WO2008035076A2

WO2008035076A2 - Blood oxygen monitor

Info

Publication number: WO2008035076A2
Application number: PCT/GB2007/003566
Authority: WO
Inventors: Geoffrey Mathews; Veronica Hickson
Original assignee: The Electrode Company Ltd
Priority date: 2006-09-20
Filing date: 2007-09-21
Publication date: 2008-03-27
Also published as: GB0618547D0; EP2063763A2; WO2008035076A3

Abstract

Calibration apparatus for use in calibrating monitoring apparatus used in determining the oxygen content of blood, the calibration apparatus including a wavelength sensitive device to receive data relating to at least one radiation wavelength received from the sensor and a processor for determining an adjustment factor calculated from the data concerning the at least one radiation wavelength of radiation which can be used to calibrate the monitoring apparatus.

Description

Blood Oxygen Monitor

The present invention is concerned with a monitor and a method for the determination of the amount of oxygen in a patient's blood and particularly, but not exclusively, with a monitor and a method for the compensation of pulse oximeter sensor errors in making an accurate determination of blood oxygen levels

The application of oxygen to patients in medical care can be lifesaving. The most common means of determining the patient's oxygen requirement is pulse oximetry. Pulse oximetry is used in medicine every day to assess the amount of oxygen in a patient's blood. The measurement is based on the amount of haemoglobin in the arteries which carry oxygen (O₂Hb) . In scientific terms this is defined as the percentage of total haemoglobin (tHb) that is saturated with oxygen (SaO₂) , but more often this value is referred to as SATs, or the SATs value. The percentage of haemoglobin, as measured using pulse oximetry (Spθ2) , which carries oxygen, is calculated using the following equation:

SpO₂ (%) = 100 ( (O₂Hb) / (O₂Hb + tHb))

A typical known pulse oximetry system consists of a sensor, such as a probe/transducer, that is applied to the patient and generates a signal and a monitor, which contains the means for processing and displaying the signal. Sensors are generally, but not always, detachable from the monitor to enable cleaning, maintenance and replacement, and can be either disposable or reusable. The disposable sensors are intended for single use only and the re-useable sensors often do not last longer than twelve months. The manufacturer's income is often derived from the sale of sensors rather than the sale of monitors. Thus, the rearward compatibility of pulse oximeter sensors, that is the ability of pulse oximeter sensors to work with older monitors, is an important feature in the pulse oximeter market.

The optical properties of blood such as the wavelength dependent attenuation, varies depending on how much oxygen it contains. The pulse oximeter sensor shines typically two beams of light through a patient's finger or earlobe for example, and a photodiode detects the transmitted and/or reflected light. The monitor then relates this transmitted or reflected signal to the oxygen saturation of the blood. However, there are many other parts of a finger (or ear lobe etc) , which will absorb light and this absorption also needs to be considered in determining the blood oxygen content.

In conventional spectrometry, the distance through a test medium though which light passes is usually a fixed value and the test medium is usually modelled as being homogenous to minimize complicating factors. In a laboratory, for example, when the haemoglobin content of blood is analysed, the blood is lysed, that is the blood cells are broken up to release the haemoglobin, and the length of the light path through the sample being analysed is known because the blood is captured in a cuvette. In medical situations where pulse oximetry is used on an individual, the length of the light path is not known and there are many constituents other than blood which interact with the light as it passes into, through and out of the tissue, e.g. skin, bone, muscle, connective tissue etc and these factors need to be taken into account if an accurate reading is to be obtained.

The problem of not knowing the length of the light path is overcome in known measuring devices by using light at two wavelengths which have different optical properties for haemoglobin and oxygenated or oxy- haemoglobin. These wavelengths are generally in the ranges of 650nm to 670nm namely the red part of the spectrum, and 880nm to 960nm, the infrared (IR) part of the spectrum, with the precise value being determined by the manufacturer. The problem of light attenuation due to constituents other than blood is overcome by recording the pulsatile component of the signal after the light has passed through the tissue and the general assumption is that the pulsatile component is derived from arterial blood. Historically, the origin of the pulsatile signal, as used in pulse oximetry, has been explained in terms of arterial pulsations. Conventionally, this has been thought to be caused by the arteries expanding during systole and therefore containing more blood during systole than during diastole. This leads to more light being absorbed when there is the maximum amount of blood in the light path during systole.

