DE102020104266B3 - Device and method for analyzing blood - Google Patents

Abstract

The invention relates to a whole blood spectroscopy device comprising the following components: light source (1) for emitting radiation in the wavelength range between 450 to 700 nm, lens (2), measuring cuvette (4), integrating sphere (6), spectrometer (8), data processing - and control unit (9), display unit (10), pump (11), valve (12) for metering the dye solution (16) and reference liquid (18), valve (14) for metering the sample (5), storage container (15) for the dye solution (16) and for the reference liquid (18), tubing set (19), sample port (20) and container for collected liquid (21). Furthermore, the invention relates to a method for the spectroscopic determination of parameters in whole blood .

Description

The invention relates to a device based on the principle of spectroscopy and a method for analyzing blood parameters, in particular blood parameters such as total hemoglobin, hemoglobin derivatives, hematocrit, bilirubin and glycated hemoglobin in blood.

ist ein Zusammenhang zwischen der Konzentration c einer Substanz in einem Medium und der spektroskopisch messbaren Abschwächung der Intensität / gegenüber der Anfangsintensität I₀ gegeben, wobei ε ein von der Substanz und der Wellenlänge abhängiger Koeffizient ist und d die optische Pfadlänge durch das Medium. Die Nutzung dieses Zusammenhangs zur Konzentrationsbestimmung wird aber durch einige Faktoren gestört und begrenzt.About the Lambert-Beer law $$\mathrm{I.}={\mathrm{I.}}_0\cdot e^{-\epsilon \cdot c\cdot d}$$

there is a relationship between the concentration c of a substance in a medium and the spectroscopically measurable attenuation of the intensity / compared to the initial intensity I ₀ , where ε is a coefficient dependent on the substance and the wavelength and d is the optical path length through the medium. However, the use of this relationship to determine the concentration is disturbed and limited by a number of factors.

If the concentrations are too high, there are deviations from this law, also due to saturation effects. The detection limit can be reached at very low concentrations. The spectral bandwidth and the deviation of the wavelength used can falsify the result. The measurement of non-homogeneous samples of natural origin, such as blood, is a particular challenge. In whole blood, most of the hemoglobin and its derivatives are found in red blood cells. The surrounding blood plasma has a lower refractive index, which leads to scattering effects at the boundary layer. Therefore, whole blood samples are often subjected to hemolysis to dissolve red blood cells prior to the determination of blood parameters such as total hemoglobin, hemoglobin derivatives, hematocrit, bilirubin, and glycated hemoglobin. Ultrasound or reagents are used for hemolysis. The need for hemolysis is seen as a disadvantage, since compared to direct measurement in whole blood, time is lost and the blood samples are irreversibly changed and are therefore no longer available for further measurements. If the hemolysis does not take place completely, scattering effects can also occur, which lead to an error in the determination of the blood parameters. Another disadvantage is that additional consumables (e.g. reagents) and possibly other components (e.g. ultrasound generator) are required for the hemolysis, which means higher maintenance and additional costs.

One method of being able to optically determine hemoglobin derivatives without prior hemolysis is in US 6 262 798 B1 described. An empirically determined correction matrix is used to correct the extinction of a whole blood sample measured at 7 wavelengths and then to be able to determine the concentrations using Lambert Beer. For this purpose, the disposable cuvettes used must be manufactured very precisely so that their optical path length is exactly known. The sample is then disposed of together with the cuvette and is not available for the analysis of other parameters. Due to the measuring principle, the sensor cannot be expanded to include additional parameters without modifications to the hardware.

In the scriptures US 10 088 360 B2 , US 10 088 468 B2 , US 10 151 630 B2 , US 2017/0 227 397 A1 , US 2019/0 017 993 A1 and US 2019/0 033 287 A1 various aspects, such as the compensation of the thermal expansion of the achromatic lenses, as well as the kernel-based algorithms for evaluating the data and an optical spectrometer for whole blood analysis are placed under protection. A disadvantage of this modular system is that the cuvettes to be used have to be coded with a fixed optical path length due to manufacturing tolerances, which must first be determined and stored in the factory for each individual cuvette. Other cuvettes cannot be used because no optical path length is assigned to them.

