Development of a low-cost pulse oximeter

simulator for educational purposes

Kevin Machado- Gamboa Andres Gonzalez-Vargas

Dept. Automática y Electrónica Dept. Automática y Electrónica
Universidad Autónoma de Universidad Autónoma de
Occidente Occidente
Cali, Colombia Cali, Colombia
kevin.machado@uao.edu.co amgonzalezv@uao.edu.co

Abstract— Medical instrumentation devices are crucial to

determine health conditions in humans. Teaching biomedical II. THE PULSE OXIMETER TECHNIQUE
instrumentation requires a perfect combination of practice and In order to understand how pulse oximetry devices work, it is
theory. Therefore, the use of physiological signals simulators such necessary to check the process that occurs when we introduce
as a %SpO2 becomes important as a teaching strategy. The
a skin tissue inside the pulse oximeter sensor. Thus, the
objective of this project is to develop and implement an oxygen
saturation simulator, for use in the biomedical instrumentation following concepts are required:
courses. The development process began by establishing the
requirements of the simulator and designing a graphical user A. Types of hemoglobin in the blood and diodes wavelengths
interface to control the simulator parameters. Subsequently, a
circuit capable of materializing the simulated signals from the Oxygen is transported by hemoglobin molecules in the
interface was developed, together with a probe or artificial finger blood [5-6]. The pulse oximetry technique in current devices
that was to be introduced inside the objective pulse oximeter. starts with the use of light-emitting diodes, each with a
Finally, the systems were integrated into the simulator specific wavelength [1,3,4-6]. This concept is explained in
[Fig. 1]. Studies have identified the safe range at which
Keywords— Pulse oximetry, low cost, education, biomedical wavelengths can be absorbed by the skin bed of the index
instrumentation, simulation finger in the human body [Fig. 1]. When hemoglobin in the
blood is carrying oxygen at its maximum saturation, the
percentage of SpO2 is 100% and this is shown by the red
I. INTRODUCTION
curve O2Hb. Conversely, when hemoglobin is not carrying
Medical instrumentation technology is important for the oxygen, the percentage of SpO2 is 0%, which is shown by the
diagnosis and management of health condition in human blue curve HHb [6]. This fact assumes that by varying oxygen
beings. Among the large amount of equipment available in saturation from 100% to 0% the curve should move from
hospitals and medical institutions, the pulse oximeter is one of O2Hb to HHb.
the most common [1,3]. The data obtained through this device
allows the monitoring of blood oxygen saturation levels
(%SpO2) and heart rate (HR) [1].
In a Biomedical Engineering program, bioinstrumentation
courses are vital to understand the design and construction of
medical devices such as the pulse oximeter. In many
educational institutions, students who build prototypes of
biomedical devices, face the problem of scarcity in the number
of simulation devices, given their high costs. The simulation
of physiological signals such as the SpO2 and HR, allows the
student to establish the correct functioning of these devices,
improve their design, correct errors and refine the skills to
optimize their results [2].
This project shows the construction of a low cost device Fig. 1. Absorption spectra for the O2Hb, HHb, COHb and MetHb species. [6]
that simulates different levels of SpO2 and HR, and
additionally intends to serve as a tool for the reinforcement of B. Reading of photoplethysmographic signals in the circuit
important concepts of the pulse oximetry technique.
After the student has managed to make a circuit that can
generate photoplethysmographic signals with red and infrared

978-1-5386-8372-9/18/$31.00 ©2018 IEEE

components, the digital processing of both signals is performed to III. USER INTERFACE VISUAL ELEMENTS
calculate the level of oxygen saturation. According to the
literature [3-5], the SpO2 is calculated from the reading of the A. Software characteristics
values of the so-called AC and DC components of the red and First, the characteristics of the operating system to use for
infrared signals. In [Fig. 2] the signals corresponding to both developing the system should be established. Due to the
components are shown after a process of separation of the signal educational nature of the project, the use of a programming
captured by the receiver photodiode in the sensor of a pulse language with the following characteristics was taken into
oximeter. For each of these, there is a current value "Imin" and account:
"Imax", from which the concentration of oxygen in the blood is
calculated indirectly. 1. Open source,
2. Object oriented,
3. It allows the development of customized user interfaces,
4. It has the ability to send signals from the interface to the
circuits in real time (data logger) and,
5. It has an ease to use and simple structure (computational
power or speed)

