WO2007118716A1

WO2007118716A1 - Method and circuit for processing signals that are indicative of the spectral features of light

Info

Publication number: WO2007118716A1
Application number: PCT/EP2007/003445
Authority: WO
Inventors: Winfried Willemer; Jürgen KRIEG; Carl Sandhagen
Original assignee: Opsolution Spectroscopic Systems Gmbh
Priority date: 2006-04-19
Filing date: 2007-04-19
Publication date: 2007-10-25
Also published as: EP2010875A1; DE112007000950A5; WO2007118716A8

Abstract

The invention relates to a method and to an electronic circuit for processing signals which as such are indicative of the features or characteristics of the spectral distribution of light which is supplied for spectrometric measurements combined with an LED light source. The problem of the invention is to provide solutions which allow the informative value of spectrometric measurements to be increased. For this purpose, a reference curve for the spectral emission is calculated based on a reference measurement and at least two additional measurements on the same reference, said spectral emission being used as a reference when carrying out measurements of practically any objects. This curve is determined in such a manner that it corresponds to a curve of the spectral emission at a temperature that is outside or far outside the temperature range of the actual temperatures of the LED during the measurements.

Description

METHOD FOR COMPENSATING FLAKES OF LIGHT SOURCE TEMPERATURE IN SPECTROMETERS

The invention is directed to a method and an electronic circuit for processing signals indicative of features or the characteristic of the spectral distribution of light which is provided for a spectrometric measurement in conjunction with an LED light source.

DE 103 53 703 A1 discloses a mobile miniature spectrometer by means of which, in particular, in vivo substance analyzes of human tissue can be carried out. This miniature spectrometer is provided with an interface via which the respective measurement data can be transmitted to a central evaluation unit and evaluated there.

For performing spectrometric measurements, it is possible to use LEDs as light sources. These LEDs can be selected for spectrometric measurements such that the differences in the spectral emission of LEDs, caused by differences in the temperature of the emitting layer, for very wide temperature ranges (eg 5 ° C to 40 ⁰ C) in a very good approximation can be represented by a temperature compensation curve (TKK), which scales according to the exact temperature difference and in wavelength is moved. This applies primarily to the differences of the logarithm of the spectral intensities (or linearly dependent quantities) as a function of the wavelength.

From the above relationship, there are some possibilities for correcting any disturbing influences. In the German Patent Application DE 10 2005 063 263.7, which is based on the applicant, is explicitly illustrated when measuring on an object whose spectral characteristic can be represented by one or more basic functions whose amplitude is e.g. may be dependent on a substance concentration in the measurement object, when using a least-squares method or an equivalent evaluation of these

Temperature compensation curve as another basic function to consider. The content of that patent application is incorporated by reference herein in its entirety into the present application.

Such an adjustment method may u. be used only to a limited extent if no detailed a-priori knowledge is available on the true trace. Another problem is that the adaptation, i. finding the parameters of the basis functions, e.g. depend on substance concentrations to be determined, on the one hand takes longer due to the greater number of basic functions to be adapted and on the other hand is associated with greater uncertainty.

This problem is solved according to the invention by calculating, on the basis of a reference measurement and at least two further measurements on the same reference, a reference curve for the spectral emission which is then used as a reference when carrying out measurements on almost any objects. This curve is determined to correspond to a spectral emission curve to a temperature that is outside or well beyond the temperature range in which the temperatures of the LED are actually measured.

Depending on the available spectral ranges and the position of the exact measurement task, it may be useful to determine such reference curves both for a temperature below (cold reference) and for a temperature above (hot reference) of the temperatures during the measurements.

The further method steps when carrying out a measurement can then be carried out both with reference to the cold reference and with reference to the hot reference, the resulting results of the spectral measurement can be compared with each other for verification and averaged to reduce the measurement uncertainty.

The use of such an extreme reference ensures that the uncorrected measurement spectrum, almost independent of the measurement object, becomes very similar to a measurement spectrum which arises during a measurement at the reference which was also used to generate the extreme reference. However, according to the underlying method, each measurement (represented logarithmically) at a reference may be mapped to a measurement at a different temperature by subtraction (or addition, depending on the exact formal representation) of a scaled and shifted LED typical temperature compensation curve.