In the publication REDMER, B. [et al.]: Determination of hemoglobin derivatives in unaltered whole blood samples using Support Vector regression in the spectral range from 450 to 700 nm. Proceedings of SPIE, Vol. 11247, February 14, 2020 . DOI: 10.1117 / 12.2544806 suggests the use of support vector machines (SVM) for the evaluation of diffuse transmission and reflection spectra in the wavelength range between 450 nm and 700 nm for the spectroscopic examination of whole blood samples for the purpose of determining blood parameters such as hemoglobin derivatives and hematocrit.

When publishing VINCELY, V. [et al.]: Extracting broadband optical properties from uniform optical phantoms using an integrating sphere and inverse adding-doubling. Proceedings of SPIE, Vol. 10486, February 19, 2018 . DOI: 10.1117 / 12.2291950 describes the use of an integrating sphere and the application of Inverse Adding-Doubling (IAD) algorithms to determine the optical properties when measuring the reflection and transmission spectra.

In the EP 0 807 813 B1 A device is proposed for determining blood gas values, in particular oxygen saturation, in whole blood conveyed extracorporeally. The device comprises two monochromatic light sources of different wavelengths and two hollow spaces protruding into one another, one of which is spherical with a diffusely highly reflective surface and the second can accommodate the whole blood sample, which is guided like a tube. The light from the two light sources should essentially only impinge on the surface of the cavity carrying the whole blood sample and be reflected from there into the second cavity, possibly also transmitted and then measured independently of the angle.

Due to the monochromatic light sources used here, it is not possible to record broadband spectra of the sample with this method.

In the DE 699 10 006 T2 describes a sample cell for the spectroscopic determination of fluid samples, in particular whole blood samples. The sample cell comprises two mutually movable parts, the position of which allows defined positions for calibration, measurement and cleaning of the sample cell. The surface distance of the sections, which is set with the aid of tolerance plates and spacers, defines the optical path length d within an interval. By adapting the flow path and thus the flow rate of the sample can be varied and thus samples of different optical density (lysed or whole blood) can be measured. The adjustment of the optical path length by the mechanical change of the sample cell requires an extremely high precision and thus a considerable technical effort regarding both the production and the handling of the necessary tolerance plates and spacers.

It would be very desirable to have a spectroscopic measuring system which allows whole blood analyzes using cuvettes that do not have to be individually coded beforehand, but in which the coding takes place via the device in which the cuvettes are used. The EP 2 246 692 B1 contains the calibration and characterization of a measuring cuvette with regard to possible contamination by reference measurement with a dye solution.

It is therefore the object of the invention to provide a device and a method which allows the spectroscopic analysis of non-hemolyzed whole blood samples using freely selectable cuvettes with an optical path length that is not predetermined.

In particular, it is the object of the invention to provide a device and a method which allow the spectroscopic analysis of non-hemolyzed whole blood samples to determine the parameters total hemoglobin, hemoglobin _{derivatives (e.g. oxygenated hemoglobin (O 2} Hb), reduced hemoglobin (HHb), carboxyhemoglobin (COHb ) and methemoglobin (MetHb)), hematocrit, bilirubin and glycated hemoglobin using freely selectable cuvettes with non-predetermined optical path lengths.

• • Verwendeter Spektralbereich 450 bis 700 nm (sichtbares Licht)
• • Messen der im Transmissionsspektrum enthaltenen gestreuten und ungestreuten Anteile
• • Erfassen der optischen Pfadlänge der Messküvette
• • Erfassen von Verunreinigungen in der Messküvette
• • Vermessung der nicht vorbehandelten Blutprobe in einer Messküvette
• • Korrektur der gemessenen Spektren mit den Daten der Messküvette
• • Extraktion eines Merkmalsvektors Vaus dem gemessenen Spektrum
• • Erzeugung eines generierten Merkmalsvektors V_R mit Hilfe von Mustererkennung nach dem Verfahren Support Vector Regression (SVR)
• • Vergleich des Merkmalsvektors V mit dem generierten Merkmalsvektor V_R aus einer Datenbank

The inventors have now found that parameters in non-hemolyzed whole blood can be determined using freely selectable cuvettes with an optical path length that has not been predetermined if the device according to the invention described below and a spectroscopic method with the following features are used.

