Proceedings of the 25'" Annual International Conference of the IEEE EMBS
`Cancun , Mexico• September 17-21, 2003
`
`Measurement Site and Photodetector Size Considerations Jin Optimizing Power
`Consumption of a Wearable Reflectance Pulse Oximeter
`Y. Mendelson, Ph.D., C. Pujary, B.E.
`Department of Biomedical Engineering, and Bioengineering Institute
`Worcester Polytechnic Institute, Worcester, MA 01609, USA
`
`A bstruct- Site selection and power consumption play a
`crucial role in optimizing the design of a wearable pulse
`oximeter for long-term telemedicine application. In this study
`we investigated the potential power saving in the design of a
`renectance pulse
`oximeter
`taking
`into
`consideration
`ltl-vivo
`measurement
`site
`and
`sensor
`configuration.
`experiments suggest that battery longevity could be extended
`considerably by
`employing a wide annularly
`shaped
`photodetector ring configuration and performing SpO2
`measurements from the forehead region.
`
`Keywords- pulse oximeter, wearable sensors, telemedicine
`
`I. INTRODUCTION
`
`is a widely accepted
`Noninvasive pulse oximetry
`method
`for monitoring arterial hemoglobin oxygen
`saturation (SpO2). Oxygen saturation
`is an
`important
`physiological variable since insufficient oxygen supply to
`vital organs can quickly lead to irreversible brain damage or
`result in death.
`
`Pulse oximetry
`spectrophotometric
`is based on
`measurements of changes in blood color. The method relies
`on the detection of a photoplethysmographic (PPG) signal
`produced by variations in the quantity of arterial blood
`associated with periodic cardiac contraction and relaxation.
`
`Pulse oximeter sensors are comprised of light emitting
`diodes (LEDs) and a silicon photodetector (PD). Typically,
`a red (R) LED with a peak emission wavelength around 660
`nm, and an infrared (IR) LED with a peak emission
`wavelength around 940 nm are used as light sources. SpO2
`values are derived based on an empirically calibrated
`function by which the time-varying (AC) signal component
`of the PPG at each wavelength
`is divided by
`the
`corresponding time-invariant (DC) component which is due
`to light absorption and scattering by bloodless tissue,
`residual arterial blood volume during diastole, and non(cid:173)
`pulsatile venous blood.
`
`because of the relatively thin skin covering the skull
`combined with a higher density of blood vessels. On the
`contrary, other anatomical locations, such as the limbs or
`torso, have a rnuch lower density of blood vessels and, in
`addition,
`lack a dominant skeletal structure
`in close
`proximity to the skin that helps to reflect some of the
`incident
`light. Therefore,
`the AC components of the
`reflected PPGs from these body locations are considerably
`smaller. Consequently, it
`is more difficult to perform
`accurate pulse oximetry measurement from these body
`locations without enhandng cutaneous circulation using
`artificial vasodilatation.
`
`transmission or
`Sensors used with commercial
`reflection pulse oximeters employ a single PD element,
`typically with an active area of about l2-l 5mm2• Nonually,
`a relatively small PD chip is adequate for measuring strong
`transmission PPGs since most of the light emitted from the
`LEDs is diffused by the skin and subcutaneous tissues
`predominantly in a forward-scattering direction. However,
`in reflection mode, only a small fraction of the incident light
`is backscattered by the subcutaneous layers. Additionally,
`the backscattered light intensity reaching the skin surface is
`normally distributed over a relatively large area surrounding
`the LEDs. Hence, the design of a reflectance-mode pulse
`oximeter depends on the ability to fabricate a sensor that has
`improved sensitivity and can detect sufficiently strong PPGs
`from various
`locations on
`the body combined with
`to process
`the
`sophisticated digital signal algorithms
`relatively weak and often noisy signals.
`
`To improve the accuracy and reliability of reflection
`pulse oximeters, several sensor designs have been described
`based on a radial arrangement of discrete PDs or LEDs. For
`example, Mendelson et al [l]-[2] and Konig et al [3]
`addressed the aspect of unfavorable SNR by developing a
`reflectance sensor prototype consisting of multiple discrete
`PDs mounted symmetrically around a pair of R and IR
`LEDs. Takatani et al [41-[5] described a different sensor
`configuration based on IO LEDs arranged symmetrically
`around a single PD chip.
