`
`APL_MAS_ITC_00015798
`
`ill 164
`
`1
`
`Design of pulse oximeters
`
`10.2.1 Simulators using blood
`
`Several simulators have been proposed that need whole blood to teSt the
`functionality of the pulse oximeter These simulators are all based on the concern
`of being able to simulate the absorbance of human tissue (normally tile finger)
`between the LEDs and the photodiode of the pulse oximeter under test. Since few
`substances have been found that simulate the optical properties of blood, these
`types of systems typically provide the most accurate simulation.
`10. 2. 1 . 1 Reynolds system. The system described in section 10 . 1
`.2 function.
`equally well as a simulator to test the functionality of pulse oximeters. In fact this
`system has been used to compare ten commercially available oximeters (Reynolds
`et at 1992), and has been used to evaluate the effects of dyshemoglobins on pulse
`oximeter accuracy (Reynolds et al 1993a,b), However, this in vitro test system iN
`not practical in a hospital setting where most pulse oximeters are used. The
`system requires a laboratory setting, is not portable, uses oxygenated whole blood
`and needs a CO-oximeter for comparison. However. this instrument is generally
`considered the gold standard for calibrating and testing a pulse oximeter over its
`complete range.
`10. 2. 1. 2 Vellfors system. The Vegfors system is similar to the Reynolds system
`but with a focus on the artificial finger or 'finger phantom' used. Vegfors et d
`( 1993) describe a system where their artificial fingers consist of silicone rubber
`tubes inserted in plastic Delrin cubes. The tubing system chosen was based on its
`characteristics of tubing diameter, wall elasticity, and blood flow velocity to
`simulate normal physiological characteristics of blood in motion, Delrin was used
`because it has similar optical scattering properties to human tissue. Figure ] O.3
`shows three models. Two different finger models, one with one tube and anothei
`with five tubes were tested along with a third artificial finger consisting of 15
`silicone rubber tubes mounted in silicone rubber in the form of glue. The „bject
`was to develop an optical model which simulated the arterial bed of the human
`finger containing blood vessels and surrounding tissue. The results of these
`different finger configurations detennined that physical dimensions of the
`artificial bed are of minor significance for pulse oximeter readings.
`10.2. 1 . 3 Single wedge system . Several other
`less complicated simulators using
`whole blood have been proposed. In one system, proposed by Yount { 1989), a
`light-absorbing wedge shaped vessel containing blood of known oxygen saturation
`level is placed in the pulse oximeter's optical path. If the wedge (figure I 0.4) is
`moved repetitively back and forth perpendicular to this optical path, either
`manually or with the aid of a mechanical device, both the pulse rate and shape of
`the pulse can be altered. Pulse rate can be simulated by changing the frequency at
`which the wedge is moved across the optical path. The shape of the pulse can be
`changed by altering the speed at which the wedge is moving.
`
`1 0.2.1.4 Dual wedge system. In another arrangement o f the system, two wedges
`are used. One is filled with 10093 oxygen saturated blood and the other with
`completely unsaturated blood. The wedges are placed as shown in figure 10.5 und
`by varying the position along the optical path of this arrangement, virtually any
`saturation level can be obtained. Note however that with this second arrangement,
`
`181
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0182
`
`APL_MAS_ITC_00015799
`
`1
`
`1
`
`11
`
`Calibration
`
`165
`
`an additional external device is needed to obtain a pulsatile variation in the
`simulator. figure 10.6 shows the polarization filter system proposed by Yount
`(1989) to achieve this pulsatile variation needed. A pair of polarizing disks
`simulate the changes in transmittance expected by the pulse oximeter. A stepper
`motor controls the motion of one disk. This changes the angle of polarization
`between the two disks, and therefore the amount of light transmitted. By varying
`the rate of angle change, this system can simulate both the shape and pulse rate
`seen by the pulse oximeter. This particular system also has several glass windows.
`This allows for multiple samples to be loaded on the same disk so different
`oxygen saturation levels can be simulated by rotating the appropriate sample into
`the probe.
`
`, .
`
` 2
`
`Blood flowIl i
`
`.f
`
`Blood flow /*'
`
`0,0,
`
`Blood flow ~-1~
`
`Figure 10.3 Block diagram of various artificial fingers as proposed by Vegfors et at (1993).
