JPWO1995005120A1 - Non-invasive blood glucose level measurement method and measuring device - Google Patents

Abstract

(57) [Abstract] This publication contains application data prior to electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] Technical Field: Non-invasive Blood Glucose Measurement Method and Device This invention relates to a non-invasive blood glucose measurement method and device that uses spectroscopic techniques to measure blood glucose levels non-invasively. More specifically, it relates to a non-invasive blood glucose measurement method and device that combines wavelength modulation and intensity modulation of light to non-invasively measure glucose in the bloodstream or biological tissues of patients suspected of having diabetes.

BACKGROUND ART Various methods and devices have been proposed for measuring glucose concentrations in vivo or ex vivo using spectroscopic techniques.

For example, International Application No. WO81100622 discloses a method and apparatus for measuring the infrared absorption of glucose in body fluids using a CO2 laser beam as an irradiation light source. According to this method, the absorption spectrum of serum or urine is measured using two wavelengths, λ1 and λ2, through transmission and reflection, i.e., the backscattering effect. In this case, wavelength λ2 is the characteristic absorption wavelength of the substance to be measured (e.g., glucose), and wavelength λ1 is the wavelength at which absorption occurs virtually independently of the concentration of the substance to be measured. This wavelength λ.

Measurement data is obtained by taking the ratio of the absorption values at λ1 and λ2. The absorption band of the substance being measured is 940 to 950 cm' (10.64 to 10.54 μm) at wavelength λ1 and 1090 to 1095 cm' (9.17 to 9.13 μm) at wavelength λ2.

Swiss Patent No. CH-612271 also discloses a non-invasive testing method for detecting biological substances in a sample or through the skin using an attenuated total reflection (ATR) prism. In this method, a waveguide (ATR prism) is placed directly on the sample to be analyzed (e.g., lips or tongue) and infrared light is guided through it. The refractive index of the waveguide is greater than that of the sample medium (the optical skin layer), and infrared light passes through it along a total internal reflection path, thereby interacting with the skin layer. This interaction involves a leakage attenuation component at the reflection site (see J. Hormones & Metabolic Res., Supplement, Ser. (1979), pp. 30-35). At a specific infrared wavelength that is readily absorbed by glucose, the light propagating through the ATR prism is attenuated according to the glucose concentration in the optical skin layer. This attenuation is then determined and processed to generate glucose detection data.

U.S. Pat. No. 3,958,560 also discloses a non-invasive detection device for detecting glucose in a patient's eye.

Specifically, the device disclosed in this U.S. patent consists of a contact lens-type sensor device including a light source that irradiates infrared light from one side of the cornea and a detector that detects the amount of transmitted light on the opposite side. When infrared light is irradiated onto the measurement site, the irradiated infrared light passes through the cornea and aqueous humor and reaches the detector. The detector converts the transmitted light into an electrical signal and supplies it to a remote receiver. A subsequent reader at the receiver outputs the glucose concentration in the patient's eye as a function of the specific changes that occur as the irradiated infrared light passes through the eye.

British Patent Application No. 2035575 discloses a detection device for determining substances, namely CO2, oxygen, and glucose, in a portion of a patient proximate to the bloodstream.

This detector is characterized by comprising a light-receiving means for detecting attenuated light backscattered or reflected from the patient's body, i.e., from the subcutaneous tissue, and uses ultraviolet or infrared light as the irradiating light.

On the other hand, the following devices use similar detection techniques to measure or monitor blood flow and bioactive parameters and components such as oxygenated hemoglobin or reduced oxyhemoglobin:

U.S. Patent No. 3,638,640 discloses a method and apparatus for measuring oxygen and other substances in blood and biological tissues. The apparatus comprises an illuminating light source and a detector placed on the patient's body. When placed at the patient's ear, the detector measures the intensity of light passing through the ear. When placed on the forehead, the detector measures the intensity of light reflected from the blood and subcutaneous tissue. The illuminating light ranges from the red portion of the visible spectrum to the near-infrared wavelengths, i.e., 660 nm, 715 nm, and 805 nm. The number of wavelengths used simultaneously in this method is equal to the sum of at least one wavelength specific to each substance (including the substance of interest) present in the test site, plus one.