`
`in either
`SpO2 measurements can be performed
`transmission or reflection modes. In transmission mode, the
`sensor is usually attached across a fingertip or earlobe such
`that the LEDs and PD are placed on opposite sides of a
`pulsating vascular bed. Alternatively, in reflection pulse
`oximetry, the LEDs and PD are both mounted side-by-side
`facing the same side of the vascular bed. This configuration
`enables measurements from multiple locations on the body
`where transmission measurements are not feasible.
`
`interested
`long been
`The U.S. military has
`in
`combining noninvasive physiological sensors with wireless
`communication and global positioning to monitor soldier's
`vital signs in real-time. Similarly, remote monitoring of a
`located in a dangerous
`person's health status who is
`environment, such as mountain climbers or divers, could be
`beneficial. However, to gain better acceptability and address
`the unmet demand for long tenu continuous monitoring,
`Backscattered light i,1t?.nsity can vary significantly
`several technical issues must be solved in order to design
`between different anatomical locations. For example, optical
`more compact sensors and instrumentation that are power
`reflectance from the forehead region is typically strong
`0-7803-7789-3/03/$17 .00 ©2003 IEEE
`3016
`
`Authorized licensed use limited to: Kansas State University. Downloaded on July 11,2020 at 15:02:02 UTC from IEEE Xplore. Restrictions apply.
`
`1
`
`Masimo Ex. 2003
`Apple v. Masimo
`IPR2020-01714
`
`

`

`efficient, low-weight, reliable and comfortable to wear
`before they could be used routinely in remote monitoring
`applications.
`For
`instance,
`real-time
`continuous
`physiological monitoring from soldiers during combat using
`existing pulse oximeters is unsuitable because commercial
`oximeters involve unwieldy wires connected to the sensor,
`and sensor attachment to a fingertip restrains normal
`activity. Therefore, there is a need to develop a battery(cid:173)
`efficient pulse oximeter
`that could monitor oxygen
`saturation and heart rate noninvasively from other locations
`on the body besides the fingertips.
`
`To meet future needs, low power management without
`compromising signal quality becomes a key requirement in
`optimizing the . design of a wearable pulse oximeter.
`However, high brightness LEDs commonly used in pulse
`oximeters requires relatively high current pulses, typically
`in the range between 100-200mA. Thus, minimizing the
`drive currents supplied to the LEDs would contribute
`considerably toward the overall power saving in the design
`of a more efficient pulse oximeter, particularly in wearable
`wireless applications. In previous studies we showed that
`the driving currents supplied to the LEDs in a reflection and
`transmission pulse oximeter sensors could be lowered
`significantly without compromising the quality of the PPGs
`by increasing the overall size of the PD [6]-[8]. Hence, by
`maximizing the light collected by the sensor, a very low
`power-consuming sensor could be developed,
`thereby
`extending the overall battery life of a pulse oximeter
`intended for telemedicine applications. In this paper we
`investigate the power savings achieved by widening the
`overall active area of the PD and comparing the LEDs
`driving currents required to produce acceptable PPG signals
`from the wrist and forehead regions as two examples of
`convenient body locations for monitoring SpO2 utilizing a
`prototype reflectance pulse oximeter.
`
`II. METHODOLOGY
`A. Experimental setup
`
`To study the potential power savings, we constructed a
`prototype reflectance sensor comprising twelve identical
`Silicon PD chips (active chip area: 2mm x 3mm) and a pair
`of R and IR LEDs. As shown schematically in Fig. l, six
`PDs were positioned in a close inner-ring configuration at a
`radial distance of 6.0mm from the LEDs. The second set of
`six PDs spaced equally along an outer-ring, separated from
`the LEDs by a radius of 10.0mm. Each cluster of six PDs
`were wired in parallel and connected through a central hub
`to the common summing input of a current-to-voltage
`converter. The analog signals from the common current-to(cid:173)
`voltage converter were subsequently separated into AC and
`DC components by signal conditioning circuitry. The analog
`signal components were then digitized at a S0Hz rate for 30
`seconds intervals using a National Instruments DAQ card
`installed in a PC under the control of a virtual instrument
`implemented using Lab VIEW 6.0 software.
`
`Fig. I. Prototype reflectance sensor configuration showing the relative
`positions of the rectangular-shaped PDs and the LEDs .
