THE NOVAMOUSETM

PULSE OXIMETER

SHERAZAD MAHOMEDALLY,

NICOLE SERAGO, AND POOJA SHAH

DEPARTMENT OF BIOMEDICAL ENGINEERING

VANDERBILTUNIVERSITYSCHOOL OF ENGINEERING

APRIL 22, 2003

ADVISORS: BOB ALLEN & BEN SCHNITZ

MICRONOVA TECHNOLOGY

INSTRUCTOR: DR. PAUL KING

ASSOCIATE PROFESSOR OF BIOMEDICAL AND MECHANICAL ENGINEERING

VANDERBILTUNIVERSITYSCHOOL OF ENGINEERING

TABLE OF CONTENTS

  1. ABSTRACT…………………………………………………………………………....
  2. INTRODUCTION
  3. SIGNIFICANCE OF MICE……….…….……………………………...………..
  4. MARKET POTENTIAL…...……………………………………………...………
  5. PULSE OXIMETRY...…………………………………………….………………
  6. PROBLEM TO BE SOLVED……………………………………………………..
  7. GOALS…………………………………………………………………………….
  8. CURRENT DEVICES…………………………………………………………….
  9. METHODOLOGY
  10. PULSE OXIMETRY THEORY…..………………………………………………
  11. RATIO OF RATIOS………………………………………………………………
  12. PROJECT SCOPE AND CIRCUITRY…………………………………………..
  13. DESIGN SPECIFICATIONS………....………………………………………….
  14. RESULTS
  15. EXPERIMENTAL DESIGN………………………………………….………….
  16. DESIGN PERFORMANCE……….…………………………………….……….
  17. PROBLEMS ENCOUNTERED/SOLUTIONS………………………….……….
  18. SAFETY ANALYSIS……………………………………………………..………
  19. ECONOMIC ANALYSIS…………………………………………………………
  20. CONCLUSIONS…………………………………………………………….…………
  21. RECOMMENDATIONS………………………………………………….…………..
  22. REFERENCES…………………………………………………………...……………

APPENDIX*

A.INNOVATION WORKBENCH…………..………………………………………

B.DESIGNSAFE REPORT…………………………………………………...…….

1. ABSTRACT

This design project is the creation of the NovaMouseTM pulse oximeter, which will be used to measure oxygen saturation and blood pressure of mice. The measurement of the oxygen saturation level is important because an insufficient supply of oxygen can result in death within minutes during surgery. The miniature size of mice that allow for ease of maintenance is also what causes the largest drawback during research involving mice. The main goal of this project is to build a pulse oximeter that is small, has the capability to be wireless in the future, and can reach heart rates up to 800beats per minute. One photo sensor is used for all measurements because the two light-emitting diodes are pulsed at different intervals. The received red and infrared signals have a non-linear relationship and are then mathematically preprocessed and normalized so that a ratio of the two can theoretically be a function of only the concentration of oxyhemoglobin and reduced hemoglobin in the arterial blood. This concentration ratio known as the ratio of ratios (R) should be constant since oxygen saturation is essentially constant for measurements taken over such a short period of time. The circuit designed has a frequency range of 0.1 hertz to 250 hertz and a gain of 625. From the Bode plots, the lower cutoff for the infrared light-emitting diode is not seen. The Bode plot of the red light-emitting diode illustrates the stair step characteristic of a Butterworth filter. These results are seen because the bandwidth is so small (only two magnitudes). The circuit is ready to be miniaturized and tested on mice.

2. INTRODUCTION

2.1 SIGNIFICANCE OF MICE

The field of biomedical research on mice is rapidly growing. The popularity of mice in research is increasing due to their ease and quickness of breeding, the ability to breed transgenic mice, and their miniature size, which reduces their maintenance costs in laboratories. This design project is the creation of the NovaMouseTM pulse oximeter, which will be used to measure oxygen saturation and blood pressure of mice.

