`
`S
`
`:
`
`~
`
`
`
`EB
`
`
`:
`:
`
`; AND
`
`
`
`
`—_——
`Volume 14 Number 6 August 1998
`
`en a
`
`379-380 BriefReviews
`J. S. Gravenstein
`Original Articles
`381-384 INFLUENCE OF THE REFERENCE GAS OF PARAMAGNETIC
`OXYGEN ANALYZERS ON NITROGEN CONCENTRATIONS
`DURING CLOSED-CIRCUIT ANESTHESIA
`Janu FP. A. Hendrickx,MD, Andre A, J. van Zundert,MD,
`PhD, and Andre M. de Wolf,MD
`385-391 AVAILABILITY OF RECORDS INANOUTPATIENT =
`PREANESTHETIC EVALUATION CLINIC
`Gordon L, Gibby, MD, and Wilhem K. Schwab, PhD
`393-402 IN VIVOEVALUATION OF A CLOSED LOOP MONITORING
`STRATEGY FOR INDUCED PARALYSIS
`Deepak Ramakrishna, MS, Khosrow Belibehani, PhD, PE,
`Kevin Klein, MD, Jeffrey Mokhtar, MS,
`
`WolfW.vonMaltzalin,PhD,PE,RobertC.Eberhart,PhD,
`
`and Michael Dollar, MS
`403-412 REFLECTANCE PULSE OXIMETRY—PRINGIPLESAND =~
`OBSTETRIC APPLICATION IN THE ZURICH SYSTEM
`Volker Konig, Renate Huch, and Albert Huch
`413-420 ACCURACY OF VOLUME MEASUREMENTS IN MECHANIGALLY
`VENTILATED NEWBORNS:
`A COMPARATIVE STUDY OF COMMERCIAL DEVICES
`Kai Roske, Bertram Foitzik, Roland R. Wauer, and
`Gerd Schimatisch
`421-424 AN EVOLUTIONARY SOLUTION TO ANESTHESIA AUTOMATED
`RECORD KEEPING
`Alvin A. Bicker, PhD, John S, Gage, MD,and
`PaulJ. Poppers,MID
`425-431 EVALUATION OF APITOT TYPE SPIROMETER IN HELIUM/
`OXYGEN MIXTURES
`Soren Sondergaard, MD, Sigurbergur Karason, MD,
`Stefan Lundin, MD, and Ola Stenqvist, MD
`433-439 DETECTION OF LUNG INJURY WITH CONVENTIONAL AND
`NEURAL NETWORK-BASED ANALYSIS OF CONTINUOUS DATA
`Jukka Rasénen, MD, and Mauricio ¢A. Leon,MD
`Algorithm
`441-446 AREAL-TIME ALGORITHM TO IMPROVE THE RESPONSE TIME
`OF A CLINICAL MULTIGAS ANALYSER
`Lawdy Wong, MSc, Ruth Hamilton, MB, ChB,
`Eileen Palayiwa, PhD, and Clive Halin, D. Phil
`Book Review
`L. G6. HENSON AND A. C. LEE (EDS): SIMULATORS IN
`ANESTHESIOLOGY EDUCATION
`Richard Morris
`449-450 INFORMATION FORCONTRIBUTORS =
`
`~J
`
`447
`
`KLUWER ACADEMIC PUBLISHERS
`
`OFFICIAL JOURNAL OF THE EUROPEANSOCIETY
`FOR COMPUTING AND TECHNOLOGY
`IN ANAESTHESIA AND INTENSIVE CARE, AND OF
`THE SOCIETY FOR TECHNOLOGY IN ANESTHESIA
`
`(cid:44)
`
`JUL 3.0 1999
`
`
`HOY Hiikana iin,
`(cid:36)(cid:83)(cid:83)(cid:79)(cid:72)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)
`AppleInc.
`APL1014
`(cid:36)(cid:51)(cid:47)(cid:20)(cid:19)(cid:20)(cid:23)(cid:3)
`U.S. Patent No.8,989,830
`(cid:56)(cid:17)(cid:54)(cid:17)(cid:3)(cid:51)(cid:68)(cid:87)(cid:72)(cid:81)(cid:87)(cid:3)(cid:49)(cid:82)(cid:17)(cid:3)(cid:27)(cid:15)(cid:28)(cid:27)(cid:28)(cid:15)(cid:27)(cid:22)(cid:19)
`
`CODEN JCMOEH
`ISSN 1387-1307
`
`APPLE 1014
`
`1
`
`APPLE 1014
`
`

