JP7729075B2 - Measuring equipment - Google Patents

Description

The present invention relates to a device for measuring biological indicators.

In recent years, the number of diabetes patients has been increasing worldwide, creating a demand for non-invasive blood glucose measurement that does not require blood sampling. Various optical sensing methods have been proposed, including those using near-infrared, mid-infrared, and Raman spectroscopy. Of these, the mid-infrared region is a fingerprint region where glucose absorption is high, allowing for greater measurement sensitivity than the near-infrared region. In addition to blood glucose levels, biomarkers such as hemoglobin concentration, blood lipids, blood proteins, and blood tumor DNA can also be measured non-invasively in the mid-infrared region.

In order to measure glucose concentrations accurately in the mid-infrared region, a method has been proposed that uses the wave numbers of the glucose absorption peaks of 1035 cm ⁻¹ , 1080 cm ⁻¹ , and 1110 cm ⁻¹ (see, for example, Patent Document 1).

General biometric measurements using mid-infrared light are affected by individual differences in digestion, absorption, and metabolism from before to after a meal, as well as differences in the measurement environment, making it difficult to achieve accurate and stable measurements.

The present invention aims to provide a non-invasive biomarker measurement device with improved measurement stability and accuracy.

In one aspect, the measurement device comprises:
a light source in the mid-infrared region;
a photodetector that detects light output from the light source and reflected by the object to be measured;
an information processing device that determines a biological index of the measurement object based on the output of the photodetector;
Equipped with
the light source emits light having a first wavelength selected from a wave number of 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less, and light having a second wavelength different from the first wavelength selected from a wave number of 950 cm ⁻¹ or more and 990 cm ⁻¹ or less;
The information processing device determines a biomarker of the measurement object from a first absorbance of the light of the first wavelength and a second absorbance of the light of the second wavelength.

Non-invasive measurement of biomarkers with improved measurement stability and accuracy is achieved.

FIG. 1 is a schematic diagram of a measurement device according to an embodiment. FIG. 1 is a diagram illustrating a hardware configuration of an information processing device. FIG. 2 is a diagram illustrating the operation of the measuring device. FIG. 2 is a diagram illustrating the operation of the measuring device. FIG. 2 is a diagram illustrating the operation of the measuring device. FIG. 10 is a diagram illustrating an example of control timing of the measurement device. FIG. 10 is a diagram in which the mean values of the coefficients of determination are mapped to the sense wavelength and the base wavelength. FIG. The absorbance of each data point around 990 cm ⁻¹ is overlaid in the graph. This is a graph in which the absorbances determined from one sense wavelength and two base wavelengths are overlaid. FIG. FIG. 10 is an enlarged view of the first mapping range in region B. FIG. 10 is an enlarged view of the second mapping range in region B. The absorbance of each data point around 1160 cm ⁻¹ is overlaid in the graph. FIG. 10 is an enlarged view of area C. The absorbance of each data point around 1740 cm ⁻¹ is overlaid in the graph. FIG. 10 is a diagram showing the range of high correlation between estimated blood glucose level changes and typical blood glucose level changes in humans.

In this embodiment, biomarkers such as blood glucose levels are measured using at least two wavelengths within a specific wavenumber range. In the field of infrared spectroscopy, the term "wavenumber" is generally used, but since wavenumber k is expressed as the number of waves contained per unit length (1/λ), in the following explanation, "wavenumber" and "wavelength" will be used synonymously.

At least one of the two or more wavelengths may be used to normalize the measurement data obtained by the measurement optical system. By using the normalization wavelength, it is possible to correct constantly changing measurement conditions, such as variations in the contact state between the measurement optical element and the measurement site on the object being measured.

An appropriate range for selecting two or more wavelengths is identified to ensure stable measurement results regardless of individual differences. Specifically, measurement data in the mid-infrared range is collected from a wide range of subjects, with different meal contents and at different measurement times, and the appropriate wavelength range is determined based on the correlation between the collected data and the changes in typical biomarkers from before to after a meal.

A typical transition in a biomarker, for example, blood glucose levels, is a series of changes in which blood glucose levels are low before a meal, peak 40 minutes after a meal, and then return to pre-meal levels 180 minutes later. Two or more wavelengths are selected for measurement from wavelength ranges that show a high correlation between the collected data and this typical transition. This allows for stable, accurate, non-invasive measurement of biomarkers, regardless of individual differences.

<Device configuration>
Fig. 1 is a schematic diagram of a measurement device 1 according to an embodiment. The measurement device 1 includes a measurement optical system 10 and an information processing device 30. In Fig. 1, a data logger 25 for recording data is disposed between the measurement optical system 10 and the information processing device 30, but the measurement results obtained by the measurement optical system 10 may also be input directly to the information processing device 30.

The measurement optical system 10 includes a light source device 11, a switch 12, a measurement optical element 17, and a photodetector 18. Depending on the design and arrangement of the optical path, optical elements such as multiplexers/demultiplexers 13 and 14 may be inserted between the light source device 11 and the measurement optical element 17. The multiplexers/demultiplexers 13 and 14 may be dichroic prisms, half mirrors, polarized beam splitters/combiners, etc. A light-guiding member 15 or 16 may be connected to at least one of the input and output sides of the measurement optical element 17. Optical fibers, optical mirrors, etc. may be used as the light-guiding members 15 and 16. The dashed lines within the measurement optical system 10 indicate the optical paths from each light source to the light-guiding member 15.

The light source device 11 outputs light of at least two wavelengths selected from a specific wavenumber range. The specific wavenumber range is determined in advance so that biomarkers can be measured stably regardless of individual differences or differences in measurement conditions. Details on determining the wavenumber range will be described later.

