].
The causes of sleep apnea could be partially explained by the throat muscles, which help keep the airway stiff while
being awake, but are more relaxed during sleep. Normally, the relaxed throat muscles still allow the airway to be open, but
during obstructive sleep apnea, the airway may be blocked or narrowed. Reasons for this include: the tongue and tonsil‟s size is
larger than the opening into the airway; obesity, which contributes to the extra soft fat tissue around the windpipe wall, which
makes a narrower airway. Another reason is the aging process, which limits the ability of brain signals to keep the throat muscles
stiff during sleep, so the airway will more likely narrow or collapse.
During apnea, or lack of respiration, oxygen levels in blood drop to dangerous levels, and trigger the brain to interrupt
the person during sleep. Then, the upper airway muscles tighten and open the windpipe, so normal breaths are resumed. The high
frequency of dropped oxygen levels reduce sleep quality and increase the release of stress hormones, which cause heart problems
such as high blood pressure, heart attack, stroke, and irregular heartbeats. Also, since untreated sleep apnea develops changes in
how the body uses energy, obesity and diabetes are more likely to develop [2]. Other symptoms of untreated sleep apnea are
impotency, emotional problems, stroke, angina, headaches, job impairment, and motor vehicle crashes. Therefore, those most at
risk for OSA are male, overweight, and over the age of forty. However, OSA can affect anyone at any age [1]. Due to the
severity and scope of OSA, it motivates the goal to develop a canine model for inducing OSA and better studying its effects.
CURRENT TREATMENTS OF OSA
Even though there is no cure for sleep apnea, there are many current treatments (for OSA). There are some non-surgical
treatments. Firstly, positional therapy teaches patients to sleep on the side or stomach, rather than on the supine (back) position.
Apneas tend to be worse when sleeping on the back as gravity makes it more likely for the tongue and other tissue to collapse
and block the airway. Hence, not sleeping on the back may reduce the number of apneas. Secondly, avoiding alcohol and other
CNS depressants, such as surgical anesthetics and Vicodin, relax the throat muscles, and for an OSA patient, this relaxes the
throat muscles even further and endangering their sleep. Thirdly, weight loss treats OSA since the additional fat around the neck
may make the airway narrower, making obstructions more likely to occur. For some overweight people, especially those with
mild cases, losing weight can be an effective treatment. Next, oral appliances, or dental appliances, are intended to treat apnea by
keeping the airway open. It is a small device that is very similar to an orthodontic retainer. It is worn in the mouth while
sleeping to help prevent soft throat tissues from collapsing and obstructing the airway. Some of the devices hold the lower jaw
forward during sleep, while other appliances directly affect tongue position [3].
For more severe cases of OSA, the Continuous Positive Airway Pressure (CPAP) machine and surgery are more
effective than the previously mentioned treatments. The CPAP machine is the most common tool used by physicians to treat
patients with sleep apnea. The machine consists of a mask, and an air pump. The mask is placed over the mouth and nose, and
the pump is turned on which pumps air into the oral and nasal cavity. This continuous stream of air keeps all air ways open, and
prevents any possible occlusions. Last but not least, surgery, which should be the last resort for treating sleep apnea, is called the
uvulopalatopharyngoplasty, or UPPP procedure. It is intended to enlarge the airway by removing or shortening the uvula and
removing the tonsils, adenoids, and part of the soft palate or roof of the mouth. A tracheotomy, which is the surgical creation of
a hole in the trachea or windpipe below the site of obstructions, is the most effective surgery for OSA. Unacceptable to most
people, it is generally reserved for serious apnea that has failed other treatment. The hole is plugged during the day for normal
breathing and unplugged during sleep so obstructions are bypassed. The site must be cleaned carefully daily to prevent infections
[4].
SIBHI SUMMER 2009 GOAL (10 WEEKS)
The goal of this project for SIBHI 2009 is to model, not to treat, OSA. The plan is to develop a device that will mimic
the effects of obstructive sleep apnea by occluding the airway. The device contains a computer-controlled actuator capable of
closing and opening a valve mounted to a canine‟s trachea tube. In order to develop a canine model for studying sleep apnea, two
mechanisms are included: adjustable degrees and frequencies of airway occlusion, and detection of sleep stages. This summer,
we have accomplished the first mechanism, and have ideas about the second mechanism of detecting sleep states.
LONG-TERM GOAL (BEYOND SUMMER 2009)
The SIBHI 2009 Summer Goal is embedded into the overall objective as proposed by Dr. Rawashdeh, Dr. Hammond,
Dr. Das, and Dr. Shanley. The project plan is divided into three parts: (1) development of a sleep-phase model, (2) development
of a computer data acquisition and control system, and (3) system evaluation.
The sleep phase model will be developed from data from human studies, provided by and interpreted by Beaumont researchers.