By appropriately processing the signals obtained from the absorption of various wavelengths using electronic circuits, quantitative data relating to the concentration of the substance being measured can be obtained without being affected by variations in the detector, variations in the intensity, direction, or angle of the radiation, or variations in blood flow at the test site.

British Patent Application No. 2075668 discloses a spectrophotometric device for measuring and monitoring in vivo, yet non-destructively, metabolic functions of living organs, such as changes in the redox activity of hemoglobin or cytochrome, or fluctuations in blood flow in organs such as the brain, heart, or kidneys.

This spectrophotometric device uses irradiating light in the wavelength range of 700 to 1300 nm, which is effective in penetrating biological tissue to a depth of several millimeters below the skin surface.

Thus, Figure 14 of this British patent application illustrates a reflectance measurement device comprising a light source directing light into a waveguide (fiber optic bundle) applied to the living body. The waveguide is applied to the living body so that the radiation is directed obliquely toward the skin surface, and the directional radiation penetrates through the skin and into the body, where it is reflected or backscattered from the tissue to be analyzed some distance from the light source. A portion of the energy is absorbed and incident on a first detector, which is placed on the skin but spaced from the light source. A second detector is provided in conjunction with the light source, and this second detector detects a backscattered reference signal. The analytical signal from the first detector and the reference signal from the second detector are output to a calculation circuit, the output of which provides data related to the analytical information.

In measuring glucose concentration and other concentrations using near-infrared spectroscopy, the quality of the spectral data obtained by the near-infrared spectrometer is largely determined by the performance of the hardware that makes up the spectrometer. Currently, the highest-performance near-infrared spectrometers have a signal-to-noise (S/N) ratio of approximately 10' to 10'. In contrast, with conventional methods that simply measure the absolute intensity of the spectrum, a spectral signal S/N ratio of approximately 10' to 10' is required to measure, for example, a physiological blood glucose concentration of 100 mg/dL with practical accuracy, which is close to the limit of what a spectrometer can measure.

For this reason, spectroscopic methods for measuring sugar content, glucose, and other concentrations generally suffer from inferior measurement sensitivity, precision, accuracy, and stability compared to chemical analytical methods that use reagents to analyze the concentrations of these substances. Furthermore, high-performance near-infrared spectrometers with high SZN ratios are complex and expensive.

However, rather than simply measuring the absolute intensity of the spectrum as in conventional near-infrared spectroscopy, if a reference method were possible that measured the one-degree change in glucose from the physiological glucose concentration of 100 mg/dL, and if that change could be measured with two- or three-digit accuracy, it would be possible to determine, for example, how much the blood glucose level of a patient suspected of having diabetes has changed from the standard value, which could be useful for managing the patient's blood glucose level.

The objective of the present invention is to provide a method for measuring blood glucose levels that uses a modulation technique that combines wavelength modulation and intensity modulation to non-invasively, easily, and reliably measure the deviation from the standard blood glucose level of a patient suspected of having diabetes, regardless of individual patient differences.

Another object of the present invention is to provide a compact, inexpensive blood glucose measuring device that, with a simple configuration equipped with wavelength modulation means and intensity modulation means, can non-invasively, easily, and reliably measure the deviation from the standard blood glucose level of a patient suspected of having diabetes, regardless of individual patient differences.

DISCLOSURE OF THE INVENTION The present invention features a method for wavelength-modulating light and intensity-modulating it to multiple intensities, irradiating the test site where blood glucose levels are to be measured with this wavelength- and intensity-modulated light, detecting the intensity of the transmitted or reflected light from the test site and the intensity of the light incident on the test site for each of the intensity-modulated light beams, determining the ratio between the two, detecting the rate of change in this ratio relative to the wavelength change caused by the wavelength modulation, and extracting a derivative spectrum of the glucose absorption spectrum at the test site based on this rate of change. Blood glucose levels at the measurement site are then detected based on these derivative spectra.

This allows the test site to be irradiated with light that is wavelength-modulated over a narrow modulation width Δλ centered around the wavelength of interest λ and intensity-modulated. The intensity modulation of the incident light at the test site changes the penetration depth into the skin, extracting information related to the glucose concentration in the area where bodily fluids containing blood components are present. Quantification is also performed using the differential absorption spectrum of glucose in the area where bodily fluids containing blood components are present, allowing for easy and reliable detection of glucose concentrations regardless of individual patient differences.

