`
`
`
`SKIN REFLECTANCE PULSE OXIMETRY: IN VIVO
`MEASUREMENTS FROM THE FOREARM AND CALF
`Y. Mendelson, PhD, and M. J. McGinn, MSc
`
`From the Department of Biomedical Engineering, Worcester Poly-
`technic Institute, 100 Institute Rd, Worcester, MA 01609,
`Received Sep 18, 1989, andin revised form Jan 11, 1990. Accepted for
`publication Jan 17, 1990.
`Address correspondence to Dr Mendelson.
`
`Mendelson Y, McGinn MJ. Skin reflectance pulse oximetry: in vivo
`measurements from the forearm andcalf.
`
`J Clin Monit 1991;7:7-12
`
`ABSTRACT. This study describes the results from a series of
`human experiments demonstrating the ability to measure arte-
`rial hemoglobin oxygen saturation (SaQ.) from the forearm
`and calf using a reflectance pulse oximeter sensor. A special
`optical reflectance sensor that includes a heating element was
`interfaced to a temperature controller and a commercial Data-
`scope ACCUSAT pulse oximeter that was adapted for this
`study to perform as a reflectance pulse oximeter. The reflec-
`tance pulse oximeter sensor was evaluated in a group of 10
`healthy adult volunteers during steady-state hypoxia. Hy-
`poxia was induced by gradually lowering the inspired fraction
`of oxygen in the breathing gas mixture from 100 to 12%.
`Simultaneous SaO2 measurements obtained from the forearm
`and calf with two identical reflectance pulse oximeters were
`compared with SaOQ, values measured by a finger sensor that
`was interfaced to a standard Datascope ACCUSATtransmit-
`tance pulse oximeter. The equations for the best-fitted linear
`regression lines between the percentreflectance, SpO.(r), and
`transmittance, SpO.(t), values in the range between 73 and
`100% were SpOo(r) = — 7.06 + 1.09 SpOo(t) for the forearm
`(n = 91, r = 0.95) and SpO2(r) = 7.78 + 0.93 SpOa(t) for the
`calf (n = 93, r = 0.88). The regression analysis of the forearm
`data revealed a mean + SD error of 2.47 + 1.66% (SaOz =
`90-100%), 2.35 + 2.45% (SaQ2 = 80-89%), and 2.42 +
`1.20% (SaO2 = 70-79%). The corresponding regression anal-
`ysis of the calf data revealed a mean + SDerror of 3.36 +
`3.06% (SaO2z = 90-100%)}, 3.45 + 4.12% (SaOz = 80-89%),
`and 2.97 + 2.75% (SaOQ»s = 70-79%). This preliminary study
`demonstrated the feasibility of measuring SaOQ, from the
`forearm and calf in healthy subjects with a heated skin reflec-
`
`tance sensor and a pulse oximeter.
`KEY WORDS. Blood gas analyses. Monitoring: oxygen. Mea-
`surement techniques: pulse oximetry; optical plethysmog-
`raphy; reflectance oximetry. Equipment: pulse oximeters.
`
`Transmittance pulse oximetry has become a widely
`used technique for noninvasively monitoring changes in
`arterial hemoglobin oxygen saturation (SaQz). The
`techniqueis based on the spectrophotometric analysis of
`the optical absorption properties of blood combined
`with the principle of photoplethysmography.
`In transmittance pulse oximetry, which is based on
`tissue transillumination, sensor application in adults is
`limited to several specific locations on the body, such as
`the finger tips, ear lobes, and toes. In infants, additional
`monitoring sites such as the palms and the feet have
`been used.
`Recently, a new reflectance pulse oximeter has been
`introduced into the market. The oximeter, which is
`manufactured by Ciba-Corning (Ciba Corning Diag-
`nostics, Medfield, MA), uses a special optical reflectance
`sensor for specific application to the forehead. Among
`the advantages of this technique, as advertised by the
`
`APPLE 1015
`
`1
`
`APPLE 1015
`
`
`
`
`
`NSOOx<2 TEMPERATURE
`
`TRANSDUCER
`
`A
`
`OPTICAL SHIELD
`
`PHOTODIODES
`OPTICALLY
`
`
`
`CLEAR ERRARSTe
`
`NAQANANY
`
`SOS
`
`
`Rie TR [LEDS
`
`PHOTODIODES
`BRASS RING
`
`THERMOFOIL
`HEATER
`
`SILICONE
`RUBBER
`
`Fig 1. (A) Frontal and (B) side views of the heated skin reflec-
`tance pulse oximeter sensor. See text for explanation. R & IR
`LEDs = red and infrared light-emitting diodes.
