2007 International Conference on Convergence Information Technology
`
`Optimum Place for Measuring Pulse Oximeter Signal in Wireless Sensor-Belt
`or Wrist-Band
`
`
`Miia Määttälä, Antti Konttila and Esko Alasaarela
`Optoelectronics and Measurement Techniques Laboratory, University of Oulu, Oulu, Finland
`miia.maattala@ee.oulu.fi
`Wan-Young Chung
`Division of Computer & Information Engineering, Dongseo University, Busan, South-Korea
`
`
`
`
`Abstract
`
`
`This study was done in order to solve the optimum
`place for integrated pulse oximeter in case of a belt
`around human chest or wrist so that it would provide
`reliable oxygen saturation (SpO2) readings for non-
`invasive constant health monitoring in modern wireless
`applications. In the study four spots on the wrist and
`on the chest were chosen and measurements from these
`spots were done by using a special device with an
`adjustable angle between light detector and the light
`sources. Then resulted signals were analyzed by
`calculating
`the average amplitude, normalized
`amplitude
`and
`signal-to-noise
`–ratio
`(SNR).
`Considerably clear signals were achieved both from
`wrist and chest and best resulted SNR was 27dB from
`the chest. The highest normalized signal amplitude was
`detected from the wrist appearing as big as 5.9% of the
`total measured signal, which was relatively high
`compared to other values. Results were promising and
`showed that an easy portable monitoring device is
`possible but also many problems, that should be
`overcome, were detected.
`
`
`1. Introduction
`
`
`the
`in
`is used universally
`Pulse oximetry
`supervision of critically ill patients in intensive care
`units and operating theatres to provide a reliable
`oxygen saturation reading [1]. Signs and symptoms of
`decreased ability
`to ventilate are, for example,
`cyanosis, dyspnea, tachypnea, decreased level of
`consciousness, increased work of breathing and loss of
`protective airway reflexes [2]. If some of these
`symptoms occur, patient assesment will determine the
`need for continuous oxygen saturation monitoring.
`Nowadays measurements are normally done by using a
`finger probe which can be considered reliable and
`
`practical in case the patient is hospitalized or lying
`steadily. Though, technologies and needs for different
`kind of healthcare are constantly changing. That is why
`it is extremely important to be able to offer solutions
`which can be used also when supervising moving
`patients in different kinds of environments. Especially,
`different modern home healthcare
`and
`sport
`applications recall for wireless, reliable and easy-to-use
`–methods so that testing could be done continuously
`also in real-life situations and not only in hospitals. In
`addition, elderly people healthcare
`is changing
`markedly because of internet and high technology
`telemedicine. It is already possible to see patient’s
`physiological state or some parts of it by transferring
`information through sensor networks and servers
`straight to doctor’s mobile phone or work station. That
`would make it easy for old people to stay home alone
`even though they have some disease which requires
`continuous supervision.
` In this study, pulse oximeter device, which was
`built in the University of Oulu (see Fig 5), was used in
`order to find the best place on human’s chest or wrist
`to provide the highest accuracy for pulse oximeter
`signal with lowest noise and disturbances. This study
`aims to optimize integration of pulse oximetry to
`wireless sensor belt and wrist band development.
`
`2. Principle and instrumentation
`2.1. Pulse oximeter principle
`
`
`joined
`Haemoglobin consists of four subunits
`together. When an oxygen molecule binds to one
`subunit, the other subunits become more likely also to
`bind oxygen. Haemoglobin saturation curve
`is
`characterized in to its S-form by this feature of
`haemoglobin. Shift to the right in the curve indicates
`that oxygen is bound less tightly – less is taken up in
`the lungs but is more easily released in tissues. The
`
`0-7695-3038-9/07 $25.00 © 2007 IEEE
`DOI 10.1109/ICCIT.2007.63
`
`1856
`
`Authorized licensed use limited to: Fish & Richardson PC. Downloaded on July 21,2023 at 03:46:10 UTC from IEEE Xplore. Restrictions apply.
`
`1
`
`APPLE 1055
`Apple v. Masimo
`IPR2022-01291
`
`