The differential spectrum is preferably integrated and averaged in response to the repetition of the wavelength modulation.

In this way, by integrating and averaging the differential spectrum, the noise component can be reduced in proportion to the square root of the number of integrations, thereby improving the signal-to-noise (S/N) ratio.

The present invention also includes wavelength-modulated light generating means for generating wavelength-modulated light, intensity modulation means for intensity-modulating the intensity of the wavelength-modulated light from the wavelength-modulated light generating means to a plurality of intensities, beam splitter means for separating the optical paths of the wavelength-modulated and intensity-modulated light incident from the intensity modulation means, light collecting means for collecting light that passes through one optical path separated by the beam splitter means and is incident on a test site where blood glucose level is to be tested and is transmitted through or reflected from the test site, first light detecting means for detecting the intensity of the light collected by the light collecting means, and first light detecting means for detecting the intensity of light that passes through the other optical path separated by the beam splitter. The apparatus is characterized by comprising: second light detecting means; ratio detecting means for detecting the ratio between the output of the first light detecting means and the output of the second light detecting means; differential spectral signal detecting means for receiving a ratio signal corresponding to the ratio from the ratio detecting means and detecting a rate of change of the ratio signal relative to a wavelength change due to the wavelength modulation to detect a differential spectral signal of the absorption spectrum of glucose at the test site; and calculating means for calculating the blood glucose level at the measurement site based on the differential spectral signal detected by the differential spectral signal detecting means for each of the intensity-modulated light beams of multiple intensities.

In this way, the wavelength-modulated light generating means is provided to modulate the wavelength of light, and the wavelength-modulated light is then intensity-modulated and incident on the test site to detect the derivative spectrum of the glucose absorbance spectrum. This allows high-quality derivative data to be obtained in real time without the need for computer processing. Furthermore, the repetitive scanning speed is faster than that of general spectrometers that scan a wide wavelength range, and glucose concentration measurement data can be obtained by short-time photometry without being affected by optical system drift.

The wavelength-modulated light generating means is preferably a wavelength-tunable semiconductor laser.

This allows the use of a semiconductor laser developed for optical fiber communications as the tunable semiconductor laser, effectively maximizing the benefits of the tunable semiconductor laser. This significantly simplifies the configuration of the means for modulating the wavelength of the measurement light, resulting in a simple, compact, non-invasive blood glucose measurement device.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is an illustration of a unimodal spectrum and its first and second derivative spectra.

FIG. 2 is an explanatory diagram of the generation of a derivative spectrum by wavelength modulation spectroscopy.

FIG. 3 shows the absorbance spectrum of an aqueous glucose solution.

FIG. 4 shows the first derivative spectrum of FIG.

FIG. 5 shows the difference absorbance spectrum relative to the water standard.

FIG. 6 shows the differential derivative spectrum.

FIG. 7 shows the differential derivative spectrum.

FIG. 8 shows the differential derivative spectrum.

FIG. 9 shows the differential derivative spectrum.

FIG. 10 is a graph showing the relationship between glucose concentration and first derivative absorbance.

FIG. 11 is an explanatory diagram of the structure of the skin and its optical properties.

FIG. 12 is an explanatory diagram of the incident light intensity and the light penetration depth.

FIG. 13 is a block diagram of one embodiment of a non-invasive blood glucose measuring device according to the present invention.

BEST MODE FOR CARRYING OUT THE INVENTION In order to explain the present invention in more detail, the present invention will be described with reference to the accompanying drawings.

Below, derivative spectroscopy and wavelength modulation methods for obtaining derivative spectra, which are necessary for understanding this invention, are described in sections [1] and [2], respectively. Furthermore, first-order derivative spectra, validation and optimal wavelength selection for glucose quantification, skin diffuse reflectance spectra, and intensity modulation spectroscopy are described in sections [3], [4], and [5], respectively. Furthermore, the configuration of a noninvasive blood glucose measurement device is described in section [6].

[1] Derivative Spectroscopy Wavelength modulation is commonly used to obtain derivative spectra. This wavelength modulation method is described by E.C. Otsuno-Haver in "Potential Clinical Applications of Derivative and Wavelength Modulation Spectrometry" in Clinical Chemistry, Vol. 25, No. 9, 1979, pp. 1548-1553.