• • Spectral range used 450 to 700 nm (visible light)
• • Measurement of the scattered and unscattered components contained in the transmission spectrum
• • Acquisition of the optical path length of the measuring cuvette
• • Detection of impurities in the measuring cuvette
• • Measurement of the blood sample that has not been pretreated in a measuring cuvette
• • Correction of the measured spectra with the data from the measuring cuvette
• • Extraction of a feature vector V from the measured spectrum
• • Generation of a generated feature vector V _R with the help of pattern recognition according to the Support Vector Regression (SVR) method
• • Comparison of the feature vector V with the generated feature vector V _R from a database

It turned out, in a way that was not to be expected, that the precision, accuracy and robustness of the method according to the invention and the device according to the invention are more than sufficient to determine parameters in non-hemolyzed whole blood using freely selectable cuvettes with an optical path length that has not been predetermined . In particular, it unexpectedly turned out that when using the method according to the invention and / or the device according to the invention, the measurement of the transmission spectra alone is sufficient.

In one aspect, the object of the invention is achieved by a spectroscopic device for measuring whole blood, which comprises the following components: light source ( 1 ) for the emission of radiation in the wavelength range between 450 and 700 nm, lens ( 2 ), Measuring cuvette ( 4th ), Integrating sphere ( 6th ), Spectrometer ( 8th ), Data processing and Control unit ( 9 ), Display unit ( 10 ), Pump ( 11 ), Valve ( 12th ) for dosing dye solution ( 16 ) and from reference liquid ( 18th ), Valve ( 14th ) for dosing the whole blood sample ( 5 ), Storage container ( 15th ) for the dye solution ( 16 ) and for the reference liquid ( 18th ), Hose set ( 19th ), Sample port ( 20th ) and container for collected liquid ( 21 ).

In a special embodiment of the spectroscopic device, reference liquids ( 18th ) and dye solution ( 16 ) in two separate storage containers ( 15th ) and ( 17th ) and are housed via two valves separate valves ( 12th , 13th ) metered in. A schematic representation can be found in .

In a further special embodiment, the lens ( 2 ) a collimation lens and a pinhole ( 3 ) attached in front of the collimation lens. This is in shown.

In a further embodiment, an optical waveguide ( 7th ) between the integrating sphere ( 6th ) and the spectrometer ( 8th ) to be appropriate.

With the broadband light source ( 1 ), which realizes the spectral range 450 to 700 nm, it can be, for example, a tungsten-halogen lamp, xenon lamp or white LED.

The reference liquid can be a rinsing solution that contains protein remover, for example. Furthermore, the rinsing solution can additionally contain a dye with known absorption properties.

1. i. Bestimmung eines Referenzspektrums I₀ im Rahmen einer Transmissionsmessung
2. ii. Bestimmung des Spektrums eines Farbstoffs I_F im Rahmen einer Transmissionsmessung
3. iii. Ermittlung eines Korrekturfaktors für eine Messküvette (4), wobei die Funktion der Messküvette derart ist, dass Farbstoff, Referenz-Flüssigkeit als auch Vollblut-Probe in die Messküvette injiziert und dort jeweils optisch mittels einer Vollblut-Spektroskopie-Vorrichtung untersucht werden
4. iv. Bestimmung eines Vollblut-Probenspektrums I_P im Rahmen einer Transmissionsmessung
5. v. Korrektur des gemessenen Vollblut-Probenspektrums I_P mit dem Korrekturfaktor der Messküvette aus Schritt iii
6. vi. Extraktion eines Merkmalsvektors Vaus dem korrigierten gemessenen Vollblut-Probenspektrum I_P
7. vii. Vergleich des Merkmalsvektors V mit einem generierten Merkmalsvektor V_R aus einer Datenbank, wobei aus dem Vergleich die Parameter des Vollbluts bestimmt werden.

In a further aspect, the object of the invention is achieved by a method for the spectroscopic determination of parameters in whole blood comprising the following steps:

1. i. Determination of a reference spectrum I ₀ as part of a transmission measurement
2. ii. Determination of the spectrum of a dye I _F as part of a transmission measurement
3. iii. Determination of a correction factor for a measuring cuvette ( 4th ), the function of the measuring cuvette being such that dye, reference liquid and whole blood sample are injected into the measuring cuvette and are each examined there optically by means of a whole blood spectroscopy device
4. iv. Determination of a whole blood sample spectrum I _P as part of a transmission measurement
5. v. Correction of the measured whole blood sample spectrum I _P with the correction factor of the measuring cuvette from step iii
6. vi. Extraction of a feature vector V from the corrected measured whole blood sample spectrum I _P
7. vii. Comparison of the feature vector V with a generated feature vector V _R from a database, the parameters of the whole blood being determined from the comparison.

