JP2007259967A - Internal component measuring instrument - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an internal component measuring instrument capable of highly precisely measuring internal components without being affected by external factors lowering the measurement accuracy. <P>SOLUTION: CPU section 19B of an arithmetic device 19 is provided with an absorbance spectrum acquisition section 25 acquiring absorbance spectra from a reflected light from the inside of a living body 14, an internal component calculation section 23 calculating the internal components based on the acquired absorbance spectra and the absorbance spectra stored in an absorbance spectrum database DB stored in the storage section 19b, and a determination/analysis section 24 determining and analyzing the absorbance spectra acquired by the absorbance spectrum acquisition section 25 and the absorbance spectra stored in the absorbance spectrum database DB. The internal component calculation section 23 outputs calculation values of the internal components when a group of a major analysis result acquired by the determination/analysis of the determination/analysis section 24 includes the absorbance spectra acquired by the absorbance spectrum acquisition section 25. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an in-vivo component measuring apparatus that non-invasively measures an in-vivo component concentration such as a blood glucose level (glucose).

  As a health management and medical use, a method of noninvasively analyzing in-vivo components such as glucose, protein, lipid, water, urea and the like without drawing blood is attracting attention. When near-infrared light is used in this analysis method, since the absorbance spectrum of water in the near-infrared region is small, it is possible to analyze an aqueous solution and to have an advantage that near-infrared light easily propagates in the living body.

  On the other hand, the near-infrared signal level is very small compared to the mid-infrared signal level, and the absorption signal of the target body component such as glucose is less than that of other body components such as water, lipids and proteins. Since it is sensitive to changes in concentration, it has been difficult to accurately analyze the target body component using the peak position and peak height.

  In recent years, it has been proposed to use multivariate analysis such as PLS regression analysis in order to improve these deficiencies in near infrared spectroscopy. In this case, practical quantitative analysis using near-infrared light is possible even if the absorption signal in the near-infrared region has a low S / N ratio or changes in the concentration of other body components. become. For example, a method for obtaining a glucose concentration in a target using near infrared spectroscopy has been proposed (Patent Document 1).

  In this method, near-infrared light is projected onto the subject's skin, and light reflected from the skin is received by the optical fiber bundle. A spectrum analysis of the received reflected light is performed, and a first wavelength region having an OH group absorption peak derived from glucose molecules (for example, 1550 to 1650 nm), a second wavelength region having an NH group absorption peak (for example, 1480 to 1550 nm), an absorption signal is detected from a third wavelength region (for example, 1650 to 1880 nm) having a CH group absorption peak. The glucose concentration is determined by multivariate analysis using these absorption signals as explanatory variables.

  In addition, a method for obtaining the concentration of the target component in the medium based on a probability statistical simulation has been proposed (Patent Document 2).

  In this method, the optical path group in the medium is analyzed by a stochastic statistical simulation such as the Monte Carlo method. In addition, a data table showing changes in diffuse reflectance when the absorption coefficient and equivalent scattering coefficient that are optical characteristics of the medium are changed within a predetermined range is created, and then smoothing processing of the diffuse reflectance is performed by a regression analysis method. The correction data table is created by executing.

Next, the measured spectrum obtained by irradiating the medium with light such as near infrared light in the wavelength range of 1000 to 2500 nm and detecting the emitted light therefrom is used as a reference spectrum provided from the correction data table. To determine the concentration of the target component in the medium. Moreover, if a spectrum change caused by a concentration change of a component other than the target component in the medium is calculated from the correction data table, a multivariate analysis such as a principal component regression analysis (PCR) or a multiple regression analysis (MLR) can be used to calculate an actual spectrum. It is said that the concentration of the target component can be obtained.
JP-A-10-325794 JP 2003-50200 A

  Even when the methods disclosed in Patent Documents 1 and 2 as described above are used, there are the following problems.

  In other words, the skin of a living body irradiated with near-infrared light generally has a non-uniform structure, and it is known that there are individual differences in skin thickness and skin structure. Often, the concentration of the subject's target component in the subject is different from the subject's target component concentration measured in the evening of the same day.

  Thus, fluctuations in the daily concentration of the in-vivo component of the subject and other in-vivo components that affect the in-vivo component concentration cause a decrease in the estimation accuracy of the in-vivo component.

  In order to solve these problems, it is desirable to create a calibration curve using more data in order to obtain stable reliability of the estimation accuracy of the target in-vivo component. However, it means a significant increase in the time required for data collection. Further, when glucose, that is, a blood glucose level is selected as the target body component, the glucose absorption signal is very weak. Therefore, even if the amount of data is increased, there is a possibility that the estimation accuracy cannot be sufficiently improved due to the influence of noise.

  Therefore, a method for preparing a calibration curve for quantitative analysis that can estimate the concentration of an in-vivo target component such as glucose with stable accuracy, for example, a plurality of near-infrared absorbance spectra and a reference absorbance spectrum of a living body. Obtain multiple differential absorbance spectra, which are the difference between them, synthesize a second reference absorbance spectrum different from the reference absorbance spectrum for each of the differential absorbance spectra, obtain multiple synthesized absorbance spectra, and use multiple synthesized absorbance spectra It has been proposed to create a calibration curve by multivariate analysis.