Given to the analysis of our requirements listed above and

the hardware and software achievements driven by [10], we
chose to work with the Raspberry Pi (RPi) and the Python
programming language, using libraries such as SciPy, NumPy,
Fig. 2. Illustration on photoplethysmographic signals in the red and infrared RtGraph, PyQt5 and integrating other software such as Qt
light components [8] Designer. The development of the user interface was carried
out using the free version of the Qt Designer software to
design a first interface model.
C. Values in photoplethysmographic signals and the
ratio relationship B. The curve of Ratio vs SpO2
Commonly, the current output of the receiver photodiode The relationship between the oxygen saturation value
that generates the signals in [Fig. 2] is transformed into (%SpO2) and the ratio (R-value) generated by the red and
voltage using a current-voltage converter [7-8]. The disposed infrared components is shown in [Fig. 4]. The value of R (red
values Imin and Imax for both red (Roj) and infrared (Iroj) dot on the blue line) changes with each SpO2 value chosen.
components become Amin and Amax respectively and are
related to each other in [eq. (1)]

(1)

D. R Ratio and SpO2 relationship

The value of the AC and DC components of the signal
received after crossing a skin tissue with a certain amount of
oxygen, is the input of a function that transforms and / or
Fig. 4. Ratio vs SpO2 design in Python
correlates this relationship in percentage values of SpO2 that
can be represented graphically in a cartesian plane such as in
[Fig. 3]. Normally, all pulse oximeters have a second or third C. Relationship of the simulated signals
order polynomial curve integrated [5,7-8] The amplitude relationship between the
photoplethysmographic signals for red and infrared, which
determines R, is shown in [Fig. 5]. In general, the curve that
represents the infrared (blue color) is greater when the SpO2 is
100% and decreases as the percentage of saturation decreases,
until the curve that represents red (red color) ends up being
higher.

Fig. 3. Theoretical (Lambert-Beer) and real calibration curve (in arterial blood
samples) for the HP M1190A adult sensor. [8]
 IV. BUILDING THE OPEN SOURCE SIMULATOR
Hardware design is partially based on the methodology in
in [3].

A. The fundamental aspects to build a simulator

The fundamental aspects necessary for the implementation
of the simulator were obtained in the literature [1,3-8]. The
device must comply with the following considerations:
• A SpO2 simulator must be able to capture and
Fig. 5. Amplitude signal relationship in python identify the red and infrared light signals from the
pulse oximeter, which are between 660nm and
990nm [3,9].
D. Absorption spectra for hemoglobin (HHb) • A simulator must be able to independently emit
and oxyhemoglobin O2Hb signals in the red and infrared light spectrum (660nm
The absorbance ratio of oxyhemoglobin and and 990nm), at the same frequency with which the
deoxyhemoglobin for different wavelength values between pulse oximeter activates its pair of red and infrared
650 nm and 950 nm, is illustrated in [Fig. 6]. diodes [3,9].
• A simulator must be able to independently control the
level of red and infrared light intensity that is emitted
to the photoreceptor of the pulse oximeter [3,9].
Subsequently, a block diagram was developed [Fig. 8] for
the system.

Fig. 6. Absorbance curve versus wavelength for hemoglobin HHb (red line)
and oxyhemoglobin O2Hb (blue line)

E. Integration of the graphic interface elements

Finally, the previous graphic elements were added to the
interface. The final version of the interface is shown in [Fig. 7]

Fig. 8. Block diagram of the simulator circuit based on the methodology of

S.J. Lorandi, G. LaMura and A.J. Kohen in [1]