`

`
`
`JOURNAL OF CLINICAL MONITORING
`AND COMPUTING
`ee
`OFFICIAL JOURNAL OF THE EUROPEAN SOCIETY FOR COMPUTING AND TECHNOLOGY IN ANAESTHESIA AND INTENSIVE GARE (ESCTAIC)
`AND OF THE SOCIETY FOR TECHNOLOGY IN ANESTHESIA (STA)
`
`
`Volume 14 Number 6 August 1998
`
`THE PARENT JOURNALS
`International Journal of Clinical Monitoring and Computing,
`founded by OmarPrakash in 1984; Reed Gardner, senior
`editor; and the
`founded in 1985 by N. Ty
`Journal of Clinical Monitoring,
`Smith, Allen K. Ream, and J. S. Gravenstein.
`
`EDITORS
`Jeffrey M. Feldman, MD
`Allegheny University Hospitals, MCP
`Dept. of Anesthesiology
`3300 Henry Avenue
`Philadelphia, PA 19129, U.S.A.
`E-mail: Feldman]@auhs.cdu
`J. S. Gravenstein, MD, Coordinating Editor
`UF/College of Medicine
`Dept. of Anesthesiology
`P.O. Box 100254
`Gainesville, FL 32610, U.S.A.
`E-mail: jgravens(@anest2.anest.ufl.edu
`Ilkka Kalh, MD, PhD
`City Maternity Hospital
`Dept. of Anaesthesiology
`P.O. Box 61
`FIN-00971 Helsinki, Finland
`E-mail:
`ilkka.kalli@iki.Fi
`Publisher
`Kluwer Academic Publishers
`P.O, Box 990
`3300 AZ Dordrecht, The Netherlands
`E-mail: Loes.Wils@ wkap.nl
`
`
`EDITORIAL BOARD
`Paul G. Barash, MD
`~ Gerard Bashein, MD, PhD
`Charlotte Bell, MD
`Jan E. W. Beneken, PhD
`J. A. Blom, PhD
`J.G. Bovill, MD, PhD
`Peter Carstensen
`William P. Coleman, PhD
`Jeffrey B. Cooper, PhD
`John Doyle, MD
`Ronald Dueck, MD
`Peter C. Duke, MD
`
`ISSN 1387-1307
`All rights reserved
`
`© 1998 Kluwer Academie Publishers
`No part of the material protected by this copyright notice may be
`reproduced or utilised in any form or by any means, electronic or
`mechanical, including photocopying, recording or by any informa-
`tion storage and retrieval system, without written permission from
`the copyright owner.
`
`Printed in the Netherlands
`
`Printed on acid-free paper
`
`nif
`
`W CTA. pine
`Es T
`JU
`L 3 0 1999
`giro
`$
`Cea
`“MAUIOUN, Wi
`vane AY
`
`A. Dean Forbes
`
`Harvey L- Edmonds, MD
`John H, Eichhorn, MD
`TammyY. Euliano, MD
`M.Fisher, MD
`Wolfgang Friesdorf, MD
`Yasuhiro Fukui, PhD
`Leslie Geddes, PhD
`Nikolaus Gravenstein, MD
`A. Gerson Greenburg, MD, PhD
`G. Hedenstierna, MD
`Wolfgang Heinrichs, MD
`..C. G. Gallandat Huet, MD
`Kazuyuki Ikeda, MD
`Ville Jantti, MD
`Gavin N.C. Kenny, MD
`Samsun Lampotang, PhD
`J. W. R. McIntyre, MD
`James H. Philip, MD
`Michael L. Quinn, PhD
`Pradeep Ramaaya, MD
`Ira J. Rampil, MD
`Gunter Rau, MD
`Michael F. Roizen, MD
`J.-A. Romand, MD
`William B. Runciman, MD
`Keith Ruskin, MD
`David A. Sainsbury, MD
`H. Schillings, MD
`John W. Severinghaus, MD
`M. Michael Shabot, MD
`DeanForrest Sittig, PhD
`Bradley E. Smith, MD
`Richard Teplick, MD
`Kevin K. Tremper, MD
`Jorge Urzua, MD, PhD
`Karel H. Wesseling, PhD
`Dwayne Westenskow, PhD
`J.C. Wyatt, PhD
`
`VOLUME 14, 8 ISSUES, 1998
`Issues include:
`Brief Reviews, Editorials, Clinical Articles, Research Ar-
`ticles, Educational Articles, Book Reviews, Correspondence,
`Selections from the Anesthesia Patient Safety Foundation
`Newsletter.
`
`Kluwer Academic Publishers, Dordrecht, Boston, London
`
`Kluwer Academic Publishers Online
`Visit our website at http://www.wkap.ol
`
`The Journal of Clinical Monitoring and Computing is indexed in Index
`Medicus, Current Contents/Clinical Medicine, Excerpta Medica, Current
`Awareness in Biological Sciences, and the Biomedical Enginecring Citation
`Index, and is indexed/abstracted in Engineering Index, Science Citation
`Index, Expanded Research Alert, Medica Documentation Service.
`
`Il
`(cid:44)(cid:44)
`
`7
`
`2
`
`

