Sensing Device
The present invention relates to a device and associated method for sensing a concentration of a substance in a sample.
There are many devices known that are designed to perform a test to check or measure concentration of a constituent component of a fluid. Such devices may be used, for example, in chemical or food processing, or for monitoring or diagnosis of body fluids in health related applications. The majority of existing devices require the fluid under test to have direct physical contact with a sensor or sensing element. This gives rise to possibilities of contamination, or, in health-related applications, an increased risk of infection. Most of these instruments also have the disadvantage of operating over a limited concentration range, or are limited to use for specific substances only. Some instruments rely on human judgement for identifying substances, for example by comparing results against colour-coded charts, and are thus prone to error.
It is an object of the present invention to provide a sensing device and method that obviate or mitigate one or more of the aforementioned disadvantages. It is a further object to provide a sensing device which is portable and has a simple design and construction, so as to be relatively inexpensive and usable in a wide range of applications.
According to a first aspect of the present invention there is provided a device for determining the concentration of a predetermined substance in a sample, the device comprising :-
(i) a light source adjustable to emit sequentially light at a plurality of wavelengths or a plurality of light sources each capable of emitting light at mutually different wavelengths, and
(ii) a sensor which generates an output signal which is dependent on the intensity of light incident thereon,
wherein the at least one light source is arranged to direct light towards the sensor via a sample and wherein the device is configured to irradiate the sample sequentially with light of a first and then a second wavelength, optionally followed by additional irradiations at different wavelengths, each irradiation generating an output signal at the sensor, the concentration of the substance in the sample being determined from the set of output signals.
According to a second aspect of the present invention there is provided a method of measuring the concentration of a substance in a sample , the method comprising:
(i) irradiating the sample with light having a first wavelength,
(ii) measuring the intensity of light transmitted by the sample at a sensor location
(iii) repeating at least once steps (i) and (ii) at a different wavelength whereby to generate a set of measured intensities,
(iv) determining the concentration of the substance in the sample from the set of measured intensities using information known about sets of intensities derived from samples having known concentrations of the substance.
Reference to "light" is intended to include not only visible light but also electromagnetic radiation having wavelengths above or below the visible spectrum. Furthermore, although the light source(s) may be capable of emitting monochromatic light, the invention is not limited thereto. Thus, reference to a wavelength is intended to include a band of wavelengths, the only restriction being that each irradiation of the sample must be at a different wavelength or wavelength band.
Where a single light source is used, it may be one which is inherently adjustable, i.e. the light source is tuneable to emit light of a required wavelength. Alternatively, the light source may be externally adjustable. Thus, in one embodiment, the light source is one which emits light over a fixed spectrum, a plurality of interchangeable filters being provided in the light path between the light source and the sample, each filter transmitting different wavelengths.
Preferably, the device comprises a plurality of light sources. More preferably the light sources are high intensity light-emitting diodes (LED) or laser light such as laser diodes.
Any light detector device (CCD, Photodiode, Photodiode array) can be used, (preferably the TAOS TCS230)
The output signal may be, for example, an electrical current or voltage. However, in a preferred embodiment, the key characteristic of the output signal is frequency, the frequency of the output signal corresponding to the intensity of light incident on the sensor. The sensor may comprise a light to frequency converter. Alternatively, the sensor may comprise a photodiode, operable for generating a current in dependence on light incident thereon and a current to frequency converter for generating the output signal.
It will be appreciated that the photodiode will have a response characteristic, generating a current in dependence on the wavelength and intensity of incident light. The frequency of each output signal will depend on the current generated by the photodiode, which in turn depends on the intensity and wavelength of the incident light. Thus, the output signal frequencies do not generally correspond to the incident light frequencies (i.e. it is not generally possible to determine the wavelength or colour of the incident light directly from the output signal). Nevertheless, for a plurality of known light sources (or a single known light source adjustable in a known manner), it is possible to determine an output signal "optical signature" (OS) (derived from the sensor output signals) for different (known) concentrations of the substance. By comparing the output signals obtained from a sample under test (from which the OS of the substance in the sample can be derived) with the known OS's, it is possible to determine the concentration of the substance in the test sample.