`
`One limitation of these wedge systems is that i f blood is used as the medium
`in the wedge, the samples either need to be prepared shortly before use or steps
`need to be taken to stabilize the blood.
`
`182
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0183
`
`APL_MAS_ITC_00015800
`
`166
`
`Design of pulse oximeters
`
`Blood
`
`Light path
`
`'4 Wedge movement
`%
`22*GRY~hlj/
`3
`/5
`'~ ? ilitli ftilll/il/t
`-4/Wi7 Tl·41 I i- ' . . . . i ' :ie
`l
`
`Constant absorber
`
`Figure 10.4 Block diagram of a wedge system as proposed by Yount ( I 989).
`
`0% oxygenated blood
`
`Light path
`
`.7-r-9»»~-w~"--ip--*-44
`
`7%~ 1 -
`
`Constant absorber
`
`100% oxygenated blood
`
`1
`
`Figure 10.5 Block diagram of the dual wedge system as proposed by Yount (1989),
`
`10. 2. 1 . 5 Bulb device. Volgyesi
`( 1989) proposed a simple mechanical design to
`simulate a pulsing finger. Figure 10.7 shows the tube and bulb type device. It
`requires a 0.5 to I mL blood sample for each saturation level to be tested. A
`piece of silicone rubber tubing is placed inside a disposable plastic test tube which
`contains a blood specimen. The operator then manually squeezes the bulb at
`regular intervals which causes the silicone rubber tubing and the blood in the
`annular space between the silicone rubber tubing and the test tube to deform or
`pulse. Samples o f heparinized blood are externally altered to different saturation
`levels so different levels of oxygen saturation can be tested. With a variety of
`oxygen saturation level samples prepared in individual test tubes, the pulse
`oximeter can be applied to the device. After the operator is able to rhythmically
`squeeze the bulb for a consistent plethysmograph (rate and amplitude), a reading
`is recorded from the pulse oximeter and the sample is sent to a CO-oximeter for
`a comparison reading. The main advantage of this system is its simple
`implementation. The disadvantage is that the pulsatile nature of the system is
`operator dependent and samples of known oxygen saturation levels of blood need
`to be prepared.
`
`183
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0184
`
`APL_MAS_ITC_00015801
`
`Drobe
`
`Support
`Motor
`
`CO
`
`Calibration
`
`167
`
`LEDs
`
`Samples
`
`rn I
`1 1'
`11
`
`Side view
`
`- Metal outer disk
`Fixed polarized filter
`
`Rotating polarized filter
`
`Metal outer disk
`
`Glass window in disk
`
`9
`
`Top view of metal disk
`
`Figure 10.6 Schematic diagram of polarization system (adapted from Yount 1989).
`
`Tubing
`
`Air
`
`2 of
`
`Bulb
`
`- Plastic cap
`
`Silicone
`» 3
`rubber 2~
`tubing
`
`~> Annular spacers
`
`1- lti'
`
`Blood
`
`/ 12*&5/
`
`Plastic test tube
`
`Pulse oximeter probe
`
`Silicone rubber plug
`
`Figure 10.7 Block diagram of tube and bulb device.
`
`184
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0185
`
`APL_MAS_ITC_00015802
`
`168
`
`Design of pulse oximeters
`
`1 0.2.2 Nonblood simulators
`
`Nonblond simulators. like simulators that use blood, are also based on the concept
`of being able to simulate the absorbance of human tissue (normally the finger)
`between the LEDs ancl the photodiode of the pulse oximeter under test. These
`devices use colored materials to simulate blood. These simulators use a variety of
`mechanical and electrical devices to achieve the desired variations in absorbance.
`The more difficult aspect is simulating the scattering properties of whole blood.
`One of the most successful studies in this area (Marble et al 1994) used a
`combination of nondairy creamer mixed with solutions of red and green dye.
`
`10. 2. 2. ] Billb device. The bulb device described in section 10 . 2 . I above can also
`be used with liquids having differing optical absorbance properties corresponding
`to oxyhemoglobin. A commercial version of this device is currently being
`marketed by Nonin under the trade name finger phantom. This product
`INonin
`1995 ) pr{,vides three translucent white arti.ficial fingers that simulate ariel-ial
`blood at naminally 80%. 90%, and 97% saturation levels. The operator gently
`presses the finger phantom about once every second to generate a pulse. The
`typical infrared percent modulation when squeezed is 0 to 5%.