`

`from the slight change in the colour provoked by a beat
`of the heart pushing arterial blood into the finger [6].
`Pulse oximetry utilizes
`the plethysmographic
`principle in combination with the optical absorption
`characteristics
`of
`haemoglobin
`[7].
`Photoplethysmoraphy
`(PPG)
`is an electro-optic
`technique to measure the pulse wave of blood vessels.
`In pulse oximeter, which is the measuring apparatus for
`PPG, motion artefacts can limit the accuracy of the
`measured PPG signal during movement [8]. The
`typical pulse oximetry sensor contains two LEDs that
`emit red and infrared light into a pulsatile tissue bed.
`The scattered light is collected with a photodiode
`positioned on an opposite surface (forward scattering
`method) or an adjacent surface (reflection method).
`The “pulse” comes from the time-varying amount of
`arterial blood in the tissue during the cardiac cycle.
`Then
`the signals which are collected by
`the
`photodetector, crate a plethysmographic waveform due
`to the resulting cycling light attenuation. The relative
`modulation of the collected red and infrared light
`signals, referred to as the modulation ratio R, is used to
`estimate arterial oxygen saturation, SpO2, based on an
`empirical calibration relationship expressed within the
`oximeter [9]. Figure 2 shows a model of a pulse
`oximetric assumption in which pulsatile signal change
`is due to arterial blood volume change. In the figure
`alternative current and direct current can be calculated
`by using two equations given below [7].
`
`
`AC
`
`=
`
`Rd
`
`−
`
`Rs
`
` and
`
`DC
`
`=
`
`Rd
`
`Rs
`
`+
`2
`
`.
`
`
`
`oxyhaemoglobin dissociation curve is the relationship
`between the partial pressure of oxygen in the blood and
`the percentage of oxygen bound to haemoglobin
`compared to the maximum. Factors such as decreasing
`carbon dioxide concentration, increasing pH and
`decreasing temperature will shift the curve toward the
`left. A left-shifted curve implies that the haemoglobin
`molecules will be more saturated at lower partial
`pressure of oxygen.
` Figure 1
`shows
`the
`oxyhaemoglobin dissociation curve in three different
`cases [3].
`
`
`
`
`Figure 1. Oxyhaemoglobin dissociation curve.
`
`
`the arterial oxygen
`Clinical measurement of
`saturation of haemoglobin has been dominated lately
`by pulse oximetry – a non-invasive technology that is
`common practice under anaesthesia
`in operating
`theatres worldwide [4]. Dual-wavelength illumination
`of arterial blood results in an absorption contrast that
`depends upon the proportion of haemoglobin that is
`chemically combined with oxygen. Pulse oximeters
`differentiate between optical absorption by blood and
`other anatomical constituents by observation that
`pulsating arterial blood induces dynamics into the
`absorption characteristics of well-perfused peripheral
`sites [5]. The sensor is based on the fact that the colour
`of blood varies depending on the oxygen it contains. In
`particular the haemoglobin molecules reflect more red
`light when they are oxygenated than when reduced
`while its behaviour is the opposite when the light is
`infra-red. The oximeter shines two beams of light
`through a finger or earlobe or etc and those beams are
`finally received in the photodetector. By comparing the
`light intensity that is received for each wavelength, the
`oximeter is able to derive the light that is being
`absorbed by the blood and, consequently, the oxygen
`saturation. Moreover, the heart rate can be estimated
`
`
`Figure 2. Conceptual tissue model for pulse oximetry.
`
`2.2. Measurement device introduction
`
`
`When designing our device, we aimed to an easily
`and diversifiedly adjustable prototype. To reach this
`goal, programmable electronics was used to get the
`system adjustable straight from the computer. In
`addition, separate transmitter and receiver units were
`used
`to make different kinds of measurement
`geometries possible. The instrumentation which is used
`
`Authorized licensed use limited to: Fish & Richardson PC. Downloaded on July 21,2023 at 03:46:10 UTC from IEEE Xplore. Restrictions apply.
`
`1857
`
`2
`
`