The hardware parts of the system are designed to be used in adult dogs that are prepared with a permanent tracheotomy, or the
surgically created opening in the neck leading directly to the trachea (or the “windpipe”), which is maintained open by a hollow
tube. The animals will be implanted with a telemetric instrument that transmits up to seven channels of hemodynamic and
neurologic data. A block diagram of the proposed computer control and acquisition system is shown in Figure 1.
Transceiver Transceiver
Embedded Microprocessor
Valve Driver
Data Acquisition
EEG
Subsystem
EMG
M
Implanted Tele.
Instrument
Electromechanical
Valve
User Interface PC
Control and User Animal Attached I/O
Interface System
Figure 1: Proposed System Block Diagram
On the right hand side is the small portable input/output (I/O) system that is attached to the dog‟s collar. It periodically
collects the data wirelessly from the implanted telemetric system. Collected data is conditioned and organized into frames by an
embedded microprocessor unit and transmitted through a wireless transceiver to the PC-based control and user interface system.
Data is collected and stored on the PC, as shown on the left hand side of Figure 1, for determining the sleep phases as well as for
post experimental processing. Based on experimental parameters determined by the investigators, airway occlusion events are
executed by sending wireless commands back to the collar‟s I/O system, which then controls a silent electromechanical
occlusion valve placed on the tracheotomy tube. The occlusion is released by the computer control when the dog arouses from
sleep, hence mimicking OSA episodes experienced by humans. There will be a fail-safe mechanism to relieve the obstruction if a
maximal respiratory effort occurs.
PREVIOUS CANINE MODELS FOR OSA
The precursor to our proposal for inducing sleep apnea in a canine model is found in the research of Kimoff et al. in 1994 [5].
They mimicked OSA in dogs, and their conclusion was that modeling OSA was potentially a very influential way of studying
OSA in full. However, the project had some setbacks that are planning to resolve in the revival of the canine model:
It will be a high fidelity system using the latest sensing devices, high bandwidth wireless communication, and high
computational power available to process data.
The sensors and telemetry unit are implantable and thus the animal can live unrestrained by tethers or jackets, and the
sensors are maintenance-free, and protected from infection and physical damage.
The implantable device has significant processing resources on board that will be utilized for pre-processing of data as
well as for onboard decision making. Redundant decision making will reduce the probability of operational errors.
The implantable device to be used has several additional channels for future expansion of sensor types, and different types
of sensors are under development.
The developed hardware attached to the animal will be completely wireless and small in size (i.e., attached to a standard
dog collar).
Silent electro-mechanical valves will be used to replace the large and noisy servos used in the Kimoff system.
The proposed system will allow gradual partial or total occlusion of the airway as opposed to the two-state open/shut
Kimoff system.
CURRENT”/SIBHI 2009 GOAL
In order to induce sleep apnea in an animal we must develop a system that is able to control the amount of airflow that enters the
trachea.
By occluding the airway we are able to place the animal into a sleep apnea state
The system will consist of an electronic computerized device capable of controlling airflow using a motor or actuator.
DESIGN IDEAS
As the system is mounted to a canine, it must remain quiet so that the dog is not discomforted by the device.
The system will be adjustable (user can change the level of airflow and duration of state)
The unit should be small enough that it can be powered by a battery pack. This requires the package to be efficient as well as use
parts that require very little current to run.
The system should be small enough to attach to a collar so that the canine can have the mobility to move freely in its
environment and not be impaired from doing typical tasks. In addition the size must not irritate the animal so that it would not try
to remove the device from its neck. Overall, small size and enhanced mobility are essential.
The method that we use must not be invasive so that operational costs and surgeries are minimized.
The system will be a complete wireless solution, allowing all functions of air occlusion as well as data to be accessed with out
having to run wires to the animal.
The system must also be inexpensive to produce that way researchers can have multiple systems testing different sleep sates of
the animals.
The system must remain durable and rugged. The longevity of the project may require years of collecting samples and test sleep
states in order to produce accurate diagnoses.
The user interface will allow the researchers to adjust the amount of airflow in and out of the trachea by imputing values that
corresponds to properties of the state. Then the effects of inducing OSA can be monitored and data recorded by a computer.
Medical researchers would not have to watch the animal 24/7 as the data can be compiled and stored remotely.
A fail safe mechanism will allow the unit to release air to the animal in the event that actuator fails or does not respond to user
inputs, allowing the animal to breath and be unharmed.
POSSIBLE WAYS TO OCCLUDE THE AIRWAY
By implanting a balloon into the trachea we can impede airflow into the lungs by a method of inflating and deflating.
How ever there are several disadvantages. First, uniformity of adjustment is difficult. Say we adjust the balloon so that the
airway is 50% closed. It would be tricky to produce a consistent 50% closure every time, since temperature changes affect
volume air, consequently the size of the balloon would also be affected. If the air is cold going into the trachea, the balloon
would not inflate as large. If the air coming into the trachea is warm, the balloon would be slightly larger for the same volume
due to the physical properties of gases. Secondly, the use of the balloon requires an air pump be used. It would require the use of
an airflow meter in order to track the volume of air being pumped into the balloon, thus raising the overall complexity, precise
adjustability of the balloons size, as well as contribute more weight to the design.