`10. 2. 2. 2 Wedge device. The wedge device described in section 10 . 2 . 1 above can
`also be used with liquids other than blood having optical absorbance properties
`corresponding to those of the human finger.
`10. 2. 2. 3 Polvester resin device. Figure 10. 8 shows a simple test object proposed
`by Munley er al ( 1989). This device consists of a piece of polyester resin that is
`f<,rmed in the shape of a finger. The resin is adapted to allow a core to be placed
`inside the artificial finger. At the end of the core, in the area exposed to the pulse
`oximeter LED's light path, a slotted piece of suitably colored Plexiglas is placed.
`As the device handle is rotated, the slot allows varying levels of LED light to
`reach the pulse oximeter photodiode. Speed of rotation of the crank will
`determine the pulse rate that
`the oximeter
`reads. Changing the color
`characteristics of Plexiglas will change the oxygen saturation reading that the
`pulse oximeter registers. This device was also shown to produce similar oxygen
`saturation readings among multiple devices of the same make and model of pulse
`oximeter.
`10.2, 2. 4 Colored colloid simulator. Leuthner ( 1994 ) proposed the pulse oximeter
`development system shown in figure 10.9. A transparent bag is filled with a
`colored colloid solution. The color determines the extinction coefficients at the
`two wavelengths of interest. This system uses a water-gelatin mix which is heated
`and colored with red and black ink. To simulate different oxygen saturation
`levels, multiple bags with varying ratios of red and black dye need to be
`prepared. The bag is positioned between two acrylic disks. The disks and bag are
`then rotated by a stepper motor under microcontroller control. With this
`configuration, both the DC and AC absorbance ratio can be adjusted. Increasing
`the angle between the two plates increases change of absorbance over each
`rotation for an increase in relative AC signal. The simulated pulse shape is
`determined by speed of the disk rotation and the pulse rate is determined by the
`rotation frequency. A constant absorber material is placed on top of the disk to
`simulate the constant light absorbance of fingers of different people. In practice,
`
`185
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0186
`
`APL_MAS_ITC_00015803
`
`Calibration
`
`169
`
`it can vary by a factor of four. Generally a piece of white paper of varying
`thickness is used as the constant absorber. Two optic fibers are integrated into an
`artificial finger which then plugs into the finger probe of the pulse oximeter. The
`other ends of fibers are connected opposite each other near the rotating plates. I f
`testing is done using different waveforms, the angular velocity of the rotation has
`to change and as such is controlled through the stepper motor via microcontroller
`control. The whole system is enclosed in a box to prevent disturbances from
`ambient light.
`
`1
`
`Plexiglas
`
`r\«\IN
`
`/1
`
`Handle
`
`11
`
`Poiyester resin finger
`
`Figure 10.8 Polyester resin system proposed by Munley et al (1989).
`
`The physical behavior of this system can be almost totally described using
`Beer's law, but the system cannot be used for finding the calibration table of a
`pulse oximeter. The main reason is that the scattering effect in whole blood is not
`present in this system. However this system can be used for a rough calibration
`table of a new instrument and to test an existing pulse oximeter for the response it
`gives when different colored bags are used.
`
`Screws to adjust
`AC/DC absorbance ratio
`
`l
`
`Probe clip
`
`Photodiode I
`
`t ~==rzccc:A>'
`
`..,1
`1
`
`LEDs
`
`Constant absorber
`
`Plastic bag
`with absorbing -
`colloid
`
`Acrylic disk
`
`l
`
`Stepper motor
`
`1 11
`Ef
`
`PI
`
`Figure 10.9 Leuthner's (1994) colored colloid disk system.
`
`186
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0187
`
`APL_MAS_ITC_00015804
`
`Design of pulse oximeters
`
`170
`10. 2.2.5 Liquid crystal retarder simulator. Zhou et al ( 1992 ) developed a device
`for generating test signals for pulse oximeters based on a voltage-controlled
`liquid-crystal light valve. In the first system, the pulse oximeter's LEDs are
`separated by an optical filter, modulated by a light valve. and recombined before
`detection by the probe's photodiode. The newer system does not require
`wavelength separation and its associated hardware as shown in figure 10.10, The
`transmittance characteristics are varied by taking advantage of the intrinsic
`wavelength dependence of a twisted-nematic liquid-crystal retarder (LCR).
