Signal processing algorithms
`
`133
`
`we do a large number of calculations of the Ratio of Ratios for each period, and
`then do the best fit calculation to the line y = Rx + b to fit the optimum value of
`R for that period, taking into account the constant I) which is caused by DC drift.
`To determine the Ratio of Ratios exclusive of the DC offset we do a linear
`regression.
`It
`is preferred to take points along the curve having a large
`differential component, for example, from peak to valley. This will cause the mx
`term to dominate the constant b:
`
`R
`
`: ”Ext-w "2 xii-"'1'
`fl
`,,
`HEIJ“ -(2Ij)"
`
`(9.33)
`
`where n = # of samples,j = sample #, x = IRdIIR /dt, y = IIRdIR /dt.
`Prior sampling methods typically calculate the Ratio of Ratios by sampling
`the combined AC and DC components of the waveform at the peak and valley
`measurements of the waveform. Sampling a large number of points on the
`waveform, using the derivative and performing a linear regression increases the
`accuracy of the Ratio of Ratios, since noise is averaged out. The derivative form
`eliminates the need to calculate the logarithm. Furthermore doing a linear
`regression over the sample points not only eliminates the noise caused by patient
`movement of the oximeter,
`it also decreases waveform noise caused by other
`sources.
`
`9.4 GENERAL PROCESSING STEPS OF OXIMETRY SIGNALS
`
`The determination of the Ratio of Ratios (ROS) requires an accurate measure of
`both the baseline and pulsatile signal components (Frick et al 1989). The baseline
`component approximates the intensity of light received at the detector when only
`the fixed nonpulsatile absorptive component
`is present
`in the finger. This
`component of the signal is relatively constant over short intervals and does not
`vary with nonpulsatile physiological changes, such as movement of the probe.
`Over a relatively long time, this baseline component may vary significantly. The
`magnitude of the baseline component at a given point in time is approximately
`equal to the level identified as RH (figure 9.2). However, for convenience, the
`baseline component may be thought of as the level indicated by RL, with the
`pulsatile component varying between the values of RH and RL over a given pulse.
`Typically, the pulsatile component may be relatively small in comparison to the
`baseline component and is shown out of proportion in figure 9.3. Because the
`pulsatile components are smaller, greater care must be exercised with respect to
`the measurement of these components. If the entire signal, including the baseline
`and the pulsatile components, were amplified and converted to a digital format
`for use by microcomputer, a great deal of the accuracy of the conversion would
`be wasted because a substantial portion of the resolution would be used to
`measure the baseline component (Cheung et al 1989).
`In this process, a substantial portion of the baseline component termed offset
`voltage V05 is subtracted off the input signal V1. The remaining pulsatile
`component is amplified and digitized using an ADC. A digital reconstruction is
`then produced by reversing the process, wherein the digitally provided
`
`information allows the gain to be removed and the offset voltage added back.
`
`_
`'
`
`|
`
`151
`
`151
`
`

`

`
`
`134
`
`Design of pulse oximeters
`
`including the baseline and
`This step is necessary because the entire signal,
`pulsatile components is used in the oxygen saturation measurement process.
`Feedback from the microcomputer is required to maintain the values for
`driver currents IO, V0S and gain A at levels appropriate to produce optimal ADC
`resolution (figure 9.4). Threshold levels L1 and L2 slightly below and above the
`maximum positive and negative excursions L3 and L4 allowable for the ADC
`input are established and monitored by the microcomputer (figure 9.5). When the
`magnitude of the input
`to and output from the ADC exceeds either of the
`thresholds L1 or L2, the drive currents ID are adjusted to increase or decrease
`the intensity of light
`impinging on the detector. This way,
`the ADC is not
`overdriven and the margin between L1 and L3 and between L2 and L4 helps
`assure this even for rapidly varying signals. An operable voltage margin for the
`ADC exists outside of the thresholds, allowing the ADC to continue operating
`while the appropriate feedback adjustments to A and VOS are made. When the
`output from the ADC exceeds the positive and negative thresholds L5 or L6, the
`microcomputer responds by signaling the programmable subtractor to increase or
`decrease the voltage VOS being subtracted. This is accomplished based on the level
`of the signal received from the ADC. Gain control
`is also established by the
`microcomputer in response to the output of the ADC (Cheung at al 1989).
`
`with. + Vos = V1
`A
`V03
`
` Microcomputer
`
`
`
`
`
`Limits
`adj
`
`5 Motion
`';
`artifact
`
`5 Peak
`E detect
`
`Analysis.
`features
`display
`
`
`Figure 9.4. A functional block diagram of the microcomputer feedback illustrating the basic
`operation of the feedback control system. The DC value of the signal is subtracted before digitizing
`the waveform to increase the dynamic range of conversion. The removed DC value is later added to
`the digitized values for further signal processing (Cheung at al 1989).
`
`A program of instructions executed by the Central Processing Unit of the
`microcomputer defines the manner
`in which the microcomputer provides
`servosensor control as well as produces measurements for display. The first
`segment of the software is the interrupt level routine.
`
`9.4.] Start up software.
`
`The interrupt level routine employs a number of subroutines controlling various
`portions of the oximeter. At
`the start up, calibration of
`the oximeter
`is
`
`
`
`152
`
`152
`
`