Using a flap to control air is a viable solution as it would allow us to open and close the tube coming out to the trachea
in a very small, confined space. Small movements are advantageous to power consumption, as an adjustment to the airflow can
be done with small movements of the flap and require the actuator/motor‟s duty cycle to be shortened. This would in turn
decrease the amount of battery power needed to move the actuator and prolong runtime. One disadvantage to this design is the
noise generated by air traveling around a flat surface. The sharp edges can cause turbulence and creating noise that may be
audible under certain conditions.
Another way to occlude the airway is to use a soft tube that that attaches to the opening on the collar and is pinched by
the actuator. This works well as the actuator can be accurately positioned. However a downside to this design is slightly
increased size requirement due to the size of the tub and the bracket that is required to hold the actuator and tube into position.
Also the tube itself, depending on the material, can be worn down over time and need replacing. The tube method is probably the
least costly and easiest to implement of the designs that were tested.
By using a cone and or a marble at the end of the actuator, we are able to occlude the end of the tube with a minimal
amount of space requirements. Also a positive aspect to using a cone or a marble is that it lends itself to being one of the quietest
methods to filling up the opining of the tube. This is because turbulence produced by various shapes with sharp edges and or flat
surfaces produce more sound. With a sphere or cone noise is at its lowest when the air is able to travel smoothly with least
resistance.
THE METHODS ATTEMPTED AND SELECTED
The preliminary design that incorporates a tube that attaches to the trachea implant (which is implanted in a
tracheastomy, a surgery where patients have their trachea removed and a new one implanted—then an exterior vent is added) and
allows the occlusion of air via pinching a ½ inch silicone or pvc tube. How ever our design took too much space and the
longevity of the tube did not seem sufficient as the structure of the tube would inevitably fail.
The design chosen that worked best uses the flap method whereby the actuator moves a flat disc in a linear motion to
open and close airflow to the tube. This method works best because it requires less space, requires less parts (less possible failure
points), and is the most accurate to control airflow in and out of the tube. See Fgure 2 on page 7 for the types of shapes
proposed.
HARDWARE
-Linear Actuator:
Model Used: Firgelli PQ12 [22] (For datasheet)
Used to control the airflow in and out of the tube
It would linearly/horizontally close the end of the tube
-Microcontroller
Model Used: Dragon 12P with Freescale HCS12 Microprocessor [23] (For datasheet)
Software:
C Language, CodeWarrior Special Edition for the HCS12 board
-Wireless Communication
Model Used: MiniBee Serial-to-Serial transmitters
-User Interface
Hyperterminal for the time being, eventually the user interface will be a GUI (Graphical User Interface)
OCCLUSION OF AIRWAY
By occluding the airway, we are able to place the animal into a sleep apnea state. The system will consist of an
electronic computerized device capable of controlling airflow using a motor or actuator. During brainstorming for our design,
there are several stipulations. The device for the collar-mounted airflow control has to be:
Quiet (the servo cannot be loud, since the canine can hear it and want to chew it off);
Adjustable, so that the user can change the level of airflow and the amount of duration;
Battery powered, which means that the unit should be small enough and efficient enough (not use too much current) for
it to be battery powered;
Collar mounted, where the unit is small enough to attach to a collar so that the canine can have the mobility to move
freely in its environment and not b e impaired from doing typical tasks. In addition, the size must not irritate the animal
so that it would not try to remove the device from its neck. Overall, small size and enhanced mobility are essential;
Require little surgery;
Wireless, so the medical researchers can be at a distance from the dog and allow it to sleep normally, and so they do not
have to watch the animal 24/7 as the data can be compiled and stored remotely;
Reasonably priced;
Durable, since the dog will most likely be monitored for years;
Equipped with a fail-safe mechanism, in order to allow the unit to release air to the animal in the event that the actuator
fails or does not respond to user inputs, allowing the animal to breath and be unharmed.
In designing a linear actuator to occlude the airway tube, we have thought of different shapes for occluding the airway:
Balloon, pinching, or flap using the sphere, cone, or flap. First, by implanting a balloon into the trachea we can impede airflow
into the lungs by a method of inflating and deflating. However, there are several disadvantages. First, uniformity of adjustment is
difficult. For example, the users may adjust the balloon so that the airway is 50% occluded. It would be difficult to produce a
consistent 50% closure each time, since temperature changes affect volume air, and consequently, the balloon would not inflate
as large or small; if the air into the trachea is warm, the balloon would be slightly larger for the same volume due to the physical
properties of gases. Secondly, the use of the balloon requires an air pump to be used. It would require the use of an airflow meter
in order to track the volume of air being pumped into the balloon, thus raising the overall complexity, precise adjustability of the
balloon‟s size, as well as contribute to more weight to the design.