The sample may be in the form of a solid, liquid or gas. With a solid, some of the incident light will be absorbed and some will be reflected subject to the nature of the solid's substances. Thus, depending on the substance concentrations, the intensity of
the reflected light can give an "Optical Signature" OS that can be used to measure the concentration of that substance. With a gas the same principle applies provided that a constant gas condition is maintained during monitoring - i.e. gas pressure kept constant, gas flow rate kept constant etc. The device enables continuous monitoring of a static or moving fluid as long as the flow rate/pressure are constant.
It is an advantage that the device may be used to test for the presence of a wide range of component substances in a sample, provided that the OS's for each substance is known over the required concentration range in the sample environment.
It is a further advantage that the light to frequency conversion provides signals which are convenient for direct transistor-transistor logic (TTL) connection to a microcontroller. The use of a frequency signal avoids any difficulties associated with scaling of signals, and reduces signal interference problems. It is also found to produce, for each OS, a distinct output having good accuracy and repeatability and avoids the need for further signal processing or signal analysis.
It will be appreciated that, the exact wavelengths and number of irradiations required to generate a suitable OS will vary according to the substance whose concentration is to be determined and the nature of the sample. Thus, the device may be configured for irradiation at different sets of wavelengths according to user supplied information on the substance to be determined and the nature of the sample.
In a preferred embodiment, the device further comprises a processor for receiving the output signals, and for determining the concentration of the substance in the sample. The processor may be programmable and have a memory for storing program instructions and data. The data may include the OS's for known concentrations of the substance, the processor being configured to perform a comparison of the OS for the sample under test with the stored OS's to determine the concentration of the substance on an interpolation basis. Alternatively, the processor may comprise a non-linear algorithm for determining the concentration of the substance in the test sample. For example, the algorithm may be a trained neural network (TNN) algorithm. It will be
understood that the memory requirements for a trained neural network are far less demanding than the storage of multiple OS's. In addition, interpolation may not provide accurate results, since the intensity changes induced by the substance at each irradiation wavelength may not necessarily be linear with respect to concentration.
Preferably, the device further comprises a sample holder positioned in the light path between the light source(s) and the sensor, into which the sample can be placed. In one embodiment the sensor and the light source(s) are mutually spaced, the sample being placed there between such that the light incident on the sensor has passed through the sample. In alternative embodiments, however, the light source(s) and sensor are arranged such that light incident on the sensor is reflected from the sample.
In the case of a liquid sample, the sample holder is conveniently a transparent vial made from, for example optical glass or Perspex.
In an alternative embodiment, a sample flowpath is provided through the device so that the concentration of the substance can be monitored on a continuous or intermittent basis without the need for obtaining multiple discrete samples. Such an embodiment is particularly useful for monitoring water quality (eg. outflow from sewage plants, or effluent flows from industrial processes).
Preferably, the device comprises a light proof housing, such that ambient light does not affect the response of the sensor.
In a preferred embodiment the sample is a biological sample (eg. blood or urine) and the substance whose concentration is to be determined is glucose or protein. A particularly useful application of the invention is for monitoring blood sugar levels in diabetics.
Other applications of the invention include the analysis of foodstuffs for concentrations of protein, sugar or other substances. This application is particularly suitable for use as an aid to dietary control.
Embodiments of the invention will now be described by way of example only with reference to the accompanying drawings in which:
Figure Ia is a schematic representation of a first embodiment of a device according to the present invention;
Figure Ib is a schematic representation of a second embodiment of a device according to the present invention;
Figure 2 is a schematic representation of a sensor for use in the device of Figures Ia or
Ib;
Figure 3a is a graph showing the spectral response of a sensor of the type shown in
Figure 2, used in experimental testing;
Figure 3b is a graph showing the spectral response of a second embodiment of sensor of the type shown in Figure 2, used in experimental testing;
Figure 4 is a graph showing results of experiments using the sensor of Figures 2 and 3; and
Figure 5 is a flow chart depicting the process of determining a concentration using the device of Figure Ia;
Figure 6 is a schematic representation of a third embodiment of a device according to the present invention;
Figure 7 is a schematic representation of part of the device of Figure 6; and
Figure 8 is a a flow chart depicting the process of determining a concentration using the device of Figure 6.