`
`cludes an array ofsix identical photodetectors arranged
`symmetrically in a hexagonal configuration surround-
`ing two pairs of red (peak emission wavelength, 660
`nm) and infrared (peak emission wavelength, 930 nm)
`light-emitting diodes (LEDs)
`[1].
`In another related
`study, we showedthat by locally heating the skin under
`the sensor to a temperature above 40°C, it is possible to
`achieve a four- to fivefold increase in the magnitude of
`the pulsatile component detected from the forearm, and
`thus significantly improvethe detectionreliability of the
`reflectance photoplethysmograms [2]. The new optical
`reflectance sensor designed for this study combines the
`two features described above.
`
`SENSOR DESIGN
`
`The temperature-controlled optical reflectance sensor
`used in this study is shown in Figure 1. The major fea-
`ture of the optical layout design is the multiple photo-
`diode array, which is arranged concentric with the
`LEDs. This arrangement maximizes the amount of
`backscattered light that is detected by the sensor. The
`technical details related to the design and geometric
`
`8 Journal of Clinical Monitoring Vol 7 No 1 January 1991
`
`company, are better reliability in critical care situations
`such as peripheral circulatory shutdown,less interfer-
`ence from ambient light, and better accuracy because
`measurement from the foreheadis relatively unsuscep-
`tible to motion artifacts.
`Currently,
`there are no commercially available re-
`flectance pulse oximeters for monitoring SaO2 from lo-
`cations other than the forehead. Therefore, the objective
`of this work was to investigate the feasibility of moni-
`toring SaO2 with a skin reflectance pulse oximeter from
`two alternative and convenient locations on the body:
`the ventral side of the forearm and the dorsalside of the
`calf. Besides extending the clinical application of pulse
`oximetry, it appears also thatreflectance pulse oximetry
`from peripheral tissues may have potential advantage in
`the assessment of local blood oxygenation after skin
`transplantation and regeneration following microvascu-
`lar surgery.
`In this article, we describe preliminary in vivo evalua-
`tion of a new optical reflectance sensor for noninvasive
`monitoring of SaO. with a modified commercial trans-
`mittance pulse oximeter. We present the experimental
`evaluation of this sensor in a group of 10 healthy adult
`volunteers and compare SaO2 measured with thereflec-
`tance pulse oximeter sensor, SpO(r), with SaQz mea-
`sured noninvasively from the finger by a standard trans-
`mittance pulse oximeter sensor, SpO2(t).
`
`REFLECTANCE PULSE OXIMETRY
`
`The principle of reflectance, or backscatter, pulse ox-
`imetry is generally similar to that of transmittance pulse
`oximetry. Both techniques are based on the change in
`light absorption of tissue caused by the pulsating arterial
`blood during the cardiac cycle. The pulsating arterioles
`in the vascular bed, by expanding and relaxing, mod-
`ulate the amountof light absorbed by the tissue. This
`rhythmic change produces characteristic photoplethys-
`mographic waveforms, two of which are used to mea-
`sure SaO, noninvasively.
`Recently, we showed that accurate noninvasive mea-
`surements of SaOQ>z from the forehead can be made with
`an unheated reflectance pulse oximeter sensor [1]. The
`major practical limitation of reflectance pulse oximetry
`is the comparatively low-level photoplethysmograms
`recorded from low-density vascular areas of the skin.
`Therefore, the feasibility of reflectance pulse oximetry
`depends on the ability to design an optical reflectance
`sensor that can reliably detect sufficiently strong reflec-
`tance photoplethysmograms from various locations on
`the skin.
`In order to partially overcome this limitation, we
`have developed an optical reflectance sensor that in-
`
`2
`
`
`
`configuration of the optical components were described
`recently by Mendelson etal [1].
`The heater consists of a ring-shaped (dimensions:
`30-mm outside diameter; 15-mm inside diameter)
`thermofoil resistive heating element (Ocean State Ther-
`motics, Smithfield, RI). The thermofoil heater was
`mounted between the surface of the optically clear
`epoxy, which was used to seal the optical components
`of the reflectance sensor, and a thin (0.005 mm) match-
`ing brass ring, which facilitates better thermal conduc-
`tion to the skin. A miniature (dimensions: 2 x 5
`x 1 mm) solid-state temperature transducer (AD 590,
`Analog Devices, Wilmington, MA) was mounted on
`the outer surface of the brass ring with the thermally
`sensitive surface facing the skin. The entire sensor as-
`sembly was potted in room-temperature vulcanizing
`silicone rubber to minimize heat losses to the surround-
`ing environment. The assembled sensor weighs approx-
`imately 65 g. The sensor measures approximately 38
`mim in diameter and is 15 mm thick. The heater assem-
`bly was separately interfaced to a temperature controller
`that was used to vary the temperature of the skin be-
`tween 35 and 45°C in 1 + 0.1°C steps.