The second option for occluding the airway tube would be pinching the tube. We developed a code to move the linear
actuator to move front and back by changing the voltage (by rotating) an in-built potentiometer. The actuator has both built-in
feedback potentiometer (A-Pot) and A-motor. This actuator potentiometer (A-pot) tells us the current position of the motor and
gives some value ranging from 537 to 957. We calibrated the in-built potentiometer (on the Dragon12Plus board) to the value
range of A-pot (which has a range from 0-1024). In the experiment we used to adjust a desired voltage in the in-built pot. This
move the A-motor in a direction so that the feedback potentiometer (A-pot) will continuously compare with the desired voltage.
When the two values match, the motor stops. Next, we developed a code that when the user enters a number from 0%-100%
through the hyperterminal, the actuator will move from 0% position (open) to 100% position (closed). The pinching method was
better than the balloon method, but it was just as a prototype for a better method, since the tube will eventually wear out from the
pinching, and there is extra resistance against the actuator in order for it to close 100%.
Before exploring the third option of the flap for occluding the airway, it is important to know the accuracy of the
actuator. See below for Table 1 with results.
Table 1: Accuracy of the Actuator
What this table tells us is that….the actuator is accurate
enough for our purposes.
Thus, we explored a third option of closing the tube with a flap, sphere (hemisphere), cone, or different shapes, along
with different materials. Rubber would be the ideal material, since it provides the best closure and traps the air the best. Below is
Figure 2 illustrating the three main ideas we had for occluding the airway: with a sphere, a cone, and a flap.
Figure 2: Options for occluding the airway
By comparing the ratio of the different shapes‟ radii to their volume as it is changing by occluding the airway (from 0%-
100%).
Figure3: Percentage of Occlusion based on Shape
Table 2: Comparisons between mechanisms to occlude the tube
Using a flap to control air is a viable solution as it would allow us to open and close the tube coming out to the trachea
in a very small, confined space. Small movements are advantageous to power consumption, as an adjustment to the airflow can
be done with small movements of the flap and require the actuator/motor‟s duty cycle to be shortened. This would in turn
decrease the amount of battery power needed to move the actuator and prolong runtime. One disadvantage to this design is the
noise generated by air traveling around a flat surface. The sharp edges can cause turbulence and creating noise that may be
audible under certain conditions.
DESIGN OVERVIEW OF SUMMER 2009
Below is a Figure 4 for the prototype completed summer 2009:
Figure 4: Summer 2009 Design Figure : Picture of Design Overview in the Lab
DESIGN OVERVIEW FLOWCHART OF CONTROL CODE
START
Get
Actuator
Feedback
Position
Convert the Feedback position from Analog to Digital in terms of Counts
Get Desired actuator position from User
through Hiperterminal
Use formula to convert the desired position to Counts
NO YES
Is Desired position
within actuator limits
NO
Is Desired position Is Desired position
greater than highest lesser than lowest
Count count
YES
YES
Use highest Count Use lowest Count
Command Actuator
YES NO
Is Desired actuator
position less than current
position
Move the actuator back Move the actuator Forward
NO
Is Desired actuator position =
current position
YES
STOP
Actual Program Code in Appendix 1.1
ETHICS OF ANIMAL RESEARCH
Ethics is an exhaustive topic, but the short summary of the ethics for this project is summed up in the answers of two
short questions: (1) Why will this research be involving an animal, and (2) Why will it be using a dog in particular? It is ethical
to use an animal for this research because animal research advances animal and human welfare, as well as significantly
contributing to the knowledge of behavior. Animal research helps explain the central nervous system (CNS). In addition, human
subjects and alternatives to live subjects have been proposed. By law, it is required to test therapeutic products on animals before
used on humans. Also, drug testing is done first with animals before it is used on humans [6]. In response to the second question
of the reason for using a canine, is that the dog‟s physiology is quite similar to humans and so the information gained is
applicable to the human condition. Secondly, the adult dog is a good body size for our purposes. Rats would be too small,
whereas something like pigs, which are on average from 150-200 pounds and gain a pound a day, would not be feasible for this
project. Thirdly, canines are sociable and are able to adapt to their surrounding environments. Lastly, dogs are able to last for the
duration of the project, which could be months or years [7].