In conventional pulse oximetry, light comes from two light emitting diodes (LEDs) in the sensor, and the light is detected by a photo diode also in the sensor. The signal of interest is the pulsatile component of the signal detected by the photodiode, originating from the red and IR LEDs. The signal is usually normalised to remove the influences of variations in individual LED intensities and photo diode efficiencies. The ratio of the signal obtained from the red and IR LED during the pulsatile (ac) and non-pulsatile (dc) component of the blood flow is typically plotted against blood oxygen saturation and is known as an R-curve. The red/IR ratio (R) can be expressed as:

JR = red/IR = (acRED/dcRED) / (acIR/dcIR)

Typically the value of this ratio runs from approximately 0.4 when the blood is 100% saturated, to approximately 1 at about 80% saturation to approximately 3.4 at about 0% saturation as shown in figure 1. The precise values depend on the wavelengths used by the, manufacturer. These figures can currently only be obtained by breathdown trials with volunteers. During these trials, volunteers are subjected to various air/oxygen concentrations in order to reduce their blood oxygen concentration levels to specified amounts. The absorption of red and IR light having specified wavelengths, at various levels of blood oxygen saturation, is then measured and an R-curve is constructed. This R-curve can then be used with the known wavelengths for future determinations of blood O₂ levels.

The range of oxygenation of interest is generally from 100% down to 70%. The venous return is often at about 70% saturation, and at this level there is minimal delivery of oxygen to the living tissues. Some manufacturers have done breathdown studies below this value for special applications such as for use in special care baby units.

The light absorption of the oxygenated and deoxygenated haemoglobin of the blood is significantly wavelength-dependent. The relationship between R and SATs value strongly depends on the specific emission characteristics of the LEDs. Sensor accuracy depends on the value of R corresponding to the actual LED wavelengths used in the sensor.

The emission spectra from a LED is not a single emission at a particular wavelength, but is comprised of a band of emissions. Usually the emissions towards the centre of the band have the highest energy value, but this does not necessarily apply if the spectra is not symmetrical . In identifying the properties of an LED for use in pulse oximetry the wavelength that represents the average energy content of the emission spectra is the important value, as it is this value that will determine the size of the pulsatile signal detected by the photodiode . This average value is generally used in the manufacture of pulse oximeter sensors in the selection of LEDs. The average wavelength value of an LED spectra can also be used for predicting the effect of wavelength errors on the accuracy of pulse oximetry. Some types of LEDs are capable of producing more than one emission band at different wavelengths to the main emission. Often this secondary emission (s), as shown in figure 2, is small. However, the effect of the secondary emission (s) varies with SATs and its presence can have noticeable effects on pulse oximetry leading to erroneous measurements. This is because the average emission is no longer the central value of the main emission band but a combination of all the emission bands. The transmission of light energy as recorded by a photodiode such as in a pulse oximeter sensor will therefore be a combination of all the peaks present . Pulse oximetry is highly dependant on the way in which light is scattered by blood cells, particularly blood cells in moving blood. The effect of scattering is dependant on wavelength, the oxygenation state of the haemoglobin and also motion of the blood, particularly the pulsatile movement of blood in the arteries etc. Blood cell orientation, size, shape and flow all change the scattering of the light, in a way, which is wavelength and oxygenation specific.

To date it has not been possible to accurately model the behaviour of light in pulse oximetry due to the number of factors that can affect the passage of light through blood and tissue. This has resulted in the continued necessity for the verification of pulse oximetry systems with human volunteers. Furthermore, no current test technique for pulse oximetry based on simulation principles is able to confirm the accuracy of the match between the R-curve in the monitor and the optical characteristics of the sensor.

Pulse oximetry is a marriage of monitor and sensor technology. Monitor calibration is achieved by carrying out the clinical breathdown trials with a sensor of a particular specification. This data is stored in the monitor in software form. The installation of software in a monitor is not a process that is particularly prone to errors, and software does not usually deteriorate with age. In contrast, sensor accuracy is dependant on maintaining during production the specification of the optical components as used during the original breathdown trials. However, unlike the installation of software in a monitor, the process of installing LEDs with the correct spectral properties in a sensor is relatively difficult to maintain in mass production. Also LEDs are prone to ageing in a way that causes changes in the spectral properties of the LEDs with consequent alterations in sensor accuracy.

Assuming that the original breathdown trials were done correctly, the accuracy of a pulse oximeter system is largely dependant on the sensor in use having the correct wavelengths. The accuracy of the sensor is proportional to the wavelength error. Systematic errors resulting from inaccurate sensors and biological errors are cumulative, and if the cumulative sensor and biological error are greater than acceptable limits, then the system does not function as intended and the user will be potentially misled regarding the oxygen content of the patient's blood.