2.2 MARKET POTENTIAL

NovaMouseTM will be marketed to universities and hospitals nationwide to aid in research and data acquisition. Approximately 50,000 devices are projected to be sold in 2004 with one million mice as potential targets. This accounts for around 10-20% of the market share. It is difficult to pin down the exact market since the FDA and NIH do not keep records on mice. [1]

2.3 PULSE OXIMETRY

Pulse oximetry is used to measure blood oxygen saturation. Heart rate is also easily attainable using a pulse oximeter. The measurement of the oxygen saturation level is important because an insufficient supply of oxygen can result in death within minutes during surgery. Also, pulse oximetry is very important in Intensive Care post-surgery to monitor recovery. Pulse oximetry readings are standard measurements in human and veterinary surgical monitoring. This shows that there is a need for such measurements in surgical procedures on mice.

These measurements would also be useful for research on free ranging mice. For example, oxygen saturation measurements could be used on researching the effects of alcohol and carbon monoxide poisoning as well as comparing cardiac output to oxygen saturation. [2] The additional heart rate measurements would also be useful in such studies. An example of the need for this measurement can be found here at Vanderbilt. At the VanderbiltMedicalCenter, Dr. Timothy Blackwell, M.D. would like a pulse oximeter to use on the mice in his investigation of pulmonary fibrosis. [3] Fibrosis, or scarring of the lung tissue, results in permanent loss of the tissue’s ability to transport oxygen. A pulse oximetry measurement would show decreased oxygen saturation if fibrosis was present. [4]

2.4 PROBLEM TO BE SOLVED

The miniature size of mice that allow for ease of maintenance is also what causes the largest drawback of research involving mice. Conventional methods for obtaining physiological measurements, such as blood gases and heart rate, do not work on mice. The NovaMouseTM pulse oximeter is a device to accurately measure blood oxygen saturation and heart rate in mice.

Prior to this product, any time data acquisition of pulse oximetry was desired, a mouse was killed in the process since the only way to acquire that data was via sticking rods through the mouse’s hands and feet. This device will spare the mouse’s life allowing for the observance of several drug administrations and the reusing of mice. NovaMouseTM would allow for reduction of costs. Not only would the device itself be less expensive than competitor’s products by a factor of ten, but the survival of the mice would also allow for reuse of the expensive transgenic mice. [1]

2.5 GOALS

The main goal of this project is to build a pulse oximeter for a mouse with the following characteristics. First, the sensor must be small enough to accurately obtain pulse oximetry measurements from the proximal end of the tail of a mouse as well as monitor the heart rate of the mouse. Second, the pulse oximeter must be compatible so in the future it can be wireless. This will help with ease of movement of the mouse during studies that require measurements to be taken over extended periods of time in which the mouse should be in its natural environment. Next, the pulse oximeter must be able to accurately detect heart rates of up to 800 beats per minute. Also, the device will be reusable and the packaging design will be built on a flexible substrate using chip and wire. The sensor will have a cuff format that will be located in a hard to reach place so that the mouse doesn't eat it off. The sensor will also be discrete, meaning that there won't be any wires attached.

To achieve these goals, a variety of literature was used. An introduction to the aim of the project and the probable components was attained through meetings with MicroNova Technology staff, including Ben Schnitz and Bob Allen. Further research was done through the use of biomedical handbooks, internet searches, and patent searches to verify pulse oximetry designs and parameters. To determine the possible uses and a definite market for the device, a personal interview was conducted with Dr. Blackwell. Dr. Blackwell also was helpful in providing information about previous devices and their downfalls. With this information, websites of individual companies, including Nonin Medical, Inc. were searched.

2.6 CURRENT DEVICES

Recent devices do not allow for movement of the mouse and are incapable of detecting heart rates high enough to monitor that of mice. One example is a pulse oximeter designed by Nonin Medical, Inc. This device, shown in Figure 1, only works on large rodents and is not useful for a mouse. The Velcro is flimsy and it is difficult to attach to the mouse. Also, although this oximeter can read a human and large rodent’s blood saturation, it cannot read a mouse’s due to a heart rate of between 450 and 800 beats per minute. Even though this device allows for some movement of the mouse, the wires can still hinder normal behavior of a free ranging mouse. On the other hand, NovaMouseTM will try to overcome all these disadvantages posed by the Nonin device. [5]

3. METHODOLOGY

3.1 PULSE OXIMETRY THEORY

In order to design NovaMouseTM, it is important to look at the theory behind pulse oximetry. The NovaMouseTM pulse oximeter works by measuring the hemoglobin in the arteries found in the mouse’s tail via optical sensing. The device will be secured to the proximal end of the tail where a central artery and two veins are found.