In more detail, there are many subtopics to this ethical consideration. Firstly, dogs have greatly advanced research and
finding cures for diseases. Tibetan Terriers are contributing to a canine DNA bank in the continuous research on studying the
genetic bases of diseases that affect both dogs and humans. These dogs can be affected by a neurological disease called neuronal
ceroid lipfuscinosis, with the human equivalent called Battens disease [8]. Another example of dogs playing a key role in
research are the Portuguese water dogs that also play a key role in genetics research. Geneticists found that small dogs shared a
snippet of DNA near the IGF1 gene, which helps control growth, on chromosome 15. Geneticists are hypothesizing that this
snippet of DNA suppresses the IGF1 gene, which keeps a dog small [9]. Furthermore, similarities in dog and human breast
cancer pre-malignant lesions have been found, which are considered to carry risk for developing breast cancer in both canines
and humans [10]. These are only three of many examples that dogs have advanced medical research [11].
There are also many legal restrictions and regulations on the use of animals for research science. Legally, there are
many organizations that ensure ethical treatment of animals such as The United States Department of Agriculture (USDA) and
the Public Health Service (PHS). In each of these organizations, there is an emphasis on seeking alternatives to live animals, and
if using live animals, minimizing their overall use. There are also guidelines for ensuring that animal use is humane, minimizing
pain, distress, or comfort. One of the best ways to ensure this is by setting the earliest possible endpoint for the experiment, for
personnel to receive adequate training prior to performing a procedure, and to use proper handling techniques for animals [12].
The U.S. Department of Health and Human Services has set up an extended policy for animal research guidelines [13].
FUTURE WORK
The first thing would be to develop a device for detecting different sleep phase, and detecting when the animals reaches
Rapid Eye Movement phase (REM) before inducing an OSA state. In addition, miniaturize the system to make it collar mounted
and install a fail safe mechanism so that if the pressure falls below a certain value, then a valve will rupture and the dog will be
able to breath normally. Lastly, for the device to collect data so that researchers can study the effects of OSA.
BASICS OF SLEEP STATES
The next steps beyond the accomplishments of summer 2009 would involve several things, one of which is the detection
of sleep state, until the ideal solution of the implantable electroencephalogram (EEG) may be used in order to record brain waves
therefore, corresponding sleep states. An alternative method for sleep state detection that is non-invasive, less expensive, and as
effective as an EEG, would be advantageous as well. Firstly, a background on sleep states is needed.
Sleep has been separated into 5 sleep phases, or separated into three parts as Light sleep (Stages 1 and 2), Deep sleep
(stages 3 and 4), and rapid-eye movement, or REM, sleep (stage 5). Stages 1-4 are known as non-REM sleep. During an average
night, the person goes through 3-6 full cycles of sleep phases. See Figure 5 which shows the different cycles through sleep, and
its frequency [14].
Figure 5:The Stages of Sleep and Its Cycles
The best way to detect sleep phase is through brainwaves, since there are different brain waves associated with different
phases of sleep and alertness. See Figure 6 below [15].
Figure 6: Brain Waves. Figure 7: Sleep phases
The figure above shows the different brain waves. The frequency of beta waves, represented of an awake or aroused
state, ranges from 15-40 cycles a second, and is present in an alert/working state. Next, the alpha wave represents non-arousal,
when a person is relaxed or reflecting; their frequency ranges from 9-14 cycles per second, meaning they‟re slower and higher in
amplitude. Next, theta brainwaves have a frequency between 5-8 cycles a second, when one is drowsy or idling. The final
brainwave state is delta, which are the slowest brainwaves, with a frequency of 1.5-4 cycles per second. Stage 1 is characterized
by non-REM sleep, with theta brain waves, and stage 2 is caused by non-REM and theta brain waves as well, along with a
decrease in muscle tone and cerebral activity. Stage 3, non-REM sleep, is a precursor to Stage 4, which is the deep state of sleep,
both with delta brain wave activity. In deep sleep, metabolic rate and temperature decrease and bodily functions are at their
lowest level. Lastly, Stage 5 is rapid eye movement, or REM, sleep, characterized by beta waves. The signs are similar to
alertness, but the body is disconnected from the brain: the brain is dreaming while the body is almost “paralyzed” [15].
CURRENT PRODUCTS FOR DETECTING SLEEP STATE
There are several products that were considered for detecting sleep states, each evaluated for their advantages,
disadvantages, price. Firstly, the AxBO Alarm Clock claims to “wake you when you‟re ready.” A wristband detects body
movement and other factors during sleep, and detects whether or not a person is in deep sleep or dream stage (or “REM sleep”).
Wirelessly, the alarm clock detects these signals and goes off within a half hour window of ideal wake up time, so that the person
would feel most refreshed since they are not being woken up during the deepest parts of sleep. However, due to cost ($241), we
are looking for other options [16].
Secondly, the Watch-Pat 100, where PAT stands for peripheral arterial tone, is a finger sensor. It provides a new
screening, diagnostic, treatment assessment and patient follow-up possibilities in the medical management of sleep-related
breathing disorders. Many OSA suffers use this, since this device is worn on the wrist and uses a non-invasive finger mounted
pneu-optical probe to measure the PAT signal. The recorded signals are stored in a removable memory card to be downloaded to
a computer for automatic analysis by medical personal. However, since we are eventually using canines as our subject, they are
not able to wear finger sensors [17]. However, the finger sensor highlights the importance of the assessment of cardiovascular
function in the diagnosis of obstructive sleep apnea, because some of the most severe ripple effects of OSA affect the heart [18].