`
`SUBJECTS AND METHODS
`
`Data Acquisition
`
`Each of the two heated optical reflectance sensors were
`separately interfaced to a temperature controller and a
`commercially available ACCUSAT (Datascope Corp,
`Paramus, NJ) pulse oximeter [3].
`Two of the three ACCUSATpulse oximeters were
`modified to function as reflectance pulse oximeters. The
`modification, which was described in a separate study
`[1], included the adjustmentof the red and infrared LED
`intensities in the reflectance sensors so that the reflec-
`tance photoplethysmograms were approximately equal
`to transmittance photoplethysmograms measured by a
`standard transmittance sensor from an averagesize adult
`finger tip.
`The third ACCUSATtransmittance pulse oximeter
`was used as a reference to measure SpO.(t) from the
`finger tip. The specified accuracy of this transmittance
`pulse oximeter is 2.0% and +4.0% for SaO2 values
`ranging between 70 and 100%and 60 and 70%, respec-
`tively [3]. The three pulse oximeters were adapted to
`provide continuous digital readouts of the AC and DC
`components of the red and infrared photoplethysmo-
`grams.
`Readings from each ofthe three pulse oximeters were
`acquired every 2 seconds through a standard RS-232C
`
`Mendelson and McGinn: Skin Reflectance Pulse Oximetry
`
`9
`
`serial port interface using an AT&T 6300 personal com-
`puter. The conversions of the reflectance red/infrared
`(R/TR) ratios measured by the two reflectance pulse ox-
`imeters to SpO.(r) were performed by using the cali-
`bration algorithm obtained in a previous calibration
`study in which measurements were made with a similar
`nonheated sensor from the forehead [1].
`
`In Vivo Study
`
`The ability to measure SpO2(r) from the forearm and
`calf was investigated in vivo during progressive steady-
`state hypoxia in humans.
`Measurements were acquired from 10 healthy non-
`smoking male adult volunteers of different ages and skin
`pigmentations. The study was performed in compliance
`with the University of Massachusetts Medical Center’s
`review guidelines on human experimentation. Each
`volunteer was informed of the complete procedure as
`well as the possible risks associated with breathing hy-
`poxic gas levels. Each volunteer received monetary
`compensation for participation in this study. The sub-
`ject distribution included 1 East Indian, 3 Asians, and 2
`darkly tanned and 4 lightly tanned Caucasians. Their
`ages ranged from 22 to 37 years old (mean + SD, 27.5
`+ 4.9 years). Measured blood hematocrits were in the
`range of 40 to 50.5% (mean + SD, 45.7 + 3.2%).
`All instruments were allowed to warm up forat least
`30 minutes before the study. The transmittance sensor
`of the pulse oximeter was attached to the index finger.
`The reflectance sensors were attached to the ventral side
`of the forearm and the dorsal side of the calf by using a
`double-sided transparent adhesive ring. In cases where
`an abundance of hair prevented intimate contact be-
`tween the sensors and the skin,
`the contact was im-
`proved by loosely wrapping the sensor and the limb
`with an elastic strap. The temperature of each reflec-
`tance sensor was set to 40°C and remained unchanged
`throughout the entire study.
`A standard lead-I electrocardiogram and end-tidal
`carbon dioxide levels were continuously monitored by
`a Hewlett-Packard 78345A patient monitor (Hewlett-
`Packard, Andover, MA). Each subject was placed in a
`supine position. A face mask was tightly fitted over the
`subject’s nose and mouth, and the subject was instructed
`to breathe spontaneously while we administered differ-
`ent gas mixtures of nitrogen and oxygen. Theinspired
`gas mixture was supplied by a modified Heidbrink anes-
`thesia machine (Ohio Medical Products, Madison, WI).
`The breathing circuit of the anesthesia machine was
`equipped with a carbon dioxide scrubber (soda lime).