`

`
`
`This material may be protected by Copyright law (Title 17 U.S. Code)
`
`403
`403
`
`
`
`REFLECTANCE PULSE OXIMETRY — PRINCIPLES AND
`OBSTETRIC APPLICATION IN THE ZURICH SYSTEM
`Volker Konig, Renate Huch, and Albert Huch
`
`Fromthe Perinatal Physiology Research Department, Department
`of Obstetrics, Zurich University Hospital, CH-8091 Zurich, Switzer-
`land.
`
`Received Nov13, 1997, and in revised form Jun 23, 1998, Accepted
`for publication Aug 4, 1998.
`
`Address correspondence to Volker Kénig, Perinatal Physiology Re-
`search Department, Department of Obstetrics, Zurich University
`Hospital, CH-8091 Zurich, Switzerland.
`E-mail: vkg@fhk.usz.ch
`
`Journal ofClinical Monitoring and Computing 14; 403-412, 1998,
`© 1998 Kluwer Academic Publishers. Printed in the Netherlands.
`
`Konig V, Huch R, Huch A. Reflectance pulse oximetry — principles
`and obstetric applicationin the zurich system.
`J Clin Monit 1998; 14: 403-412
`
`ABSTRACT. Transmission and reflectance are the two main
`modes of pulse oximetry. In obstetrics, due to the absence of a
`transilluminable fetal part for transmission oximetry, the only
`feasible option is the reflectance mode, in which sensor and
`detector are located on the same surface of the body part.
`However, none of the reflectance pulse oximeters developed
`for intrapartumuse are fully satisfactory, as indicated by the
`fact that none have entered routine use. We have designed,
`developed, constructed and tested a reflectance pulse oximeter
`with the possibility to adjust the electronic circuits and signal
`processing in orderto determine the effects of various param-
`eters on signal amplitude and wave-form and to optimize the
`sensitivity and spatial arrangementof the optical elements.
`Following an explanation of the principles of reflectance
`pulse oximetry, we report our experience with the design,
`development, construction and field-testing of an in-house
`reflectance pulse oximetry system for obstetric application.
`
`KEY WORDS. Oxygen saturation, reflectance pulse oximetry,
`intrapartumfetal monitoring.
`
`INTRODUCTION
`
`Pulse oximetry is the combination of spectrophotome-
`try and plethysmography. It permits rapid noninvasive
`measurement of arterial oxygen saturation with the
`added advantages of simple sensor application and direct
`measurement, requiring neither calibration nor pre-
`adjustment, Pulse oximeters are thus in widespread and
`fast-increasing use, ¢.g. in intensive care, anesthetics and
`neonatology [1]. All these applications employ “trans-
`mission” pulse oximetry, so called because the light
`used to determine blood oxygensaturation is “trans-
`mitted” from a light emitter on one side of the body
`part to a light receiver on the otherside; suitable sites
`are the fingers in adults or hands and feet in neonates or
`children, whichare said to be “transilluminated.”
`In obstetrics, fetal oxygen status during labor is a
`crucial parameter. However, no transilluminable fetal
`part is available. The only optionin this case is reflec-
`tance oximetry [2], using a sensor with its light emis-
`sion and detection elements on the same surface of
`the bodypart. Various types of such a reflectance pulse
`oximeter have been developed for intrapartumuse at
`various locations. However, for a wide variety of reasons,
`all are still experimental and not infull routine use [3-7].
`Basically a reflectance measurement can be achieved
`using planar sensors — which can be produced,
`for
`example, by modifying conventional transmission sen-
`sors — and a sensitive modern pulse oximeter. However,
`
`3
`
`