Thirdly, the ultrasonic oscillosensor is a non-invasive method for detecting sleep state. The sensor is placed underneath
a mattress and detects the vibration (in Hertz) of a person or canine. Based on these movements, the sleep state is determined.
This provides a unique advantage to the conventional systems such as the electroencephalogram (EEG) and the pressure sensor,
since it does not require a professional to implant an ultrasonic oscillosensor, as with an EEG, and it is not constrictive, as with
the pressure sensor. This sensor can noninvasively detect the vibration of a patient by placing it under a bed frame, and fuzzy
logic plays a primary role in the recognition [19]. Though we are very interested in father exploring this option and how it could
be integrated in our project, we are not able to find it on the market, but only in research papers. However, we could find
ultrasonic sensors, which are slightly different than ultrasonic oscillosensors.
Our fourth idea for sleep state detection was the Piezo Eye Film, which is contrary to the previous devices which mainly
depended on body movement variation, since it directly detects rapid eye movement, a characteristic of the dreaming stage of
sleep. (Note, sleep apnea during this stage of sleep could be more common or impactful than in other stages.) The piezo eye film
is placed above the radius of the eyeball. The film sensor generates electricity as the eye moves, therefore detecting REM sleep.
However, it poses some disadvantages for our current project, since it is not feasible for dogs, and it is constrictive, since double-
stick tape is placed on the eyelids. They are also relatively expensive and fragile in clinical use [20].
Fifthly, we explored the option of the Kvasar Dreammask. It is similar to the piezo eye film, however, the Kvasar
dreammask uses an infra-red light emitting diode (IR LED) which shines light on the eyelid and is detected by the photo sensor.
When the eye is not moving, the light reflects back, but is disrupted during REM sleep since the eye is moving. It uses pulse
width modulation. It is user adjustable for the number, intensity, and type of signal, as well as the dream-alarm function in order
to write down dreams at optimal times. It is optimal for lucid dreaming, but is useful for sleep state detection as well. However,
this idea is not on the market, and would be too expensive as well for the project‟s budget [21].
There are more options for detecting sleep state, and after going through all of them, we decided to try to make our own
method of detecting sleep phase: the REM Eyeliner. Due to time constraints, this idea has not been developed yet. This idea
provides an advantage over other methods, since it is non-invasive, within our budget, wireless, non-constrictive, and a novel
contribution to science. Red eyeliner will be applied to the dog‟s face, along with reference points, and an algorithm will be
constructed using MATLAB in order for it to recognize whether the dog is in REM sleep or not.
IN MORE DETAIL
There are proposed steps for REM detection through use of MATLAB:
First, mark the reference points on the face with a red market: along the eye (as if applying an eyeliner); three dots on
the eyelid, either in a triangle or a straight line, or to mark a grid so MATLAB can use it as a matrix; and two dots
above the middle of each eyebrow, so that if the face moves, this does not equate to REM state, since the reference will
be the same.
Secondly, the distance between each point will be calculated, as well as the elevation of the dots on the eyelid. This is in
order for MATLAB to detect the change when the cornea moves beneath the eyelid, the elevation of the eyelid will
change.
Thirdly, these distances will be captured in an image or frame.
Fourthly, the video in MATLAB will capture the eye movement. Some ideas include MATLAB capturing images every
five seconds or so, and detect changes in movement that way. However, images are large files and there may be a
problem in storing them.
Fifthly, this step will account for detecting the movement; as the eyelid moves up and down, since during REM sleep,
the eyes sometimes become half open, the eyeliner will move with respect to the eye movement. As the cornea moves
inside the eye, the grid on the eyelid where the cornea is moving underneath it, will be distorted. As the face moves, the
reference points will move as well.
Sixthly, MATLAB will collect all this data from movements, and from combining previous medical data on REM and
non-REM sleep, we will program the classification of movements as REM or non-REM sleep.
The seventh step would be combining this sleep data with the heart data, in order to give the most accurate results of
whether the canine is in REM or non-REM sleep, or what stage of sleep they are in. There will be a method for
combining this data as well.
The eighth step would involve the indication of the sleep stage, as a result of combining the REM and heart data.
The current sleep state will also be indicated on the screen or somewhere so the user may be notified. Lastly, once the sleep stage
is determined, the tracheal valve may be regaled to induce or not induce sleep apnea, and observe and study its effects.