Wavelength Modulation Spectroscopy The concepts of spectroscopy and derivative spectroscopy are closely related. They are based on measuring the change in intensity or absorbance with wavelength.

First, we will explain derivative spectroscopy. Derivative spectroscopy involves plotting the intensity or absorbance spectrum versus wavelength, taking its first or higher derivative with respect to wavelength. The goals of derivative spectroscopy are (a) to correct for baseline shifts and (b) to improve the ability to detect subtle spectral features.

FIG. 1 shows a unimodal spectrum and its first and second derivative spectra.

The peak maximum point P-8 of the original spectrum corresponds to the zero-crossing point P a1 in the negative derivative and the central peak point P e in the second derivative. The peak maximum point P 4 v a1 and minimum point P-11 of the negative derivative spectrum correspond to the maximum slope points P and 1°P l of the original spectrum, respectively, and also correspond to the zero-crossing points PO2 and PO3 of the second derivative, respectively.

There are various methods for obtaining a derivative spectrum, including the following:

First, if the spectral data is digital and computer-processable, the derivative spectrum can be calculated by numerical differentiation using software techniques.

Second, derivative spectra can be acquired in real time using hardware techniques by taking the time derivative of the spectrum at a constant scan rate.

This is based on the fact that, if the wavelength scanning rate dλ/di is constant, the derivative of intensity I with respect to wavelength, dl/dλ, is proportional to the derivative of intensity I with respect to time, dl/di, as can be seen from the following equation:

That is, by using an electrical differentiator, the following calculation can be performed:

Third, the derivative spectrum can be obtained by the wavelength modulation method described below.

In the wavelength modulation method, light periodically modulated with a narrow modulation width Δλ centered around a specific wavelength λi is irradiated onto a sample, and the transmitted or reflected light is detected by a detector. The ripple or AC component is isolated from the output signal or measured electrically. The modulation width Δλ represents the spectral band.

When the modulation frequency is sufficiently smaller than the modulation wavelength, the AC component of the photoelectric signal at the modulation frequency generates an AC signal, or differential spectrum D, whose amplitude is proportional to the slope of the spectrum within the modulation wavelength width.

The wavelength modulation techniques mentioned above include (a) vibrating the monochromator slit, mirror, diffraction grating, or prism, (b) inserting a vibrating or rotating refractive mirror into the light beam inside the monochromator, (c) using a continuously variable wavelength filter, (d) vibrating or tailing an interference filter, and (e) vibrating a Fabry-Perot interferometer.

Another possible wavelength modulation technique is to use a continuously variable length semiconductor laser.

A method of varying the oscillation wavelength by controlling the angle of a reflective diffraction grating installed externally to a semiconductor laser is well known. This method allows for wavelength variation while narrowing the spectral linewidth. Implementation is simple when jumps between longitudinal modes are acceptable rather than continuous variation.

To continuously change the lasing wavelength, a single-mode filter that synchronizes with the tuning wavelength over an even narrower bandwidth can be added, allowing single-mode oscillation at any wavelength. These lasers are called external cavity tunable semiconductor lasers.

Furthermore, the June 15, 1987 issue of Nikkei Electronics (No. 423), pages 149-161, describes a wavelength-tunable semiconductor laser being developed for coherent optical communications. This publication describes, for example, a semiconductor laser based on a distributed Bragg reflector single-mode laser that controls wavelength using a three-electrode structure. It also describes a wavelength range of 3.1 nm that smoothly and continuously changes wavelength while maintaining the same longitudinal mode. Furthermore, if longitudinal mode changes are acceptable, the wavelength range is approximately 5 nm.

[2] Wavelength Modulation for Obtaining a Derivative Spectrum In wavelength modulation, when the modulation width Δλ (= λ2 - λ1) is sufficiently smaller than the spectral bandwidth, the AC component ΔI (= ]2-1 +) of the photoelectric signal at the modulation frequency generates an AC signal ΔI/Δλ with an amplitude proportional to the spectral slope within the modulation wavelength width, i.e., a differential spectrum D, expressed by the following second equation. Extracting only the amplitude of the AC signal can be done in real time using an appropriate electrical system.