`

`404 Journal of Clinical Monitoring and Computing Vol 14 No6 August 1998
`
`such instruments come with a “black-box” microproces-
`sor-controlled mode of operation making constructional
`adjustments to the electronic circuits and signal process-
`ing virtually impossible. As a result, it becomes difficult
`to determine the effect of various parameters on signal
`amplitude and wave-form, optimize sensorsensitivity to
`light intensity and the arrangement of the optical ele-
`ments, and hence assess the dependenceof arterial oxygen
`saturation measurement on key physical, technical and
`above all physiological variables. This was the aim
`driving our decision to design, develop and construct
`in-house a system dedicated to obstetric applications.
`Following a brief review of the principles of pulse
`oximetry, we report our experience with the develop-
`ment of the new device,
`together with some field-
`testing results.
`
`Ep
`
`d(t)
`
`absorption coefficient of reduced hemoglobin
`(function of wavelength)
`time function of mean pulsatile change in artery
`thickness, with amplitude d = d(diastole) —
`d(systole)
`
`transmitted
`Measurement may be impaired by light
`directly from the light source to the receiver or light
`which does not pass througharterially perfused tissue.
`If this “direct light” Igi, is taken into account, Equation
`(1) changes to:
`
`I(t) = Ihissue * exp(—(SO2 - Expo +
`(1 _ $O2) -Enb) * d(t)) + lair
`
`(ia)
`
`where
`
`
`
`Tessue = Io° exp(—Erisue . s)
`
`PRINCIPLES OF PULSE OXIMETRY
`
`Sigial recording
`
`Light is absorbed on passing through matter. The de-
`gree of absorption depends onthe nature of the trans-
`illuminated material and the wavelength of the light
`employed. All optical techniques for determining arte-
`rial oxygen saturation use the marked difference in the
`absorption of red light between oxygenated and re-
`duced hemoglobin.
`The absorption of light passing through bone or
`nonpulsatile tissue is constant over time. Oxygenated
`and reduced hemoglobinin the arterial vascular bed, on
`the other hand, cause changes in absorption timed by
`the heart rate due to the pulsatile variation in artery
`thickness. The total intensity ofthe light after passing
`through tissue can be measured, for example, as the
`photocurrent I(t) of a photodiode, andis obtained from
`the Lambert-Beerabsorptionlawas:
`
`I(t) = Ip + exp(—Etisue 8) + exp(—(SO2 + Enpo +
`(1 — SOz) - Evy) - d(t))
`
`(1)
`
`where
`intensity of incident light
`Ip
`Etissue Mean absorption coefficient of tissue (function of
`wavelength)
`mean thickness of transilluminatedtissue
`5M
`SO oxygen saturation to be determined (= HbO/
`(HbO + Hb), i.e. ratio of oxygenated hemoglo-
`bin concentration to sum of oxygenated and
`reduced hemoglobin concentrations)
`Epo absorption coefficient of oxygenated hemoglobin
`(function of wavelength)
`
`Since the pulsatile component of the absorption ts at
`most a few percent,
`i.c, the exponent of the second e
`function in Equation (1)
`is very small, we can use the
`approximation:
`
`exp(x) =1+x for |x| <1
`
`to obtain the veryclose approximation:
`
`I(t) =
`.
`.
`Leissue * a _ (SOo *Enpo + (1 - $O2) *EHb * d(t)) + Tair
`
`(1b)
`
`This light intensity is measured in the photodiodes and
`can be broken downelectronically into two compo-
`nents, a time-independent signal
`
`DC = leissue + Laie
`
`with amplitude equal to the value of this signal
`
`de = DC = Trissue F lair
`
`(2)
`
`(3)
`
`and a signal which varies in time with the pulsatile
`changeinartery thickness
`
`AC = lrissue * (SO2 *Enbo + (1 ~ SOz) "ey tb) . d(t)
`
`(4)
`
`with amplitude
`
`ac = I(diastole) — I(systole) =
`Iissue * (SO2 - Ero + (1 — SO2) + Eu) «a.
`
`Theratio between the ac and de amplitudesis then
`
`4
`
`