`Polarizers are used to generate optical density variations that can be made to
`resemble blood perfused tissue. The intensity transmitted through the optical
`system can be adjusted by varying the voltage on the LCR. To simulate a pulsatile
`change in transmittance, the attenuation is initially made a constant DC value. A
`small AC voltage is then superimposed on top of the DC voltage to provide a
`pulsatile component. The transmittance at both the red and IR wavelengths varies
`depending on the voltage amplitude applied to the LCR. This allows the AC/DC
`ratios to be controlled by adjusting the amplitude of the voltage applied to the
`LCR. The polarizers are required because the angle of polarization strongly
`affects the range of variation of the red/IR ratio and its sensitivity to the applied
`voltage. Zhou et at are continuing work on this concept to provide the capability
`of simulating the shape of the plethysmographic waveform applied to the LCR.
`
`AC Voltage i DC Voltage
`
`Light path ~ ~ ~U
`
`UFixed polarizing filter
`
`Rotated polarizing filter
`
`Figure 10. 10 Diagram of the liquid crystal retarder (LCR) system proposed by Zhou et d
`(1992)
`tisSILe model, A device based on the same general principles as
`10.2. 2. 6 Aovagi
`the wedge system been proposed by Aoyagi etal ( 1994). Figure 1 0.1 I shows that
`a static tissue model having absorption characteristics similar to a human finger is
`inserted into a pulse oximeter probe. A blood model having blood absorption
`characteristics similar to a specified oxygen saturation level is moved within the
`tissue model to simulate pulsatile motion and pulse rate. By altering the geometry
`of the blood model and/or the rate of motion of the blood model in and out of the
`tissue model, both the pulsatile waveform and pulse rate can be simulated.
`10.2. 2.7 Optoelectronic device. A number of
`relatively simple easy-to-use
`simulators have begun to appear on the market based on optoelectronic
`
`1
`
`1
`
`187
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0188
`
`APL_MAS_ITC_00015805
`
`Calibration
`
`171
`
`principles. Figure 10.12 shows a block diagram for one of these types of
`simulators. First, the user selects the parameter(s) to be simulated. The pulse
`oximeter probe is then attached to the device and a signal is received from the
`pulse oximeter probe's LEDs by the simulator. Pulse separator and timer
`circuitry convert the red and infrared light pulses from the pulse oximeter probe
`into electric signals. These signals are modulated with the appropriate level of
`AC/DC ratio (under computer control) and then converted back to light pulses,
`via the LED bar, to the probe's photodiode. Finally, the pulse oximeter responds
`to the converted light pulses as it would to light pulses modulated by living tissue.
`LEDs
`
`Pulse oximeler probe
`
`Blood model -_-1_-1 NO' 4*-f.*'.64
`
`11I
`
`Photodede
`
`~%-2177~1//1 1/ll//1 ,
`
`Tissue model
`
`1
`
`Figure 10.11 Block diagram of system as proposed by Aoyagi et al (1994).
`
`These systems can test the probe and oximeter over the complete specified
`range of the oximeter. Also, simulation of a wide range of conditions is possible.
`The modulated signal can vary plethysmographic amplitude and wave shape to
`simulate a variety of ambient light conditions, motion arti facts, and arrhythmias.
`At least one system (Clinical Dynamics 1995) also includes a probe analyzer
`capability which independently tests LED and photodiode continuity and
`sensitivity. These types of simulators are primarily used by pulse oximeter
`manufacturers during final assembly and checkout of their products. In addition,
`their capability to generate automatic test sequences help document JCAHO (Joint
`Commission on Accreditation of Healthcare Organizations) testing requirements.