In a preferred embodiment of the method according to the invention, I ₀ (step i) and I _F (step ii) are determined with a measurement.

In a special embodiment of the method according to the invention, incorrect measurement results can be caused by severe contamination and / or inferior and / or non-original measuring cuvette ( 4th ) and / or reference liquid ( 18th ) and / or dye solution ( 16 ) can be displayed by determining the optical path length d.

Ideally, the dye used does not absorb or fluoresce within the spectral range used to determine the blood parameters. In this way the dye in the reference liquid ( 18th ) and does not affect the recording of the reference spectrum I ₀ . If this is not the case, a calibration spectrum I _F with dye solution ( 16 ) in the measuring cuvette ( 4th ), which is then recorded with the stored extinction coefficient ε, the concentration of the dye solution ( 16 ) c and the one for the measuring cuvette ( 4th ) determined optical path length d is corrected using Lambert-Beer's law in order to obtain the original characteristic of the light source ( 1 ) to restore the reference spectrum I _0.

und $$I_0=\frac{1}{T}\cdot I_F$$

ist dabei der Zusammenhang zwischen dem gemessenen Kalibrierspektrum I_F und dem erforderlichen Referenzspektrum I₀ gegeben, wobei T die Transmission ist.over $$T=\frac{\mathrm{I.}}{{\mathrm{I.}}_0}=e^{-\epsilon \cdot c\cdot d}$$

and $${\mathrm{I.}}_0=\frac{1}{T}\cdot {\mathrm{I.}}_{\mathrm{F.}}$$

the relationship between the measured calibration spectrum I _F and the required reference spectrum I _{0 is} given, where T is the transmission.

It is advantageous to choose a dye with an excitation maximum in the near-infrared spectral range (NIR) for two reasons: Due to the emission spectrum of the light source used and the sensitivity of the semiconductor-based detector the result in the NIR is a particularly advantageous high signal-to-noise ratio (SNR) for the measurement. In addition, the light emitted by excited dyes (fluorescence) is shifted to higher wavelengths and would overlap the spectral range used for the measurements if the excitation wavelength is too low. For example, dyes from the group of cyanines, such as cyanine 7 or indocyanine green, come into consideration, the exact optical properties depending on the solvent used.

The concentration of the dye c is chosen so that a linear modulation of the spectrometer ( 8th ) is guaranteed and the change in transmission ΔT becomes a maximum due to an optical path length Δd deviating from the desired value. This is the case when the absorption coefficient μ _{a of} the dye solution, which is the product of the molar extinction coefficient ε and concentration c, is chosen so that the transmission T at the desired path length d of 100 μm is between 30% and 45%, preferably between 35% and 42%, particularly preferably between 36% and 41%, very particularly preferably 36.8% (1 / e). With a maximum expected deviation of the optical path length Δd of 10 μm, the change in the transmission ΔT is greatest and lies approximately between 3.5% and 4%. As a result, the optical path length d can be determined particularly precisely. This connection is in to see. The relationship between signal deviation and transmission is shown for the maximum expected deviation of the optical path length Δd from 10 μm with a desired path length d of 100 μm, resulting in an optical path length d of 90 μm and 110 μm.

Does the spectrometer ( 8th ) an analog-to-digital converter (ADC) with 16-bit resolution and if the maximum modulation is limited to 90% of the value range, the detected signal changes by approximately 2 ¹⁶ · 0.9. 0.035 ≈ 2064 counts. This means that the noise of the detector may amount to approximately 206 counts for the desired accuracy of 1 µm. Modern spectrometers ( 8th ) have an SNR of 300: 1 and better, which guarantees this (2 ¹⁶ · 0.9 / 300 ≈ 197).

For a precise prediction of the blood parameters, the optical path length d of the measuring cuvette ( 4th ) must be known to an accuracy of about 1 µm. Deviations lead to changes in the transmission behavior of the sample ( 5 ) and must therefore be taken into account arithmetically. When manufacturing optical plastic parts by injection molding, tolerances of up to ± 10 µm are to be expected. Decisive for an exact absorption spectroscopic determination of the optical path length d is the skillful choice of the dye and its concentration c.

If the determined optical path length d is outside a specified confidence interval, this can indicate particularly severe contamination or an inferior, possibly non-original measuring cuvette ( 4th ) or reference liquid ( 18th ) or dye solution ( 16 ) Clues. In order to avoid incorrect results, the measurement can be stopped in this case and a warning can be displayed to the user.