In the example shown in Figure 1, three light sources 111, 112, and 113 are used, each outputting light of a different wavelength, but this example is not limiting. A single tunable light source or a light source that emits light over a wide wavelength range may also be used as the light source device 11. Examples of wide wavelength range light sources that can be used include lamp light sources, light emitting diodes (LEDs), and super luminescent diodes (SLDs). Similar measurements can also be made using a combination of a wide wavelength range infrared lamp and a wavelength filter, or a combination of a wide wavelength range infrared lamp and a spectrometer such as an FTIR (Fourier transform infrared spectrophotometer).

When using a light source with a wide wavelength range, a wavelength filter that extracts the desired wavelength may be provided on the output side of the light source device 11 as needed. Alternatively, multiple light-receiving elements may be used in the photodetector 18, with a wavelength filter placed on the input side of each light-receiving element. The number of wavelengths used for measurement is not limited to three, and may be two, four, or more. When using individual light sources 111, 112, and 113, a portion of the output light from each light source may be monitored to correct for output fluctuations.

The switch 12 switches the wavelength of the light incident on the measurement optical element 17. In the example of Figure 1, switching elements 121, 122, and 123 are provided on the output side of light sources 111, 112, and 113, respectively, but this is not limited to this example and may also control the on/off of each light source. If a single wavelength-tunable light source is used, the switch 12 may also be a selector for the light source wavelength.

There are no restrictions on the specific configuration of the switch 12, as long as it can change the wavelength of the light irradiated onto the measurement site. It may be a shutter that selectively blocks the light emitted from the light sources 111, 112, and 113, or a switch that turns each of the light sources 111, 112, and 113 on and off.

Light of the wavelength selected by the switch 12 is guided to the measurement optical element 17 via the multiplexer/demultiplexer 13, 14 or light-guiding member 15 as necessary. The measurement optical element 17 has a contact surface 171 that comes into contact with the object being measured. During measurement, the contact surface 171 of the measurement optical element 17 is pressed against the object being measured, and picks up the return light from the measurement site, i.e., the light that has been optically absorbed at the measurement site.

Light incident on the measurement optical element 17 is attenuated at the interface between the measurement site and the contact surface 171 by an amount corresponding to the optical absorption spectrum of the medium being measured. The attenuated light emerging from the measurement optical element 17 contains information about the interior of the medium being measured. For non-invasive measurements using light in the mid-infrared region, an ATR (Attenuated Total Reflection) prism can be used as the measurement optical element 17. An ATR prism is suitable for spectroscopy in the mid-infrared region, where the optical absorption of glucose and glycogen is observed.

The ATR method is a measurement method in which infrared light is incident on a high-refractive index ATR prism and utilizes the "evanescent field" that appears when total reflection occurs at the interface between the prism and the medium being measured. When the ATR prism is pressed against the object being measured and measurement is performed, the evanescent field that has been extruded is absorbed by the object being measured.

In the case of infrared light, the light penetrates from the ATR prism to the measurement site to a depth of only a few microns, and does not reach the capillaries that are located at a depth of several hundred microns. However, it is known that components such as plasma in blood vessels seep into skin and mucosal cells as tissue fluid (interstitial fluid). By detecting components such as glucose present in this tissue fluid as optical absorption spectrum intensity or absorbance, it is possible to measure biomarkers.

To measure the optical absorption spectrum intensity using the seepage of magnetic fields, it is desirable to use measurement sites such as the oral mucosa, which has no epidermis, or the earlobes or lips, which have thin epidermis.

Light emitted from the measurement optical element 17 is detected by the photodetector 18 as an optical signal containing biological information. When a light-guiding member 16 is used, the optical signal from the measurement optical element 17 may be received by the photodetector 18 via the light-guiding member 16. The photodetector 18 converts the received optical signal into an electrical signal and outputs it to the data logger 25. The information recorded in the data logger 25 is supplied to the information processing device 30, where it is processed to estimate the desired biological indicator. As mentioned above, the functionality of the data logger 25 may be incorporated into the information processing device 30.

When using the output of a single photodetector 18 as shown in Figure 1, the signal from the photodetector 18 can be processed in synchronization with the switching timing of the switching elements 121-123, allowing the single photodetector 18 to appropriately measure absorbance at multiple different wavelengths.

The information processing device 30 has, as functional blocks, a data processing unit 31, a biological index estimation unit 32, a switching control unit 33, and a timing control unit 34. The data processing unit 31 calculates the absorbance of the measured object for each wavelength from the light intensity of each wavelength detected by the photodetector 18. The absorbance A at wave number k is expressed as follows:
A(k)=-log ₁₀ (I/I ₀ ) (1)
Here, _I0 is the intensity of the light incident on the measurement optical element 17, i.e., the intensity of the light emitted from the light source. I is the intensity of the light emitted from the measurement optical element 17 and detected by the photodetector 18.

The biomarker estimation unit 32 estimates biomarkers based on the absorbance at each wavelength using a prediction formula described below. The switching control unit 33 controls the operation of the switch 12 in the measurement optical system 10. The timing control unit 34 controls the timing of recording and capturing the detection results of the photodetector 18 in synchronization with the wavelength switching. The switching control unit 33 may be provided external to the information processing device 30. In that case, the external switching control unit and the information processing device 30 may be connected by cable or wirelessly, and the switching control for the measurement optical system 10 and the timing of data capture in the information processing device 30 may be synchronized.