`The inspired oxygen concentration was adjusted be-
`tween 12 and 100% and was monitored continuously
`
`3
`
`
`
`1.6
`
`bot N
`
`oO
`
`2 p>
`
`
`
`R/IRREFLECTANCERATIO
`
`10 Journal of Clinical Monitoring Vol 7 No 1 January 1991
`
`throughout the study with an IL 408 (Instrumentation
`Laboratories, Lexington, MA) oxygen monitor, which
`was inserted in the inspiratory limb of the breathing
`circuit.
`Steady-state hypoxia was gradually induced by low-
`ering the inspired fraction of oxygen in the breathing
`gas mixture. Initially, the inspired oxygen concentra-
`tion was changed in step decrements, each step pro-
`ducing approximately a 5% decrease in SpOz.(t) as
`determined from the display of
`the ACCUSAT
`transmittance pulse oximeter. The inspired oxygen was
`maintained at each level for at least 3 minutes until the
`pulse oximeter readings reached a steady level (e.,
`SaQ> fluctuations of less than +3%). When the inspired
`oxygen level reached 12%,
`the process was reversed.
`Thereafter, the inspired oxygen level was increased in a
`similar stepwise manner to 100%. Data were recorded
`during both desaturation and reoxygenation.
`All subjects tolerated the procedure well without ad-
`verse reactions. None of the subjects showedelectrocar-
`diographic abnormalities before or after the study. Each
`subject was studied for approximately 1 hour.
`
`Data Analysis
`
`To avoid operator biases, the data from each pulse ox-
`imeter were acquired automatically by the computer
`and later subjected to the samestatistical tests.
`For each step change in inspired oxygen, readings
`from the three pulse oximeters were averaged consecu-
`tively over a period of 20 seconds. Averaged readings
`from the 10 subjects were pooled and a least-squares
`linear regression analysis was performed. Student’s ¢ test
`determined the significance of each correlation; p <
`0.001 was considered significant.
`Although the correlation coefficient of the linear re-
`gression (r) provides a measure of association between
`the SpO2(r) and SpO.(t) measurements,it does notpro-
`vide an accurate measure of agreement between the two
`variables. Therefore, the measurement accuracy was es-
`timated on the basis of the mean and standard deviations
`of the difference between the readings from the trans-
`mittance and reflectance pulse oximeters. The mean of
`the difference between the pulse oximeter measure-
`ments, which is often referred to as the bias, was used to
`assess whether there was a systematic over- or underes-
`timation of one method compared with the other. The
`standard deviation of the bias, which is often referred to
`as the precision, represents the variability or random
`error. Finally, we computed the mean errors and stan-
`dard deviations of each measurement. The mean error
`is defined as the absolute bias divided by the corre-
`sponding SpO.(t) values.
`
`¥ = 0.87x + 0.04
`
`+ --
`
`FORARM
`¥ = 1.02x ~- 0.05
`
`_4- CALF
`
`0
`0. 4
`Q. 8
`le
`1.6
`R/IR TRANSMITTANCE RATIO
`
`Fig 2. Comparison of red/infrared (R/IR) ratios measured by the
`modified reflectance pulse oximeter (y axis) and the standard trans-
`mittance pulse oximeter (x axis) during progressive steady-state
`hypoxia in 10 healthy subjects. The solid line represents the best-
`fitted linear regression line for the forearm measurements. The bro-
`ken line represents the best-fitted linear regression line for the calf
`measurements.
`
`
`
`RESULTS
`
`Normalized R/IR ratios and SpOa(r) values measured
`by the reflectance pulse oximeters from the forearm and
`calf of the 10 subjects were compared with the nor-
`malized R/IR ratios and SpO.(t) values measured simul-
`taneously by the transmittance pulse oximeter from the
`finger. A total of91 and 93pairs of data points measured
`simultaneously from the forearm andcalf, respectively,
`were used in the regression analysis, which provided the
`estimated slopes and intercepts of the linear regression
`lines. Each pair of data points represents a different hy-
`poxic level.
`Regression analysis of the normalized R/IR ratios
`measured from the reflectance pulse oximeters from the
`forearm and calf (y axis) versus the normalized R/IR
`ratios measured simultaneously by the transmittance
`pulse oximeter from the finger tip (x axis) is shown in
`Figure 2. The equationsfor the best-fitted linear regres-
`sion lines were y = ~ 0.05 + 1.02x (ry = 0.94, SEE =
`0.08, p < 0.001) for the forearm and y = 0.04 + 0.87x
`( = 0,88, SEE = 0.11, p < 0.001) forthe calf.