`

`Signal processing algorithms
`
`[35
`
`performed. After calibration, period zero subroutine is executed which includes
`five states, zero through four (figure 9.6).
`Period zero subroutine is responsible for normal sampling
`
`State 0: Initialize parameters
`State 1: Set drive current
`State 2: Set offsets
`State 3: Set gains
`State 4: Normal data acquisition state.
`
`Probe set-up operations are performed during the states zero to three of this
`subroutine. During these states probe parameters including the amplifier gain A
`and offset voltage V05 are initialized, provided that a finger is present in the
`[.u'obe. State 4 of the intemrpt period zero subroutine is
`the normal data
`acquisition state. The signals produced in response to light at each wavelength are
`then compared with the desired operating ranges
`to determine whether
`modifications of the driver currents and voltage offsets are required. Finally state
`4 of the period zero subroutine updates the displays of the oxirneter. Sequential
`processing returns to state 0 whenever the conditions required for a particular
`state are violated (Cheung at a! 1989).
`
`High rail
`
`L3
`
`
`
`>\\\m L1
`\
`
`
` L6
` L2
`
`L4
`
`
`
`
`
`
`Low rail \\\V
`
`Figure 9.5 A graphical representation of the possible ranges of digitized signal, showing the
`desired response of the I/O circuit and microcomputer at each of the various possible ranges
`(Cheung et al 1989).
`
`9.5 TRANSIENT CONDITIONS
`
`The relative oxygen content of a patient's arterial pulses and the average
`background absorbance remain about the same from pulse to pulse. Therefore.
`the red and infrared light that is transmitted through the pulsatije flow produces a
`regularly modulated waveform with periodic pulses of comparable shape and
`amplitude and a steady state background transmittance. This regularity in shape
`helps in accurate determination of the oxygen saturation of the blood based on the
`maximum and minimum transmittance of the red and infrared light.
`Changes in a patient‘s local blood volume at the probe site due to motion
`artifact or ventilatory artifact affect
`the absorbance of light. These localized
`
`
`
`— 1
`
`53
`
`153
`
`