In this embodiment, two or more wavelengths suitable for estimating biomarkers are set in the measurement device 1, and the biomarker estimation unit 32 stably and accurately estimates biomarkers based on the absorbance at the two or more wavelengths.

Figure 2 is a hardware configuration diagram of information processing device 30H. Information processing device 30H is a hardware reconfiguration of the functions of information processing device 30 in Figure 1. Information processing device 30H has a CPU (central processing unit) 301, random access memory (RAM) 302, read-only memory (ROM) 303, auxiliary storage device 304, input device 305, display device 306, external interface (I/F) 307, and communication interface (I/F) 308, which are interconnected by bus 309.

The CPU 301 and the main memory devices RAM 302 and ROM 303 are connected directly to the main bus, and other devices may be connected to the main bus via an internal interface. In Figure 2, for simplicity, the devices are shown as being interconnected by bus 309.

The data processing unit 31, biometric estimation unit 32, switching control unit 33, and timing control unit 34 in Figure 1 can be realized by a CPU 301, RAM 301, and ROM 302. A memory-equipped ASIC (application specific integrated circuit) or programmable logic device (PLD) may be provided separately from the CPU 301, and the functions of the switching control unit 33 and timing control unit 34 may be allocated to the ASIC or PLD.

ROM 303 stores programs, parameters, etc. necessary for processing executed by CPU 301. Prediction formulas used to estimate biomarkers may be stored in ROM 303. RAM 302 can be used as a work area for calculation processing by CPU 301.

Data supplied from the data logger 25 and biometric indicators estimated by the CPU 31 may be recorded in the ROM 303 or in the auxiliary storage device 304.

The input device is a user interface such as a touch panel or keyboard. The display device 306 displays the results of data processing by the CPU 31 and estimated biometric indicators. The external interface 307 is used to connect to the data logger 25 and the light source device 11 and switch 12 of the measurement optical system 10. The communication interface 308 is used for data communication with a network or external server. The biometric indicators estimated by the CPU 31 may be transmitted to the subject, etc. via the network or external server.

Figures 3, 4, and 5 show examples of the operation of the measurement optical system 10. In Figure 3, light of the first wavelength output from light source 111 is selected. If switching elements 121-123 are shutters, switching element 121 opens and switching elements 122 and 123 close, so that only light emitted from light source 111 is guided to measurement optical element 17. If switching elements 121-123 are on/off switches (or current injection switches) for light sources 111-113, light source 111 is turned on and light sources 112 and 113 are turned off, and light of the first wavelength is output from light source 111.

Light of the first wavelength passes through the multiplexer/demultiplexer 13, 14 and the light-guiding member 15 and enters the measurement optical element 17. As described above, the contact surface 171 of the measurement optical element 17 is pressed against the measurement site of the object to be measured, and the optical signal of the first wavelength that has been attenuated at the measurement site is detected by the photodetector 18 via the light-guiding member 16.

In Figure 4, light of the second wavelength output from light source 112 is selected. When switching elements 121-123 are shutters, switching element 122 opens and switching elements 121 and 123 close, so that only the light emitted from light source 112 is guided to measurement optical element 17. When switching elements 121-123 are on/off switches (or current injection switches) for light sources 111-113, light source 112 is turned on and light sources 111 and 113 are turned off, and light of the second wavelength is output from light source 112.

The light of the second wavelength passes through the multiplexer/demultiplexer 13, 14 and the light-guiding member 15 and enters the measurement optical element 17. The optical signal of the second wavelength, which is attenuated at the interface between the measurement optical element 17 and the object being measured, is detected by the photodetector 18 via the light-guiding member 16.

In Figure 5, light of the third wavelength output from light source 113 is selected. If switching elements 121-123 are shutters, switching element 123 opens and switching elements 121 and 122 close, so that only the light emitted from light source 113 is guided to measurement optical element 17. If switching elements 121-123 are on/off switches (or current injection switches) for light sources 111-113, light source 113 is turned on and light sources 111 and 112 are turned off, and light of the third wavelength is output from light source 113.

The light of the third wavelength passes through the multiplexer/demultiplexers 13 and 14 and the light-guiding member 15 and enters the measurement optical element 17. The optical signal of the third wavelength, which is attenuated at the interface between the measurement optical element 17 and the object being measured, is detected by the photodetector 18 via the light-guiding member 16. In this way, the optical absorption spectrum intensity of the object being measured for light of each wavelength is detected.

Figure 6 is a diagram showing an example of the control timing of the measuring device 1. The operation timing of the switching elements 121, 122, and 123 and the timing of light incidence on the photodetector 18 are shown on the time axis. A value of "1" on the vertical axis indicates that light is being irradiated onto the object being measured, i.e., the shutter is open or the light source is ON. A value of "0" indicates that no light is being irradiated, i.e., the shutter is closed or the light source is OFF.

When switching element 121 is open and switching elements 122 and 123 are closed, measurement light of the first wavelength is incident on photodetector 18. When switching element 122 is open and switching elements 121 and 123 are closed, measurement light of the second wavelength is incident on photodetector 18. When switching element 123 is open and switching elements 121 and 122 are closed, measurement light of the third wavelength is incident on photodetector 18.

When all switching elements 121-123 are closed (off section in the figure), no measurement light is incident on the photodetector 18, but background light, such as light emitted from the object being measured or light emitted from the measurement optical system 10, can be detected.