`

`to form the pulse oximeter is built by PC, LabView 8.2
`–program, National Instrument Data Acquisition Card
`6211, voltage supply and the sensor part which was
`built in our university.
`
`
`Figure 3. Instrumentation
`
`
`
`
`
`
`
`In executing pulse oximetry, the controlling signals
`for the photo diodes are generated by the Data
`Acquisition Card. The photo diodes are
`time
`multiplexed by a frequency of 312.5 Hz by turns and
`when one is emitting light, it simultaneously transmits
`carrier wave whose frequency is 20 kHz. If the volume
`of blood in the measurement spot is changed, an
`amplitude modulation to the carrier wave is triggered
`and this modulation is demodulated by using coherent
`expression. The signal which reaches the receiver is
`high-pass filtered, rectified synchronously by using a
`demodulator.
`The phase shift of the carrier waves at the photo
`diodes is 180o. After low-pass filtering the voltage at
`the red photo diode is negative and for the infrared
`light diode positive. Because of this, no multiplexing is
`needed because differentiating these two different
`signals is based on the fact that one is negative and the
`other one positive.
`Now we have a curve which has minimum and
`maximum peaks which change their values by the
`change of the blood volume at the measurement spot.
`The peaks which are obtained at the frequency of 312.5
`Hz are identified by using the LabWiev-program
`created for this task. Minimum and maximum peaks
`are collected both to their own curves and from these
`two curves the amplitude modulation, caused by the
`blood volume change, can easily be seen. After this the
`program finds AC and DC components of red and
`infrared lights and calculates the ratio R which is used
`to determine the actual oxygen saturation value.
`The pulse oximeter execution principle is presented
`in figure 4.
`
`
`
`
`1858
`
`Figure 4. Pulse oximeter execution principle
`
`In our measurements we concentrated on studying
`SNRs and signal amplitudes instead of finding out
`actual SpO2-readings. That was done because of
`looking for an optimum spot that would provide strong
`noiseless signal from which actual SpO2 reading could
`later be calculated in our sensor belt or wrist band
`development.
`All measurements were done by using a special
`pulse oximeter device which was built
`in
`the
`University of Oulu. Device was built so that the
`adjusting angle, α, can be changed as big or as small as
`needed due to the location of the sensor on patient’s
`skin. By using this adjustable probe, it was possible to
`obtain visible pulse oximeter signal from almost every
`spot on patient’s skin by adjusting the angle and
`simultaneously checking
`the signal on computer
`screen. For each patient this angle in all measurement
`spots was a little bit different so it had to be specified
`separately every time.
`
`
`Figure 5. Measurement setup and principle from
`patient’s skin by using a special device with an
`adjustable angle between light sources and a detector
`
`
`
`
`
`
`
`
`3. Experiments
`
`
`In the experimental part of the study we only
`concentrated on measuring spots around human chest
`and wrist in order to find the most optimal places for
`the integrated SpO2 sensor. Before real measurements,
`some preliminary tests were accomplished in order to
`find the spots which would give the most clear
`saturation curves. Four spots from both wrist and chest
`were chosen (figures 6-9) and measurements from
`those areas were done. Measurement setup was built in
`a dark room to avoid external light interference. Also
`
`Authorized licensed use limited to: Fish & Richardson PC. Downloaded on July 21,2023 at 03:46:10 UTC from IEEE Xplore. Restrictions apply.
`
`3
`
`

`

`
`Figure 9. Measurement spots on the inner wrist
`
`
`
`
`
` Figures 10 and 11 are demonstrating how
`normalized amplitudes and SNRs were resulted in the
`measurements. Measurements from the wrist area were
`easier to accomplish because unfortunately chest area
`is extremely sensitive for disturbances because of a
`constant movement caused by breathing. Still all 8
`measurement areas were resulting close to each others
`and gave similar results. The basic rule was that big
`amplitudes were achieved by using wrist spots and
`higher SNR by using chest spots.
`
`
`Normalized amplitude
`
`Wrist 1
`
`Wrist 2
`
`Wrist 3
`
`Wrist 4
`
`Chest 1
`
`Chest 2
`
`Chest 3
`
`Chest 4
`
`Spot
`
`Figure 10. Normalized amplitude of the pulse
`oximeter signal in all eight measurement spots
`SNR
`
`Wrist 1
`
`Wrist 2
`
`Wrist 3
`
`Wrist 4
`
`Chest 1
`
`Chest 2
`
`Chest 3
`
`Chest 4
`
`Spot
`
`
`
`
`
`01234567
`
`Normalized amplitude (%)
`
`30
`
`25
`
`20
`
`15
`
`10
`
`05
`
`SNR (dB)
`
`
`
`
`
`Figure 11. The signal-to-noise –ratio of the pulse
`oximeter signal in all eight measurement spots
`
`
`
`The signal analysis showed that if the device is
`wanted to be integrated in a chest belt, the optimum
`place for integration would be chest 1 -spot. It
`
`target vibration was tried to eliminate as well as
`possible. The age distribution of the measured patients
`was 23-29 and their body structures were normal. 20%
`of the patients were Asian and 80% were Finnish but
`skin colour did not seem to make any affect on the
`results. More problems occurred if the thickness of the
`fat tissue in the measurement spots was bigger than
`others and that is why the spots were chosen locating
`near bones so that the distance between the sensor and
`the bone was small enough to provide good returning
`signal for the photo detector if the probe without too
`many reflections and refractions.
`
`
`
`Figure 6. Measurement spots on the chest
`
`
`
`Figure 7. Measurement spots on the back
`
`
`
`
`
`
`
`
`Figure 8. Measurement spots on the outer wrist
`
`
`
`
`Authorized licensed use limited to: Fish & Richardson PC. Downloaded on July 21,2023 at 03:46:10 UTC from IEEE Xplore. Restrictions apply.
`
`1859
`
`4
`
`