  However, it is difficult to realistically reproduce a complete environmental state using simulation. In addition, when the subject measures the probe used for optical measurement, for example, without accurately contacting the measurement unit, the influence of noise becomes very large, and measurement accurate measurement values cannot be obtained. There is a fear that the obtained measured value is also judged to be a correct value and the blood glucose level is calculated.

  The present invention has been made in view of the above-described points, and the object of the present invention is to be able to accurately measure in-vivo components without being affected by external factors that reduce measurement accuracy. It is in providing a component measuring device.

  In order to achieve the above object, according to the first aspect of the present invention, a measurement probe for irradiating a living body to be measured with infrared light, and an absorbance for acquiring an absorbance spectrum from reflected light or transmitted light from the living body. Stored in the spectrum acquisition unit, the absorbance spectrum database storing the absorbance spectrum data of the living body obtained in advance under a plurality of conditions, the absorbance spectrum acquired by the absorbance spectrum acquisition unit, and the absorbance spectrum database Discriminant analysis that performs discriminant analysis using an in-vivo component calculation unit that calculates an in-vivo component based on the absorbance spectrum, an absorbance spectrum acquired by the absorbance spectrum acquisition unit, and an absorbance spectrum stored in the absorbance spectrum database The in-vivo component calculation unit includes the discriminant analysis unit. Groups of major analytical results obtained by another analysis, and outputs the calculated value of the body component when it contains absorbance spectrum obtained by the absorbance spectrum acquisition unit.

  According to the first aspect of the present invention, since the internal component can be calculated after removing the absorbance spectrum affected by an external factor that lowers the measurement accuracy, the effect of being able to accurately measure the internal component. There is.

  The invention of claim 2 is characterized in that, in the invention of claim 1, the absorbance spectrum data stored in the absorbance spectrum database is absorbance spectrum data obtained by simulation.

  According to the second aspect of the present invention, a signal such as a weak glucose signal can be reliably reproduced, and an error between devices that occurs when an absorbance spectrum is acquired from a living body and a database is created. In addition, a database can be constructed from absorbance spectrum data that is not affected by the noise component being measured.

  In the invention of claim 3, a measurement probe that irradiates a living body to be measured with infrared light, an absorbance spectrum acquisition unit that acquires an absorbance spectrum from reflected light or transmitted light from the living body, and a plurality of previously obtained plural Based on the absorbance spectrum database storing the absorbance spectrum data of the living body under the above conditions, the absorbance spectrum acquired by the absorbance spectrum acquisition unit, and the absorbance spectrum stored in the absorbance spectrum database, And a discriminant analysis unit for performing discriminant analysis using the absorbance spectrum acquired by the absorbance spectrum acquisition unit and the absorbance spectrum stored in the absorbance spectrum database. Based on the result of discriminant analysis Characterized in that it comprises a probe adjusting section for adjusting the contact pressure to the living body.

  According to the invention of claim 3, by adjusting the contact pressure to the living body based on the discriminant analysis result of the discriminant analysis unit, it is influenced by external factors that reduce the measurement accuracy such as a change in the skin state of the living body. Therefore, it is possible to obtain an absorbance spectrum that can be determined to be good by discriminant analysis, and as a result, there is an effect that the measurement of in-vivo components can be performed with high accuracy.

  The present invention can calculate an in-vivo component after removing an absorbance spectrum affected by an external factor that lowers measurement accuracy, or can externally reduce measurement accuracy such as a change in a skin state of a living body. Absorbance spectra that can be determined to be good by discriminant analysis can be acquired without being influenced by factors, and as a result, there is an effect that the measurement of in-vivo components can be performed with high accuracy.

  FIG. 1 (a) shows the configuration of the in-vivo component measuring apparatus of the present invention. In order to perform non-invasive quantitative analysis of in-vivo components of a subject to be measured, a halogen lamp or the like as shown in the figure. Light source 1, a diffusion plate 2 that diffuses near-infrared light emitted from the light source 1, a pinhole 3 that passes near-infrared light diffused by the diffusion plate 2, and a near-red light that passes through the pinhole 3 A lens 4 that collects external light, a light incident body 5 that receives near-infrared light collected by the lens 4, a light-emitting side optical fiber 6 that connects one end of the light incident body 5, and a light-emitting side optical fiber 6 The light irradiation part which comprises the probe 8 for a measurement which connects the front-end | tip to the surface of the biological body 14 like a test subject's skin is connected.

  Further, in order to obtain an absorbance spectrum from near-infrared light (reflected light) reflected in the living body 14, a light-receiving optical fiber 10 having one end connected to the measurement probe 8 and the other end of the light-receiving optical fiber 10 are connected. Collected by the connected light emitting body 12, a lens 15a that collects the reflected light emitted from the light emitting body 12, a shutter 21a that opens and closes the optical path between the light emitting body 12 and the lens 15a, and the lens 15a. There is provided a biological signal acquisition optical system including a diffraction grating 16 that receives and splits the reflected light reflected by the reflecting mirror 22 and a light receiving element 17 that receives and detects the reflected light dispersed by the diffraction grating 16. is there.