There are three main approaches to ensuring sensor accuracy during manufacture. As a first approach, many sensor manufacturers purchase and screen LEDs to a narrow range of wavelengths and program into the monitor a single calibration R-curve that corresponds to this range. As a second approach, US4621643 describes the use of a resistor-encoding scheme in which several calibration curves are programmed into pulse oximeter monitors to span a broad range of LED wavelengths needed for high-volume sensor manufacturing. As a third approach, a liberal range of wavelengths with a single curve can be used resulting in degraded accuracy performance, particularly at low oxygen saturations.

In principal, the first two approaches should ensure accurate sensors, whilst the third option is potentially cheaper in manufacturing terms. However, the implications for accuracy resulting from a mismatch between the sensor spectral properties and the R-curve are the same for all three approaches. Any sensor with wavelengths that do not accurately match the R-curve will introduce a degree of error.

In US5823950 techniques are disclosed to enable a known wavelength to be obtained by varying drive currents. This is dependant on minimal variation in the relationship between drive current and shift in LED wavelength. However, another variable with drive current is intensity. Often it is desirable to vary drive current to obtain a particular intensity. If drive currents are varied in order to obtain a particular wavelength, then a compromise on intensity has to be accepted. In addition, this document discloses the use of a special sensor with apparatus enabling the measurement of wavelength in order to enable the drive current to be adjusted to obtain the correct wavelength, the aim being to use specific selected wavelengths in measurement of blood parameters. It presumes that it is possible to alter the drive current to obtain the required selected wavelength at a usable intensity. This sensor is not rearward compatible and is expensive, and the described filter technique for measuring wavelength is not reliable if the bandwidths of the LED spectra are variable.

The present invention seeks to overcome the disadvantage in pulse oximetry of the problems associated with a mismatch of the R-curve and LED spectra by providing a monitor and an associated method for calibrating a pulse oximeter system such that substantially any pulse oximeter sensor may be used with the monitor. The monitor is arranged to calibrate itself so that one or more R-curves can be appropriately corrected or calculated which are wavelength specific for that sensor.

In accordance with this invention as seen from a first aspect there is provided a calibration apparatus for use in calibrating a monitoring apparatus used in determining the oxygen content of blood, said calibration apparatus including a wavelength sensitive device to receive data relating to at least one radiation wavelength received from said sensor and a processor for determining an adjustment factor calculated from said data concerning said at least one radiation wavelength of radiation which can be used to calibrate said monitoring apparatus.

The adjustment factor is the creation of a new R curve for the device or the recalibration of, or recalculation of data for existing R-curves to create a correction factor for the device.

Preferably, the calibration apparatus includes a processor, which validates data received from said sensor.

It is envisaged that there further comprises memory means for storing the calibration data.

Preferably, the wavelength sensitive device is calibrated each time the monitoring apparatus is used.

It is preferred that said data includes calculated R-values, absorption values of radiation at said at least one wavelength and SATs values corresponding to said R- values .

According to an embodiment of the invention the monitoring apparatus includes a calibration apparatus according the first aspect of the invention, said monitoring apparatus including a wavelength sensitive device for measuring at least one wavelength of radiation emitted by said sensor, memory means for storing data associated with parameters being sensed at said at least one wavelength, and a processing arrangement that uses data received from said sensor and makes adjustments to the operation of the monitoring apparatus based on the measured wavelength.

It is preferred that the monitoring apparatus includes a first processor of said processing arrangement which validates said data received from said sensor and a second processor of said processing arrangement which calibrates said monitoring apparatus.

Preferably, said processing arrangement validates said data received from said sensor and calibrates said monitoring apparatus .

Preferably, said processing arrangement processes any secondary emissions of said at least one LED and calibrates said monitoring apparatus.

Preferably, said received data is stored by said memory means so that adjustments can occur at a future date. It is envisaged that said wavelength sensitive device is calibrated each time said apparatus is used.

It is preferred that said sensor comprises at least one light emitting diode (LED) and at least one photodiode .

Said parameters may be the intensity of the radiation emitted by said at least one light emitting diode as measured by said at least one photodiode.

It is preferred that at least one LED comprises one LED arranged to emit red radiation and one LED arranged to emit infra-red (IR) radiation.

Preferably said at least one LED comprises one LED to emit wavelength other than red or infra-red (IR) radiation.

It is envisaged that said data comprises calculated R-values, corresponding SATs values and corresponding absorption values .

It is preferred that said adjustment of the monitoring apparatus involves the determination of the deviation of the SATs values from the expected SATs values using the R-values referenced from the stored data which correspond to the measured wavelength.

It is preferred that the adjustment further comprises the step of determining the deviation of the R- curve from the expected R-curve. Preferably said adjustment involves the construction of an R-curve which relates to the at least one measured wavelength, using the absorption values corresponding to said at least one measured wavelength.