`
`10.2.3 Electronic simulators
`
`Electronic simulators have limited usefulness since they only simulate electronic
`signals to and from the probe. Usually these relatively simple devices are
`provided by the pulse oximeter manufacturer and only check a small number of
`values. These devices typically plug into the probe port on the pulse oximeter and
`
`188
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0189
`
`APL_MAS_ITC_00015806
`
`172
`
`Design of pulse oximeters
`
`use the drive current of the probe LEDs to generate a simulated photodiode
`signal back to the pulse oximeter using the device. Figure I (). 13 shows an
`example of such a device. In remote mode. the LEDs just drive an amplifier and
`the output shows up on the sensor output. This is useful for simple continuity
`testing. In local mode, these devices are able to electronically simulate a discrete
`number of simulated oxygen saturation levels, pulse rates and plethysmographic
`waveform strengths. In addition the calibration resistor value reading capability
`of the pulse oximeter can be checked. These simulators are good for functional
`checks of the pulse oximeter's internal circuitry, but because they bypass ihe
`pulse oximeter's probe, are of limited usefulness.
`
`>X\LEDs©4
`
`1
`Pulse separator
`and timer
`1
`RED
`IR
`1
`r-_-- DC multip'lers I
`1
`Conputer control
`1
`L--74 AC multipliers
`1
`1
`44,4- - .
`
`Userinput
`variables
`
`Pulse oximeter
`components
`
`0 Simulator
`
`components
`
`IR Switch
`
`i\\\\\\\I
`\Photodio{le
`\\\\\\:\
`1 Photodiode
`
`1
`
`R + IR
`Timing
`
`Amplifier --- LED BAR
`
`Figure 10.12 Block diagram electro-optic simulator system developed by Merrick and Haas
`(1994).
`
`LED output monitor
`
`10.3 STANDARDS
`
`Although the pulse oximeter has been on the market since 1977 (Santamaria and
`Williams 1994), surprisingly little standardization has been documented to this
`point. Statements like 'machines and probes are interchangeable with less than
`0.5% difference', 'warm-up time factor of 0.5% to 1.0%' and 'the low perfusion
`light on the Ohmeda oximeter indicates the oximeter's microprocessor has low
`confidence level in the data' can be found in the literature. Several standards do
`exist, but their value from the designer's point of view is limited at best.
`
`189
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0190
`
`APL_MAS_ITC_00015807
`
`Calibration 173
`
`0%
`
`Local
`
`Remote
`
`1
`
`DC Level
`
`AC Level
`
`Pulse
`Conditioning
`
`Pulse
`Generator
`
`Sensor
`Output
`
`LED drive
`inputs from
`oximeter
`n
`
`I Amplifier
`
`L
`
`- Power --- +V
`Conversior
`
`CC
`
`Calibration
`Resistor
`
`Resistor input
`
`Figure 10.13 Block diagram of an electronic simulator that replaces the pulse oximeter probe 1
`(used with permission (Nellcor 1994) Pulse oximeter tester Model SRC-2).
`
`10.3.1 ASTM F1415
`
`The ASTM F1415 standard (ASTM 1992) contains requirements for the pulse
`oximeter designer in regard to marking and documenting the system, electrical
`safety concerns, electromagnetic interference and alarms. No specific information
`is provided regarding specific design requirements of the parts of the system
`discussed in the preceding chapters. In addition, no specific information is
`provided in regard to calibration or testing of these devices.
`
`10.3.2 ISO 9919
`
`This standard mentions a few requirements regarding calibration. These include
`requiring manufacturers to provide:
`
`1. The calibration range of the pulse oximeter.
`2. Whether the pulse oximeter is calibrated to display functional or fractional
`saturation.
`3. The accuracy and range of HbO2 saturation level displayed.
`4. Whether the calibration was functional or fractional saturation.
`5. Test methods for calibration need to be available from manufacturer upon
`request.
`
`The ISO 9919 (International Organization for Standardization 1992) also offers
`this disclaimer in Annex L:
`
`Values derived from the pulse oximeter are not a measurement of blood or tissue oxygen
`tension and therefore pulse oximetry provides no direct indication of oxygen delivery to or
`
`consumption by, tissues. At present there is no widely accepted direct in vitro calibration
`
`1
`
`190
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0191
`
`APL_MAS_ITC_00015808
`
`1
`
`It
`
`174
`
`Design of pulse oximeters
`
`method for pulse oximeters. The only accepted in viiro test method for correlation of the
`reading from a pulse oximeter (Sp02) is bench-type oximetry employing more than two
`wavelengths of light or other methods using blood samples drawn from human subjects
`
`Although work ts progressing on the development of direct in \,it,·o calibration methods ,
`
`present techniques still require the use of human subjects. To include test methods in
`standards that require the use of human subjects, has, through past experience, been found to
`be unacceptable, and therefore in vit'o test methods are not included in this International
`Standard.