D = ΔI/Δλ = 12 1 + /λ2 - λ, ... (2) Generally, when measuring low concentrations of glucose, the DC component is larger than the AC component. Removing this meaningless DC component has the advantage of effectively and efficiently utilizing the dynamic range of the A/D converter used in blood glucose measurement devices, as described below. This also has advantages in subsequent mathematical processing.

Wavelength modulation involves periodic scanning back and forth with a narrow modulation width Δλ, allowing for faster, repeated scanning than conventional spectrometers, which scan a wide wavelength range. This facilitates integration and averaging. Furthermore, noise components are reduced in proportion to the square root of the number of integrations, so increasing the number of integrations can improve the signal-to-noise (S/N) ratio. Furthermore, short-time photometry is highly effective in suppressing drift in the spectrometer's optical system.

Although wavelength modulation is limited to a narrow Δλ range, it offers the advantage of obtaining high-quality derivative spectra in real time without requiring computer processing. This makes wavelength modulation suitable for routine analysis of well-characterized samples, such as in quality control and clinical analysis.

In contrast, when performing numerical differentiation on the digital original spectrum, the numerical accuracy and quality of the intensity itself become an issue.

Differentiation increases high-frequency noise, so differentiating a poor-quality spectrum significantly reduces the signal-to-noise ratio.

Furthermore, in measuring low-concentration samples, if the numerical precision and number of significant digits of the intensity Ii of the raw spectrum itself are large, meaningful changes in the derivative spectrum of the target component will not be obtained. In other words, the signal-to-noise ratio must be quite large.

[3] Verification of glucose quantification from first-order derivative spectra. If the spectral data is digital, the derivative spectrum can be calculated by numerical differentiation. Therefore, we numerically differentiated the absorbance spectrum obtained with a Fourier transform spectrometer to obtain a first-order derivative spectrum, and verified the quantitative ability of the wavelength modulation method to quantify glucose concentration.

The samples used were water and aqueous glucose solutions of 1000 mg/dL, 3000 mg/dL, and 5000 mg/dL.

Because it is difficult to observe detailed changes between samples by comparing absorbance spectra or first-order derivative spectra, we calculated the change relative to the reference water, i.e., the difference absorbance spectrum and the difference first-order derivative spectrum, so that the differences could be visually observed. The differentiation was performed from the long wavelength side to the short wavelength side.

First, consider the glucose absorption band located between the water absorption peak at 1.43 μm and 1.93 μm. Figure 3 shows the absorbance spectrum, and Figure 4 shows its first-derivative spectrum. Figure 5 shows the difference absorption spectrum. In the difference absorbance spectrum in Figure 5, glucose absorption is observed between 1.55 and 1.85 μm. Furthermore, an eight-figure characteristic is observed between 1.35 and 1.45 μm. This is due to a shift in the water absorption peak at 1.43 μm, which is caused by hydration. The center wavelength of the wavelength modulation can be selected from the wavelength ranges of 1.6 to 1.67 μm and 1.75 to 1.85 μm, which are at or near the zero-crossing point where interference is not present and where the absorption band slope is steep and the effect of interference is minimal.

As can be seen from the differential first-order derivative spectrum shown in Figure 6, it is clear that glucose concentration can be quantified from the first-order derivative spectrum. Figure 10 shows the relationship between first-order derivative absorbance at a wavelength of 1.555 μm and glucose concentration.

The 1.5 μm wavelength band is easy to implement because it can utilize wavelength-tunable semiconductor lasers developed for fiber optic communications. By applying wavelength modulation to wavelength-tunable semiconductor lasers, the characteristics of wavelength-tunable semiconductor lasers can be effectively and fully utilized.

Above the water absorption peak of 1.93 μm, there are absorption bands for glucose at 21 μm, 2.27 μm, and 2.33 μm. Note the slopes around these absorption peaks. As can be seen from the differential spectrum shown in Figure 7, the peaks are between 2.06 and 2.1 μm.

2.1 to 2.24 μm.

2.24 to 2.27 μm.

2.27 to 23 μm.

2.3 to 2.32 μm.

Can be selected from 2.32 to 2.38 μm.