Referring to Figure Ia, a measuring device 10 has a substantially lightproof enclosure 12. Inside the enclosure 12 is a set (in this case four) of selectable light sources in the form of high intensity light emitting diodes (LEDs) 14a, 14b, 14c, 14d, and a switch 15 for selectably connecting one of the LEDs to a power supply (not shown). Instead of LEDs lasers such as a laser diode can be used. Also inside the enclosure 12 is a liquid sample 16 contained within a sample holder 18, a sensor 20 and a controller 22 having a memory 23. An interface 24 allows for power to be provided from an external supply
(not shown), and data signals to be transmitted between the controller and an external processor or instrument, such as an oscilloscope or digital data display. In an alternative embodiment, the interface 24 includes a display viewable on the exterior of the enclosure 12.
The light sources 14a, 14b, 14c, 14d each emit light at a different wavelength or range of wavelengths. The sample holder 16 is a phial or sample bottle which is substantially transparent to light at the wavelengths emitted by the light sources 14a, 14b, 14c, 14d.
The sensor 20 is a TSL230 or TCS230 light to frequency convertor (available from TAOS Inc.) that will be described in more detail below with reference to Figure 2. The sensor 20 includes a photodiode element or array 26 and provides a variable frequency output signal. The controller 22 is a high performance 8-bit microcontroller having an externally programmable read-only memory (EPROM). Alternatively a 32 bit microprocessor with external RAM or memory card interface can be used. An algorithm is programmed into the memory 23 of the controller 22 for analysing output signals from the sensor 20 and calculating a concentration (as will be described in detail below).
In use, as shown in Figure Ia, the switch 15 is positioned to connect the LED 14b to the power supply, so that the LED 14b emits light which passes through the sample 16 in the sample holder 18. The light is modified by absorption, refraction or reflection as it passes through the sample 16. The modified light is incident on the photodiode 26. The sensor 20 produces an output signal which is read by the controller 22 and stored in the memory 23.
The controller 22 controls operation of the switch 15 so that it is positioned to provide power to each of the other LEDs 14a, 14c 14d in turn. As the light emitted by each LED is of a different wavelength, the degree of modification of the light by the sample 16 will be different. This means that the sensor 20 will produce a different output signal for each LED, and store the outputs in the memory 23. Alternatively, the switch 15 may be operated manually.
The controller 22 analyses the output signals stored in the memory 23 and calculates a concentration value. To perform the calculation, the controller 22 uses an algorithm programmed into the memory 23. The calculated concentration value is provided as an output to the interface 24.
Figure Ib shows a variation of the embodiment of Figure Ia (equivalent components are given the same reference numeral as Figure Ia) in which, instead of a set of LEDs, a single LED 14 is provided. The LED 14 emits light, covering a wide range of wavelengths. For example the LED 14 may be a white LED covering substantially the entire visible spectrum. A set of selectable filters 27 (only one shown) is provided between the LED 14 and the sample 16. Tlie filters 27 are arranged so that any one filter of the set is positionable between the LED 14 and the sample 16. Each filter of the set passes light of a different wavelength. A mechanism (not shown) is provided for positioning the filters 27 in response to instructions received from the controller 22. In a further variation (not shown) the filters are mounted on an indexed carousel which is manually operable to select each of the filters in turn.
Use of the device of Figure Ib is similar to that of Figure Ia, except that instead of changing the light source to obtain different output signals, the filter 27 is changed.
Referring to Figure 2, the sensor 20 is a TSL230 in the form of an 8-pin dual inline package, or TCS230 The sensor 20 includes in addition to the photodiode or photodiode array 26 a current to frequency converter 28. The photodiode 26 generates a current in dependence on the light incident thereon. The sensitivity of the photodiode can be set by high or low settings on a pair of input pins SO, Sl, by varying the active area of the photodiode surface. The current generated by the photodiode may be converted to a frequency output by the converter 28 using a scale factor which can be set by high or low settings on a pair of pins S2, S3. The sensor generates an output signal in the form of a pulse train or square wave having a frequency which is directly proportional to the current generated by the photodiode, which in turn depends on the incident light intensity. Instead of setting a scale factor the output can be automatically scaled by the microcontroller 22 as can the photodiode sensitivity.