`
`10.3.3 Other smndards
`
`American Society of Anesthesiologists. Standards for Basic Intra-Operative
`Monitoring, 1986 (0696-ASA).
`American Society of Anesthesiologists. Standards for Post-Anesthesia Care, 1989
`(0697-ASA).
`European Committee for Standardization. Drafting European norm for pulse
`oximeters.
`
`REFERENCES
`
`Ackerman S W and Weith P 1995 Knowing your pulse oximetry monitors Med. Electron. 26 ( 1 )
`82--6
`ASTM 1992 Standard Specification for Pulse Oximeters F/415-92 (Philadelphia PA: American
`Society for Testing and Materials)
`Aoyagi T, Fuse M, Shindo Y and Keto M 1994 Apparatus for calibrating pulse oximeters US
`patent 5,278,627
`Cheung PW, Gauglitz KF, Hunsaker SW, Prosser SJ, Wagner DO and Smith RE1993
`Apparatus for the automaticcalibration ofsignalsemployed in oximetry USpatria 5.259,381
`Clinical Dynamics 1995 Technical sales brochm·e (Wallingford, CT: Clinical Dynamics)
`International Organization for Standardization 1992 Pulse Oximeters for Medical Use-
`Requiremenfs ISO9919: 1992(E)
`Lcuthner T 1994 Development system for pulse oximetry Med. Biol. Eng. CompuL 32 596-8
`in vitro
`Marble D R, Bums D H and Cheung P W 1994 Diffusion-based model of pulse oximetry :
`and in vivo comparison Appl Op' 33 1279-85
`Merrick E B and Haas P 1994 Simulation for pulse oximeter US Patent 5, 348,005
`Munley A J , Sik M J and Shaw A 1989 A test object for assessing pulse oximeters Lancet 1048-9
`Nellcor 1994 Pulse Oximeter Tester Model SRC-2 (Pleasanton. CA: Nellcor)
`Nonin Medical 1995 Nonin finger phantom Technical Nok (Plymouth , MN : Nonin Medical)
`Moyle JT B 1994 Pulse Oximelry ( London: BMG)
`Payne J P and Severinghaus J W ( eds) 1986 Pulse Oximetry (New York : Springer)
`Pologe J A 1989 Functional saturation versus fractional saturation: what does the pulse oximeter
`read J. Ck Monit. 5 288-9
`Reynolds KJ, del<ock JP, Tarssenko L and Moyle JTB 1991 Temperature dependence of LED
`and its theoretical effect on pulse oximetry Brit. J. Anaesthesiot. 67 638-43
`Reynolds KJ,Moyle J TB, Gale LB, Sykes MK and Hahn CEW 1992 In vitro performance
`test system for pulse oximeters Med. Biol. Eng. Comput. 30 629-35
`Reynolds KJ, Moyle JTB. Sykes MK and Hahn CEW 1993a Responses of 10 pulse oximeters
`to an in vitro test system Brit. J. Anaestliesiot. 6% 265-9
`Reynolds KJ, Palayiwa E, Moyle JTB, Sykes MK and Hahn CEW 1993b The effects of
`dyshacmoglobins on pulse oximetry J Clin. Monit. 9 81-90
`Santamaria T and Williams J S 1994 Device focus: pulse oximetry Med. Device Res. Rep. 1 ( 2) 8-
`I 0
`Severinghaus J W, Naifeh K H and Koh S O 1989 Errors in 14 pulse oximeters during profound
`hypoxia J. Clin. Monit. 5 72-81
`Vegfors M, Lindberg L G, Oberg P A and Lennmarken C 1993 Accuracy of pulse oximetry al
`various haematocrits and during hacmolysis in and in vitromodel Med. Biol Eng. Comput. 3 1
`135-41
`
`191
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0192
`
`APL_MAS_ITC_00015809
`
`Calibration
`
`175
`
`Volgyesi G A 1992 Method of testing the accuracy of pulse oximeters and device therefor US
`patent 5,166,517
`Wukitsch M W, Petterson M T, Tobler D R and Pologe J A 1988 Pulse oximetry: analysis of
`theory , technology, and practice J Clin. Monit. 4 290-301
`Yount J E 1989 Device and procedures for in vitro calibration of pulse oximetry monitors US
`patent 4,834,532
`Zhou GX, Schmitt JM and Walker E C 1992 Electro-optical simulator for pulse oximeters Med.