Preferably said sensor is fitted to a patient's finger.

It is preferred that said wavelength sensitive device comprises a spectrometer or interferometer.

According to a further embodiment there is provided a method of calibrating a monitoring apparatus for use with a sensor in determining the oxygen content of blood, said method comprising the steps of:

i. measuring at least one wavelength emitted by said sensor;

ii. storing data associated with parameters being sensed at said at least one wavelength; and,

iii. validating data received from said sensor such that said monitoring apparatus is calibrated to operate at the measured wavelength.

It is envisaged that said received data is stored by said memory means for future calibrations.

A preferred arrangement relates to a pulse oximeter system comprising a monitoring apparatus as previously discussed. Embodiments of the present invention will now be described by way of example only and with reference to the accompanying drawings, in which:

Figure 1 is a graph showing a typical R-curve;

Figure 2 illustrates a secondary emission in the spectrum output from an LED;

Figure 3 is a schematic representation of the sensor and patient's finger;

Figure 4 is a schematic diagram illustrating the typical attributes of the absorption of light in passing through a patient's finger.

Figure 5 is a schematic representation of the spectral sensitive monitor;

Figure 6 is a graph showing a series of calculated R-curves for different wavelengths of the red LED;

Figure 7 is a graph illustrating the steps for calibrating a sensor in accordance with an embodiment of the invention; and,

Figure 8 is a graph illustrating the attenuation of various haemoglobin groups as a function of wavelength.

Referring to figures 3 and 5 of the drawings there is provided a sensor 100 and a monitor 300 for displaying signals obtained by the sensor 100. The sensor 100 comprises one red light emitting diode (LED) 110 and one infrared LED 120. In use, the emitted red and IR radiation is passed through a finger 130 and the intensity of the red and IR radiation transmitted there through is determined using a photodiode 140. The absorption A of the red and IR radiation is influenced by tissue 210, venous blood 220, arterial blood 230 and the pulse added volume of arterial blood 240 as illustrated in figure 4.

Conventional monitors assume that there is an accurate match between the spectral properties of the LEDs and the associated R-curve programmed in the monitor for the determination of the patients SATs value. Referring to figure 5, in the first embodiment of the present invention, a wavelength sensitive device 310, such as a spectrometer or an interferometer, is incorporated into the monitor 300, thus creating a single device, a spectral sensitive monitor. Typically, before the sensor 100 is used the wavelength sensitive device 310 is calibrated using the spectral lines of a Neon-Argon (Ne- • Ar) light source 320. The Ne-Ar source 320 is found to produce a series of spectral lines that extend over the red and IR wavelengths used in oximetry and thus serve to calibrate the device for the required measurements. The calibration period further involves connecting the sensor 100 to the monitor 300 by plugging the plug end of the sensor into the socket in the monitor and placing the patient end of the sensor 100 containing the LEDs and photodiode on a light guide 330 which extends from within the monitor 300. The spectra from the LEDs 110, 120 are fed via the light guide to the 'wavelength sensitive device 310 within the monitor 300. The spectral properties of the sensor 100, the average wavelengths of the red and IR in particular, are then used to calculate an R-curve and/or minimize any discrepancy between the spectra of the LEDs and the pre-programmed R-curve within the monitor 300 using a processor 340 within the monitor 300. After the sensor 100 has been tested it can be removed from the light guide 330 and placed on the patient's finger 130 and used to obtain data to enable the calculation of the SATs values.

This testing of the sensor 100 during the calibration period takes place on either turning on the system and or when replacing the sensor 100.

Additionally, in order to detect any instability, the calibration period for a sensor may be repeated at intervals such as 24 hours, 100 hours, and 500 hours. Such repeat tests are either set by default, on a random or timed basis, or by the user on demand.

When pre-determined calibration periods occur, an indication that a re-calibration is due is given. The user can then decide to carry it out immediately or defer it to be done later if for example, it is a critical time for the patient being monitored. The user can also request repeat tests at any time.

The data on the specific sensor in use can be stored and used by the pulse oximeter system. This data can be compared at the next test of the sensor 100. If the sensor 100 has altered by the next test, a warning is given that the sensor is unstable and the user can decide to discontinue using the sensor, or continue to use the sensor and retest at substantially the same time or a shorter time than would otherwise have been the case. Additionally, the data on each sensor 100 tested is stored by the system in memory with an identification code. This identification code is manually entered by the user, or the sensors can have an in built identification code on for example a chip or an identification resistor, so that the monitor 300 can identify which type of, and which individual sensor 100 is connected to it and therefore the data set that is to be used with the sensor 100.