`

`Konig et al: The Zurich Obstetric Reflectance Pulse Oximeter
`
`405
`
`r= ac/de a (Tsesne/ tissue + lair):
`(6)
`(SO- Enno + (1 — SO2) > erm) >
`this ratio r is
`In the case that “direct light” Ii, = 0,
`independent ofthe incident light intensity Ip and of the
`absorptionin the nonpulsating tissue valuelrissue?
`
`6
`
`However, despite various theoretical models [9, 10],
`the scatter coefficients of the various tissue types are
`not known with sufficient accuracy to permit exact
`calculation. Experimental calibration thus has to be
`performed bydirectly comparing the pulse oximeter
`readings witharterial blood sample values.
`
`r= ac/de =
`(SO2 + Enbo + (1 —SO2)- env) +d
`
`for lair =0.
`
`6a
`
`(s )
`
`Transmission pulse oximetry
`
`is then dependent only on the oxygen
`This ratio r
`saturation SQ> to be determined, the knownabsorption
`coefficients Epo and ey, and the meanpulsatile change
`d in the thickness of the arterial vessels in the trans-
`illuminated region.
`To eliminate this dependence on d, the measurement
`is performed at two wavelengths with maximally dif-
`fering absorptioncoefficients. On the assumption that
`the d values are the same for both wavelengths, we
`obtain a variable
`
`R
`
`ll
`
`ll
`
`Cred /Tir = (ac/de),.4/(ac/de);,
`
`(SO+ Ern0 + (1 — SO2) > Erb) sea /
`(SO- E140 + (1 — SO2) + Erb) ie
`
`(7)
`
`(7a)
`
`from which the unknown SQz is readily calculated
`without knowing theincident light intensity or tissue
`thicknesses.
`Calculation assumesthe following physical prerequi-
`sites:
`
`e No light must be measured that has not passed
`through the pulsatile vascular bed e.g. light passing
`directly fromlight source to receiver(Igir).
`e The pulsatile changes in artery thickness must be the
`same for both wavelengths,
`i.c. both wavelengths
`must transilluminate the sametissue region.
`© Valid measurement assumes that the pulsatile signal
`originates only from varying absorption byarterial
`oxygenated and reduced hemoglobin. Theresults are
`falsified by othercauses of pulsatile changesin optical
`thickness, e.g. hemoglobin derivatives, circulating
`pigments, pulsatile changes in thickness produced
`mechanically in nonarterially perfused tissue by car-
`diac action, and, above all, venous pulsation,
`e To simplify descriptionof the principle behind meas-
`urement and its limitations,
`the Lambert-Beer law
`was assumedvalid for the passage of light through
`tissue. However,as light is not only absorbedin tissue
`but also scattered, the lawis of limited applicability
`[8]. The exact absorption coefficients must be cor-
`rected by taking the scattering effect into account.
`
`The optical elements are located on opposite sides of a
`bodypart. The sensorsare applied mainlyto the fingers
`and toes. Ears and nose are used onlyrarely due to poor
`perfusion. In neonates the sensoris applied around the
`hand orfoot. This arrangement largely ensures that the
`optical paths are the same for both wavelengths. Never-
`theless, incorrect sensor attachment can give spurious
`results, e.g. if some of the transmitted light reaches the
`receiving diodes around the outside of a fingeras “direct
`light.”
`Signal magnitudes are an important determinant of
`measurement accuracy: in normal fingertips, the ratio
`of the signal due to absorption in pulsating blood (ac)
`to the signal due to absorption intotal tissue (dc), r =
`ac/de, is 0.02—0.05.
`
`Reflectance pulse oximetry
`
`In this method the light backscattered in the bodyis
`used to determine oxygen saturation. The optical ele-
`ments are thus located on the same plane on the same
`body surface. Reflection originates from nonhomo-
`geneityin the optical path, i.e. at the interfaces between
`materials with different refractive indices. This means
`that on physiological grounds,strong reflections can be
`expected on the entry of light into bone. The trans-
`illuminated tissue must also be well perfused to obtain
`as strong a signal as possible, Not all bodyparts are as
`well perfused as the fingers or hands, but an ac/dc ratio
`of (.001—0.005 can be achieved on the forehead. Perfu-
`sion is also good over the sternum. One method of
`signal enhancement is to heat the measurement site to
`induce hyperperfusion, which can safely be performed
`up to 42°C. A rubefacient, e.g. nicotinic acid (Rubri-
`ment), can also be applied to the measurement site.
`Theprincipal physical limitations are the following:
`
`© The sensor design must eliminate “direct light,”i.c.
`light passing directly from the light sources to the
`photodiodes orthatis onlyscattered in the outerpart
`of the skin.
`e The measured AC signals are some 10 times weaker
`
`5
`
`