At a reference, this is used in the context of calibration measurements), the determination of displacement and scaling can be done successively: first, it is determined by determining the correlation of the predetermined temperature compensation curve with the uncorrected trace of the reference for different shifts the maximum correlation shift (maximum amount of Pearson's see correlation coefficient) is determined. In the next step, the best scaling can then be determined using a least squares method.

Alternatively, or in combination with this approach, it is also possible to determine the reduction of the computational effort, the determination of the shift leads to the best correlation, via a Fourierverfahren. This can be done in particular by making use of the fact that a shift in the spatial space in the Fourier space corresponds to unambiguous phase changes of the individual frequency components. Specifically, this can be done so that a Fourier decomposition of the TKK as well as a Fourierzelegung of the measurement signal, which is here the logarithm of the ratio of the intensity signal of the extreme reference and the intensity signal of the actual measurement, determined. The phases of essential frequencies are compared, from which the displacement of the measurement signal is determined relative to the TKK whose phase is detected in the original signal and in the shifted signal and evaluated to determine the shift.

Now that the relationship between scaling and displacement of the required correction term is linear, as already assumed and used in the generation of the extreme reference, the relationship between scaling and displacement can be unambiguously made from three measurements on a reference at different temperatures. From two of these measurements, the unshifted temperature compensation curve is determined as the logarithm of the ratio I ₁ / I _{2 of} the respective spectral intensities or of a quantity which depends linearly on them: TKK (lambda) = log ₁₀ (Ii (lambda) / I ₂ (lambda)) = log ₁₀ (li (lambda)) - log ₁₀ (I ₂ (lambda))

(Formula C l)

If a measurement with respect to the quantity Ii is now carried out at a temperature corresponding to that for measurement 2, the same situation is obtained as in the determination of the temperature compensation curve, i. the necessary correction term is exactly the unshifted temperature compensation curve. (By subtracting this curve from the trace, a constant spectrum with the value 0 is ideally obtained.) That is, to completely determine the linear dependence of the necessary scaling on the shift, it is sufficient to perform another measurement on a reference for the explicit displacement and scaling are determined.

By further explicit determinations of scaling and displacement from measurements at the reference at further temperatures, the metrological uncertainty of the determination of the linear relationship can be reduced.

With the help of this linear relationship it can now be computationally determined which values would result in the reference measurement at any temperature. This is used in the determination of the extreme reference.

As already mentioned above, in the actual measurement by reference to an extreme reference, it is ensured that, independently of the measurement object, an uncorrected measurement of the spectrum, ie the logarithm of the ratio of the values of the extreme reference to the measured intensities (or linearly dependent on the intensities Values), is very similar to an uncorrected measurement on a reference. This is because the uncorrected measured values (in this case always the logarithm of the ratio of intensities) result additively from the values associated with a measurement at the measurement temperature at a reference (with reference to the extreme reference) and the measured values at the object as measured at the time of measurement Object with respect to a reference that would have been recorded at exactly the measurement temperature result. Last size depends on the object, the first size can be influenced by setting the virtual temperature of the extreme reference, so that the influence of the object on the shape of the uncorrected measurement can be made very small depending on the choice of the extreme reference.

Since the curve belonging to the logarithm of the ratio of the first measurement at the reference and the extreme reference is identical to a shifted and scaled TKK (logarithm of the ratio of the first reference measurement and the second reference measurement), it is possible to use the extreme Reference exclusively for determining the displacement to use and subsequently as an uncorrected measurement to understand the logarithm of the ratio of the intensities of the first reference measurement and the measurement. The shift must then be corrected for the displacement of the extreme reference against the TKK.

The corrected measurement signal then results additively from the uncorrected measurement signal and the corresponding shifted and scaled TKK.

A possibly remaining residual error from the in each case small incorrect determination of the displacement due to the signature of the measurement object can be further reduced by forming the difference between the uncorrected measurement signal and the corrected measurement signal (difference signal). This signal is now even more free from the influence of the measurement object.