`
`B. In Vivo Experiments
`
`A series of in vivo experiments were performed to
`quantify and compare the PPG magnitudes measured by the
`two sets of six PDs. The prototype sensor was mounted on
`the dorsal side of the wrist or the center of the forehead
`below
`the hairline. These representative regions were
`selected as two target locations for the development of a
`wearable telesensor because they provide a flat surface for
`mounting a reflectance sensor which for example could be
`incorporated into a wrist watch device or attached to a
`soldier's helmet without using a double-sided adhesive tape.
`After the sensor was securely attached, the minimum peak
`currents flowing through each LED was adjusted while the
`output of the amplifier was monitored continuously to
`assure that distinguishable and stable PPGs were observed
`from each set of PDs and the electronics were not saturated.
`
`Two sets of measurements were acquired from each
`body location. In the first set of experiments we kept the
`currents supplied to the LEDs at a constant level and the
`magnitude of the PPGs measured from each set of six PDs
`were compared. To estimate the minimum peak currents
`required to drive the LEDs for the near and far-positioned
`PDs, we performed a second series of measurements where
`the driving currents were adjusted until the amplitude of the
`respective PPG reached approximately a constant amplitude.
`
`III. RESULTS
`
`Typical examples of reflected PPG signals measured by
`the inner set of six PDs from the forehead and wrist for a
`constant peak LED current (R: 8.SmA, IR: 4.2mA) are
`plotted respectively in Fig. 2.
`
`The relative RMS amplitudes of the PPG signals
`measured by the six near (N) and far (F) PDs, and the
`combination of all 12 PDs (N+F) are plotted in Fig. 3(a) and
`3(b) for a peak R LED drive current of8.SmA and a peak IR
`LED drive current of 4.2mA, respectively. Analysis of the
`data revealed that there is a considerable difference between
`the signals measured by each set of PDs and amplitude of
`the respective PPG signals depends on measurement site.
`
`3017
`
`Authorized licensed use limited to: Kansas State University. Downloaded on July 11,2020 at 15:02:02 UTC from IEEE Xplore. Restrictions apply.
`
`2
`
`

`

`!_Ii.~ 1 :, ;;;1;
`
`,~
`
`0
`
`10
`S
`Time (s)
`
`IS
`
`O
`
`5
`10
`Time ( s)
`
`15
`
`(c) Wrist (_!R)
`
`23.40
`
`25 -
`
`~ 20 ·
`
`s
`'E 15
`~
`(3 10
`
`9.50
`
`>
`•
`
`<
`
`·\c----,,-- -,cc,- ..-,,
`Time(,)
`
`. 1,
`
`1 l . 1WiAt)Ji1,Nv~vN
`,.
`
`N
`
`F
`
`N+F
`
`Fig. 4. Relative LED peak dri,·ing currents required 10 maintain a constant
`PPG amplitude of0 .840V RMS for the near (N), far (F) and
`combinition (N+F) PD configurations. Measurements were
`obtained from the forehead .
`
`15
`
`r
`
`'
`
`Tlme(s:)
`
`Fig. z. Raw PPG signals measured from the forehead (a and b) and wrist
`(c and d) for constant LED driving currents.
`
`(a)
`
`j D Wrist
`
`C Forehead
`
`0.58
`
`0.47
`
`0.28
`
`0.27
`
`0.12
`
`N
`
`F
`
`N+f
`
`{b)
`
`(cid:143) Wrist
`
`C Forehead
`
`1.25
`
`0.8
`
`0.7
`.. 0.5
`E 0.6
`
`'t>
`~
`Q. 0.4
`E
`,( 0.3
`0
`Q. 0.2
`Q.
`0.1
`
`0
`
`1.8
`1.6
`E 1.4
`~ 1.2
`::,
`"" ii
`E 0.8
`,(
`0 0.6
`IL
`Q. 0.4
`0.2
`
`0
`
`N
`
`F
`
`N+F
`
`Fig. 3. PPG signal amplitudes measured by the near (N), far (F) and
`combination (N+F) PDs from the wrist and forehead for
`constant Rand IR LED drive currents corresponding to 8.5mA
`(a) and 4.2mA (b), respectively.
`
`Fig. 4 compares the relative peak LED currents required to
`maintain a constant AC RMS amplitude of approximately
`0.840(±0.0 I 7)V for the N, F and (N+F) PDs measured from
`the forehead.