`

`provided SNR almost as great as 27dB, which tells that
`signal recorded from this spot is really clear. Promising
`SNR, 26dB, was also recorded from the wrist 3 –spot
`which offered also relatively big amplitude, 0.8V, and
`normalized amplitude, 5.9% of the total measured
`signal. In case of chest 1 –spot normalized amplitude
`was only 1.9% of the total.
`
`4. Errors
`
`
`Few of many possible facts that may have caused
`errors in the measurement results given above are
`perfusion problems, target movement and wrong
`target, detecting errors, optical interference and broken
`fibers. Perfusion problems are caused usually by a
`significant decrease in peripheral vascular pulsation,
`such as in hypothermia, vasoconstriction, hypotension,
`during cardiopulmonary bypass or cardiac arrest and
`they may result in a plethysmographic signal which is
`insufficient to be processed reliably by the oximeter
`[3]. Patient movement reduces the reliability of a pulse
`oximeter severely. Even small patient movements,
`such as shivering, can cause powerful artifacts that
`swamp the pulsatile absorption signal [10]. Also any
`dislocation of the nail bed for example can affect the
`transmission of light through the digit in case the
`measurement is done through a finger. Dark nail polish
`and bruising under the nail can also severely limit the
`transmission of light and result in an artificially
`decreased SpO2-value [11]. Detecting errors happen
`because pulse oximeters are unable to differentiate
`between oxygen and carbon monoxide bound to
`haemoglobin. Readings in the presence of carbon
`monoxide will be falsely elevated. Pulse oximetry
`should never be used in suspected cases of carbon
`monoxide exposure and an arterial blood gas reading
`would also be good to obtain. In addition they should
`never been used in a cardiac arrest situation because of
`the extreme limitations of blood flow [11]. Bright
`external light sources are known to affect pulse
`oximeters and all pulse oximeters share this sensitivity.
`This occurs because these instruments use optical
`means to make their measurements. Consequently, to
`obtain accurate measurements, potential sources of
`optical interference must be controlled. Because pulse
`oximeters’ optical components are located in the probe,
`proper probe application and use are key factors in
`reducing optical
`interference. Optical interference
`occurs when bright light from an external source
`reaches the photodiode or when light reaches the
`photodiode without passing
`through a pulsatile
`arteriolar bed [3]. In case of intra-vascular fibre optic
`oximetry sensors, optical fibres which are used to
`guide the light into the target may suffer several
`
`damages before reaching the target. This may cause
`severe measurement errors [7].
`
`5. Discussion
`
`
`This study concentrated on finding out the most
`suitable spot for the integrated pulse oximetry sensor
`which would make measurements not only when the
`patient is lying down but also while he or she is
`moving. That forces us to take into account the
`principal factor which
`limits both
`the practical
`accuracy and general applicability of pulse oxymetry –
`poor photoplethysmographic signal [12]. It is caused
`usually by low perfusion states or artifact corruptions
`arising from ambient light or subject movement [13],
`which lead usually to interpretation errors and false
`alarms [14].
`These days especially motion artifacts are tried to
`minimize because many applications are wanted to
`design for moving patients. Also measurements which
`were done in this study were affected by these artefacts
`for example by moving chest or shivering wrist. In
`case the interface between the measurement probe and
`the skin was moving, the signal immediately showed
`the movement by temporarily fading or peaks. That is
`why all measurements from all spots had to be taken
`many times from each patient so that reliable results
`could be recorded. Also the person executing the
`measurements was affecting the results by giving some
`small vibrations to the probe.
`Before the sensor can be reliably integrated to a
`chestbelt or wrist device, more experiments with
`moving patients should be done in order to see how
`movement in these spots affects the signal and can
`results be trusted. Also many signal processing steps
`still have to be completed in order to eliminate motion
`artifacts Many problems still have to be overcome but
`this study proves
`that wireless
`integrated pulse
`oximeter implanted in a chest belt or wrist band is
`possible to achieve with many benefits and practical
`applications.
`
`6. References
`
`[1] A. Jubran, “Advances in respiratory monitoring during
`mechanical ventilation”, Chest, vol. 116,1999, pp 1416-1425.
`
`[2] S.L. Schutz, “Oxygen Saturation Monitoring by Pulse
`Oximetry”, AACN Procedure manual for Critical Care,
`Fourth Edition, W.B.Saunders.
`
`[3] J. G. Webster, “Design of Pulse Oximeters”, Taylor &
`Francis Group, LLC, 1997.
`
`
`Authorized licensed use limited to: Fish & Richardson PC. Downloaded on July 21,2023 at 03:46:10 UTC from IEEE Xplore. Restrictions apply.
`
`1860
`
`5
`
`