This signal can now again be subjected to the temperature correction procedure. This has multiple benefits: a) the shift determined therefrom can be used to correct the original uncorrected measurement signal; b) the comparison of the displacements associated with the uncorrected measurement signal and the difference signal (ideally these are the same) can be used as a measure of the influence of the signature of the measurement object on the displacement determined above; c) the corrected difference signal ideally gives a constant and, if stability not only the relative spectral intensities of the LED but also the absolute one is assumed, a constant with the value 0. The deviation from the ideal case can be understood as a measure of the quality of the temperature compensation ,

The reduction of the residual errors can be continued iteratively by the difference signal taking over the role of the uncorrected measurement signal.

Without knowledge of the measurement object available in advance, it is now possible to determine the displacement of the necessary temperature compensation curve almost exactly from the measurement on the measurement object with reference to an extreme reference by a correlation method as described above, and from this displacement using the predetermine the relationship between displacement and scaling to determine the full, scaled temperature compensation curve so that the data can be corrected to produce a spectrum, as in the case of a measurement on the object with reference to a measurement at the reference of the Measuring temperature identical temperature would arise. The corrected measured values become independent of the temperature.

In particular, these methods can be used to advantage whenever it is possible to specify an extreme reference such that there are spectral regions in which, expressed in lax terms, the influence of the temperature on the shape of the uncorrected measurement influences the influence of the object to be measured Form dominates. It is not necessary for this condition to be satisfied throughout the spectral range detected.

Furthermore, because of the linearity of the relationship between scaling and displacement, even transferring the correction from the information about the displacement obtained from the measurement spectrum to an LED on the displacement and scaling of the measurement spectra associated with other LEDs is possible, as far as Calibration measurements (eg directly after the device production), the relationship between the necessary shifts of the individual temperature compensation curves is explicitly determined and at the same time it is ensured that the temperatures of the LED do not differ significantly from each other during a measurement, eg by being thermally bonded to a single copper mass or the like. That is, the applicability condition mentioned above need only be satisfied in a spectral range that belongs to a single all-used LED.

The determination of the shift required to achieve the best correlation may be based on the Fourier method. The procedure for this can be designed concretely as follows:

* ** Preferably, the calibration is performed based on the least squares approach. The subsequent measurements are then used to determine the shift to achieve the best correlation. The respectively required correction term is then derived from the shift on the basis of the parameter set determined during the calibration (necessary: ratio of shift to scaling for an LED, ratio of the shifts of the LED among one another).

The calibration is preferably carried out by the fact that in the first calibration measurement, all LEDs are at the temperature level Tl. In the second calibration measurement, all LEDs are at temperature level T2.

From the spectral distributions obtained at these two temperature levels, the temperature compensation curve is determined (see, for example, formula Cl). This temperature compensation curve can be stored as a temperature compensation curve with a shift of 0 and a scaling of 1. Basically, this means that exactly one "TKK" has to be subtracted from a "raw measurement" at T2, which is related to the calibration measurement at T1 to determine the optical density OD, so that the true OD is determined.

Third calibration measurement at T3:

the parameters are still unknown, the correlation and least-squares are determined: these parameters are determined by determining how the temperature compensation curve (TKK) must be scaled shifted so that the raw measurement at T3, that for determining the OD on the measurement is taken at T1, after subtracting the shifted, scaled TKK represents the true OD. These is known in the calibration measurement, namely zero, as a

Measurement at the same reference as measured at the first calibration measurement.

The relationship between scaling and displacement is linear, hence the relationship is completely determined by the three calibration measurements.

Further measurements at identical or different temperatures at the reference can be used to further reduce the measurement uncertainty.

The relationship between scaling and displacement belongs to the set:

Reference = intensity reading of the

Calibration measurement at T1

TKK = logarithm (of the values from

Calibration measurement Tl by corresponding values

T2)

The other LEDs can / will be evaluated simultaneously. Since the temperatures of the LEDs among each other during each measurement process are the same, even in the calibration measurements, the necessary scaling for all LED are identical to each other, only the shifts differ.