The detection results of the photodetector 18 when all switching elements 121-123 are closed may be stored in the data processing unit 31 as correction values, and the measurement results for each wavelength obtained at adjacent times may be corrected. For example, the correction value may be subtracted from the received light intensity of each of the three wavelengths of measurement light. This correction can eliminate the effects of characteristic fluctuations due to changes in the ambient temperature or the temperature of the object being measured, fluctuations in the sensitivity of the photodetector 18 due to temperature changes, and the effects of radiated light from the object being measured or the measuring instrument.

The same correction value does not necessarily have to be applied to the detection results of three wavelengths obtained within one switching cycle. For example, a correction value obtained during a certain off-interval may be applied to the detection results of two wavelengths obtained immediately before that off-interval and the detection result of one wavelength obtained immediately after that off-interval. This reduces the time lag between background light measurement and correction.

Correction does not necessarily have to be performed after obtaining the measurement results for all three wavelengths. For example, an off period can be provided between measurements at each wavelength, and measurements and corrections can be repeated in the following order: measurement at the first wavelength → measurement and correction of background light → measurement at the second wavelength → measurement and correction of background light → measurement at the third wavelength → measurement and correction of background light... With this method, the detected value for each measurement wavelength can be corrected using the correction value obtained most recently.

By switching the photodetector 18's detection interval (interval with value "1") and off interval (interval with value "0") at short intervals, the effects of disturbances are effectively eliminated. Furthermore, the contact state between the measurement optical element 17 and the measurement site may change during measurement, causing the absorbance to fluctuate. However, by providing off intervals at short intervals, fluctuations in absorbance due to changes in contact state, etc., can be absorbed, improving measurement accuracy.

Prior to actual measurement, the output intensity of each of light sources 111, 112, and 113 may be changed in stages. The linearity of photodetector 18 can be corrected by checking the change in the light intensity received by photodetector 18 in response to changes in the light source output intensity. Over a very short period of time when no changes can be assumed to occur in the measurement object or measurement environment, the output intensity of the light source and the light intensity received by photodetector 18 should be linearly proportional. However, if the linearity of photodetector 18 is poor, a region will emerge where the output intensity of the light source and the light intensity received by photodetector 18 deviate from the linear proportionality. In this case, the effects of the nonlinearity of photodetector 18 can be eliminated by using only data from the linear region obtained in advance.

<Identifying the wavelength range suitable for measurement>
Next, a description will be given of a suitable wavelength range to be used for measuring a biomarker. In the embodiment, at least a wavelength selected from the following wavelength range is used.
(A) A first wavelength selected from a wave number range of 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less, and a second wavelength selected from a wave number range of 950 cm ⁻¹ or more and 990 cm ⁻¹ or less.
(B) Two wavelengths selected from the wavenumber range of 1130 cm ⁻¹ or more and 1220 cm ⁻¹ or less. In particular, a combination of a first wavelength having a wavenumber of 1156 cm ⁻¹ to 1164 cm ⁻¹ and a second wavelength having a wavenumber of 1164 cm ⁻¹ to 1174 cm ⁻¹ , or a combination of a first wavelength having a wavenumber of 1134 cm ⁻¹ to 1146 cm ⁻¹ and a second wavelength having a wavenumber of 1170 cm ⁻¹ to 1216 cm ⁻¹ , exhibits a high correlation.
(C) Two wavelengths selected from the wavenumber range of 1700 cm ⁻¹ or more and 1760 cm ⁻¹ or less.

When wavelength range (A) is used, a biomarker can be estimated from the absorbance for a first wavelength selected from the range of at least 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less, and the absorbance for a second wavelength selected from the range of at least 950 cm ⁻¹ or more and 990 cm ⁻¹ or less.

When wavelength range (B) is used, a biomarker can be estimated from absorbances for at least two wavelengths selected from the range of 1130 ^{cm −1} or more and 1220 cm ⁻¹ or ^less . When using the absorbance obtained with light of a first wavelength having a wavenumber of 1156 cm ⁻¹ to 1164 cm ⁻¹ and the absorbance obtained with light of a second wavelength having a wavenumber of 1164 cm ⁻¹ to 1174 cm ⁻¹ , or when using the absorbance obtained with light of a first wavelength having a wavenumber of 1134 cm −1 to 1146 cm −1 and the absorbance obtained with light of a second wavelength having a wavenumber of 1170 cm ⁻¹ to 1216 cm ⁻¹ , estimation accuracy is improved.

When the wavelength range (C) is used, a biomarker can be estimated from absorbance at at least two wavelengths selected from the range of 1700 cm ⁻¹ or more and 1760 cm ⁻¹ or less.

It is not necessary to use only one of the wavelength ranges (A) to (C); two or more wavelengths may be selected by combining wavelength ranges (A) to (C). The rationale for these wavelength ranges will be explained using blood glucose levels as an example biomarker.

In this embodiment, blood glucose levels are measured stably and accurately, regardless of individual differences in metabolism from before to after a meal, or differences in the measurement environment or measurement conditions. First, to reduce variability in measurement results due to individual differences, blood glucose measurement results are collected from multiple subjects with multiple different meal contents over a wide time range, from 100 minutes before a meal to 250 minutes after a meal. A correlation is then found between the obtained measurement results and a typical blood glucose level trend (low before a meal, peaking 40 minutes after a meal, then returning to pre-meal levels 180 minutes later), and the wavelength range where the correlation is high is identified.