`

`
`
`changes often introduce artificial pulses into the blood flow causing the periodic
`pulses ride on a background intensity component of transmittance that varies as
`blood volume changes. This background intensity component variation, which is
`not necessarily related to changes in saturation, affects the pulse to pulse
`uniformity of shape, amplitude and expected ratio of the maximum to minimum
`transmittance, and can affect the reliability and accuracy of oxygen saturation
`determination (Stone and Briggs 1992).
`
`Calibration
`at start up
`
`Start up
`
`Other interrupt
`
`penods
`
`
`
`
`
`
`
`States 0 to 3
`to setup probe
`
`sam oles
`
`
`
`Drive LEDs a
`input samples
`
`Check offsets
`
`Update display
`
`
`
`
`
`Figure 9.6. Flow chart of a portion of an interrupt
`microcomputer (Cheung ct al 1989).
`
`lcvc] software routine included in the
`
`In addition, there are times when the patient’s background level of oxygen
`saturation undergoes transient changes, for example, when the patient loses or
`requires oxygen exchange in the lungs while under gaseous anesthesia. The
`transient waveform distorts the pulse shape, amplitude, and the expected ratio of
`the pulses, which in turn affects the reliability and accuracy of the oxygen
`saturation determination.
`
`With changes in the background intensity absorbance component due to
`artifacts from changes in blood volume or
`transient saturation changes,
`the
`determined saturation value is not accurate and it would not become accurate
`again until the average absorbance level stabilizes.
`an
`signals provide
`The
`saturation calculations based upon transient
`overestimation 0r underestimation of the actual saturation value, depending upon
`the trend. The transmittance of red light increases as oxygen saturation increases
`resulting in a signal value having a smaller pulse, and the transmittance of the
`infrared light decreases as saturation increases resulting in the infrared pulsatile
`amplitude increasing. For
`these wavelengths,
`the transmittance changes with
`saturation are linear in the range of clinical
`interest,
`i.e., oxygen saturation
`between 50% and 100%. The accuracy of the estimation is of particular concern
`
`during rapid desaturation. In such a case, the determined saturation based on the
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`154
`
`
`136
`Design, of pulse ()ximctcrs
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`154
`
`

`

`
`
`Signal processing algorithms
`
`137
`
`value. This
`actual
`than the
`drop
`a greater
`indicates
`detected signals
`underestimation of oxygen saturation may actuate low limit saturation alarms that
`can result in inappropriate clinical decisions.
`The pulsatile amplitude is usually quite small, typically less than 5% of the
`overall
`intensity change and any small change in overall or background
`transmittance, such as slight changes in average blood saturation, can have a
`relatively large effect on the difference in maximum and minimum intensity of
`the light
`levels. Because the Change in transmittance with changing oxygen
`saturation is opposite in direction for the red and infrared,
`this can result
`in
`overestimation of the pulsatile ratio during periods when saturation is decreasing,
`and underestimation during periods when saturation is increasing. It is therefore
`essential to compensate for the effects of transient conditions and localized blood
`volume changes on the actual
`signal,
`thereby providing a more accurate
`estimation of the actual oxygen saturation value.
`This can be achieved by using a determined rate of change from pulse to
`pulse, using interpolation techniques
`and by using the
`low frequency
`characteristics of the detected signal values.
`The transient error is corrected by linear interpolation where the determined
`maxima and minima for a first and second optical pulses are obtained, the second
`pulse following the first. The respective rates of change in the transmittance due
`to the transient are determined from the maximum transmittance point of the first
`detected pulse to the second detected pulse (Stone and Briggs 1992). The
`determined rates of change are then used to compensate any distortion in the
`detected transmittance of the first detected pulse introduced by the transient in
`accordance with the following algorithm
`
`V
`max(”)
`
`* =
`
`max(”)
`
`+ V
`[ max(n)
`
`—V
`max(”l
`
`“man: (”J ‘ ’mm (”ll
`+l X—-
`)]
`[Imnxln+ll_lmaxl’1)l
`
`(9.34)
`
`where tmax(n) is the time of occurrence of the detected maximum transmittance at
`the n maximum,
`tmin(n) is the time of occurrence of the detected minimum
`transmittance of the wavelength at the n minimum, Vmax(n) is the detected optical
`signal maximum value at the maximum transmittance of the wavelength at the n
`maximum Vmax(n)* is the corrected value, for n being the first optical pulse, and
`n + 1 being the second optical pulse of that wavelength.
`the detected
`By application of the foregoing linear interpolation routine,
`maximum transmittance value at
`tmax(”) can be corrected, using the values
`tm3x(n+l), detected at the next coming pulse, to correspond to the transmittance
`value that would be detected as if the pulse were at steady state conditions. The
`corrected maximum value and the detected (uncorrected) minimum value thus
`provide an adjusted optical pulse maximum and minimum that correspond more
`closely to the actual oxygen saturation in the patient’s blood at
`that time, not
`withstanding the transient condition. Thus, using the adjusted pulse values in place
`of the detected pulse values in the modulation ratio for calculating oxygen
`saturation provides a more accurate measure of oxygen saturation than would
`otherwise be obtained during transient operation.
`Similarly, the respective rates of change in the transmittance are determined
`from the minimum transmittance point of the first detected pulse to the minimum
`of the second detected pulse. The determined rates of change are then used to
`compensate for any distortion in the detected minimum transmittance of the
`
`155
`
`155
`
`