Similarly, between the absorption peaks of water at 0.96 μm and 1.15 μm, there is a broad absorption band of glucose at 1.06 μm. As can be seen from the differential spectrum shown in Figure 8, the absorption band can be selected from 1.07 to 1125 μm or 1.00 to 1.05 μm.

Similarly, between the absorption peaks of water at 1.15 μm and 1.43 μm, there is a broad absorption band of glucose at 1.25 μm. As can be seen from the differential spectrum shown in Figure 9, the absorption bands can be selected from 1.28 to 1.36 μm and 1.18 to 1.23 μm.

[4] Selection of Optimal Wavelength As shown in Figure 11, the skin consists of the stratum corneum 1, epidermis 2, and dermis 3, from the outside in, and has an anisotropic structure in the depth direction. To transcutaneously measure the glucose concentration in the capillary bed 4, where body fluids containing blood components are present, using the diffuse reflectance method from above the skin, the wavelength used is important and is an integral part of the measurement method.

In the near-infrared wavelength range, the long wavelength region close to mid-infrared light and the short wavelength region close to visible light can be compared relative to one another as follows:

In the long wavelength range, light is absorbed by water present in biological tissues and therefore does not penetrate deep into the body (skin), and is less susceptible to scattering and attenuation due to the small effect of light scattering.

Furthermore, since the absorption coefficient of the glucose absorption band is large, the path length may be shorter.

That is, the light penetration depth may be relatively small.

In the short wavelength range close to visible light, light is less absorbed by water and can easily penetrate deep into the skin. However, light is easily scattered and attenuated.

Furthermore, since the absorption coefficient of the glucose absorption band is small, a long path length is required to increase the measurement sensitivity.

Thus, the selection of the optimal wavelength involves many factors. The optimal wavelength for measuring glucose is preferably selected from the wavelength ranges already discussed, taking into account the characteristic absorption coefficient of glucose, the depth of light penetration into the skin, and, from a practical standpoint, the range of 1.45 to 1.58 μm, since tunable semiconductor lasers for coherent fiber optic communication applications can be used.

[5 Skin Diffuse Reflectance Spectroscopy and Intensity Modulation Spectroscopy] As mentioned above, wavelength modulation allows high-quality differential data to be obtained in real time without the need for computer processing. Differential data is essentially single-point data, and from a practical standpoint, it is important that the data be normalized and that various variables, such as sample temperature changes and component interference, be automatically compensated for. In this invention, wavelength modulation is combined with light intensity modulation to automatically compensate for these variables.

The diffuse reflectance spectrum of skin is based on the signal detected by a detector after weak diffusely reflected light that has been repeatedly absorbed and scattered within the skin and then re-emerges at the skin surface, which is collected by an integrating sphere. In relation to the anisotropic structure of skin in the depth direction, as shown in Figure 11, it is a mixture of (a) the spectral component of specularly reflected light 7 of incident light 5 at the skin surface; (b) the spectral component of diffusely reflected light 8 of incident light 5 from the glucose-free stratum corneum 1 near the skin surface and from the surface tissue; (c) the spectral component of diffusely reflected light 9 of incident light 5 from areas 4 containing blood-containing body fluids; and (d) the spectral component of transmitted light 6 of incident light 5 into deeper tissues. Generally speaking, the contribution of spectral components near the skin surface is large, while the contribution of spectral components from areas 4 containing blood-containing body fluids is small. This is a typical diffuse reflectance spectrum.

When focusing on the glucose concentration at site 4, where body fluids containing blood components are present, it is clear that glucose concentration can be measured more accurately if quantitative analysis can be performed on a spectrum from which the spectral components (a) and (b) above have been removed.

As one method for achieving this, the inventors of the present application proposed the following light intensity f:Il method in Japanese Patent Application No. 62-290821 and U.S. Patent No. 4,883,933.

The light penetration depth into the skin is ensured by varying the incident light intensity. That is, as shown in Figure 12, a stronger incident light intensity contains information from deeper areas than a weaker incident light intensity. The detection limit penetration depth is bl, and the penetration depth at a suitably weak incident light intensity rot is 7. The diffuse reflected light intensity from □ is measured, and the ratio is calculated and normalized using the following equation 3:

A+ = ]0g(1++/1.)" (3) This third equation contains spectral information only near the skin surface.