`

`406 Journal of Clinical Monitoring and Computing Vol14 No 6 August 1998
`
`than in the transmission method. The conditions
`governing the heating of the light-emitting diodes
`(LED) limit
`the potential
`for producing stronger
`signals by increasing the incident light intensity: not
`only can high uncontrolled temperatures damage
`tissue at the measurement site, but the wavelength
`of the emitted light changes as the LEDs become
`warmer. For this reason the photodiode area must be
`as large as possible.
`As in the transmission mode, the principle of meas-
`urement is the determination of absorption, except
`that this now refers to incoming reflected light, The
`light path is less well defined than in transmission
`mode, and thus may differ between the two wave-
`lengths. The effective absorption coefficients of the
`calibration inserted in the Lambert-Beer law must be
`checked and if necessary corrected by comparison
`with photometrically measured arterial blood values.
`
`neonatal head. Experimentationled to the choice ofa
`vacuumsystem using sensors cast fromsilicone rub-
`ber, with a suction grooveforfixation, a guard ring
`against direct light, and a connector for a suction
`pump. The photoelectric componentsare identical in
`both types of sensor.
`e Some metal sensors were fitted with a resistance-wire
`heating coil of maximal output 200 mW to induce
`local hyperemia. The temperature was monitored by
`a negative temperature coefhicient (NTC)resistor in-
`corporatedin the sensor unit.
`
`Numerous types of sensor meeting the above require-
`ments were built. A sensor used for intrapartum meas-
`urements — cast from silicone rubber with suction
`channel and pump connection — is shown in cross-
`section in Figure 1. It is attached to the fetal head with a °
`vacuumof approximately 100 mbar.
`
`THE ZURICH REFLECTANCE PULSE OXIMETER
`
`Electronics
`
`Sensors
`
`In constructing planar reflection sensors, i.e. sensors in
`whichthe photoelectric emitting and receiving elements
`lie next to eachotherin virtually the same plane, special
`attention was paid to the following points (Figure 1).
`
`e For maximal independence from local tissue differ-
`ences, a radially-symmetric pattern was selected for
`the photoelements. The light source — a chip with
`two LEDs for the wavelengths red = 660 nm and
`infrared = 920 nm — was placed in the center of the
`sensor and surrounded bya radial photodiode array
`for detecting the reflected light. To obtain a good
`signal at minimal
`light
`intensity,
`the area of the
`photodiodeshad to be as large as possible. After some
`preliminary experiments,
`six BX33 photodiodes
`(Siemens) were used with a mean radius of 17 mm.
`Their connections are led outwards individually, so
`that the sensor as a whole remains operational if a
`wire breaks or an individual diode is lost. This arrange-
`ment gives an external sensor diameter of 22 mm.
`e A guard ring around the LEDs acts as a barrier to
`“direct light.” The sensor must also fit snugly to the
`skin to minimize the risk of ambient light reaching
`the photodiodes.
`fixation to the
`Unlike with transmission sensors,
`measurementsite can pose problems. Aluminum sen-
`sor units are readily fixed with double-sided adhesive
`ECG rings. However,
`these rigid sensor heads are
`unsuited to the small radii of curvature of the fetal/
`
`Initially, we decided to separate signal processing in the
`analog part before the analog/digital converter (ADC),
`including signal amplification, filtering, and separation
`into DCand ACsignals, all handled electronically ; the
`post-ADC digital part was handled by software which
`input the data, averaged and evaluated amplitudes, cal-
`culated saturation andheart rate, and produced a some-
`what complicated screen display. Now, using modern
`techniques, we have a systemin preparation in which a
`fast separate microprocessor unit handles most signal
`processing and digital filtering tasks.
`Figure 2 shows the block diagram of the current
`apparatus with the following individual units:
`
`e Time control of measurement
`Froma rectified mains power supplysignal (100 Hz),
`a phase-locked loop (PLL) — connected as a frequency
`multiplicr — generates a square-wave signal of 64
`kHz. Coupling to the line frequency eliminates data
`acquisition faults due to interferences with the line
`frequency. The 64 kHz from the PLL clock drives a
`* 7-bit counter that addresses an EPROM giving a data
`cycle of 1 kHz, 64 pulses long. The output pulses of
`the EPROM control
`the entire sequence of light
`emission, signal acquisition andsignal processing.
`e LED drive
`The oppositely poled red and infrared LEDs are
`located in the output circuit of a current-stabilized
`push-pull output stage. They are triggered bydigital
`signals from the contro] unit via analog switches at
`the 1 kHz sampling rate in the equally spaced se-
`
`6
`
`

`

`Kénig et al: The Zurich Obstetric Reflectance Pulse Oximeter
`
`407
`
`
`
`
`
`(4) Vacuum-tube
`Photodiodes
`(1)
`(s)
`Signal-cable
`Red and Infrared LED's
`(2)
`
` (G3) Light barrier
`Fig. 1. Cross-section througha silicone-rubberreflectionsensorforintrapartummeasurement o ‘oxygensaturation.
`
`7
`
`