To compensate for any error or deviation in the SATs value displayed on the monitor resulting from a discrepancy between the observed spectra of the LEDs in the sensor and the LED spectra associated with the stored

R-curve, the wavelength of the radiation emitted by the red 110 and IR LED 120 is first measured using the spectrometer or interferometer 310. Information is then recalled on the wavelengths expected for the LEDs 110

120. For the red 110 and IR LED 120, the error in the R- value and the corresponding SATs value displayed by the monitor, due to the difference between the expected and the measured wavelengths, can be determined at various

SATs levels, for example, 97%, 90%, 80% and 70%, as described below.

If there is no discrepancy between the spectra being emitted from a sensor and the expected spectra stored in the monitor, there will be no error in the SATs readings displayed on the pulse oximeter monitor. If a sensor with the incorrect spectra is used, the effect of this can be quantified by comparing a calculated R-curve for the erroneous sensor, with the calculated R-curve for a sensor with the correct spectra.

R-curves for combinations of red and IR wavelengths can be calculated from blood absorption data at various levels of oxygen saturation. This data is obtained by measuring light absorption of lysed blood samples containing known proportions of haemoglobin, and oxyhaemoglobin (see figure 6) .

R-curves for the purpose of quantifying the effect of errors in the red and IR spectra are calculated first for the red and then for the IR.

R-curves showing the effect of wavelength errors for the red wavelengths can be calculated by varying the red wavelength and keeping the IR constant. For example, the R-curves for 650nm, 655nm, ββOnm, and 670nm wavelengths can be calculated using the data for these wavelengths and a constant IR wavelength such as 945nm (see figure 6) .

Similarly R-curves showing the effect of wavelength errors in the IR can also be calculated. Various R-curves for values in the IR are calculated by using the appropriate data (molar extinction coefficients) for these wavelengths, while assuming a constant value for the red wavelength such as 660nm.

Accordingly, different R-values can be determined, tabulated and stored within the system 350 for various combinations of red and IR wavelengths.

The deviation in the SATs value displayed by the monitor, resulting from the deviation in the observed spectra emitted by the LEDs and the expected spectra from the LEDs, is calculated from knowledge of the wavelength error. The calculation of the deviation in the SATs valuetermined as follows with reference to figure 7:

• Consider a pulse oximeter monitor intended for use with a sensor emitting 660nm and 945nm. • R-curve a appropriate for ββOnm and 945nm is installed in the monitor.

• If a perfect pulse oximeter sensor emitting ββOnm and 945nm is used, the red/IR ratio at 2 will be generated. • Then the SATs value at 1 on the y-axis is obtained by reading across from where the value of R at 2 intersects R-curve a.

• If a different sensor is used emitting 665nm and 945nm, then R-curve b would be appropriate. • Using R-curve b a different value of R would be obtained as at 4, which would intersect R-curve b at 3, again resulting in a SATs value at 1 on the y-axis .

• Thus the same SATs values for the 2 different sensors are obtained and there is no error in SATs reading.

• However in practice the wavelengths emitted from sensors are frequently not a perfect match with the R-curve installed in the monitor. • If the second sensor of wavelengths 665nm and 945 is used with R-curve a then the two sensors do not read the same, and there is an error in the SATs reading.

• The R-value generated by the second sensor of wavelengths 665nm and 945nm is read of the x-axis at 4.

• This R-value of 4 intersects R-curve a at 5. • This leads to a SATs value at 6 on the y-axis being obtained.

• Thus in this example where the second sensor has a 5nm error in the red with respect to R-curve a installed in the pulse oximeter monitor there is an erroneous high SATs reading.

• If no correction takes place then the results displayed on the pulse oximeter monitor will be erroneously high. • Furthermore due to the divergence of the two R- curves a and b as oxygen saturation falls, any error in the SATs reading displayed on the monitor gets larger as the oxygen content of the patient's blood falls. • The results obtained by comparing calculated R- curves such as a and b, underestimate the sensitivity of pulse oximetry to wavelength errors, and the results obtained are multiplied by a clinically derived factor to bring the results into line with clinical observations.

• Thus, the error introduced into the pulse oximetry system by a sensor having the incorrect wavelengths can be predicted at any level of oxygen saturation in the patient by calculating the R-curve or looking up previously stored data for the erroneous sensor, and comparing that R- curve with the expected or perfect R-curve.

Thus, it can be seen that the error in the measurement of the SATs levels as attributed to variations in the red and IR wavelength from the expected wavelengths, can separately be deduced at for example 97%, 90%, 80% and 70% SATs levels. The display on the monitor can then be corrected accordingly.