Next, the diffuse reflection intensity 1t2 from penetration depth b272 at incident light intensity 1112, which is stronger than In at penetration depth b2, is measured, and the ratio is calculated and normalized using the following equation (4):

Az = lOg (lOz/In2) - (4) This fourth equation contains the deeper spectral information. Then, we calculate the difference ΔA between the two.

ΔA = A, -A, = log(l/Lx) - 1ag(Lt/I, +)... (5) Equation 5 above can be considered to normalize the spectrum of the glucose-free tissue near the skin surface of the subject. This means that the spectrum is not affected by individual differences such as race, gender, and age.

The intensity of the incident light can be modulated, for example, by switching between attenuators with different attenuation ratios using a rotating disk, as will be described later.

For each cycle of modulation of the incident light intensity, normalization is performed by calculating the ratio of the intensities, and the difference in absorbance is calculated. Then, by performing an averaging process over multiple cycles, the S/N ratio can be improved.

A regression equation is created using this difference calculation and the reference concentration obtained by chemical analysis for authentic samples of glucose concentration. This regression equation is then used to quantify the glucose in unknown samples.

While the incident light intensity modulation algorithm has been described above using the spectral intensity I, it is known that the derivative intensity can also be quantitatively determined using a regression method. Therefore, in this example, to use the first derivative D = ΔA/Δλ calculated using the wavelength modulation method, the absorbance value A in Equations 3, 4, and 5 is replaced with ΔA/Δλ, and the calculations of Equations 8, 9, and 10, described below, are performed.

[6] Non-invasive Blood Glucose Measurement Device The configuration of a non-invasive blood glucose measurement device is shown in Figure 13.

The noninvasive blood glucose measurement device includes a tunable semiconductor laser 11, an attenuator 12 that periodically varies the intensity of the wavelength-modulated laser light from the tunable semiconductor laser 11, a beam splitter 14 that splits optical path 13 of the wavelength-modulated and intensity-modulated laser light incident from attenuator 12 into optical paths 13a and 13b, and an integrating sphere 18 that collects the laser light that passes through one of the optical paths 13a split by beam splitter 14, enters an examination site 17 on skin 16 where the blood glucose level is to be examined, and is transmitted through or reflected from the optical path 13a.

The noninvasive blood glucose measurement device also includes a first detector 21 that detects the intensity of the laser light focused by the integrating sphere 18, a second detector 22 that detects the intensity of the laser light Φ transmitted through the beam splitter 14, an amplifier 23 that amplifies the output of the first detector 21, an amplifier 24 that amplifies the output of the second detector 22, a logarithmic ratio amplifier 25 that outputs the logarithmic ratio of the outputs of the first and second detectors 23 and 24, a lock-in amplifier 26 that detects a differential spectral signal of the glucose absorption spectrum at the test site 17 from the rate of change of the output of the logarithmic ratio amplifier 4i' Pr2,i5 with respect to wavelength change, and a processing unit 27, comprised of a microprocessor or the like, that A/D converts the differential spectral signal detected by the lock-in amplifier 26 and processes the converted digital value to calculate the blood glucose level at the measurement site.

The wavelength-modulated laser light from the wavelength-tunable semiconductor laser 11, which has been adjusted and controlled to have a center wavelength λ11 and a wavelength modulation width Δλ, is intensity-modulated by an attenuator 12 and then split into two beams by a beam splitter 14.

Laser light L2 passing through the beam splitter 14 is used to monitor the incident light intensity and is converted into an electrical signal L by a second detector 22. The other laser light L+ is incident on the test site 17, which tests the blood glucose level. The diffused reflected light from this test site 17 is collected by an integrating sphere 18 and then converted into an electrical signal L by a first detector 21.

The electrical signals Is and I are amplified by amplifiers 23 and 24, respectively, and then input to logarithmic ratio amplifier 25. Logarithmic ratio amplifier 25 outputs a normalized absorbance signal A = log (Io/Is) (6), which is expressed by the following equation (6).

The absorbance signal A is highly accurate and not subject to drift because the electrical signals I and Io are measured simultaneously by the first detector 21 and the second detector 22 using the same laser light split by the beam splitter 14.