`

`138
`
`Design of pulse oximeters
`
`second detected pulse introduced by the transient
`following algorithm
`
`in accordance with the
`
`v . my”:
`mm
`
`-(n—l)+[V .
`lTlln
`mm
`
`
`Hm“ (n) — ’min (a — 1.)]
`_
`(n) V -
`(n
`1)]><
`[£111in(n)_r|r1|ul”_l)l
`min
`
`(9.35)
`
`where tmaxm) is the time of occurrence of the detected maximum transmittance at
`the n maximum;
`tminm) is the time of occurrence of the detected minimum
`transmittance of the wavelength at the )2 minimum; Vmin(”) is the detected optical
`signal minimum value at the minimum transmittance of the wavelength at the n
`minimum; Vmin(n)* is the corrected value, for n being the second optical pulse,
`and n — 1 being the first optical pulse of that wavelength.
`the detected
`By application of the foregoing linear interpolation routine,
`minimum transmittance value at
`t = n can be compensated using the detected
`values at the preceding pulse 1‘ = n — l, to correspond to the transmittance value
`that would be detected as if the pulse were detected at steady state conditions. The
`compensated minimum value and the detected (uncompensated) maximum value
`thus provide an adjusted optical pulse maximum and minimum that correspond
`more closely to the actual oxygen saturation in the patient’s blood at that time,
`notwithstanding the transient condition. Thus, using the adjusted pulse values in
`place of the detected pulse values in the modulation ratio for calculating oxygen
`saturation provides a more accurate measure of oxygen saturation than would
`otherwise be obtained during transient operation.
`As is apparent from the algorithms, during steady state conditions the
`compensated value is equal
`to the detected value. Therefore,
`the
`linear
`interpolation routine may be applied to the detected signal at all times, rather than
`only when transient conditions are detected. Also, the algorithm may be applied
`to compensate the detected minimum or maximum transmittance values by
`appropriate adjustment of the algorithm terms. The amount of oxygen saturation
`can then be determined from this adjusted optical pulse signal by determining the
`relative maxima and minima as compensated for the respective wavelengths and
`using that
`information in determining the modulation ratios of the known
`Lambert—Beer equation.
`The Nellcor® N—200 oximeter is designed to determine the oxygen saturation
`in one of the two modes.
`In the unintegrated mode the oxygen saturation
`determination is made on the basis of optical pulses in accordance with
`conventional pulse detection techniques. In the ECG synchronization mode the
`determination is based on enhanced periodic data obtained by processing the
`detected optical signal and the ECG waveform of the patient.
`The calculation of saturation is based on detecting maximum and minimum
`transmittance of two or more wavelengths whether the determination is made
`pulse by pulse (the unintegrated mode) or based on an averaged pulse that is
`updated with the occurrence of additional pulses to reflect the patient’s actual
`condition (the ECG synchronized mode).
`incoming
`Interrupt programs control
`the collection and digitization of.
`optical signal data. As particular events occur, various software flags are raised
`which transfer operation to various routines that are called from a main loop
`processing routine.
`The detected optical signal waveform is sampled at a rate of 57 samples per
`second. When the digitized red and infrared signals for a given portion of
`
`156
`
`156
`
`