However, the R-values calculated for the measured wavelengths of the red 110 and IR LED 120, are only accurate to the extent that the measured wavelengths correspond to those wavelengths that have been tabulated with the corresponding R-values and SATs levels. It is to be further understood that while only four discrete SATs values were chosen to determine the error associated with the LED set, a series of SATs values corresponding to the series of R-curves 400 for each combination of tabulated wavelength could have also been employed. Another limitation of this technique is that the R-curves are calculated from absorption coefficients. Calculated R- curves are known to vary from empirically derived R- curves. While in this application the calculated R-curves are multiplied by a clinically derived factor to bring them into line with empirical data, there is still a compromise. However if the calculated R-curves are substituted for R-curves derived from normalised data (as described below) , then this compromise would be substantially eliminated.

Accordingly, in a second and preferred method of minimising the errors in SATs values, as appropriate for a particular LED set 110, 120 within the sensor 100, the absorption of the red and IR radiation is determined. As previously stated, the relationship between wavelength and absorption varies with oxygen content of the blood. Similarly, the relationship between the size of the absorption, as measured by the photodiode 140 in the oximeter sensor 100, also varies with the oxygen content of the blood, and also in a predictable manner. The data to enable this prediction is obtained by doing breathdown trials with hypoxic volunteers and collecting data on the amplitude of the pulsatile absorption peak for levels of oxygenation between 100% and 70%, for wavelengths in the range of interest. For conventional pulse oximetry used to measure arterial oxygen, this would be the 650nm to 670nm and the 880nm to 960nm ranges. For example if data were gathered in Iran steps, it would then be possible to predict the size of the normalised pulsatile peaks in the same lnm steps. Then it follows that it is possible to predict the red over IR ratio, namely the R-value, from knowledge of the wavelengths and the size of the absorption peaks that they would produce.

From knowledge of the wavelengths emitted from each LED 110 120 it is possible to predict the size of the normalised absorption peak 200 that would be obtained at those wavelengths at various oxygenation values. When the size of the absorption peaks 200 for red and IR are known, then the R-value can be predicted at various oxygenation values and then it becomes possible to reconstruct the R-curve. Examples of the peaks are shown as 200 in figure 4.

In applying this technique, a suitable light source such as tungsten and/or halogen capable of producing radiation over a wavelength range between, for example 550nm and llOOran, is set up with a spectrometer that can sample the wavelength range at approximately 200 times/second. A host of volunteers are used to acquire the initial data set. For example each volunteer is desaturated by means of breathing various air/oxygen percentages. The SATs levels are typically monitored by an accurate pulse oximeter or by use of a CO-oximeter.

When the SATs levels of a particular volunteer are steady, the light absorption properties of the tissue are plotted in lnm intervals over the spectral range of the light source. Accordingly, the amplitude of the pulsatile component 200 of the heartbeat can be determined for various SATs levels, over the spectral range. Any variation in the size of the pulsatile component 200 due to variations in the optics can be removed by normalising the signal.

Within a single data set, i.e. for a given SATs level, the pulsatile light intensity 240 will vary with wavelength. Between data sets, the pulsatile light intensity 240 for a specific wavelength will vary dependant on the amount of tissue perfusion and other biological and other parameters.

The influence of tissue perfusion etc. can be removed by normalising the data. The combined data can then be tabulated so that the normalised pulse amplitude at any combination of wavelength and SATs can be read off. Thus, by measuring the red and IR wavelength of the LEDs 110 120 within the sensor 100, the R-curve for that particular combination of wavelengths, and indeed for any other combination of wavelengths within the experimental spectral bandwidth, can be calculated using the tabulated absorption data.

Conventional two wavelength pulse oximetry is unable to distinguish between oxy-haemoglobin, and carboxy- haemoglobin. This is because the attenuation for carboxy- haemoglobin approaches that for oxy-haemoglobin in the 660nm range (in figure 8) . In practice this means that a pulse oximeter system can indicate that a patient is well oxygenated when in reality the patients blood is seriously contaminated with carbon monoxide. If the lack of oxygen in the patient's blood is not detected, the safety of the patient could be compromised. However, if additional wavelengths such as 680nm is available with different attenuation to oxy-haemoglobin, as with the above-described method, then additional constituents such as dysfunctional haemoglobins for example carboxyhaemoglobin (COHb) , methaemoglobin (MetHb) can be additionally calculated as for example described in US5823950. Furthermore, additional wavelengths may also be used in adaptive noise cancellation, for example as in US5431170. The use of the correct wavelengths in a sensor used with a system for identifying any of these additional blood constituents must match the wavelength selected in the system design. The presence of any secondary emission (s) with any of the selected wavelengths will affect the accuracy of measuring these additional parameters as for Spθ₂. The invention described above can also be applied to such multi-wavelength systems such as monitoring and testing devices.