Furthermore, in order to reduce the influence of variations in the contact state between the measurement optical element 17 (for example, an ATR prism) and the measurement site, the following prediction formula is used:
y=a×x(sense)/x(base)+b (2)
where y is the blood glucose level estimated from the measurement results of the photodetector 18, x(sense) is the absorbance at the sense wavelength sensitive to the molecules contained in glucose, and x(base) is the absorbance at the base wavelength that normalizes the contact state of the ATR prism. x(sense) and x(base) correspond to the absorbance A(k) in equation (1). "a" and "b" are coefficients of a linear equation that performs sensitivity correction from the absorbance ratio [x(sense)/x(base)] to the blood glucose level value y. Since coefficients a and b vary depending on the device and individual differences, they may be updated through learning during data collection and actual measurement. For example, when the sense wavelength is 980 cm ^-1 and the base wavelength is 986 cm ^-1 , the values are a = -1900 and b = 2000.

By dividing the absorbance measured at the sense wavelength by the absorbance at the reference base wavelength, it is possible to correct for variations in the contact condition that occur with each measurement. In this case, the base wavelength and sense wavelength are not distinguished when specifying the wavelength range, so for convenience they are referred to as the base wavelength and sense wavelength. However, in reality, the found base wavelength can be used as the absorption wavelength of the object being measured, and the sense wavelength can be used as the reference wavelength. In other words, the identified base wavelength and sense wavelength are valid even if they are reversed.

As an example, the lips are used as the measurement site, and the light absorption spectrum intensity of the lips is measured. A lamp light source from a Fourier transform infrared spectrometer is used as the light source, and the light absorption spectrum intensity of the lips is measured for each wavelength.

Using the above prediction formula (2), we calculate the correlation with the typical change in blood glucose levels before and after a meal, with the period before and after one meal considered as one series for each person. Blood glucose levels are measured by multiple subjects at random times between 100 minutes before and 250 minutes after eating, with multiple different meal contents.

Figure 7 shows the coefficient of determination for each series, mapped against the sense wavelength (first wavelength) and base wavelength (second wavelength). The horizontal axis is the base wavelength, the left vertical axis is the sense wavelength, and the right vertical axis is the coefficient of determination. The coefficient of determination is expressed as the mean of the square of the correlation coefficient R and is an index of prediction accuracy. As mentioned above, the correlation here is the correlation between the blood glucose levels before and after meals obtained from each individual and their typical trends.

In the map diagram of Figure 7, lighter colors indicate stronger correlation. Regions A, B, and C with high correlation are indicated by circles. Region A is a combination of a sense wavelength of 970±20 cm ^-1 (i.e., 950 cm ^-1 or more and 990 cm ^-1 or less) and a base wavelength of 990±20 cm ^-1 (i.e., 970 cm ^-1 or more and 1010 cm ^-1 or less) (or the reverse combination). Region A is the basis for the above wavelength range (A).

Region B is a combination of a first wavelength and a second wavelength selected in the range from 1130 cm ⁻¹ to 1220 cm ⁻¹ . As will be described later, in region B, there are two combination regions in the range from 1130 cm ⁻¹ to 1220 cm ⁻¹ where correlation is particularly high. Region B is the basis for the above wavelength range (B).

Region C is a combination of a sense wavelength from 1700 cm ⁻¹ to 1760 cm ⁻¹ and a base wavelength from 1700 cm ⁻¹ to 1760 cm ⁻¹ . Region C is the basis for the wavelength range (C) above.

<Measurement in Area A>
8 is an enlarged view of region A. In the range of the first wavelength (e.g., sense wavelength) of 950 ^{cm −1} or more and 990 cm ⁻¹ or less, the coefficient of determination is particularly large at 980 cm ⁻¹ . In the range of the second wavelength (e.g., base wavelength) of 970 ^{cm −1} or more and 1010 cm ⁻¹ or less, the coefficient of determination is particularly large at 986 cm ⁻¹ .

Even when multiple people have different meal contents and measure blood glucose levels at any time between 100 minutes before and 250 minutes after eating, accurate estimation of blood glucose levels can be achieved by selecting the first wavelength in the range of 950 cm ⁻¹ or more and 990 cm ⁻¹ or less and the second wavelength in the range of 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less.

By using two or more wavelengths selected from the above range and estimating blood glucose levels based on the absorbance at each wavelength according to prediction formula (2), stable and accurate measurements are possible from before to after meals.

By using the absorbance ratio, which is the absorbance at the sense wavelength divided by the absorbance at the reference base wavelength, it is possible to correct for variations in the contact condition between the ATR prism and the lips, which varies from measurement to measurement.

9 shows an overlay of absorbance data obtained at a sense wavelength near 990 cm ⁻¹ normalized by a base wavelength near the sense wavelength. Although the normalized absorbance varies depending on the data, when viewed overall, there is a small absorbance peak near 990 cm ⁻¹ . This peak is the absorption peak due to the CH ₂ OH group of glucose. As shown in area A of the determination coefficient map in FIG. 7, it can be seen that a combination of a sense wavelength selected in the range of 950 cm ⁻¹ or more and 990 cm ⁻¹ or less and a base wavelength selected in the range of 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less, or the reverse combination, is suitable for detecting glucose.

Blood glucose levels are measured by setting a first wavelength selected from the range of 950 cm ⁻¹ or more and 990 cm ⁻¹ or less and a second wavelength selected from the range of 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less in the light source device 11 of the measurement device 1 in Figure 1. This allows for stable and accurate non-invasive blood glucose level measurement regardless of individual differences or fluctuations in the contact state between the measurement optical element 17 and the measurement site.

The wavelengths used in biomarker measurement are not limited to one sense wavelength and one base wavelength. For example, in addition to the sense wavelength and base wavelength described above, a second base wavelength may be selected from the range of 950 cm ^{−1 or} more to 990 ^{cm −1 or} from the range of 970 cm ⁻¹ or more ^to 10 ...