`

`i
`l
`
`I
`
`Signal processing algorithms
`
`139
`
`detected optical signals are obtained, they are stored in a buffer called DATBUF
`and a software flag indicating lhe presence of data is set. This set flag calls a
`routine called MUNCH, which processes each new digitized optical
`signal
`waveform sample to identify pairs. of maximum and minimum amplitudes
`corresponding to a pulse. The MUNCH routine first queries whether or not there
`is ECG synchronization,
`then the MUNCH routine obtains
`the enhanced
`composite pulse data in the ECG synchronization mode. Otherwise, MUNCH
`obtains the red and infrared optical signal sample stored in DATBUF,
`in the
`unintegrated mode. The determined maximum and minimum pairs are then sent
`to a processing routine for processing the pairs. Preferably,
`conventional
`techniques are used for evaluating whether a detected pulse pair is acceptable for
`processing as an arterial pulse and performing the saturation calculation, whether
`the pulse pair is obtained from the DATBUF or from the enhanced composite
`pulse data.
`The MUNCH routine takes the first incoming pulse data and determines the
`maximum and minimum transmittance for each of the red and infrared detected
`optical signals, and then takes the second incoming pulse data, and determines the
`relative maximum and minimum transmittance. The routine for processing the
`pairs applies the aforementioned algorithm to the first and second pulse data of
`each wavelength. Then the oxygen saturation can be determined using the
`corrected minimum and detected maximum transmittance for the second pulses of
`the red and infrared optical signals. Some of the examples demonstrate the above
`application.
`
`Example 1
`
`Figure 9.7(a) shows the representative plethysmographic waveforms in a steady
`state condition for the red and infrared detected signals. VmaxR(1) equals 1.01 V,
`and VminR(1) equals 1.00 V, for n = l, 2 and 3 pulses. VminR(n) is the detected
`optical signal minimum value at
`the minimum transmittance at
`the n pulse
`minimum. The modulation ratio for the maxima and minima red signal is:
`
`
`VmaXR(n) _ 1.01v
`_
`VminR(n)
`1.00v
`
`=1.01.
`
`For the infrared wavelength, VmaxIR(n) equals 1.01 V and VminIR(n) equals
`1.00 V and the determined modulation ratio is 1.01.
`
`Using these determined modulation ratios in the formula for calculating the
`ratio R provides:
`
`R _ lnl'VnmeJIme R00] _ m _1 00
`1n[VmaXIR(n)/VminIR(n)]
`0.01
`'
`
`A calculated R = 1 corresponds to an actual saturation value of about 81% when
`incorporated into the saturation equation. A saturation of 81% corresponds to a
`healthy patient experiencing a degree of hypoxia for which some corrective
`action would be taken.
`
`* 1
`
`57
`
`157
`
`

`

`140
`
`Design of pulse oximeters
`
`1.010V
`
`1.000V
`
`Red
`
`1 s
`
`2 s
`
`3 s
`
`Steady state saturation
`
`Infrared
`
`2 s
`
`3 s
`
`1.010V
`
`1.000V
`
`1 s
`
`(a)
`
`Decreasing saturation
`
`Vmax3(|Fl)
`
`Vmax2(lR)
`
`Vmax1(R)
`
`
`
`Vmax8(R)
`
`
`
`Vmin3(R)
`
`
`
`
`
`Vmax1(|R)
`
`vmin3(|R)
`
`Vmin2(|R)
`
`(b)
`
`
`
`
`
`Increasing saturation
`
`Vmax3(Fi)
`
`
`
`Vmin1(Fl)
`
`Vmin1(|Fl)
`
`
`(c)
`Vmin2(IR)
`vmin3(IR)
`
`Figure 9.7. Graphical representation of detected optical signals during the steady state and
`transient conditions (Stone and Briggs 1992).
`
`Example 2
`
`Figure 9.7(b) shows the representative plethysmographic waveforms for a patient
`during desaturation or decreasing saturation transient conditions for the red and
`infrared detected signals having optical pulses n = 1, 2, and 3. However, in this
`transient example, it is known at n = 1, that the actual saturation of the patient is
`very close to that during the steady state conditions in example 1. In this transient
`example, the detected values are as follows for both the red and infrared signals:
`
`
`
`158
`
`158
`
`