It is to be appreciated that while the minimising of errors has been described with reference to the wavelength sensitive device 310 being incorporated within the monitor 300, this is not essential. In a second embodiment, the wavelength sensitive device 310 could also be incorporated within an interface device (not shown) separate from the monitor 300. This device would then enable a conventional type sensor to be used with a conventional monitor via the interface device and thus obtain data free from any error caused by the mismatch of sensor spectra and R-curve, regardless of the sensor spectra and R-curve, provided the sensor and monitor were recognisable to the interface device.

In one aspect any manufacturer' s sensor could be used with any monitor. Alternatively only sensors specifically made for connecting to the interface would be recognised. In this aspect, the recognition of the sensor 100 to the interface device could be a chip in the sensor plug that contains the recognition data, or an identification resistor. Consequently, sensors from alternative manufacturers, which are not recognised by the interface, would not function. These interface sensors would have less rigorous specifications for the LEDs than present pulse oximeter sensor thus making them cheaper to manufacture.

In a third embodiment, the wavelength sensitive device 310 could be a device intended for testing the combined accuracy of a pulse oximeter sensor and monitor. In this embodiment the sensor is tested on a sensor testing device. A calibration resistor value for example may determine which of a number of R-curves in the monitor may be used. This communication could be via an infrared link or cable. The calibrator in turn is connected to the monitor via the socket that in normal mode of operation of the monitor would take the sensor. The spectral properties of the sensor, in particular the average wavelength of each LED spectra, are measured. Then using the calculated or normalised data as described above, the amplitude of the normalised pulsatile signal at various SATs levels that would be expected if the light had been transmitted through tissue containing pulsatile arterial blood such as a patient's finger, is calculated within the calibrator unit. If the sensor testing device detects any secondary emissions from either or both of the LEDs, then the user is either warned of the presence of secondary emissions and/or the effect of the secondary emissions on the amplitude of the pulsatile signal is calculated as described above. The user has the option to test the monitor at various SATs levels. An electrical signal determined by the values of the above-calculated pulsatile signal amplitudes appropriate to a SATs value selected by the user is fed to the monitor in place of the signal which would normally come from the photo diode in the pulse oximeter sensor. Also any calibration information originating from the sensor as described above is also relayed to the monitor. Thus if the spectral properties of the sensor cause the sensor to have a positive bias (namely a high reading) , then when this predicted sensor signal is fed to the monitor in an electrical form, the SATs value on the monitor will be higher then if a sensor with no error had been used. Similarly if the spectral properties of the sensor result in a low bias (namely a low reading) the predicted signal from this will result in low readings on the monitor.

The accuracy of the monitor can be tested by using spectral data representative of a sensor with no error. Data defining the correct wavelengths for the sensor under test are sent from the sensor tester to the calibrator. Similarly to the above pulsatile amplitudes are calculated from look up tables in the calibrator and an electrical signal is fed to the monitor. Then any deviation in the SATs value displayed on the monitor and the SATs test value selected by the user is due to error based in the monitor.

An advantage of this technique is that knowledge of the R-curve in the monitor is not necessary. A further advantage of this technique is that the user is provided with traceable calibration. In a fourth embodiment a wavelength sensitive device could be used to calculate the ideal R curve to be used with sensor, or a number of sensors of a specified spectral output. The spectral properties of the sensor, the average wavelength of each LED spectra in particular are measured. Alternatively LEDs could be selected with specific known wavelengths. Then by use of normalised or calculated wavelength related data, R curve values over a range of saturation values can be established. This information could then be used by a manufacturer of pulse oximeter systems as a means of establishing the R curve for a monitor for use with a sensor or group of sensors of a specified spectral output, without having to revert to breathdown trials with human volunteers . In this application the desired output from the processor associated with the wavelength sensitive device, the R curve, could be downloaded for example into a spread sheet on a computer, or a chip/processor in a pulse oximeter system etc.

From the foregoing it should be apparent that the focus of the invention is to provide:

1. The use of calculated and or normalised wavelength related data to correct errors in pulse oximetry and to test the accuracy of pulse oximeter systems. 2. A means for predicting normalised pulsatile absorption peaks corresponding to pulsing blood at particular wavelengths of radiation when passed through tissue containing pulsing blood. 3. A means of determining the R curve for a pulse oximeter sensor from knowledge of the emitted wavelengths .