CONCLUSION
In this project, we developed a method and device to occlude the airway by using an airflow control system. In addition
to the production of the actuator and the valve system, we have outlined a procedure for the placement of the valve, actuator, and
trachea tubes attached to the dog. A user/ researcher can control the degree of occlusion via hyper terminal. A work was
accomplished by developing a program code in C and communicating with a dragon board with serial interface. A wireless
transceiver is attached at the user end and at actuator attached to trachea tube to remotely control the collar mounted unit. The
user gets the feedback of current occlusion status and inputs the percentage of occlusion desired in percentage from 0 to 100
scales. The various degrees of occlusion will allow researchers to mimic accurately the condition of obstructive sleep apnea in a
canine model
ACKNOWLEDGMENT
This work was performed during the SIBHI program at Oakland University funded by NSF and NIH under the BBSI program
(Summer Institute for Bioengineering and Bioinformatics), grant no. 0552707. The authors ould like to thank the staff to the
Electronics Shop and the students in the Embedded Systems Research Lab for all their help and support.
_______________________________________________________________________________________________________
REFRENCES
[1] V. K. Somers, D. P. White, R. Amin, W. I. Abraham, F. Costa, et al, “Sleep Apnea and Cardiovascular Disease,” Circulation,
vol. 118, no. 10, p. 1-33, month 2008. [Online]. Available: Data Base Name, (if appropriate), internet address. [Accessed Jul. 21,
2009].
[2] “Sleep Apnea Information”, American Sleep Apnea Association, 2008. [Online]. Available: http://www.sleepapnea.org.
[Accessed: Jul. 21, 2009].
[3] “P. Daniel,"Sleep Disorder Help," Oral Appliance Treatment, 2009.
[Online]. Available: http://www.sleepdisorderhelp.net/treatment.html. [Accessed: Jul. 21, 2009].
[4] P. Spiegler. “Overview of Sleep Apnea Surgery,” Sleep Apnea Surgery Center, [Online]. Available:
http://www.sleepapneasurgery.com. [Accessed: Jul. 21, 2009].
[5] R. J. Kimoff, H. Makino, R. L. Horner, L. F. Kozar, F. Lue, et al, “Canine model of obstructive sleep apnea: model
description and preliminary application,” Journal of applied physiology, vol. 76, no. 4, p. 1810-1817, month 1994. [Online].
Available: Data Base Name, (if appropriate), internet address. [Accessed Jul. 21, 2009].
[6] “C.A.R.E.: APA Board of Scientific Affairs Committee on Animal Research and Ethics,” American Psychological
Association. [Online]. Available: http://www.apa.org/science/anguide.html. [Accessed: Jul. 21, 2009].
[7] “Working with Dogs in Research Settings.” [Online]. Available: www.researchtraining.org/-
moduletext.asp?intModuleID=604. [Accessed: Jul. 21, 2009].
[8] University of Missouri-Columbia, “Dogs Help Humans in Research to Fight a Terminal Disease,” ScienceDaily, Nov. 25,
2005. [Online]. Available: http://www.sciencedaily.com. [Accessed: Jul. 21, 2009].
[9] T. Tsouderos. “Portuguese water dogs play key role in genetics research,” ScienceDaily, Apr. 14, 2009. [Online].
Available: http://www.sciencedaily.com. [Accessed: Jul. 21, 2009].
[10] Purdue University,“Similarities in Dog and Human Breast Cancer Pre-malignant Lesions Found,” ScienceDaily, Nov. 16,
2009. [Online]. Available: http://www.sciencedaily.com. [Accessed: Jul. 21, 2009].
[11] “Animal Research Facilities in the United States”, The Humane Society of the United States, 2009. [Online]. Available:
http://www.hsus.org. [Accessed: Jul. 21, 2009].
[12] “Ethics and Alternatives,” Research Animal Resources. [Online]. Available: http://www.ahc.umn.edu. [Accessed: Jul. 21,
2009].
[13] “Office of Laboratory Animal Welfare: Public Health Service Policy on Humane Care and Use of Laboratory Animals,”
U.S. Department of Health and Human Services. Aug. 1, 2002. [Online]. Available: http://www.hhs.gov. [Accessed: Jul. 21,
2009].
[14] “Sleep Phases,” Sleep Information Center, 2009. [Online]. Available: http://myzeo.com.
[Accessed: Jul. 21, 2009].
[15] “What is the function of the various brainwaves?” Intelegen Inc. [Online]. Available: http://www.web-
us.com/brainwavesfunction.htm. [Accessed: Jul. 21, 2009].
[16] M. Perton. “Axbo alarm clock wakes you when you‟re ready,” engadget. Jan. 23, 2006. http://www.engadget.com.
[17] „Watch-PAT 100”, http://www.bio-medicine.org/medicine-products/Watch-PAT-100-18494-1/.
[18] T. Penzel, R. Fricke, U. Brandenburg, H. F. Becker, C. Vogelmeier. “Peripheral arterial tonometry monitors changes of
autonomous nervous system in sleep apnea”, Engineering in Medicine and Biology, Vol. 2, pp1552-1555, 2002.