`

`Signal processing algorithms
`
`141
`
`_________.___——————-—-—
`tmax(1) = 1.0 s
`vmeo) = 1.012 v
`VmuxIRU) = 1.008 V
`rm (1) = 1.2 s
`VminR(l)=1.000V
`vmmIRm) = 1,000V
`1m” (2) = 2.0 s
`vmaxRa) = 1.002 V
`Vmulea) = 1.018 V
`rm (2): 2.2 s
`VminR(2) = 0.990 V
`VminIRa) = 1.010 V
`[max (3) = 3.0 s
`Vmach) = 0.992 V
`meIR(3) = 1.028 V
`1m (3): 3.2 s
`VminR(3) = 0.980 V
`meIRo) = 1.020 V
`
`Calculating the oxygen saturation ratio R at n = 1, using the detected optical
`signal provides the following
`
`R = 1n[VmaxR(1) / VminR(1)]
`1n[VmaxIR(1) / VminIR(l)]
`
`= 1n[1.012 / 1.000] / 1n[1.008 / 1.000]
`
`=1n[1.012]/1n[1.008]
`=0.012/0.008 =15.
`
`The calculated saturation ratio of 1.5 based on the detected transmittance
`corresponds to a calculated oxygen saturation of about 65 for the patient, which
`corresponds to severe hypoxia in an otherwise healthy patient. This contrasts with
`the known saturation of about 81% and demonstrates the magnitude of
`the
`underestimation of the oxygen saturation (overestimation of desaturation) due to
`the distortion in transmittance of the red and infrared light caused by transient
`conditions.
`the distorted maximum
`Applying the correction algorithm to correct
`transmittance point of the detected red signal during the transient condition:
`
`VmaxR(1)* = VmaxR(1)—[VmaxR(1)‘ VmaxR(2)] x
`
`lfmull} — 1.1111111”
`
`“11:31:12) — rIna-110)]
`
`=1.012 - [1.012 — 1.002] X [1.0 -1.2]/[1.0 — 2.0]
`=1.010.
`
`and correspondingly for the maximum transmittance of the detected infrared
`signal
`
`VmaxIR(1)* = 1.008 — [1.008 — 1.018] x [1.0 — 1.2] /[1.0 — 2.0]
`=1.010
`
`Thus, by replacing VmaxR(n) with VmaxR(n)’ and replacing VmaxIR(n) with
`VmaxIROt)a in the calculations for determining the oxygen saturation ratio R, we
`have
`>k
`
`R- 1n[VmaxR(1)
`_
`1n[VmaXIR(1)
`
`>1:
`
`/VmiuR(1)]
`/VminIR(1)]
`
`=1n[1.010/1.00]/1n[1.010/1.00]
`= 001/001
`
`= 1.0.
`
`159
`
`159
`
`

`

`
`
`
`
`142
`
`Design ()fpulse oximeters
`
`Thus, basing the saturation calculations on the corrected maximum transmittance
`values and the detected minimum transmittance values,
`the corrected R value
`corresponds to the same R for the steady state conditions and the actual oxygen
`saturation of the patient.
`
`Example 3
`
`Figure 9.7(c) shows the representative plethysmographic waveforms for a patient
`during desaturation or decreasing saturation transient conditions for the red and
`infrared detected signals having optical pulses n = l, 2 and 3. However,
`in this
`transient example, it is known that at n = 2, the actual saturation of the patient is
`very close to that during the steady state conditions in example 1. In this transient
`example, the detected values are as follows for both the red and infrared signals:
`
`VmHXRU) = 1.022 v—v',—mx1R(1)= 1.002 v
`‘ —lmax(1) = 1.0 s
`
`VminRU) = 1.008 v
`VminIRU) = 0.992 v
`1mm (1) = 1.2 s
`va(2) = 1.012 v
`meIR(2) = 1.012 v
`zmx (2) = 2.0 s
`VminRQ) = 0.998 v
`VminIR(2) = 1.002 v
`1min (2) = 2.2 s
`anxR(3) = 1.002 v
`V,mxIR(3) = 1.022 v
`1mx (3) = 3.0 s
`
`
`VminR(3) = 0.988 v1min(3)= 3.2 s VminIRO) = 1.012 v
`
`Calculating the oxygen saturation ratio R at n = 2, using the detected optical
`signal provides the following
`
`R = 1n[anxR(2) / meR(2)]
`ln[VmaXIR(2) / me IR(2)]
`
`= ln[l.012 / 0.998] / 1n[1.012 / 1.002]
`=0.01393/0.0099 :14.
`
`the calculated saturation ratio of 1.4 based on the detected transmittance
`Thus,
`corresponds to a calculated oxygen saturation of about 51% for the patient, which
`corresponds to severe hypoxia in an otherwise healthy patient. This contrasts with
`the known saturation of about 81% and demonstrates the magnitude of the
`underestimation of the oxygen saturation (overestimation of desaturation) due to
`the distortion in transmittance of the red and infrared light caused by transient
`conditions.
`
`the distorted minimum
`Applying the correction algorithm to correct
`transmittance point of the detected red signal during the transient condition, we
`find the following:
`
`Vmin R(2)* = Vmin R(2) — [Vmin R(Z) — me R(1)] X
`
`[train (2) '- rmax {I i']
`
`= 1.008 — [0.998 — 1.008] X [2.0 —1.2]/[2.2 —1.2]
`= 1.0
`
`and correspondingly for the minimum transmittance of the detected infrared
`optical signal we have:
`
`
`
`160
`
`160
`
`