The absorbance is defined as a function of the base wavelength so that the ratio of the absorbance at the first base wavelength to the absorbance at the second base wavelength is 1. For example, if the horizontal axis is the base wavelength and the vertical axis is the absorbance at the base wavelength, the intercept is found so that the slope of the line is 1. The absorbance obtained at the sense wavelength is divided by the absorbance at the base wavelength expressed by this linear equation, and the blood glucose level is estimated using prediction equation (2).

As an example, 954 cm ⁻¹ may be set as the first base wavelength, 1202 cm ⁻¹ may be set as the second base wavelength, and the absorbance of the sense wavelength may be divided by the linear equation of the absorbance of the base wavelength.

Figure 10 is a graph showing the absorbance obtained from one sense wavelength and two base wavelengths superimposed on each other. The two base wavelengths are set to 954 cm ^-1 and 1202 cm ^-1 , and the sense wavelength is set to 992 cm ^-1 . As shown in Figure 10, the CH ₂ OH peak near 990 cm ^-1 can be extracted, making it possible to accurately predict blood glucose levels. Using three wavelengths results in a higher coefficient of determination than measurement using two wavelengths.

Instead of setting the sense wavelength and base wavelength to specific wavelengths, an average of a series of wavelengths included in the above wavelength range may be used for at least one of the sense wavelength and base wavelength. For example, the average value of absorbance measured at wavelengths between 992 cm ^-1 and 994 cm ^-1 may be used as the absorbance of the sense wavelength. This makes it possible to reduce measurement noise. Alternatively, the average value of absorbance over a predetermined range of wavelengths may be used in addition to the sense wavelength and base wavelength of specific wavelengths.

The third wavelength may be a wavelength selected from region B or region C. For example, the final predicted value may be obtained by combining the predicted values at the sense wavelength and base wavelength selected in region A with the average value of the predicted values at 1700 cm ⁻¹ to 1760 cm ⁻¹ . Since blood glucose levels are predicted from multiple perspectives, the estimation accuracy is improved.

<Measurement in Area B>
11 is an enlarged view of region B. In region B, the coefficient of determination is large for both the sense wavelength and the base wavelength in the range of 1130 cm ⁻¹ or more and 1220 cm ⁻¹ or less. It can be seen that even within region B, there are combinations of sense wavelength and base wavelength that result in particularly high coefficients of determination.

The first combination is the region located near the diagonal line in Fig. 11. If one of the base wavelength and the sense wavelength is the first wavelength and the other is the second wavelength, then
First wavelength: 1156cm ^-1 ~ 1164cm ^-1
Second wavelength: 1164cm ^-1 ~ 1174cm ^-1
The coefficient of determination is high for the combination of Figure 12. An enlarged view of the vicinity of this combination range is shown in Figure 12.

The second combination is a vertically elongated region extending below the diagonal line in Fig. 11 or a horizontally elongated region extending above the diagonal line. When one of the base wavelength and the sense wavelength is defined as the first wavelength and the other as the second wavelength,
First wavelength: 1134cm ^-1 ~ 1146cm ^-1
Second wavelength: 1170cm ^-1 ~ 1216cm ^-1
The coefficient of determination is high for this combination. Figure 13 is an enlarged view of the vicinity of the second combination range. In Figure 13, by using a sense wavelength of 1134 cm ^-1 to 1146 cm ^-1 for a base wavelength of 1170 cm ^-1 to 1216 cm ^-1 , the correlation between the measurement results and the typical transition of blood glucose levels is high.

Even when multiple people, with different meal contents, are measuring their blood glucose levels at any time between 100 minutes before and 250 minutes after eating, accurate blood glucose levels can be estimated by selecting the sense wavelength and base wavelength from either the first or second combination.

By using two or more wavelengths selected from region B and estimating blood glucose levels based on the absorbance at each wavelength according to prediction formula (2), stable and accurate measurements are possible from before to after a meal. By using the absorbance ratio, which is the absorbance at the sense wavelength divided by the absorbance at the reference base wavelength, it is possible to correct for variations in the contact state between the ATR prism and the lips, which fluctuates from measurement to measurement.

Figure 14 shows data obtained at a sense wavelength around 1160 cm ^-1 overlaid with absorbance normalized by a base wavelength near this sense wavelength. Although the normalized absorbance varies depending on the data, there is an absorbance peak near 1160 cm ^-1 . This peak is an absorption peak due to the pyranose ring of glucose. As shown in the determination coefficient map in Figure 11, it can be seen that two wavelengths selected in the range of 1130 cm ^-1 or more and 1170 cm ^-1 or less are suitable for detecting glucose.

Blood glucose levels are measured by setting a first wavelength and a second wavelength selected from the range of 1130 cm ⁻¹ or more and 1220 cm ⁻¹ or less to the light source device 11 of the measurement device 1 in Figure 1. This allows stable and accurate non-invasive blood glucose level measurement to be achieved regardless of individual differences or fluctuations in the contact state between the measurement optical element 17 and the measurement site.

The wavelengths used in biomarker measurement are not limited to one sense wavelength and one base wavelength. For example, a second base wavelength may be used in addition to the sense wavelength and base wavelength selected in the range of 1130 cm ⁻¹ or more and 1220 cm ⁻¹ or less. The second base wavelength may be selected from the same range of 1130 cm ⁻¹ or more and 1220 cm ⁻¹ or less, or may be selected from region A (range of 950 cm ⁻¹ or more and 1010 cm ⁻¹ or less) or region C (range of 1700 cm ⁻¹ or more and 1760 cm ⁻¹ or less).