`Biol. Eng. Comput. 31 534-9
`
`INSTRUCTIONAL OBJECTIVES
`
`10.1 Describe how R curves are determined through in vivo testing.
`10.2 Explain the role that LED temperature plays in oxygen saturation level determination.
`10.3 Explain why the term Sp02 is necessary when referring to oxygen saturation levels.
`10.4 Explain the reason why different R curves may be needed for a manufacturer's pulse
`oximeter system.
`10.5 Describe how oxygen saturation level is altered through an in vitro test system.
`10.6 Explain why pulse oximeters are less accurate for SPO2 saturation levels below 60%.
`10.7 Describe the operation of an optoelectronic simulator system.
`10.8 Describe the operation of an colored colloid simulator system.
`10.9 Describe the operation of polyester resin device simulator system.
`10.10 Describe the operation of a wedge simulator system.
`10.11 Describe the operation of the tube and bulb simulator system.
`10.12 Explain the limitations of the electronic simulators used for testing pulse oximeters.
`
`t
`
`1
`
`192
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0193
`
`APL_MAS_ITC_00015810
`
`1
`
`CHAPTER 11
`
`ACCURACY AND ERRORS
`Supan Tung/itkusolmun
`
`Continuous assessment of arterial oxygen saturation (Sa'2) is important in
`clinical management of critically ill patients. Pulse oximeters have been widely
`used as blood oxygen monitoring devices since the early 1980.. Currently, pulse
`oximeters can be found in virtually every operating room, recovery room, and
`intensive care unit. The advantages of pulse oximetry include noninvasiveness,
`ease of use, portability, and patient comfort. A light source generated by two
`LEDs, with wavelengths at approximately 660 nm and 940 nm, and a photodiode
`are mounted in a probe of a pulse oximeter. Circuit control. saturation
`calculation, and display are managed by a microprocessor instrument as described
`in chapter 8. Unlike earlier techniques such as the in vive eight-wavelength
`oximeter (chapter 3), no heating or arterialization techniques are required in
`pulse oximetry.
`All pulse oximeters work using absorption spectrophotometry, however,
`considerable differences exist in the way different manufacturers obtain and
`process the data. These differences occur in the light-emitting diodes, sampling
`frequency, microprocessor algorithms. and the constants used in the calculations,
`or the look-up tables. Since the technique has come into wide clinical use over the
`past decade, it is important to examine circumstances where its reliability may be
`questioned. The objective of this chapter is to describe several sources of error in
`pulse oximetry which may cause hazardous consequences to the patients.
`Recognizing the limitations described in this chapter and applying appropriate
`corrective interventions are essential to optimize the clinical use of pulse
`oximeters.
`
`11.1 EVALUATION OF PULSE OXIMETERS
`
`The gold standard measurement ot arterial oxygen saturation is the CO-oximeter ,
`described in chapter 3. A comparison of the pulse oximeters' readings and CO-
`oximeters' readings is thus required to verify the reliability of the pulse oximetry
`technique. Comparisons between pulse oximeters' arterial oxygen saturation
`values and the CO-oximeters' readings, as well as the HP eight-wavelength ear
`oximeter will be discussed in this section.
`
`176
`
`193
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0194
`
`APL_MAS_ITC_00015811
`
`r
`
`1
`
`Accuracy and errors
`
`177
`
`11.1.1 Accuracy, bias, precision, and confidence limit
`
`Accuracy is a measure of systemic error or bias; the greater the error, the less
`accurate the variable. The accuracy of a measurement is the degree to which it
`actually represents what it is intended to represent. The location of the mean
`errors reflects the accuracy of the measurement. The accuracy of pulse ox imeter
`oxygen saturations can usually be tested by comparing with the reference
`techique, CO-oximetry. Parameters frequently used to represent the degree of
`accuracy are bias, and absolute mean errors. Bias, in this case, is defined as the
`mean of the differences between the pulse oximeter readings and the CO-
`oximeter readings, which can be expressed as
`
`NI xi
`i=1
`bias =-=.F
`N
`
`(11.1)
`
`where xi is calculated by subtracting the ith CO-oximeter measurement from the
`corresponding oximeter saturation displayed by a pulse oximeter. N is the total
`number of measurements. Units are percent saturation.