`A comparison of SpO(r) readings from the refiec-
`tance pulse oximeter(y axis) and SpO2(t) readings mea-
`
`4
`
`
`
`
`
`Y = 0.93% + 7.78
`
`Mendelson and McGinn: Skin Reflectance Pulse Oximetry
`
`11
`
`
`
` DIFFERENCES
`
`100
`
`90
`
`
`
`REFLECTANCESpO9(%)
`
`g
`
`FOREARM
`y = 1.09% - 7.06
`
`-4- CALF
`
`70
`
`90
`80
`TRANSMITTANCE SpO02
`
`(%)
`
`100
`
`70
`
`80
`90
`TRANSMITTANCE SpOo (%)
`
`100
`
`Fig 3. Comparison ofpercent arterial hemoglobin oxygen satura-
`tion (SpOx2) measurements obtained from the modified reflectance
`pulse oximeter (y axis) and SpO2 values measured by a standard
`transmittance pulse oximeter (x axis) during progressive steady-
`state hypoxia in 10 healthy subjects. The solid line represents the
`best-fitted linear regression line for the forearm measurements. The
`broken line represents the best-fitted linear regression line for the
`calf measurements.
`
`oO
`
`DIFFERENCES
`
`70
`
`90
`80
`TRANSMITTANCE SpO» (%)
`
`100
`
`Fig 4. Mean differences between arterial hemoglobin oxygen sat-
`uration (SpO2) measured from the forearm by the modified reflec-
`tance pulse oximeter and the standard transmittance pulse oximeter
`measurements from the finger tip.
`
`Fig 5. Mean differences between arterial hemoglobin oxygen sat-
`uration (SpO2) measured from the calf by the modified reflectance
`pulse oximeter and the standard transmittance pulse oximeter mea-
`surements from the fingertip.
`
`Statistical Analysis of Arterial Oxygen Saturation (SaOQz) Levels
`Measured from the Forearm and Calf by the Modified Reflectance
`Pulse Oximeters
`
`Location/
`% SaOr
`
`No. of
`Data Points
`
`Mean Value (SD)
`
`Difference
`
`% Error
`
`Forearm
`90-100
`80-89
`70-79
`Calf
`90-100
`80-89
`70-79
`
`42
`37
`12
`
`43
`33
`17
`
`1.25 (2.55)
`0.52 (2.85)
`— 0.82 (1.96)
`
`2.47 (1.66)
`2.35 (2.45)
`2.42 (1.20)
`
`1.57 (4.00)
`2.22 (4.00)
`1.95 (2.42)
`
`3.36 (3.06)
`3.45 (4.12)
`2.97 (2.75)
`
`sured simultaneously from the transmittance pulse ox-
`imeter (x axis) is shown in Figure 3. The equations for
`the best-fitted linear regression lines were y = — 7.06
`+ 1.09x (ry = 0.95, SEE = 2.62, p < 0.001) for the
`forearm and y = 7.78 + 0.93x (r = 0.88, SEE = 3.73,
`p < 0.001) for the calf.
`Figures 4 and 5 show thepercentdifferences between
`SpO2(r) and SpO2(t), that is, SpOa(r) — SpOo(t), ob-
`tained from the forearm andcalf data plotted in Figure
`3, respectively. The corresponding means and standard
`deviations of the differences and errors for the forearm
`and calf measurements are summarized in the Table.
`
`5
`
`
`
`12
`
`Journal of Clinical Monitoring Vol 7 No 1 January 1991
`
`Data were summarized for three different ranges of
`SpO2(t) values between 70 and 100%.
`
`DISCUSSION
`
`Commercially available transmittance sensors can be
`used on only a limited numberofperipheral locations of
`the body. Brinkman and Zijlstra [4] and Cohen and
`Wadsworth [5] showed that instead oftissue transil-
`lumination, noninvasive monitoring of SaO. can be
`performed based on skin reflectance
`spectropho-
`tometry. More recently, we described an improved
`optical reflectance sensor that was used for measuring
`SaO> from the forehead with a modified commercial
`transmittance pulse oximeter[1].
`Measuring large reflectance photoplethysmograms
`from sparsely vascularized areas of the skin is challeng-
`ing. Differences in capillary densities between various
`locations on the body are knownto affect the magnitude
`and quality of the reflected photoplethysmograms. For
`example, estimated average capillary density of the hu-
`manforehead is approximately 127 to 149 loops/mm7,
`whereas the capillary densities of the forearm and calf
`are approximately 35 to 51 and 41 loops/mm*, respec-
`tively [6,7]. Furthermore, the frontal bone ofthe fore-
`head provides a highly reflective surface that signifi-
`cantly increases the amount of light detected by the
`reflectance sensor. Therefore,
`reflected photoplethys-
`mograms recorded from the forehead are normally
`larger than those recorded from the forearm and calf.