When the TCS230 is used the sensor 20 also incorporates built in RGB filters enabling specific frequency bands to be analysed independently. This can be used to look at other effects such as phase shift and fluorescence, enabling detailed, accurate measurement. This may also be done by placing optical (coloured) filters in front of the detector or source 14. The TCS230 is an 8 pin SOIC package with an 8x8 photodiode array. SO and Sl control frequency scaling. S2 and S3 select the photodiodes behind each filter (Red, Blue, Green and Clear. When used with the clear filter the chip is essentially the same as the TLS230 ). Using the filters helps eliminate the effect of ambient light as well as increasing the overall accuracy.
The photodiode 26 has a response characteristic, which is dependent on the wavelength of the incident light. The response for the TSL230 is shown in Figure 3 a, from which it can be seen that the exemplary photodiode has a peak response at approximately 780 nm. When light from one of the LEDs 14, 14a, 14b, 14c, 14d is incident on the photodiode 26 without any sample 16 present, the sensor 20 will generate an output signal. This output signal is stored in the memory 23 of the controller 22 to be used as a reference against which subsequent output signals are compared. This allows calibration of the device to take into account any intensity "variation from the LEDs 14a, 14b, 14c, 14d.
The response for the TCS230 light to frequency converter is shown in Figure 3b.
For each LED 14a, 14b, 14c, 14d, or for each filter 27 when used together with the LED 14, a known concentration of a substance in a sample 16 in the sample holder 18 will attenuate the light by a certain amount. Thus, for a specific set of LEDs or filters a unique "Optical Signature" (OS) can be derived for any substance. Data relating to the OS can be stored in the memory 23 of the processor 22, and used by the processor 22, in conjunction with the algorithm programmed therein, to determine the concentration of the substance in an unknown sample. It will be understood that if the device is to be used for different sample types (eg. blood, milk, alcohol, urine etc.) then a different algorithm may be required for each substance to be measured in each sample type.
Thus, the device is also programmed to accept user input on the substance to be analysed and the type of sample being analysed.
Referring to Figure 4, there is shown the results of an experiment to test the output frequency response of the sensor 20 using a TSL230 to samples containing different concentrations of glucose in an aqueous solution. Three different coloured LEDs were used as light sources (using the embodiment of Figure Ia). Line A in Figure 4 shows the response to light from a red LED (wavelength 644nm), line B to light from a white LED (470-750nm) and line C to light from a blue LED (430nm). It can be seen that line A has the highest frequency output and the greatest change in frequency with increasing glucose concentration. Lines B and C have low and unvarying responses. Thus, the OS for glucose concentrations has a high response in the red light (longer wavelength) region of the visible spectrum, but low responses in other regions.
The above discussion relates to analysing single component solutions. For many applications multiple substances will be present in the sample and it may not be possible or desirable to isolate or separate the substance of interest from other substances in the sample. Accordingly, the algorithm must enable the controller to recognise the OS of the substance that may be present in the sample, so as to determine the concentration of the substance based on a single set of output signals. In an exemplary embodiment, this is done by means of an artificial neural network program. A neural network operates by creating "connections" between processing elements (PEs) (these are the computer equivalent of neurones in the human brain), and weighting these connections with respect to a sample of input data. The weighting of the connections represents the relative strength, or importance of the connection on the output. Each PE operates on the data fed into it by applying a transfer function. Examples of transfer functions include linear, step, sigmoidal and gaussian functions. The organisation of the network, PEs and the connection weights determine the output.
The neural network may be organised in layers, having an input layer, a number of intermediate layers, and an output layer. The data to be analysed is fed into the PEs in the input layer and then via the network connections to PEs in the next, intermediate
layer, and so on through to the output layer. The network "learns", or adapts by adjusting the weighting of the connections based on "training" data supplied. A large database of training data improves the effectiveness of the network.
In the present case, the neural network is trained by providing samples of known concentrations of different substances, and submitting these to a test in the device 10. The output signals from this test are provided as training input data to the neural network, which is thereby able to "learn" how to distinguish between the OS of one substance and another when analysing a sample. The controller 22 uses the neural network as the algorithm for determining the concentration of the substance in the test sample. When training the neural network, the known concentrations of all of the main constituents of the solution that affect the sensing wavelength/s are provided to the controller 22.