In the following, the generality of the teaching is to be described as an example and the procedure is not restricted.

The measurement process is initiated by the user by taking the sample ( 5 ) to the sample port ( 20th ) is set and the "Start" control is operated. The broadband light source ( 1 ), for example a tungsten-halogen lamp, xenon lamp or white LED, is switched on and a reference _{spectrum I 0} or spectrum of a dye I _{F is} measured before the sample reaches the measuring cuvette ( 4th ) reached. Then sample spectra in the spectral range from 450 to 700 nm are measured until the sample ( 5 ) the measuring cuvette ( 4th ) completely happened. From the stable plateau, which occurs when the measuring cuvette is completely filled ( 4th ) and constant volume flow, a sample spectrum I _{p is} selected.

The light source ( 1 ) is switched off and dark spectra are recorded at the integration times that were used _{for measuring the reference spectrum I 0} or the calibration _{spectrum I F} and sample spectrum I _p. The associated dark spectrum is subtracted from all spectra and the values are normalized to counting events per second. If the calibration spectrum I _{F has been} determined, it is corrected as described above and the reference spectrum I _{0 is} calculated from it.

The quotient of the sample spectrum I _P and the reference _{spectrum I 0 is then} formed, which represents the standardized transmission. This is corrected in a manner known to the person skilled in the art for non-linear effects caused by the use of the integrating sphere ( 6th ) occur [Prahl2011 Scott A. Prahl, "Everything I think you should know about Inverse Adding-Doubling", handbook, 2011.] The correction factor is determined by the geometry of the integrating sphere ( 6th ) determines the ratio of a circular opening facing the sample and its diameter. The measured transmission is greater than its actual value. The feature vector V resulting therefrom serves as input for the regression models stored in the device for the blood parameters to be determined, each of which is represented by a feature vector V _R generated from a database. These then supply the values for the blood parameters, which the device finally displays to the user. The measured The sample spectrum I _P is determined by means of a device in the data processing and control unit ( 9 ) stored correction function is corrected depending on the previously determined optical path length d. In the simplest case, the function can be a constant factor for all wavelengths, but wavelength-dependent functions are also possible.

1. 1. Nicht-lineare Zusammenhänge können über einen Kernel-Trick transformiert werden
2. 2. Nur eine überschaubare Anzahl an Hyperparametern muss optimiert werden
3. 3. Bereits eine relativ kleine Anzahl an Beobachtungen führt zu stabilen Modellen

The regression models were previously determined from a large number of blood samples. Influencing variables relevant to the optical behavior, such as the hematocrit, the hemoglobin _{derivatives O 2} Hb, HHb, COHb and MetHb and the osmotic concentration, were specifically set and varied over the clinically relevant range. The total transmission of each sample was then measured in a laboratory setup and the associated reference values of the blood parameters were determined. The total hemoglobin (ctHb) and the proportions of the hemoglobin derivatives O ₂ Hb, COHb and MetHb were determined with an OSM 3 hemoximeter radiometer. The mean value was formed from two measurements. HHb was calculated as the difference between the sum of the other derivatives and 100%. The hematocrit was determined by centrifugation of glass capillaries according to DIN 58933-1. Support Vector Regression (SVR) was used as the algorithm. Compared to some other methods, such as neural networks, SVR offers advantages for biological samples:

1. 1. Non-linear relationships can be transformed using a kernel trick
2. 2. Only a manageable number of hyperparameters needs to be optimized
3. 3. Even a relatively small number of observations leads to stable models

A separate regression model was created for each blood parameter. This process is called training. Several so-called hyperparameters influence the adaptation of the models. Their optimal values were determined in an iterative process and the prediction errors of the models were thereby minimized. In order to avoid over- or under-fitting, a 5-fold cross-validation was also carried out. An example of a test data set made up of 38 samples from 8 individuals that were not part of the data set used for training are the results in the and for the predicted values versus their reference values for the blood parameters O ₂ Hb, HHb, COHb, MetHb, ctHb and hematocrit.

The performance of the regression models was assessed using the metrics common in machine learning: correlation coefficient (R ² ), mean square deviation (RMSE) and mean absolute deviation (MAE). The correlation coefficient R ² is greater than 0.975 for all blood parameters and confirms the excellent correlation between prediction and reference. The error in the prediction is sufficiently small with an MAE of always less than 1.97% over the entire clinically relevant range. The cross-sensitivities between the individual parameters are sufficiently low.