`Local skin heating could be used as a practical method
`for improving the signal-to-noise ratio of the reflected
`photoplethysmograms from the forearm or calf areas
`and thus reduce the measurementerrors in reflectance
`pulse oximetry.
`The approach presented in this article demonstrated
`that SaO»z can be estimated by using a heated skin
`reflectance sensor from the forearm andcalf overa rela-
`tively wide range of SaOz values. This technique may
`provide a clinically acceptable alternative to currently
`available transmittance pulse oximeters. In a previous
`study [2], we found that the ability to measure accurate
`SaOzvalues with a reflectance skin oximeteris indepen-
`dent of the exact skin temperature. We noticed, how-
`ever,
`that a minimum skin temperature of approxi-
`mately 40°C is generally sufficient to detect adequately
`stable photoplethysmograms. Furthermore, our experi-
`ence in healthy adults also has shown that at this skin
`temperature, the heated sensor can remain in the same
`location without any apparent skin damage.
`Note that despite the proven advantage oflocal skin
`heating to increase skin blood flow, reflected photo-
`plethysmogramsrecorded from the forearm and thecalf
`are considerably weaker than those recorded from the
`
`the mean errors for the SpOo(r)
`forehead. Therefore,
`measurements from the forearm andcalf are higher than
`the corresponding errors for similar SpO2(r) measure-
`ments made with an unheated reflectance sensor from
`the forehead. For comparison, relative to SaOQ2 mea-
`sured with a noninvasive transmittance pulse oximeter,
`the SEE for SpOz(r) measurements obtained from the
`forehead using a similar unheated optical reflectance
`sensor were 1.82% [1]. The SEE obtainedin this study
`using the heated reflectance sensor were 2.62%for the
`forearm and 3.73% for the calf measurements. Despite
`those differences, it is apparent that the degree of corre-
`lation obtained in this preliminary study is encouraging
`and in selected clinical applications may be acceptable.
`We conclude that reflectance pulse oximetry from the
`forearm and calf may provide a possible alternative to
`conventional transmittance pulse oximetry and reflec-
`tance pulse oximetry from the forehead. Further stud-
`ies, however, are needed in order to compare our
`reflectance pulse oximeter against SaO2 measurements
`obtained directly from arterial blood samples. Addi-
`tional work to investigate the source of variability in
`reflectance pulse oximetry is in progress.
`
`Financial support for this study was provided in part by the
`Datascope Corporation and NIH Grant R15 GM36111-01A1.
`The authors would like to acknowledgethe clinical assistance
`of Albert Shahnarian, PhD, Gary W. Welch MD, PhD, and
`Robert M. Giasi, MD, Department of Anesthesiology, Uni-
`versity of Massachusetts Medical Center, Worcester, MA.
`We also thank Paul A. Nigroni, Datascope Corporation,
`Paramus, NJ, and Kevin Hines, Semiconductor Division,
`Analog Devices, Wilmington, MA, for technical assistance.
`The skillful art work by Yi Wang is also greatly appreciated.
`
`REFERENCES
`
`1. Mendelson Y, Kent JC, Yocum BL, Birle MJ. Design and
`evaluation of a new reflectance pulse oximeter sensor.
`Biomed Instrum Technol 1988;22(4):167-173
`2. Mendelson Y, Ochs BD. Noninvasive pulse oximetry
`utilizing skin reflectance photoplethysmography.
`IEEE
`Trans Biomed Eng 1988;35(10):798-805
`3. Mendelson Y, Kent JC, Shahnarian A,et al. Evaluation of
`the Datascope ACCUSAT pulse oximeter
`in healthy
`adults. J Clin Monit 1988;4:59-63
`4. Brinkman R, Zijlstra WG. Determination and continuous
`registration of the percentage oxygen saturation in clinical
`conditions. Neth J Surg 1949;1:177-183
`5. Cohen A, Wadsworth N. A light emitting diode skin
`reflectance oximeter. Med Biol Eng Comput 1972;10:385—
`391
`6. Moretti G. Handbuch der Haut- und Geschlechtskran-
`kheiten. Berlin: Springer-Verlag, 1968:491-623
`7. Rothman §. Physiology and biochemistry of the skin.
`Chicago: University of Chicago Press, 1954:685
`
`6
`
`