When a second base wavelength is used, the absorbance is defined as a function of the base wavelength so that the ratio of the absorbance at the first base wavelength to the absorbance at the second base wavelength is 1. With the base wavelength on the horizontal axis and the absorbance at the base wavelength on the vertical axis, the intercept is found so that the slope of the line is 1. The absorbance obtained at the sense wavelength is divided by the absorbance at the base wavelength expressed by this linear equation, and the blood glucose level is estimated using prediction formula (2). Using three wavelengths results in a higher coefficient of determination than when measuring at two wavelengths.

Instead of setting specific wavelengths as the sense wavelength and the base wavelength, the average absorbance of a series of wavelengths included in region B may be used for at least one of the sense wavelength and the base wavelength. Alternatively, in addition to the sense wavelength and the base wavelength being set to specific wavelengths, the average absorbance of a predetermined wavelength range may be used. For example, the average absorbance between 1162 cm ⁻¹ and 1168 cm ⁻¹ , which has a high coefficient of determination, may be used as the absorbance of at least one of the sense wavelength and the base wavelength, or as the third absorbance. This reduces measurement noise. Since blood glucose levels are predicted from multiple perspectives (glucose components), prediction accuracy is improved.

<Measurement in Area C>
15 is an enlarged view of region C. In region C, the coefficient of determination is large in the range of 1700 cm ⁻¹ or more and 1760 cm ⁻¹ or less for both the sense wavelength and the base wavelength. If one of the base wavelength and the sense wavelength is the first wavelength and the other is the second wavelength, then
First wavelength: 1738±8cm ^-1
Second wavelength: 1748±8cm ^-1
The combination of these wavelengths has a particularly high coefficient of determination. Both the sense wavelength and the base wavelength may be selected from the range of 1743±8 cm ⁻¹ . If the first wavelength is wave1 and the second wavelength is wave2, the particularly bright range enclosed by the frame in FIG. 15 (the range with a large coefficient of determination) is
Wave2=1743cm ^-1 - (wave1-1743cm ^-1 ) ±8cm ^-1
It may also be expressed as:

Figure 16 shows the absorbance of data obtained at a sense wavelength around 1740 cm ^-1 normalized to a base wavelength near this sense wavelength. An absorbance peak is observed between 1741 ^{cm -1} and 1744 cm ^-1 . Around 1740 ^{cm -1} , there is an absorption peak due to carbonyl groups resulting from C=O bonds. Blood glucose concentrations can be regressed by detecting substances related to intracellular glucose metabolism, such as ketone bodies such as acetoacetate and 3-hydroxybutyrate, G6P, F6P, and pyruvate present in the cytosol, and intracellular acetyl-CoA (coenzyme A).

A wavelength selected from region C may be combined with a wavelength selected from region A or a wavelength selected from region B. This allows blood glucose levels to be estimated from various components in the blood, improving measurement accuracy.

Up to this point, we have mainly estimated blood glucose levels using two wavelengths. To further improve estimation accuracy, it is possible to estimate blood glucose levels based on the cross-correlation value between the absorption spectrum of the glucose solution and the absorption spectrum of the person being measured. A high cross-correlation between the absorption spectrum of the glucose solution and the absorption spectrum of the person being measured indicates that the absorption spectrum of the person being measured contains a large amount of glucose.

When estimating blood glucose levels based on cross-correlation values, it is desirable to correct for differences and variations in the degree of contact between the human body and the ATR prism to capture minute changes. For example, before calculating the cross-correlation, it is preferable to pre-process and normalize the absorption spectrum of the glucose aqueous solution and the absorption spectrum of the person being measured. The absorption spectrum of glucose (the substance being measured) may be measured in advance and stored in internal or external memory of the information processing device 30. Pre-processing may be performed by the data processing unit 31 of the information processing device 30. As an example, pre-processing is divided into three stages.

First, the absorbance value of the neighboring wavenumber is divided. The output of the divided spectrum, absorbance_d(i), is expressed by the following formula:
absorbance_d(i) = [absorbance(i)／absorbance (i-r1)]-1 (3)
Here, absorbance(i) is the absorbance at wavenumber i, and r1 is a value that specifies the wavenumber distance at which normalization is performed. When absorbance(i) and absorbance (i-r1) are the same value, the output of the division spectrum will be 0.

Next, to capture minute changes, the value of the neighboring wavenumber is subtracted from the divided value. The output of the difference spectrum, absorbance_d2(i), is expressed by the following formula:
absorbance_d2(i) = absorbance_d(i) - absorbance_d(i-r2) (4)
Here, r2 is a value that specifies how many wave numbers apart the subtraction is to be performed.

Furthermore, to reduce the variation in the data, a moving average filter is applied to the output of the difference spectrum for smoothing. The spectrum abs(i) after applying the moving average filter is expressed by the following equation:

Here, N is the range over which the moving average filter takes the average.

If the absorption spectrum of the glucose aqueous solution that has been pretreated as described above is taken as ref, and the absorption spectrum of the subject that has been similarly pretreated is taken as abs, the cross-correlation value is expressed as follows:

Here, "corr" is the cross-correlation value, "st" is the start wavelength of the wavelength range in which cross-correlation is obtained, and "en" is the end wavelength of the wavelength range in which cross-correlation is obtained. The cross-correlation value "corr" is linearly converted using the following formula to obtain the target bioindicator, which in this example is the estimated blood glucose level "estimated_glucose."
Estimated_glucose=a×(corr)+b (5)
Here, a and b are coefficients obtained by fitting using data labeled with true blood glucose levels.