`

`
`
`Signal processing algorithms
`
`143
`
`VminIR(2)* = 0.992 — [1.002 — 0.992] x 0.8
`= 1.0.
`
`Thus, by replacing VminR(n) with VminR(n)‘ and replacing Vm-mIR(n) with
`Vmian(n)' in the calculations for determining oxygen saturation ratio R we have:
`
`.
`1
`R _ lnlvmnx RUIN Vmin 12(2)
`_ —-_‘——‘_-——Ilr—
`ln[VmaxIR(2)/ VminIR(2)
`]
`
`~4<
`
`= ln[l.012/1.0]/ln[l.012/1.0]
`= 1.0.
`
`Thus, basing the saturation calculations on the corrected minimum transmittance
`values and the detected maximum transmittance values, the corrected R value
`corresponds to the same R for the steady state conditions and the actual oxygen
`saturation of the patient.
`
`9.6 ECG SYNCHRONIZATION ALGORITHMS
`
`Electrical heart activity occurs simultaneously with the heartbeat and can be
`monitored externally and characterized by the electrocardiogram waveform. The
`ECG waveform comprises a complex waveform having several components that
`correspond to electrical heart activity of which the QRS component relates to
`ventricular heart contraction. The R wave portion of the QRS component
`is
`typically the steepest wave therein having the largest amplitude and slope, and
`may be used for indicating the onset of cardiac activity. The arterial blood pulse
`flows mechanically and its appearance in any part of the body typically follows
`the R wave of the electrical heart activity by a determinable period of time. This
`fact
`is utilized in commercially available pulse oximeters to enhance their
`performance. Another advantage of
`recording ECG is
`that
`it provides a
`redundancy in calculating the heart rate from both the ECG signal and the optical
`signal to continuously monitor the patient even if one of the signals is lost (figure
`9.8).
`
`With ECG synchronization, the pulse oximeter uses the electrocardiographic
`(ECG) QRS complex as a timing indicator that the optical pulse will soon appear
`at the probe site. The R portion of the ECG signal is detected and the time delay
`by which an arterial pulse follows the R wave is determined to establish a time
`window an arterial pulse is to be expected. By using the QRS complex to time the
`oximeter’s analysis of the optical pulse signal, ECG processing synchronizes the
`analysis of oxygen saturation and pulse rate data. The established time window
`provides the oximeter with a parameter enabling the oximeter to analyze the
`blood flow only when it is likely to have a pulse present for analysis. This method
`of signal processing passes those components of the signal that are coupled to the
`ECG (i.e.,
`the peripheral pulse), while attenuating those components that are
`random with respect
`to the ECG (e.g., motion artifact or other noise in the
`signal).
`
`161
`
`161
`
`