Relative proportions of oxygenated and reduced hemoglobin in the arterial blood determine the oxygen saturation (SpO2) of arterial blood. Reduced hemoglobin is hemoglobin that is not carrying oxygen, i.e. deoxygenated blood, while oxyhemoglobin corresponds to oxygenated blood. The pulse oximeter works by measuring the difference in the absorption spectra of these two forms of hemoglobin in order to calculate the SpO2 using two light-emitting diodes, one of wavelength 700 nm in the red band and one of wavelength 950nm in the infrared, and a photodiode. Similar to the human’s finger, the light-emitting diodes are used to sense the degree of oxygen saturation in blood since the reduced hemoglobin absorbs more light in the red band than does oxyhemoglobin, while oxyhemoglobin absorbs more light in the infrared band than does reduced hemoglobin.

Because the tissue contains arterial, capillary and venous blood, as well as muscle, connective tissue and bone, the red and infrared signals received from the probe contain both a DC and a pulsatile component. While the large DC component is influenced by the absorbency of the tissue, the intensity of the light source, and the sensitivity of the detector, the small pulsatile component corresponds to the pulsatile arterial blood.

3.2 RATIO OF RATIOS

The amount of light transmitted through the tail is measured several hundred times per second at both wavelengths. One photo sensor is used for all measurements because the two light-emitting diodes are pulsed at different intervals (6). The received red and infrared signals have a non-linear relationship and are then mathematically preprocessed and normalized so that a ratio of the two can theoretically be a function of only the concentration of oxyhemoglobin and reduced hemoglobin in the arterial blood. This concentration ratio known as the ratio of ratios (R) should be constant since oxygen saturation is essentially constant for measurements taken over such a short period of time. The preprocessing MicroNova plans to do such as averaging over time and planting the device on a relatively immobile area will eliminate extra noise variables such as motion artifact and ambient light [7].

The mathematical side of the ratio of ratios includes Equation 1 that characterizes

( ∑ Hb(λR) - ∑ Hb(λIR) *R) 100 Equation 1

∑ Hb(λR) - ∑ HbO2(λR) + [ ∑ HbO2(λIR) - ∑ Hb(λIR)] R

the amount of light absorption at each frequency for oxygenated and deoxygenated blood to give the non-linear relationship referred to earlier. The values in Table 1 represent the

Table 1: Saturation of hemoglobin at both wavelengths.

∑ Hb(λR) / 0.81
∑ Hb(λIR) / 0.18
∑ HbO2(λIR) / 0.29
∑ HbO2(λR) / 0.08

saturation of hemoglobin at each wavelength. For instance, ∑ Hb(λIR) is the saturation of deoxygenated blood at 950 nm. Using these values Equation 2 is generated and R, the ratio of ratios, is used to get the oxygen saturation desired. R should be between .5 and 2.5 [1].

SpO2 = (.81-.18R) 100 Equation 2

(.29-.08R)

3.3 PROJECT SCOPE AND CIRCUITRY

Figure 1, which represents the project outline, illustrates our scope of the project as the blue parts, while the red parts represent the parts MicroNova still has to do. The



first part of the project to research and design is the biosensor. Here, the light-emitting diodes sense the blood saturation and send both the DC and pulsatile signal to the photodiode. In the preamplifier stage, the transimpedance amplifier converts the current coming from the photodiode to a voltage. Because there is a very large input impedance and negligible output impedance, a very large feedback resistor is used to set the current gain as can be seen in Figure 2. Also included is a capacitor of .1 microfarad in order to



tune out high frequency measurements and stabilize voltage spikes. Once the signal is conditioned into a voltage output, an analog to digital conversion card is used to format the signal into a digital signal that can be sent wirelessly.