The above method can be used to estimate blood glucose levels from the absorption spectrum of the subject. However, depending on the wavelength range, the spectrum of the glucose solution may not contain clues such as absorption peaks or dips, or may contain peaks or dips of fluctuating substances other than glucose, which can hinder estimation. For this reason, it may be desirable to use an appropriate wavelength range.

Figure 17 shows the range of high correlation between blood glucose level changes estimated by cross-correlation and typical blood glucose level changes in an average person. For the absorption spectrum of the subject measured at predetermined times after a meal, the correlation coefficient between the estimated blood glucose level (estimated_glucose) calculated from the cross-correlation value and typical blood glucose level changes after a meal in an average person is shown by a shading indicator. The vertical axis represents the start wavelength "st" and the horizontal axis represents the end wavelength "en."

FIG. 17 clearly shows a range (bright region) where the correlation is good. The highest correlation is obtained with the combination of st = 982.5 cm ⁻¹ and en = 997.5 cm ⁻¹ . It is believed that in this wavelength range, the absorption of the CH ₂ OH group of glucose is measured. This wavelength range coincides with the results of the wavelength range (A) suitable for the measurement described above. The use of light with a first wavelength selected from wavenumbers of 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less, and light with a second wavelength selected from wavenumbers of 950 cm ⁻¹ or more and 990 cm ⁻¹ or less in wavelength range (A) also suggests that a high correlation can be obtained in estimation using the cross-correlation value. Using these wavelength ranges enables accurate estimation of blood glucose levels. This estimation method can be applied not only to humans, but also to the estimation of blood glucose levels of animals and other bodily substances.

Wavelength selection has been explained above using blood glucose measurement as an example, but the non-invasive measurement of the present invention is not limited to blood glucose measurement. The technical concepts of the embodiments, such as wavelength selection and determination, can also be applied to the measurement of other biomarkers, such as proteins and tumor DNA in the blood.

As mentioned above, when selecting a third or fourth wavelength in addition to two wavelengths, regions A, B, and C may be combined with one another. This allows stable and accurate measurement of biomarkers regardless of individual differences or differences in the measurement environment.

REFERENCE SIGNS LIST 1 Measurement device 10 Measurement optical system 11 Light source device 111, 112, 113 Light source 12 Switcher 121, 122, 123 Switching element 13, 14 Multiplexer/demultiplexer 15, 16 Light guide member 17 Measurement optical element 18 Photodetector 25 Data logger 30 Information processing device 31 Data processing unit 32 Biological index estimation unit 33 Switching control unit 34 Timing control unit

Patent No. 5376439

Claims (8)

a light source in the mid-infrared region;
a photodetector that detects light output from the light source and reflected by the object to be measured;
an information processing device that determines the blood glucose level of the measurement object based on the output of the photodetector;
Equipped with
the light source emits light having a first wavelength selected from a wavenumber of 970 cm ⁻¹ or more and 1010 cm ⁻¹ or less, light having a second wavelength different from the first wavelength selected from a wavenumber of 950 cm ⁻¹ or more and 990 cm ⁻¹ or less, and light having a third wavelength selected from a wavenumber of 1700 cm ⁻¹ or more and 1760 cm ⁻¹ or less, or 1130 cm ⁻¹ or more and 1220 cm ⁻¹ or less,
the information processing device determines the blood glucose level of the measurement object from a first absorbance of the light of the first wavelength, a second absorbance of the light of the second wavelength, and a third absorbance of the light of the third wavelength.
Measuring equipment. a light source in the mid-infrared region;
a photodetector that detects light output from the light source and reflected by the object to be measured;
an information processing device that determines the blood glucose level of the measurement object based on the output of the photodetector;
Equipped with
the light source emits light having a first wavelength and light having a second wavelength, which are different from each other and selected from a wavenumber of 1130 cm ⁻¹ or more and 1220 cm ⁻¹ or less, and light having a third wavelength, which is selected from a wavenumber of 1700 ^{cm −1 or} more and 1760 cm ⁻¹ or less, 970 cm ⁻¹ or more and 1010 cm −1 or less, or 950 cm ⁻¹ or more and 990 cm ⁻¹ or less;
the information processing device determines the blood glucose level of the measurement object from a first absorbance of the light of the first wavelength, a second absorbance of the light of the second wavelength, and a third absorbance of the light of the third wavelength.
Measuring equipment. the wave number of the first wavelength is 1156 cm ⁻¹ or more and 1164 cm ⁻¹ or less,
3. The measuring apparatus according to claim 2, wherein the wave number of the second wavelength is not less than 1164 cm ⁻¹ and not more than 1174 cm ⁻¹ . The wave number of the first wavelength is 1134 cm ⁻¹ or more and 1146 cm ⁻¹ or less,
3. The measuring apparatus according to claim 2, wherein the wave number of the second wavelength is 1170 cm ⁻¹ or more and 1216 cm ⁻¹ or less. the information processing device estimates the blood glucose level based on a value obtained by dividing the first absorbance by the second absorbance.
The measuring device according to any one of claims 1 to 4. the information processing device determines a normalization line such that a ratio of the second absorbance to the third absorbance becomes 1, and normalizes the first absorbance using the normalization line.
The measuring device according to any one of claims 1 to 5. a measurement optical element disposed between the light source and the photodetector and configured to emit reflected light from the measurement object toward the photodetector;
The measuring device according to any one of claims 1 to 6 , further comprising: the measurement optics is an attenuated total internal reflection prism;
8. The measuring device according to claim 7 .