Figure 5 shows the steps of a method for measuring the concentration of a substance in a fluid using the device of Figure Ia.
At step 50 a sample 16 of fluid to be tested is inserted into the sample holder 18 and placed in the device 10. As previously mentioned, the method may include the initial steps prior to step 50 of establishing (from user input) the nature of the sample and th.e substance whose concentration is to be determined. Measurement is initiated at step 52 by moving switch 15 to provide power to the first LED 14a. Light from the LED pas ses through the sample 16 and causes an output signal to be generated by the sensor 20. ^At step 54 the sensor output signal is received by the controller 22, and at step 56 the output signal value is stored in the memory 23.
At step 58 a determination is made as to whether all the LEDs 14a, 14b, 14c, 14d have been used. If not, at step 60 the controller causes the switch 15 to connect power to the next LED5 and steps 54 to 56 are repeated. Once all the LEDs have been used, at step 62 the controller initiates the algorithm for calculation of the concentration of the substance. At step 64 the output signal values are read from the memory into the
algorithm input, and at step 66 the concentration is calculated. Finally, at step 68, the concentration value is output to the display interface 24.
For the embodiment of the device shown in Figure Ib, the same method steps shown in Figure 5 are applied, except that instead of connecting an LED at steps 52 and 60, a filter is placed in the light path between the LED 14 and the sensor 20. At step 58 the determination is made as to whether all filters have been used.
Referring to Figure 6, there is shown a third embodiment of measuring device 110. Components which are substantially similar or have similar functions to corresponding components of device 10 are given the same reference number but preceded by a 1.
Device 110 has a substantially lightproof enclosure 112. The enclosure includes an LCD display 125, a PC interface 127 (USB or serial), a keypad 129 and a sample holder 118. the PC interface 127 can be configured to perform all of the same functions as interface 24.
The internal make up of the device is shown in Figure 7. Device 110 includes a main circuit board 128 containing the microcontroller/microprocessor 122 and drive circuitry, a battery compartment 130, Multiple/Multi Wavelength Emitter/s (LED/Laser diode) 114, sensor 120, PC interface 127, and a secondary detector 132 offset at an angle to measure the amount of light scattered/refracted by any substance in solution.
The device 110 is controlled via the keypad 129, the desired substance and/or solution components are selected and the sample 116 is inserted in the sample holder 118. The controller 122 then automatically selects the required wavelengths and sequentially tests the sample 113 using emitter 114 storing the results in the memory. There is no need for manual switching.
A typical sample time for any sample 116 is lms, allowing for discrete or continuous measurements. The device 110 can also be interfaced with a PC via USB or Serial
(RS232). This will allow control and device management/calibration from the PC, as well as expanding the continuous monitoring capability.
In a similar way the device can be calibrated and ANN trained for new samples/substances 116.
Figure 8 shows the steps of a method for measuring the concentration of a substance in a fluid using the device of Figure 6.
At step 150 a sample 116 of fluid to be tested is inserted into the sample holder 118 and placed in the device 110. As previously mentioned, the method may include the initial steps prior to step 50 of establishing (from user input) the nature of the sample and the substance whose concentration is to be determined. If so the desired substance is entered into keypad 129 at step 151.
At step 153 measurement is initiated pulsing the correct wavelengths form emitters 114 and measuring output signal generated by the sensor 120. At step 156 the sensor output signal is received by the controller 122, and stored in the memory 123.
At step 158 a determination is made as to whether all the sources of emitter(s) 114 have been used. If not, at step 160 the controller automatically moves on to the next source or filter of emitter (s) 114, and steps 153 to 156 are repeated.
Once all the sources have been used, at step 162 the controller initiates the algorithm for calculation of the concentration of the substance ANN . At step 164 the measured stored values are read from the memory into the algorithm input along with any reference data and suitable weightings at step 165. The concentration is then calculated.
At step 167 it is then determined if a PC is connected to PC interface 127. If no, at step 168, the concentration value is output to the LCD display 125. If yes at step 170 the
data is sent to the PC. At step 172 the controller 122 determines if it is in continuous mode. If no the process ends but if yes steps 153 through 166 are repeated.