Then, the lock-in amplifier 25 extracts only the amplitude of the AC signal as shown in the following equation 7:

D = ΔA/Δλ (7) The AC component is a signal proportional to the slope of the sample spectrum at the center wavelength of the wavelength modulation.

As already mentioned, the attenuator 12 modulates the intensity of incident light to change the depth of light penetration into the test site 17 of the skin 16, and is configured to switch between two sets of attenuator units 12a, 12b or more sets of attenuator units with different attenuation ratios using a rotating disk 12c. This modulation of the intensity of incident light enables more accurate measurement of the glucose concentration in areas where body fluids containing blood components are present.

The lock-in amplifier 26 outputs an AC signal, D, given by the following equation (8), for an incident light intensity of 1 al from the attenuator 12: D = ΔAt/Δλ (8)

Furthermore, the lock-in amplifier 26 outputs an AC signal, D2 = ΔA4/Δλ (9), calculated using the following equation (9) in response to the incident light intensity 162 from the attenuator 12.

The arithmetic processing unit 27 performs A/D conversion on the AC signals D1 and D2, and performs a difference calculation according to the following equation (10) for each cycle of the modulation of the incident light intensity of the AC signals D1 and D2.

ΔD = D 2 - D + = ΔA z / Δλ - ΔA + / Δλ - (10) The arithmetic processing unit 27 uses the regression equation data stored in advance in a memory (not shown) to determine the glucose concentration at the test site from the value obtained by Equation 10.

When measuring the glucose concentration, performing multi-cycle integration averaging by alternating between attenuator units 12a and 12b of attenuator 12 improves the S/J ratio.

Furthermore, by modulating the incident light intensity over at least three levels, the optimal incident light intensity range for glucose quantification can be determined, allowing for more precise selection of the optimal attenuator 12 and method. This also eliminates the need for a standard diffuser plate, which is used for calibration in conventional diffuse reflectance methods.

Furthermore, unlike the transmission method, if the condition of infinite sample thickness is satisfied, i.e., the light that penetrates the inspection area 17 does not leak out from the bottom of the sample, then information about the thickness of the inspection area is not required.

Claims (4)

[Claims] 1. A method for non-invasively measuring blood glucose levels, comprising: wavelength-modulating light and intensity-modulating it to multiple intensities; irradiating the test site where blood glucose levels are to be measured with this wavelength- and intensity-modulated light; detecting the intensity of the transmitted or reflected light from the test site and the intensity of the light incident on the test site for each of the intensity-modulated light beams; determining the ratio of the two; detecting the rate of change of this ratio relative to the change in wavelength due to the wavelength modulation; extracting a derivative spectrum of the glucose absorption spectrum at the test site based on this rate of change; and detecting the blood glucose level at the measurement site based on these derivative spectra. 2. The method for non-invasively measuring blood glucose levels according to claim 1, wherein the differential spectrum is integrated and averaged in response to the repetition of the wavelength modulation. 3. A wavelength-modulated light generating means for generating wavelength-modulated light, an intensity modulation means for intensity-modulating the intensity of the wavelength-modulated light from the wavelength-modulated light generating means to a plurality of intensities, a beam splitter means for separating the optical paths of the wavelength-modulated and intensity-modulated light incident from the intensity modulation means, a light collecting means for collecting light that passes through one optical path separated by the beam splitter means, enters the test site where blood glucose level is to be tested, and is transmitted or reflected, a first light detecting means for detecting the intensity of the light collected by the light collecting means, and a second light detecting means for detecting the intensity of light passing through the other optical path separated by the beam splitter. A non-invasive blood glucose measurement device comprising: a ratio detection means for detecting the ratio between the output of the first light detection means and the output of the second light detection means; a differential spectrum signal detection means for receiving a ratio signal corresponding to the ratio from the ratio detection means and detecting the rate of change of the ratio signal relative to the wavelength change due to the wavelength modulation to detect a differential spectrum signal of the glucose absorption spectrum at the test site; and a calculation means for calculating the blood glucose level at the measurement site based on the differential spectrum signal detected by the differential spectrum signal detection means for each of the intensity-modulated light beams of multiple intensities. 4. The non-invasive blood glucose measuring device according to claim 3, wherein the wavelength-modulated light generating means is a tunable semiconductor laser.