`Precision is a measure of variation of random error, or degree of
`reproducibility. The dispersion of points around the mean reflects the precision
`of the measurement. Precision is often described statistically using the standard
`deviation (SD) of the differences between the pulse oximeter readings and the
`CO-oximeter readings of repeated measurements (Nickerson et al 1988) as in
`equation (11.2). Units are percent saturation.
`
`1N
`9
`12(xi -x)-
`precision = SD = 11 i=l
`1 N-1
`Some researchers frequently use a 95% confidence limit, which for a normal
`distribution is equal to 1.96 times SD:
`
`(11.2)
`
`95% confidence limit = 1.96 x SD = 2 x SD.
`
`(11.3)
`
`Example l
`
`The results from an experiment to compare pulse oximeter and CO-oximeter
`readings are shown in table 11.1. Ten measurements were made.
`
`From table 11.1,
`
`10
`24i=1
`15---1.5%
`bias =x=--10
`10
`
`194
`
`MASIMO 2053
`Apple v. Masimo
`IPR2022-01299
`
`
`
`RX-0035.0195
`
`APL_MAS_ITC_00015812
`
`178
`
`Design of pulse oximeters
`
`I 0I (ri - i)2
`i=I
`
`10-1
`
`&0.5
`= 1-= 1.51%,
`N 9
`
`precision = li
`
`Table 11.1 Comparison of pulse oximeter and CO-oximeter readings.
`
`Measurement CO-oximeter Pulse oximeter
`(i)
`readings (%) readings (g,)
`9 7
`100
`98
`99
`9
`3
`92
`9I
`98
`96
`4
`99
`97
`5
`93
`90
`6
`90
`89
`1
`98
`95
`8
`90
`9
`88
`93
`92
`10
`
`3
`1
`-1
`2
`2
`3
`1
`3
`9
`
`X, (56) .ri - .7C
`
`1.5
`-0.5
`-2.5
`0.5
`0.5
`1.5
`-0.5
`1.5
`05
`-2.5
`
`and
`
`95% confidence limit =2x 1.5 1% = 3.02%.
`
`The bias of 1.5% means that the test pulse oximeter tends to overestimate the
`oxygen saturation level (postive bias). A 95% confidence limit of 3.02% means
`thal the pulse oximeter will give an outcome in the range between 1.5 - 3.02%
`and 1.5% + 3.02%. or between -1.52% and 4.52% from the true value (the CO-
`oximeter reading) with a probability of 0.95.
`The use of bias and precision is helpful in getting a clear picture of a pulse
`oximeter's performance and how thiE compares to other units or other studies. A
`unit may be very precise. so that the results are highly reproducible with a low
`scatter, but have a high bias so that the results are not centered on the true values,
`In contrast, a unit may have a very low bias, but have poor precision. with values
`swinging widely from side to side of the true value. In clinical practice, a 95%
`confidence I imit o f less than *3% is considered acceptable for most cases.
`Other statistical terms from the regression analysis (correlation coefficient,
`positive error, intercept, and slope) are also used in several studies c Yelderman
`and New 1983, Taylor and Whitwam 1988).
`
`11.1.2 What do pulse oximeters really measure?
`
`Pulse oximeters only measure a ratio of transmitted red and infrared light
`intensities, and relate this to a look-up table of empirical oxygen saturation values
`(see chapter 9). The values in the table depend on the manufacturer's purpose of
`estimating functional or fractional oxygen saturation, but will in reality be
`neither of these unless the dyshemoglobin tdysfunctional hemoglohin) levels, and
`the pH levels in a subject's arterial blood are exactly the same as the average
`values of those used in the empirical calibration to create the look-up table. Choe
`er al ( 1989) found that the measured oxygen saturations in two instruments
`(Ohmeda Biox 3700 and Radiometer Pulse Oximet