`
`IV. DrscusSION
`
`The successful design of a practical wearable pulse
`oximeter presents several unique challenges. In addition to
`user acceptability, the other most important issues are sensor
`placement and power consumption. For example, utilizing
`disposable tape or a r,!usable spring-loaded device for
`attachment of pulse oximeter sensors, as commonly
`practiced in clinical medicine, poses significant limitations,
`especially in ambulatory applications.
`
`Several studies have shown that oximetry readings may
`vary significantly according to sensor location. For example,
`tissue blood volume varies in different parts of the body
`depending on the number and arrangement of blood vessels
`near the surface of the skin. Other factors, such as sensor-to(cid:173)
`skin contact, can influence the distribution of blood close to
`the skin surface and consequently can cause erroneous
`readings. Therefore, to ensure consistent performance, it is
`important to pay close attention to the design of optical
`sensors used in reflectance pulse oximetry and the selection
`of suitable sites for sensor attachment.
`
`The current consumed by the LEDs in a battery
`powered pulse oximeter is inversely proportional to the
`battery life. Hence, minimizing the current required to drive
`the LEDs is a critical design consideration, particularly in
`optimizing the overall power consumption of a wearable
`pulse oximeter. However, reduced LED driving currents
`directly impacts the incident light intensity and, therefore,
`could lead to deterioration in the quality of the measured
`PPGs. Consequently, lower LED drive currents could result
`in unreliable and inaccurate reading by a pulse oximeter.
`
`From the data presented in Fig. 2, it is evident that the
`recorded PPGs vary
`amplitude and quality of the
`significantly between the forehead and the wrist. We also
`observed that using relatively low peak LED driving
`currents, we had to apply a considerable amount of external
`pressure on the sensor in order to measure discemable PPG
`
`3018
`
`Authorized licensed use limited to: Kansas State University. Downloaded on July 11,2020 at 15:02:02 UTC from IEEE Xplore. Restrictions apply.
`
`3
`
`

`

`signals from the wrist. In contrast, using minimal contact
`pressure and similar LED driving currents produced
`significantly larger and less noisy PPG signals from the
`forehead. These noticeable differences are due to the lower
`density of superficial blood vessels on the arms compared to
`the highly vascular forehead skin combined with a strong
`light reflection from the forehead bone. Additionally, during
`conditions of peripheral vasoconstriction, a sensor placed on
`the forehead can maintain stronger PPGs longer compared
`to a finger sensor [9].
`
`Despite the noticeable differences between the PPG
`signals measured from the wrist and forehead, the data
`plotted in Fig. 3 also revealed that considerable stronger
`PPGs could be obtained by widening the active area of the
`PD which helps
`to collect a bigger proportion of
`backscattered
`light
`intensity. The additional
`increase,
`however, depends on the area and relative position of the
`PD with respect to the LEDs. For example, utilizing the
`outer-ring configuration, the overall increase in the average
`amplitudes of the R and IR PPGs measured from the
`forehead region was 23% and 40%, respectively. Similarly,
`the same increase in PD area produced an increase in the
`PPG signals measured
`from
`the wrist, but with a
`proportional higher increase of 42% and 73%.
`
`The data presented in Fig. 4 confirmed that in order to
`produce constant PPG amplitudes, significantly higher
`currents are required to drive the LEDs when backscattered
`light is measured by the outer PD set compared to the inner
`set. This observation was expected since the backscattered
`light intensity measured is inversely related to the separation
`distance between the PD and the LEDs [10]. In comparing
`the three different PD configurations, we found that by
`combining both PD sets to simulate a single large PD area,
`it is possible to further reduce the driving currents of the
`LEDs without compromising the amplitude or quality of the
`detected PPGs.
`
`Lastly, we used the LED peak driving currents plotted
`in Fig. 4 to estimate the expected battery life of a typical
`220mAh Lithium coin size battery assuming that a similar
`battery is used to power the optical components of a
`wearable pulse oximcter. Table I summarizes the estimated
`battery life for the different PD configurations tested in this
`study. The calculations are based on LEDs pulsed
`continuously at a typical duty cycle of approximately 1.5%.
`
`Table 1. Comparison of estimated battery life for different PD configurations .