`

`Tanaka,
`and K.
`Shimada
`I.Yoshiya, Y.
`[4]
`“Spectrophotometric monitoring of arterial oxygen saturation
`in the finger tip”, Med. Biol. Eng. Comput., vol. 18, no. 1,
`1980, pp 27-32.
`
`[5] A. B. Herzman, “Photoelectronic plethysmoraphy of the
`fingers and toes in man”, Proc. Soc. Exp. Biol. Med., vol. 37,
`1937, pp 529.
`
`[6] M. J. Moron, E. Casilari, R. Luque and J. A. Gazguez,
`“A Wireless Monitoring System
`for Pulse-oximetry
`IEEE Proceedings of
`the 2005 Systems
`Sensors”,
`
`Communications (ICW´05), 2005.
`
`
`[7] S. Takatani, and J. Ling, “Optical Oximetry Sensors for
`Whole Blood and Tissue”, IEEE Engineering in medicine
`and biology, no. 6, 1994, pp 347-357.
`
`[8] B. S. Kim and S. K. Yoo, “Motion Artifact Reduction in
`Photoplethysmography Using
`Independent Component
`Analysis”, IEEE Trans Biomed Eng, vol. 53, no.3, 2006, pp
`566-568.
`
`[9] M. Yelderman and J. Corenman, “Real time oximetry in
`Computing in Anesthesia and Intensive Care”, O. Prakash,
`Ed. Boston, MA: Martinus Nijhoff, 1983.
`
`
`
`
`[10] M. J. Hayes and P. R. Smith, “A New Method for Pulse
`Oximetry Possessing Inherent Insensivity to Artifact”, IEEE
`Transactions on biomedical engineering, vol. 48, no. 4, 2001
`pp 452-461.
`
`[11] S. L. Schutz, “Oxygen Saturation Monitoring by Pulse
`Oximetry”, AACN Procedure manual for Critical Care,
`Fourth Edition, W.B.Saunders, 2001.
`
`[12] M.J. Hayes, P.R. Smith, D.M. Barnett, M.D.L. Morgan,
`S. Singh and D.D. Vara, “Quantitative investigation of
`artefact in photoplethysmography and pulse oximetry for
`testing”, Frontiers Computer-Aided
`respiratory exercise
`Visual. Vascular Functions, vol. 263, 1997, pp 117-124.
`
`[13] N.S. Trivedi, A.F. Ghouri, N.K. Shah, E. Lai and S.J.
`Barker, “Effects of motion, ambient light, and hypoperfusion
`on pulse oximeter function”, J. Clin. Anesth., vol. 9, no. 3,
`1997, pp 179-183.
`
`[14] L. Wiklund, B. Hok, K. Stahl and A.Jordeby-Jonsson, “
`Postanaesthesia monitoring revisited: Frequency of true and
`false alarms for different monitoring devices”, J. Clin.
`Anesth., vol. 6, no. 6, 1994, pp 182-188.
`
`
`
`
`
`
`Authorized licensed use limited to: Fish & Richardson PC. Downloaded on July 21,2023 at 03:46:10 UTC from IEEE Xplore. Restrictions apply.
`
`1861
`
`6
`
`