[19] Y. Kamozaki, T. Sawayam, K. Taniguchi, S. Kobashi, K. Kondo, et al. “A New Ultrasonic Oscillosensor and Its
Application in Biological Information Measurement System Aided by Fuzzy Theory.” IEICE Transactions Information and
Systems, Vol. E90-D, No. 11, 2007.
[20] J. Silvola, A. Saarinen. “Piezoelectric film transducer for recording of oculography in elecgro-encephalogram-polygraphy”,
Medical and Biological Engineering and Computing, Vol. 50, pp 469-470, 2002.
[21] “Kvasar dreammask: Project finished”, http://brindefalk.solarbotics.net/kvasar/kvasar.html.
MISC
[22] Firgelli: Datasheet for PQ12 Linear actuator, http://www.firgelli.com/pdf/PQ12_datasheet.pdf
[23] EvbPlus: Datasheet HCS12 Dragon12 Microcontroller, http://www.evbplus.com/hcs12.html
Appendix
1.1 Complete program code with comments__ Current progress summer 09
_______________________________________________________________________________________________________
/* User input through hyperterminal to move acuator front and back without fluctuation (tolerence limit added)*/
/* These are the libraries required to run the code*/
#include // common defines and macros
#include // derivative information
#pragma LINK_INFO DERIVATIVE "mc9s12dg256b"
#include "main_asm.h" // interface to the assembly module
/* declaring subfunctions */
void show (void);
void input(void);
void before(void);
/*declaring variables */
unsigned int val,diff,diff1;
unsigned int desired,current,t;
int c1,c2,c3,i;
int highest=825; //highest digital value Actuator potentiometer
int lowest=45; //lowest digital value of Actuator Potentiometer
void main(void)
{
PLL_init(); // set system clock frequency to 24 MHz
ad0_enable(); // enable a/d converter 0
led_enable(); //enabling LED
seg7_disable(); // disabling 7 segments
lcd_init(); // enable lcd
SCI0_init(9600); // initialize SCI0 at 9600 baud
motor0_init(); // enable 8-bit pwm1 for motor
DDRB=0xff; //Port B output
DDRH=0x00; //Port H input
/* closed loop program*/
while(1)
{
outchar0(0x0c); // clear hyperterminal screen
current = ad0conv(3); //current: present digital value for A-pot;
/* this subfunction allow user to see current %of occlusion on LCD*/
before();
/* this sunfunction will allow a user to enter the percentage in 0-100 and show the entered value on hyperterminal*/
input();
desired = 45+(78*val)/10; // converts the percentage into digital number
diff1=current-desired ;
diff=desired-current;
/* this sunfunction displays the user input as percentage on LCD‟s ist line*/
show();
/* safety valve so the user input will not exceed or lower than limits (45-825)*/
if (desired>highest) //safety valve so desired won't exceed 825
{
desired=highest;
show();
}
if (desired=desired)
{
diff1=current-desired ;
if (diff1current) //user wants A-hand extended= CLOSE VALVE
{
/* adding the tolerance limit of 4% */
while(current<=desired)
{
diff=desired-current;
if (diff1<=25||diff<=25)
{
PORTB=0x00;
motor0(0);
break;
}
PORTB=0x02;
motor0(diff); // value of current (A potentiometer) increases.
show();
}
}
else if (desired=current) //if desired P = current P
{
PORTB=0x00; // initialising the port B
motor0(0); // stop the Actuator motor
show();
}
}
}
void show()
{
desired = 45+(78*val)/10;
current = ad0conv(3); //new: for potentiometer;
set_lcd_addr(0x00); // 1st line of lcd display
type_lcd("User Input:");
write_int_lcd(val);
}
void before()
{
set_lcd_addr(0x40); // 2nd line of lcd display
type_lcd("current:");
t=((10*current)-45)/78 ;
write_int_lcd(t);
}
/* sunfunction INPUT allows the user to enter the percentage in the range of( 0-100) in digits e.g. 050 or 000) through
hyperterminal*/
void input()
{
char* q1; // creating a pointer for a array
q1="Enter percentage (0-100):" ;
PORTB=0x00; //initialize PORT B
for(i=0;i<25;i++) // The message on hyperterminal asking user to input
{
outchar0(q1[i]);
}
c1=inchar0(); //sending data from hyper terminal to microcontroller
outchar0(c1); //sending data to hyperterminal from microcontroller
c2=inchar0();
outchar0(c2);
c3=inchar0();
outchar0(c3);
/*converting the ASCII code to decimal values*/
c1=c1-0x30
c2=c2-0x30;
c3=c3-0x30;
val=(c1*100)+(c2*10)+(c3*1);
/*displaying the percentage on LCD*/
set_lcd_addr(0x00); // 1st line of lcd display
write_int_lcd(val); // display user input as percentage
}
_______________________________________________________________________________________________________