`Values based on forehead measurements for a typical 220m.Ahr coin size battery.
`
`PD CONFIGURATION
`
`Near
`Far
`Near+Far
`
`BATTERY LIFE [Days]
`45.8
`20.3
`52.5
`
`Note that the estimated values given in Table 1 are very
`conservative since they rely only on the power consumed by
`the LEDs without taking into consideration the additional
`
`power demand imposed by otht:r ~omponents of a wearable
`pulse oximeter. Nevertheless, the considerable differences
`in the estimated power consumptions clearly points out the
`practical advantage gained by using a reflection sensor
`comprising a large ring-shaped PD area to perform SpO2
`measurements from the forehead region.
`
`V. CONCLUSION
`
`Site selection and LED driving currents arc critical
`design consideration
`in optimizing
`the overall power
`consumption of a wearable battery-operated reflectance
`pulse oximeter. In this study we investigated the potential
`power saving
`in a
`ring-shaped sensor configuration
`comprising two sets of photodetectors arranged
`in a
`concentric ring configuration. In-vivo experiments revealed
`that battery longevity could be extended considerably by
`employing a wide annular PD and
`limiting SpO2
`measurements to the forehead region.
`
`ACKNOWLEDGMENT
`
`the
`support by
`the
`We gratefully acknowledge
`Department of Defense under Cooperative Agreement
`DAMD 17-03-2-0006.
`
`REFERENCES
`
`[I] Y. Mendelson, J.C. Kent, B.L. Yocum and M.J. Birle, "Design and
`Evaluation of new
`reflectance pulse oximeter
`sensor," Medical
`lnslrumenlalion, Vol. 22, no. 4, pp. 167-173, Aug. 1988.
`[2] Y. Mendelson, M.J. McGinn, "Skin reflectance pulse oximetry; in vivo
`measurements from the forearm and calf/' Journal of Clinical Monitoring,
`Vol. 37(1),pp. 7-12, 1991.
`[3] V. Konig, R. Huch, A. Huch, "Reflectance pulse oximetry - principles
`and obstetric application in the Zurich system,"' Journal of Clinical
`Monitoring, Vol. 14, pp. 403-412, 1998.
`(4] S. Takatani, C. Davies, N. Sakakibara, ct al, "Experimental and clinical
`evaluation of a noninvasive reflectance pulse oximcter sensor/' Journal.
`Clinical Monitoring. 8(4), pp. 257-266, 1992.
`[5] M. Nogawa, C.T. Ching, T Ida, K. Itakura, S. Takatani, "A new hybrid
`reflectance optical pulse oximetry sensor for lower oxygen saturation
`measurement and for broader clinical application," in Proc SPIE, Vol.
`2976, pp. 78-87, 1997.
`(6] C. Pujary, M. Savage, Y. Mendelson, "Photodetector size considerations
`in the design of a noninvasive reflectance pulse oximeter for tclemedicine
`applications," in Proc. IEEE 29'' Annu. Northeast Bioengineering Con/,
`Newark, USA, 2003.
`[7] M. Savage, C. Pujary, Y. Mendelson, "Optimizing power consumption
`in the design of a wearable wireless telescnsor; Comparison of pulse
`oximetcr modes," in Proc. IEEE 29111 Annu. Norlheast Bioengineering
`Conf, Newark, USA, 2003.
`[8] Y. Mendelson, C. Pujary, M . Savage, '"Minimization of LED power
`consumption in the design of a wearable pulse oximctcr," in Proc.
`International Association of Science and Technology for Development.
`International Conj BioMED 2003, Salzburg, Austria, June 25-27, 2003.
`[9] D.E. Bebout, P.D. Mannheimer, C.C. Wun, '"Site-dependent differences
`in the time to detect changes in saturation during low perfusion;" Critical
`Care Medicine, Vol.29,no.12,p.AIIS,2001.
`[J OJ Y. Mendelson, B.D. Ochs , '"Noninvasive pulse oximetry utilizing skin
`reflectance photoplethysmography," IEEE Trans Biomet/. Eng., Vol. 35,
`no. I 0, pp. 798-805, Oct 1988.
`
`3019
`
`,I
`
`Authorized licensed use limited to: Kansas State University. Downloaded on July 11,2020 at 15:02:02 UTC from IEEE Xplore. Restrictions apply.
`
`4
`
`