Design of a Pulse Oximeter for Use in Mice

Spring 2005 Senior Design Project

Daniel J. Ford, Deanna R. Nachreiner, Robert E. Thomas

Department of Bioengineering

The Pennsylvania State University

Sponsor: Dr. Herbert H. Lipowsky

Executive Summary ׀ Background and Significance ׀ Design Criteria ׀ Design Process Overview ׀ Budget ׀ Design Evaluation ׀ Conclusion

Executive Summary:

Pulse oximetry is a method used to measure oxygen saturation levels in the blood non-invasively. This measurement is not only crucial to human patients undergoing anesthesia, but is also of great value to animals undergoing similar procedures. While pulse oximetry is relatively easy to employ on larger animals, it is often difficult to obtain a signal in a mouse. The goal of this project is to modify a pulse oximeter design such that oxygen saturation levels can be measured economically and effectively in mice. The primary approach is to build a pulse oximeter sensor using two light emitting diodes (LEDs) and two photodiodes, which will be connected to an amplification circuit and then fed to an oscilloscope for observation. Signal post-processing will take place in excel using calibrated as well as know, empirical relationships and will yield SaO₂ values. The deliverable of this project is a pulse oximeter design that is capable of detecting hemoglobin (Hb) saturation levels in mice in the range of 65 -95%; the project will be completed by April 22.

Background and Significance:

The percentage of Hb in the blood that is saturated with oxygen can tell a physician a great deal about a patient’s health status and in particular, can indicate whether the patient is receiving enough oxygen to his or her vital organs. Pulse oximetry is a quick, effective, non-invasive approach that uses light sources and photo detectors with the goal of measuring blood oxygenation. In addition to its use in humans, pulse oximetry plays a vital role in the animal research realm; the monitoring of blood oxygen levels during surgeries and anesthesia is equally as important in animals as in humans. While sensors suiting larger species of animals currently exist, a pulse oximeter small enough to use on a mouse tail and sensitive enough to detect the pulsatile signal of a mouse that is up to six times faster than that of a humans does not, though the development of such a device is crucial for the advancement of small animal research. An effective, economical, non-invasive means by which to monitor blood oxygenation in mice would benefit investigators and animal well-being alike.

The basic concept behind pulse oximeter technology is that as infrared light (850-1000 nm) and red light (600-750 nm) are being transmitted through the skin and absorbed at different levels at these two wavelengths by oxygenated and deoxygenated Hb. Deoxygenated Hb allows more infrared light to pass through and thus absorbs more red light than oxygenated Hb.  After the two wavelengths of light have been transmitted through tissue and received by a photodiode, the ratio of red/infrared light intensity is computed. The hemoglobin saturation can therefore be determined by the level of light intensity incident at the two photodiodes. This process is illustrated as a general schematic below in figure 1:

Design Criteria:

This project is a sensor that will successfully be used to measure the hemoglobin SaO₂ of a mouse. The device will attach to a mouse tail and will be capable of doing the following:

• Measure mouse hemoglobin SaO₂ levels of 65 – 95%

• Work in a range of average mouse heart rates of approximately 400-600 beats per minute

• Fit comfortably on a mouse tail of approximately 3-5 mm diameter.

Design Process Overview:

I) BioPac System Implementation

First we attempted to use a pre-existing BIOPAC pulse oximeter finger probe in conjunction with the BIOPAC software available in the Medical Instrumentation Lab to determine accurate SaO₂ levels in mice. In this method, the software system configurations would have been modified in order to suit the rapid heartbeat of a mouse; the system changes would have included increasing variables such as sampling rates, frequencies, and gains by factors of 4-8. However, issues with system compatibility and software costs prevented us from pursuing this approach.

II) Pre-Existing Probe Reverse Engineering

Following a generous donation of a pulse oximeter probe from SurgiVet, our previously-proposed primary approach was modified: instead of using the BioPac system to attempt to obtain SaO2 readings, we decided to attempt to reverse-engineer the donated probe using pin-out diagrams also secured through SurgiVet.   The V3087 Mini Clip Sensor can be seen below in figure 2.

Figure 2 (V3087 Mini Clip Sensor)

Before undertaking this task, a preliminary circuit diagram and design were completed, and transimpedance op-amps that are specifically used in photodiode amplification were ordered from Newark. Using a female 9 pin connector, we accurately identified and powered the pins on the device which controlled the LEDs. We also identified the pins contributing to the Red/IR LED alternating feature of the probe and were able to supply the probe with an appropriate signal using the lab’s signal generator. To complete the preliminary stage of this process, we constructed an amplification circuit into which we were able to incorporate power, the generated signal, and the probe; using the oscilloscope, we saw a very noisy output signal.

Unfortunately, it was then brought to our attention that the sophistication of the probe would probably prevent us from successfully removing noise and acquiring a workable signal, forcing our group to take on a third design attempt.

III) Complete Probe and Circuitry Prototype Construction

The pulse oximeter design consists of LED excitatory circuits as well as photodiode sensory circuits; the receiving circuits convert the red and infrared light currents into voltages which can then be observed using an oscilloscope. Signal voltage data is then be recorded and post-processed in excel in order to obtain blood oxygen saturation levels. The approach can be broken down as follows:

1)    Two light emitting diodes (LEDs), of red (600 to 750 nm) and infrared (940 nm) wavelengths, were directed at tissue and activated by an excitatory circuit.

2)    The light emitted from the LEDs were transmitted through the skin and detected by two photodiodes. An infrared rejection filter photodiode was then placed across from the red LED in order to detect transmitted red light and prevent infrared light interference. Similarly, a visible light rejection filter photodiode was placed across from the infrared LED with similar intentions. 

3)   The two photodiodes were then connected to a transimpedance amplification circuit that converted the current to an appropriately-enhanced voltage signal.  The general circuit diagram for each photo diode are shown below in Figure 3 (Red Photodiode) and figure 4 (Infrared Photodiode).  The circuitry when these two circuits are combined can be seen in figure 5.

        Figure 3                                                             Figure 4                                               Figure 5 (circuit prototype)

4)   The output voltage was then observed through the use of an oscilloscope.  The output voltage received when a human was tested can be seen below in figure 6.

Figure 6 (human output voltage on oscilloscope)

5)  This oscilloscope signal was then acquired in a PC through the use of an RS32 connection and Agilent waveform processing software. The data was placed into files readable by excel, and were subsequently processed.  The results can be seen below in figure 7 (Voltage Output from Red Photodiode) and figure 8 (Voltage Output from IR Photodiode).

Figure 7                                                                                                       Figure 8

6)   The design accuracy was subsequently verified through intense calibration and experimentation procedures. The results of these verification processes can be found in the Design Evaluation section.

Design Evaluation:

One of the first tests run on our device was performed by placing pieces of paper between the LEDs and their corresponding photodiode.  As more pieces of paper were placed between the clamp, the intensity of light reaching the photodiodes was expected to decrease.  The results can be seen in figure 9 (Red log fit) and figure 10 (IR log fit).

    Figure 9                                                                                    Figure 10

The curves displayed in the above figures show that our device works under Beer's Law (equation 1).  Where I=t and V_o≈I.

(equation 1)                I=I_oe^(-αct)