`

`408 Journal ofClinical Monitoring and Computing Vol14 No6 August 1998
`
`
`
`
` Vacuum
`
`
`Pump
`and
`Control
`
`Pre-
`Amplifier
`
`
`
`S&H
`
` Filter
`| Dark-Curr.
`
`Subtract.
`
`
`
`
`
`Ext. Device
`
`
`Oscillator/Control Unit
`
`(HP, Nellcor)
`
`
`
`
`Fig. 2. Schematic representation of measurement electronics.
`
`quence: infrared-dark-red-dark (overall length: 1 ms).
`Light intensity is determined by the amplitude of the
`signal driving this output stage. Input signal intensity
`canbeselected in two ways:
`— Manually: The red and infrared LED intensities
`can be manually adjusted independently using two
`potentiometers (Helipot). This permits the use of
`any desired light intensity within the limits stipu-
`lated for test and research purposes.
`— Automatically: DC voltage as the input signal
`controls the LEDs so that the DC voltages at the
`computerinput for both wavelengths are 2.0 + 0.5
`V. Outside this range the control circuit changes
`the LED currents to reset the DC voltages to 2.0 V.
`This setting is used for normal clinical applications.
`To prevent skin damage from overheating even in
`the event of electronic component failure, maximal
`LEDintensityis limited by an electronic circuit.
`Input amplifier and sample-and-hold stages
`Using operational amplifiers the photocurrent sup-
`plied by the photodiodes is converted to.a voltage
`and then amplified. Six switch positions permit am-
`plifications of 50 mV/nA to 5 V/nA. Afterwards,
`three sample-and-hold (S&H)
`stages — switched
`bythe correspondingsignals from the control stage —
`resolve the signals into the three components infra-
`red, red and dark. The dark currents are then sub-
`tracted fromthe red andinfrared signals in a subtrac-
`tion stage which also eliminates small ambient light
`components that may have reached the photodiodes.
`
`e Filters
`Using low-passfilters the discrete-time signals at the
`S&H output are reconverted to continuous-time sig-
`nals and trimmed of high-frequency components
`using cighth-order Bessel
`filters with a cut-off
`frequency of 7 Hz. The DC components are then
`separated using a low-pass filter with a cut-off fre-
`quency of 0.1 Hz. The AC componentsare passed to a
`further Bessel high-pass 0.7 Hzfilter for separation of
`slow motionartifacts, and then to a 40-fold amplifi-
`cation stage. The four signals DC;,, DC,.g, ACj, and
`ACyeq are thus available at
`the output. As oxygen
`saturation is calculated from the ratios (ac/dc);, and
`(ac/de),eq, it was essential to ensure by careful com-
`ponent selection that the amplifications for the am-
`plification and separation stages were as near as possi-
`ble identical for both wavelengths.
`Patient insulation
`the LED controller output
`For patient protection,
`stage and photodiode input stage were electrically
`‘isolated from the mains using Burr & Brownisola-
`tion amplifiers. These units were powered by an
`insulated powerpack.
`Heating
`Sensorsincorporating aresistance coil were fitted with
`a precision controller stabilizing temperature within
`the range 38.0—41.0 °C to an accuracy of 0.1 °C.
`Vacuum pump
`The sensoris fixed to the skin using a small pump
`producing a maximal -300 mbar vacuum. The pump
`
`8
`
`

`

`Konig etal: The Zurich Obstetric Reflectance Pulse Oximeter
`
`409
`
`
`
`60 min
`O sec File: LAOBLY22.PK2
`Start: 23. 1.97
`B8:19°521 Tot
`RPOXLL
`RPOX
`$o2 %
`LxxxeMMM Axxx geb. 20.03.63 US Datum 23.
`
`
`
`
`
`- aoormocooococo
`
`
`
`RPOX HR bpm
`
`
`
`File:
`
`
`
`
`
`Fig. 3. Screen display of a 10-minute measurement. Fromtop to bottom: oxygen saturation, heart rate, heart rate from HP CTGmonitor,
`uterine contraction from HP CTG monitor. Underneath the DC and AC signals for infrared andred light for the last 5 seconds (note the
`different ordinate scalesfor DC and AC).
`
`is maintained electronically via a manometer at a
`preset negative pressure. In normal medical use, ad-
`equate fixation is achieved with a vacuum of about
`—100 mbar.
`
`Softivare
`
`Apart from varioustest programs, a programfor data
`acquisition, display, calculation and storage and a pro-
`gram for subsequent data postprocessing were written
`in PASCAL.
`
`e Signal acquisition, calculation and display
`For data acquisition a 486 DOS computer was
`equipped with a 12-bit analog/digital conversion
`
`(ADC) card (Metrabyte). A counter on this card
`triggers analog/digital conversion at 400 Hz, which
`starts an interrupt programin the computerfor read-
`ing the 4 measured values DCj,, DCreq and AC;,,
`ACyea into a cyclic buffer of 5-second length. From
`the signal (AC;,+AC,.4) the maxima and minimaare
`determined for each cardiac cycle, and hence the
`instantaneous heart rate. Oxygen saturations are cal-
`culated from the amplitudes ac and de of the AC and
`DCsignals.
`Saturation is calculated from the Lambert-Beer
`law using the absorption coefficients [11]
`for
`the
`nominal wavelengths of our LEDs. However,
`as
`the actual wavelengths may deviate from these nom-
`inal values and the applicability of the Lambert-Beer
`law is limited by scattering in tissue, experimental
`
`9
`
`