Figure list

• : Schematic representation of the device according to the invention, with a collimation lens as a lens ( 2 ) and pinhole ( 3 ).
• : Schematic representation of the device according to the invention, with two separate storage containers ( 15th ) and ( 17th ) for reference liquid ( 18th ) and dye solution ( 16 ) and separate valves ( 12th , 13th ) for dosing reference liquid ( 18th ) and dye solution ( 16 ).
• : Relationship between signal deviation and transmission of the dye solution with Δd equal to 10 µm.
• : Predicted fractional proportions of (a) oxygenated hemoglobin (O ₂ Hb), (b) reduced hemoglobin (HHb), (c) carboxyhemoglobin (COHb), and (d) methemoglobin (MetHb). Each cross corresponds to a blood sample from the test data set. The perfect prediction is shown by a solid line. The linear fit of the measurement data is shown by a dashed line.
• : Predicted values for (a) total hemoglobin and (b) hematocrit.

List of reference symbols

1

Light source

2

lens

3

Pinhole

4th

Measuring cuvette

5

sample

6th

Integrating sphere

7th

optical fiber

8th

spectrometer

9

Data processing and control unit

10

Display unit

11

pump

12th

Valve

13th

Valve

14th

Valve

15th

Storage container

16

Dye solution

17th

Storage container

18th

Reference liquid

19th

Hose set

20th

Sample port

21

Collecting container

Claims (11)

Whole blood spectroscopy device comprising the following components: light source (1) for emitting radiation in the wavelength range between 450 to 700 nm, lens (2), measuring cuvette (4), integrating sphere (6), spectrometer (8), data processing and control unit (9), display unit (10), pump (11), valve (12) for dosing dye solution (16) and reference liquid (18), valve (14) for dosing the whole blood sample (5), storage container ( 15) for the dye solution (16) and for the reference liquid (18), tubing set (19), sample port (20) and container for the collected liquid (21). Whole blood spectroscopy device according to Claim 1 , characterized in that the device for the reference liquid (18) has a separate storage container (17) and a further valve (13). Whole blood spectroscopy device according to one or more of the preceding Claims 1 and 2 , characterized in that the lens (2) is a collimation lens and a pinhole (3) in front of the collimation lens. Whole blood spectroscopy device according to one or more of the preceding Claims 1 to 3 , characterized in that an optical waveguide (7) is present between the integrating sphere (6) and the spectrometer (8). A method for the spectroscopic determination of parameters in whole blood comprising the following steps: i. Determination of a reference spectrum I ₀ as part of a transmission measurement ii. Determination of a spectrum of a dye I _F as part of a transmission measurement iii. Determination of a correction factor for a measuring cuvette (4), the function of the measuring cuvette being such that dye, reference liquid and whole blood sample are injected into the measuring cuvette and each examined there optically by means of a whole blood spectroscopy device iv. Determination of a whole blood sample spectrum I _P as part of a transmission measurement v. Correction of the measured whole blood sample spectrum I _P with the correction factor of the measuring cuvette (4) from step iii vi. Extraction of a feature vector V from the corrected measured whole blood sample spectrum I _P vii. Comparison of the feature vector V with a generated feature vector V _R from a database, the parameters of the whole blood being determined from the comparison. Procedure according to Claim 5 , characterized in that I ₀ (step i) and I _F (step ii) are determined with a measurement. Method according to one or more of the preceding Claims 5 to 6th , characterized in that the transmission of the dye solution at an optical path length d of 100 µm is between 30% and 45%, preferably between 35% and 42%, particularly preferably between 36% and 41%, very particularly preferably 36.8% ( 1 / e). Method according to one or more of the preceding Claims 5 to 7th , characterized in that the dye has an excitation maximum in the near-infrared spectral range (NIR). Method according to one or more of the preceding Claims 5 to 8th , characterized in that the dye is a cyanine dye. Method according to one or more of the preceding Claims 5 to 9 , characterized in that the dye is cyanine 7.5 and / or indocyanine green. Method according to one or more of the preceding Claims 5 to 10 , characterized in that incorrect measurement results caused by severe contamination and / or inferior and / or non-original measuring cuvette (4) and / or reference liquid (18) and / or dye solution (16) are displayed by determining the optical path length d.

Images