B. Configuration of photodiodes
In order to corroborate the red and infrared signal capture
from the pulse oximeter, two receptor diodes were used. A
generic photodiode was used, which is able to measure the
wavelength range of the infrared spectrum 880/905 nm. Then
a photodiode capable of recording a range of wavelengths in
the red spectrum 653/659 nm and in the infrared spectrum
Fig. 7. Final version of the graphic interface, including specifications of the 880/905 nm was used.
bioinstrumentation course
With the above, the Masimo sensor photodiode together with
the generic one were used. Guided by the method in [2] we could
identify the light source activated from the pulse oximeter. The
diodes of the simulator were then synchronously activated and the
implemented circuit is shown in [Fig. 9]
 D. Generation of %SpO2 and HR parameters.
The simulation parameters set by the user in the graphical
interface, indicate the amplitude values of excitation signals
for the red and infrared diode of the sensor. These signals
arrive constantly to channels 1 and 3 of the multiplexer.
The formula in [eq. (2)] shows how the
photoplethysmographic curve of the infrared component (sIR)
is generated. The ampIR determines the amplitude of the wave
and the variable minIR its baseline level. This control over the
signals allows to vary the relation R between both and
Fig. 9. Polarization of photodiodes of the simulator therefore the SpO2 value.

C. Photodiode pulse preparation circuit

Following the methodology in [3], the outputs of the (2)
Masimo and the generic photodiodes were transformed into
appropriate pulses to generate synchronized signals that
E. Artificial finger for the simulator
activate the logic circuits of subsequent blocks. Fig. 10 shows
level detectors that amplify the outputs of the receiver The finger for the simulator was designed and built in 3D
photodiodes to more suitable voltages. The outputs of the printer using an ABS material. Then, the diodes of the
transistors carry the signals that indicate the frequency at simulator were coupled as shown in [Fig. 12].
which the red and infrared diodes have been activated in the
target pulse oximeter.

Fig. 10. Circuit for generation of selection signals

Outputs of the transistors [Fig. 11] are received by a circuit

that is responsible for directing the activation signals to the red
and infrared diodes of the simulator

Fig. 12. Developed process of the artificial finger sensor

F. Housing for the system

A simple box model was designed to introduce the circuit
and it was built in a laser printer using transparent acrylic,
item 1 shown in [Fig. 13]. The signals with the SpO2 and HR
characteristics determined by the user in the interface were
implemented by the Raspberry Pi 3, item 2 in [Fig. 13]. Item 3
in [Fig. 13] corresponds to the artificial finger and items 4 and 5
Fig. 11. Transistor outputs. The blue line corresponds to transistor Q1 are the power sources for the circuit and the Rpi.
(commercial photodiode). The yellow line corresponds to transistor Q2
(Masimo photodiode)

By combining both outputs with the selection inputs A

and B of the multiplexer circuit CD4051BE, we can identify
the activation of the infrared light source when A and B
acquire the logical level one and the activation of the red light
source when A is a logical one and B is logical zero.
 .

Fig. 13. Components of the UAOspo2sim simulator

Fig. 15. Change in the waveform of the photoplethysmographic signal
Finally, for easy handling of the simulator RealVNC observed in the monitor at a frequency of 176 beats per minute
software was used in its free version to manipulate from a
laptop the user interface implemented in the Raspberry.

V. RESULTS
B. Error estimation of %SPO2
A. Simulation ranges A total of 156 simulations of different %SpO2 values
The SpO2 values that can be simulated by the developed between 100 and 50 were performed in descending and
system are in the range between 50% and 100% SpO2. The ascending order. The absolute error (EAb) [eq. (3)] was
system performance tests were developed considering the calculated as the difference between the ‘observed value’
%SpO2 and HR readings of a Mindray D6 vital signs monitor. (VOb) and the ‘true value’ (VTe). The relative error (ERe) was
Prior to this, the reading of the oxygen saturation values of the calculated as the quotient between absolute error (EAb) and the
monitor with the finger probe AFP-1 of a SpO2 Metron model ‘true value’ (VTe) [eq. (4)]. EAb distribution is displayed in
DAEG analyzer was corroborated. The signal displayed on the [Fig. 16].
Mindray monitor's pulse oximeter at the moment of using our
simulator is very similar to a real human
photoplethysmographic signal [Fig. 14]. (3)

(4)

The mean and standard deviation for EAb were 0.4872 and
1.2826 respectively, and the mean and standard deviation for
Ere were 0.0078 and 0.0194, respectively. The maximum
relative error was 0.0755. These values indicate the device has
a very good precision and accuracy.

The graphic interface designs are available online, with free

access for any user in the link:
https://github.com/LatinAmericanProgramer/PyQtGraph-real-
time-plotting-RTgraph-modified-version.