During the calibration measurement, the LED for at least one LED

For the others, it is sufficient to determine the displacement associated with a particular calibration measurement.

The measurement then only requires the determination of the shift required to arrive at the best correlation for an LED, which becomes the remaining parameters derived according to the parameter set determined during the calibration:

Scaling to shift the first LED = constant

Shift of the second / third / .. LED to the first LED =

constants

In the method described concretely, first the displacement is determined with reference to the extreme "calculated" reference and then recalculated to the displacement with respect to T1. In principle, this concept can also be carried out "with reversed roles" of the temperatures T1, T2, T3. In particular, it is also possible to extrapolate to "extra cold." Furthermore, it is also possible to perform an averaging using two extrapolated references.

Claims

claims

1. A method for processing signals indicative of features or the characteristic of the spectral distribution of light provided as part of a spectrometric measurement in conjunction with an LED light source, in which:

a reference curve for the spectral emission is generated on the basis of a first reference measurement and at least two further measurements on the same reference, and the reference curve is then used as a reference in the evaluation of test measurements,

wherein the reference curve is determined to correspond to a spectral emission curve to a temperature outside the temperature range in which the temperatures of the LED are actually present during assay measurements.

2. The method according to claim 1, characterized in that, depending on the available spectral ranges and depending on the position of the exact measurement task reference curves both for a temperature below (cold reference) and for a temperature above (hot reference) determines the temperatures during examination measurements become.

3. A method according to claim 2, characterized in that the performance of an examination measurement is carried out both with reference to the cold reference and with reference to the hot reference, and the resulting results of Spectral measurement for verification compared and averaged to reduce the measurement uncertainty.

4. The method according to at least one of claims 1 to 3, characterized in that it is ensured by the use of an extreme reference that the uncorrected measurement spectrum almost independent of the measurement object such a measurement spectrum is very similar, which in a measurement at the reference, which was used to generate the extreme reference arises.

5. The method according to at least one of claims 1 to 4, characterized in that each measurement (shown logarithmically) to a reference by subtraction (or addition, depending on the exact formal representation) of a scaled and shifted LED-typical temperature compensation curve to a measurement at a different temperature is mapped.

6. The method according to at least one of claims 1 to 5, characterized in that in the context of a calibration or preconfiguration, the determination of displacement and scaling successively by first by determining the correlation of the predetermined temperature compensation curve with the uncorrected measurement curve of the reference for various Displacements the shift of maximum correlation (maximum amount of Pearson see correlation coefficient) is determined and then in the next step via a least-squares method, the best scaling is determined.

7. The method according to at least one of claims 1 to 6, characterized in that the method for determining the displacement of the correction curve belonging to the measurement is iteratively applied to the difference signal uncorrected and corrected trace or on

Difference signals in which each of the preceding difference signal takes over the role of the uncorrected measurement signal to minimize the influence of the signature of the measurement object.

8. Electronic circuit for carrying out the aforementioned method.

9. A mobile spectrometer comprising: a light source device comprising as such at least one LED,

an operating circuit for controlling the LED,

a spectrometer device for detecting the spectral characteristic of a light to be analyzed which has been generated in interaction with that light source device, and

- An electronic circuit for processing the signals generated by the Spektrometereinrichtung, the electronic circuit is constructed and configured such that by this a signal processing under recourse to at least one reference curve is generated in accordance with the method according to at least one of claims 1 to 7 was done.

10. Postprocessing system for generating data for a spectral distribution of light on the basis of a standardization approach, with a computer device for accomplishing a data processing process, wherein the computer device is designed such that it processes the normalization approach, wherein the standardization approach includes a normalization by recourse to a reference curve , where that reference curve is determined according to the Method according to at least one of claims 1 to 7 is generated.

11. Postprocessing system according to claim 10, characterized in that the output data provided for processing via a data transfer network, in particular Internet are made available for access.

12. Postprocessing system according to claim 11, characterized in that the output data provided for processing via a mobile radio system in the data transfer network, are fed.