In order to wirelessly transfer the information serial transmission is utilized since it requires only one transmitter. Moreover, it is accomplished using amplitude shift keying modulation. This method is preferred to other types of modulation because 1) it is the easiest to implement, 2) it is easy to read – does a signal exist or not, and 3) it does not require any sort of interpretation of the signal.

Although the one drawback is that there can be lots of error since zero acts as a data point, there are ways to fix the error. For one, there can be several check points where one can check for 5 points before and after the point recorded to make sure the signal received includes zeros that are part of the signal and not zeros that mean the signal is dead. Although this decreases the sampling rate, that is acceptable because the pulse oximeter for mice doesn’t need a high sampling rate. Another advantage is that the distance between the mouse and receiver is short; thus there is less noise [2].

After the signal is received, it is reconstructed back into an analog signal and then sent first through an active high pass filter of gain = 6.25 to get rid of signals below .1 Hertz and then a two-pole low pass filter (8) to rid signals above 250 Hertz. This gives the desired frequency range of .1-250 Hertz. Finally, the signal is passed through yet another low pass filter with a gain of 100 to get amplification and get rid of all frequencies passed the highest, important frequency before data acquisition. If this anti-aliasing process was not done, we would use the Nyquist theorem of sampling frequency for biological situations and sample at a frequency of 5 times the highest frequency as opposed to five times the highest, important frequency. Once the data is collected, the signal is processed in LabView in order to monitor the blood saturation.

3.4 DESIGN SPECIFICATIONS

In order to be able to detect a high heart rate, the frequency range must be between 0.1 Hertz and 250 Hertz. Therefore, the circuit must eliminate all frequencies above and below this range including random noise. Also, power to the operational amplifier and light-emitting diodes will be provided via inductive coupling and voltage regulators. But right now, we are modeling the dual operational amplifiers with a power supply of 3.6 volts. This type of operational amplifier can be operated with a power supply from 2 volts to 44 volts. However, a voltage of 1.7 volts is needed for the light-emitting diodes, so right now the function generator is being used as the source. Also there is a very large gain in the transimpedance amplifier with a gain of 625 in the active filtration. As for the device, it will be designed to fit a mouse with the following dimensions: 3 millimeter diameter at the base with 2 centimeters before the tail begins to taper.

4. RESULTS

4.1EXPERIMENTAL DESIGN

Figure 1 below shows the experimental setup for our pulse oximeter design. The function generator provides the light-emitting diode with a voltage of 1.7 volts and the power supply provides the circuit with a voltage of 3.6 volts. The light-emitting diode and the photodiode are put in a box so that ambient light cannot affect the signal. The function generator is set to a frequency of 0.1hertz to begin with and is incremented all the way to 250 hertz. The output signal is seen on the oscilloscope and the output voltage is recorded.

4.2DESIGN PERFORMANCE

Currently, the characterization of the light-emitting diodes has been completed. This means that Bode plots have been obtained using the red and the infrared light-emitting diodes by pulsing them from the frequency range as mentioned above. The results obtained are shown in graphs 1 and 2 below. Looking at the graphs, it can be noted that the since the bandwidth is very small, the plot appears as a stair step for the red light-emitting diode.

Graph 1: Bode plot for red LED.Graph 2: Bode plot for infrared LED.

4.3PROBLEMS ENCOUNTERED/SOLUTIONS

During this project several problems were encountered. First, the red light emitting diode had a voltage range of 1.8 volts to 4 volts and the infrared light-emitting diode had a voltage range of 1 volt to 1.5 volts. Since there was no overlap in the voltage range, there was no way to power our light-emitting diodes. To solve this problem, light emitting diodes with the same or overlapping voltage ranges were found at 1.7 volts. Thus, a voltage divider will be used to make sure that the battery voltage does not go below 1.7 volts; otherwise the light-emitting diodes will not work.

Another problem was that several operational amplifiersstopped working due to static electricity and therefore operational amplifiers had to be switched and static wristbandswere carefully used. The light-emitting diodes had inconsistent performance. Sometimes they worked and other times they didn’t. Several different types of light-emitting diodes were used. Similarly, there were problems with the filters. Several different techniques of trouble shooting were used.