C. Development cost
Fig. 14. Simulation with 70% SpO2 at 60 beats per minute. On the left the
computer for handling the graphical interface and on the right the pulse
The list of all the materials used for the development of the
oximeter where the results are observed. simulator is shown in [Table I]. The development cost was
approximately 143 US dollars.
Heart rate presented good correspondence with the simulation.
However, when simulating high frequencies (> 120 bpm) there A search was performed to compare the cost of similar
was a change in the waveform of the photoplethysmographic products in the market, which is showed in [Table II]. The
signal observed in the monitor [Fig. 15], despite the fact that the lowest priced product simulates just 80, 90 & 97% saturation
value corresponded to the simulated one. levels. Unlike our simulator, none of the products
commercially available contain any graphical user interface or oximeter of the vital signs monitor shows a waveform similar
explanatory visualisations of the physiological values. to the real photoplethysmographic signal produced in the pulse
oximetry technique. The user interface developed has
elements required for learning the pulse oximetry technique
Table I. List of materials used in the development of the and can be used in undergraduate bioinstrumentation courses.
simulator In future work, the device will be used as part of the regular
Items Cost (US Dollars) laboratory work carried out by the students and their opinions
Electronic components 18.00 will be gathered to understand what features can be improved.
Power supply (adaptor) 8.88
ABS housing 17.22
Raspberry Pi 3B 60.35 REFERENCES
Module CA/D PCF8591 9.59
Artificial finger design & print 9.59 [1] E. D. Chan, M. M. Chan e, M. M. Chan, “Pulse oximetry:
Understanding its basic principles facilitates appreciation of its
Housing Design & print 9.59 limitations,” Respiratory medicine 107, no. 6 (2013): 789-799. [online],
Available in: http://dx.doi.org/10.1016/j.rmed.2013.02.004
Circuit board design & print 9.59
[2] F. Lateef, "Simulation-based learning: Just like the real thing." Journal
Total 143.00 of Emergencies, Trauma and Shock 3, no. 4 (2010): 348.
[3] S. J. Lorandi, G. LaMura y A. J. Kohen, “Simulador optoelectrónico
Table II. List of the cost of similar products available in para el ensayo de oxímetros de pulso,” en XVIII Congreso Argentino de
the market Bioingeniería SABI 2011 - VII Jornadas de Ingeniería Clínica Mar del
Plata, 28 al 30 de septiembre de 2011.
Items Cost (US Dollars) [4] Webster, John G., ed. “Design of pulse oximeters,” CRC Press, 1997.
BC Biomedical Fingersim Kit SpO2 [5] E. D. Chan, M. M. Chan e, M. M. Chan, “Pulse oximetry:
FingerSim Starter Kit 475.00 Understanding its basic principles facilitates appreciation of its
CONTEC MS100 SpO2 Simulator 875.00 limitations,” Respiratory medicine 107, no. 6 (2013): 789-799. [online],
Available in: http://dx.doi.org/10.1016/j.rmed.2013.02.004
FLUKE Prosim SPOT light SPO2
[6] D. E. Pérez, “El Transporte de Oxigeno por la Hemoglobina y su
pulse oximeter analyzer 4500.00 Patología,” Real Academia Nac. Medicina, 1986.
OxSim II Classic Optical Pulse [7] L. Santiago, and R. T. A. C. Americas. "Pulse oximeter fundamentals
Oximetry Simulator 1500.00 and design." Free-scale Semiconductor Inc. application note document
Metron Finger probe AFP-1 model No AN4327 Rev 1 (2011): 4327.
daeg 1500.00 [8] S. Kästle, F. Noller, S. Falk, A. Bukta, E. Mayer, D. Miller, “A New
Family of Sensors for Pulse Oximetry,” Hewlett-Packard Journal,
Article 7. February 1997
[9] K. A. Ruiter. “Light transmission simulator for pulse oximeter,”
VI. CONCLUSIONS US7346378 B2. Cesionario original Pronk Technologies Inc. [online]
available in: https://www.google.ch/patents/US7346378
This article described the development of a device to [10] S. Sepúlveda, P. Reyes, A. Weinstein. "Visualizing Physiological
simulate oxygen saturation values between 50 and 100 Signals in Real Time." In Proc. of the 14th Python in Science Conf, pp.
%SpO2 and different heart rate values between 40 and 250 190-194.
beats per minute. The simulated signal observed in the pulse