4. A means testing a pulse oximeter system by feeding electrical signals proportional to calculated pulsatile signals appropriate to various SATs values and wavelengths of the sensor.

Although the invention is described with reference to individual embodiments , it is to be understood that the invention encompasses combinations of any of the embodiments described. Further, it is to be understood that the invention is covered not only by specific features described, but also by functional equivalents as covered by the claims .

Claims

1. Calibration apparatus for use in calibrating monitoring apparatus used in determining the oxygen content of blood, said calibration apparatus including a wavelength sensitive device to receive data relating to at least one radiation wavelength received from said sensor and a processor for determining an adjustment factor calculated from said data concerning said at least one radiation wavelength of radiation which can be used to calibrate said monitoring apparatus.

2. Calibration apparatus according to claim 1 wherein said processor validates data received from said sensor.

3. Calibration apparatus according to claim 1 or 2, further comprising memory means for storing the calibration data.

4. Calibration apparatus according to any preceding claim, wherein said wavelength sensitive device is calibrated each time the monitoring apparatus is used.

5. Calibration apparatus according to any preceding claim, wherein said data includes calculated R-values, absorption values of radiation at said at least one wavelength and SATs values corresponding to said R- values .

6. Monitoring apparatus including a calibration apparatus according to any preceding said monitoring apparatus including a wavelength sensitive device for measuring at least one wavelength of radiation emitted by said sensor, memory means for storing data associated with parameters being sensed at said at least one wavelength, and a processing arrangement that uses data received from said sensor and makes adjustments to the operation of the monitoring apparatus based on the measured wavelength.

7. Monitoring apparatus according to claim 6, wherein a first processor of said processing arrangement validates said data received from said sensor and a second processor of said processing arrangement calibrates said monitoring apparatus .

8. Monitoring apparatus according to claim 6, wherein said processing arrangement validates said data received from said sensor and calibrates said monitoring apparatus.

9. Monitoring apparatus according to claim 6, wherein said wavelength sensitive device indicates the presence of any secondary emissions from said sensor.

10. Monitoring apparatus according to claim 6, wherein said processing arrangement processes any secondary emission of said sensor and calibrates said monitoring apparatus.

11. Monitoring apparatus according to any of claims 6 to 10, wherein said received data is stored by said memory means so that adjustments can occur at a future date.

12. Monitoring apparatus according to any preceding claim, wherein said wavelength sensitive device is calibrated each time said apparatus is used.

13. Monitoring apparatus according to any preceding claim, wherein said sensor comprises at least one light emitting diode (LED) and at least one photodiode.

14. Monitoring apparatus according to claim 6, wherein said parameters are the intensity of radiation emitted by said at least one light emitting diode as measured by said at least one photodiode.

15. Monitoring apparatus according to claim 6, wherein said at least one LED comprises one LED arranged to emit red radiation and one LED arranged to emit infrared (IR) radiation.

16. Monitoring apparatus according to claim 6, wherein said at least one LED comprises one LED to emit wavelength other than red or infrared (IR) radiation.

17. Monitoring apparatus according to any preceding claim, wherein said data comprises calculated R-values, corresponding SATs values and corresponding absorption values .

18. Monitoring apparatus according to claim 17 as appended to claim 11, wherein said adjustment of the monitoring apparatus involves the determination of the deviation of the SATs values from the expected SATs values using the R-values referenced from the stored data which correspond to the measured wavelength.

19. Monitoring apparatus according to claim 18, further comprising the step of determining the deviation of the R-curve from the expected R-curve.

20. Monitoring apparatus according to claim 17, wherein said adjustment involves the construction of an R-curve which relates to the at least one measured wavelength, using the absorption values corresponding to said at least one measured wavelength.

21. Monitoring apparatus according to any preceding claim, wherein said sensor is fitted to a patient's finger.

22. Monitoring apparatus according to any preceding claim, wherein said wavelength sensitive device comprises a spectrometer or interferometer.

23. A method of calibrating a monitoring apparatus for use with a sensor in determining the oxygen content of blood, said method comprising the steps of:

i. measuring at least one wavelength emitted by said sensor;

24. A method according to claim 23, wherein said received data is stored by said memory means for future calibrations .

25. A pulse oximeter system comprising a monitoring apparatus as claimed in any of claims 6 to 22.

26. A pulse oximeter system for determining the oxygen content of blood using the method steps of claims 23 or 24.

27. The use of wavelength related data to produce calibration data to correct errors in a pulse oximeter system according to claim 25 or 26.

28. The use of wavelength related data to produce calibration data to test the accuracy of a pulse oximeter system according to claim 25 or 26.

29. Monitoring apparatus substantially as herein described with reference to the accompanying drawings.