`

`410 Journal of Clinical Monitoring and Computing Vol 14 No6 August 1998
`
`calibration of the measured saturation valuesis essen-
`tial.
`Saturation and heart rate can be averaged over 1-9
`cardiac cycles and are displayed every second. Analog
`heart rate and uterine contraction signals from any
`CTG monitor with analog output
`(e.g. Hewlett
`Packard model 8040) can be input to the ADC every
`second and likewise displayed.
`The measurement display (Figure 3) shows, along the
`bottom, the 4 measured values DC;,,DCyq and AC},
`ACyea, over the last 5 seconds of measurement. This
`serves to monitor signal quality during acquisition.
`Poor-signal periods can be marked and excluded from
`data processing. Along the top, measured arterial
`oxygensaturation andheartrate values per second are
`displayed cyclically over a 10-minute interval, with
`the CTG heart rate and contraction input under-
`neath. Any time point can be marked for subsequent
`identification and all values and comment stored ina
`file at any time.
`e Data analysis
`The values from a stored file can be redisplayed in
`measurement mode using an evaluation program.
`Timeintervals can be marked with the arrow keys or
`mouse. Means and standard deviations — including
`the CTG data — are then calculated and displayed,
`
`
`MEASUREMENTS AND DISCUSSION
`
`Clinical application of any newinstrument or measure-
`ment system presumes:
`
`— mechanicalreliability, accuracy andcalibration,
`— feasibility in the clinical situation, including accept-
`ance by both medical personnel and patients,
`the ability not only to determine physiological pa-
`rameters not previously measured in both physiolog-
`ical and pathological situations, but also to evaluate
`the diagnostic significance of such parameters,in this
`case oxygensaturation.
`Forthe first two more technical points is to say, that
`our instrument required calibration before clinical
`use, together withfield tests of the suction device and
`long-term oximeter performance during birth.
`For
`these points controls and clinical
`trials were
`performed in our own unit and with colleagues in
`Copenhagen (DK), Graz (A), Oulu (SF) and Berlin
`(D). The majorinvestigations comprised:
`
`Calibration
`
`To calibrate a pulse oximeter, an oxygen saturation
`value must be assigned to the measured variables
`
`R = trea/Tir = (ac/de),.4/(ac/de),,
`
`cf (7)
`
`on the basis of an experimental or theoretical relation-
`ship. Initially we used the absorption coefficients of
`Zijlstra et al [1i]. The general problems ofcalibrating a
`pulse oximeter have been discussed elsewhere [12, 13].
`For fine calibration we performed the following inves-
`tigations:
`
`e Tests in the arterial oxygen saturation range 88—
`100% were conducted in 14 healthy adult voluntcers
`breathing normal air and then air with approxi- ~
`mately 80% normal oxygen content for 10 minutes
`in each case. The reflection sensor was fixed to the
`forehead or sternumwith an adhesive ring. Owing to
`the invasive nature of arterial catheterization, refer-
`ence values were provided by a MINOLTA PUL-
`SOX 8 transmission pulse oximeter attached to the
`index finger. Data analysis [13] showed a 4.5% differ-
`ence in oxygen saturation between the MINOLTA
`and the preliminary results of ourreflectance system
`based onthe absorptioncoefficents of Zijlstra et al.
`e At lowersaturation levels, measurements were per-
`formed in cyanotic children before surgery. The chil-
`dren had arterial lines, permitting direct comparison
`with arterial blood readings[14].
`@ Lowsaturations in vive can also be measured in the
`fetal lamb [15]. We used this method to compare our
`pulse oximeter readings directly with arterial values
`in the oxygensaturation range 10-80% [16].
`
`Preliminary evaluation of these data shows that our
`previous calculations of oxygen saturation have to be
`corrected in the 10-100% saturation range by a factorof
`1.045.
`
`Fixation
`
`Thefirst experiments in sensorfixation to the head and
`other parts of the human body were performed in
`adults [17] and neonates[18].
`Attachmentis simple in practice, even during birth.
`After rupture of the membranes, the sensor canbe fixed
`to the fetal head once the cervix has dilated to at least 2
`em. Initial fixation takes 30-60 seconds and allowsfull
`freedom of movement[19].
`Approximately 100 mbaris the most suitable pressure
`
`10
`
`10
`
`

`

`at which to maintain the sensor as it ensures good
`sensor-skin contact and reliable continuous fetal oxygen
`saturation values. Measurements at twodifferentsites of
`the head of the fetus have shownno significant differ-
`ences in the oxygen saturation values,
`indicating no
`effect of suction on the local blood supply. The same we
`have observedin the case of caput succedaneum[20]. °
`
`Long-term performance
`
`Vacuumfixation did not impair oxygen measurement,
`even overfixation periods of 4.3 hours [20], Suction
`marks had disappeared normally within 10 to 20 mi-
`nutes after removal of the sensor for neonates [18] and
`for fetuses.
`
`Heating
`
`in the
`AC signals at 41.0°C are twice as strong as
`unheated state [21]. However, the silicone rubber sen-
`sors required for vacuumfixation in fetal monitoring
`cannot be fitted with a