`

`144
`
`Design of pulse oximetcrs
`
`ECG
`electrodes
`
`ECG
`amiifie
`
`R-wave
`
`
`detection
`
`
`Satu ration
`
`caiculation
`
`
`Ensemble
`Optical pulse
`
`
`processing
`averaging
`
`routine
`circuitry
`
`
`
`
`
`
`
`Digital
`display
`
`Figure 9.8. Block diagram illustrating the ECG processing components, its subcomponents and
`their relationship in an oximeter.
`
`9. 6. I Nellcorw system
`
`C—LOCK ECG synchronization enhances the signal-precessing capabilities of
`Nellcor'y systems such as the N300 pulse oximeter and the N—lUOf) multifunction
`monitor. This improves tltc quality of the optical signal in certain clinical settings
`in which the performance of a conventional pulse oximeter may deteriorate, e.g.
`when a patient is moving or has poor peripheral pulses. Consequently. C~LOCK
`signal processing extends the range of clinical situations in which pulse oximetry
`may be used. Patient movement and poor peripheral pulses present similar
`problems
`for a conventional pulse oximeter: performance may deteriorate-
`because the oximeter is unable to distinguish between the true optical pulse signal
`and background noise. C-LOCK ECG synchronization improves signal quality in
`these difficult signal-detection settings (Goodman and Core-omen 19901
`The digital optical signal is processed by the microprocessor of the Nellcor
`N—IOOO Pulse Oximeter in order
`to identify individual optical pulses and to
`compute the oxygen saturation from the ratio of maximum and minimum pulse
`levels as seen by the red wavelength compared to the pulse seen by the infrared
`wavelength.
`Noninvasive pulse oximeters process optical signals which are prone to
`motion artifacts caused by the muscle movement proximate to the probe site. The
`spurious pulses induced in the optical signals may cause the pulse oximeter to
`process the artifact waveform and provide erroneous data. This problem is
`particularly significant with infants, fetuses, or patients that do not remain still
`during monitoring. Another problem exists in circumstances where the patient is
`in poor condition and the pulse strength is very Weak. In continuously processing
`the optical data, it can be difficult to separate the true pulsatile component from
`the artifact pulses and noise because of low signal
`to noise ratio.
`Inability to
`reliably detect the pulsatiie component in the pulsatile signal may result in a lack
`of the information needed to calculate oxygen blood saturation.
`By incorporating the patient‘s heart activity into the pulse oximeter,
`problems clue to motion artifact and low signalvto—noise ratio can be solved.
`Processing of the signals that occur during a period of time when the optical
`pulses are expected to be found. increases the likelihood that the oximeter will
`process only optical waveforms that contain the pulsatile component of arterial
`blood, and will not process spurious signals.
`
`_ 1
`
`62
`
`162
`
`

`

`Signal processing algorithms
`
`145
`
`The software incorporated into the microprocessor for processing the ECG
`signals and displaying the calculated ECG pulse rate receives the digitized version
`of diagnostic ECG signal
`(DECG) and filtered ECG signals (FECG). The
`microprocessor calculates the amplitude of the ECG waveform and controls the
`AGC (automatic gain control) amplifier, so that DECG and FECG will fall within
`the voltage range limits of the electronic circuitry used to process these signals.
`The microprocessor regularly searches a status input latch at a rate of 57
`cycles per second. The output of detected R wave (DRW) sets the latch to a
`logical
`1 when the R wave is detected. Depending on the
`status,
`the
`microprocessor selects the next operation and resets the DRW latch to 0. At this
`first
`level,
`the microprocessor counts the time interval beginning from the
`detection of an R wave pulse until the occurrence of the next logical
`l at the
`status input latch. Based on this time interval,
`the pulse oximeter displays the
`pulse rate. After averaging several time intervals and establishing a regular ECG
`pulse rate, the microprocessor will change to the second level of processing.
`After
`the detection of an R wave pulse,
`the microprocessor
`separately
`analyzes the digital optical signal and correlates the period of time by which an
`optical pulse follows the detected R wave pulse to establish the time window
`during which the optical pulse is likely to occur. During this second level,
`the
`pulse oximeter just calculates and displays the time period or pulse rate between
`DRW pulses.
`The third level of processing starts after a time window has been established.
`On detecting an R wave pulse, the microprocessor activates the time window so
`that only optical
`signals detected within the time window following the
`occurrence of an R wave pulse will be evaluated for acceptance or rejection and
`for use in calculating and displaying vital measurements
`such as oxygen
`saturation, pulse flow, and pulse rate. The evaluation of a detected pulse is made
`in conjunction with a preselected confidence factor that is associated with the
`quality of the optical signals. The higher the op