  Furthermore, in order to correct fluctuations in near-infrared light caused by changes in ambient temperature, positional relationships of optical components, etc., a reference light-emitting side optical fiber 7 having one end connected to the light incident body 5 and the reference light-emitting side optical fiber 7, a reference probe 9 connected to the other end, a reference plate 20 for reflecting near-infrared light emitted from the reference probe 9, and a reflected light reflected by the reference plate 20 with one end connected to the reference probe 9. A reference light-receiving optical fiber 11, a light emitting member 13 to which the other end of the reference light-receiving optical fiber 11 is connected, a lens 15b that collects reflected light emitted from the light emitting member 13, and a light emitting member. A shutter 21b that opens and closes an optical path between the lens 12 and the lens 15b, a reflecting mirror 22 shared by the absorption spectrum acquisition unit, and a living body The is provided a reference signal acquisition optical system including a diffraction grating 16 and the light receiving element 17 for sharing the degree optics.

  Whether reflected light received by the light receiving element 17 is reference light or measurement light is selected by opening and closing the shutters 21a and 21b.

  The light receiving signal output from the light receiving element 17 is A / D converted by the A / D converter 18 and then input to the arithmetic device 19, and the absorbance spectrum is obtained from the light receiving signal input by the arithmetic device 19. By analyzing the absorbance spectrum, the blood sugar level and the like can be calculated.

  Here, as shown in FIG. 1 (b), the arithmetic unit 19 is composed of a computer such as a personal computer having a storage unit 19a and a CPU unit 19b, and the storage unit 19a has an absorbance spectrum data group to be described later. A database creation operation program for creating an absorbance spectrum database DB and an actual operation program for actually measuring the in-vivo components of the subject, and the CPU 19b executes these programs, As will be described later, an in-vivo component calculation unit 23 that estimates and calculates in-vivo components, a discriminant analysis unit 24 that performs discriminant analysis, an absorbance spectrum acquisition unit 25 that acquires an absorbance spectrum from a light reception signal of the light receiving element 17, and an absorbance spectrum database DB For this purpose, the database creation unit 26 is realized as a function.

  FIG. 1 (c) shows an end face of the measurement probe 8. On this end face, the end faces of the strands of the light-emitting side optical fiber 6 are arranged at regular intervals on a circumference having a predetermined radius. In addition, the end face of the light receiving side optical fiber 10 is arranged at the center of the circle.

  Here, the skin tissue of a living organism including human beings is usually formed by three layers of an epidermis layer 30 including a stratum corneum, a dermis layer 31, and a subcutaneous tissue layer 32 as shown in FIG. 2 to 0.4 mm, the thickness of the dermis layer 31 is about 0.5 to 2 mm, and the thickness of the subcutaneous tissue layer 32 is about 1 to 3 mm. In the dermis layer 31, capillaries and the like are developed, and blood glucose concentration ( The glucose concentration in the dermis layer 31 changes following (blood glucose level). On the other hand, the subcutaneous tissue layer 32 is mainly adipose tissue, and water-soluble biological components such as glucose are unlikely to exist uniformly in the subcutaneous tissue layer 32. Therefore, in order to accurately measure the blood glucose concentration (blood glucose level), it is necessary to selectively measure the near-infrared spectrum of the dermis layer 31.

  Therefore, in order to selectively measure the near-infrared spectrum of the dermis layer 31, it is preferable that the light receiving interval L is 2 mm or less, so that the end face of each strand of the light emitting side optical fiber 6 of the measuring probe 8 is set to a predetermined value. The radius is 0.65 mm. In the figure, X indicates the optical path of near infrared light propagation in the living body 14.

Here, the absorbance spectrum data stored in the absorbance spectrum database DB includes the following.
1) When using the data of the absorbance spectrum group obtained by actual measurement,
2) When using the data of the absorbance spectrum group calculated using a simulation as described later,
3) Or a differential absorbance spectrum group in which a standard absorbance spectrum is set (referred to as a center spectrum) from the absorbance spectrum groups obtained by actual measurement or simulation, and the center spectrum is subtracted from the absorbance spectrum group. There are cases where data of

  In addition, when using the data of a differential absorbance spectrum group, the following treatment is performed.

  3-1) When the data of the differential absorbance spectrum group is used as it is as the absorbance spectrum database DB, the absorbance spectrum obtained by subtracting the center spectrum from the measured absorbance spectrum is subjected to discriminant analysis with the absorbance spectrum database to obtain a good absorbance spectrum. The blood glucose is estimated for those determined to be.

  3-2) Before performing blood glucose estimation, one measured spectrum is added to each of the differential absorbance spectrum groups and used as an absorbance spectrum database (the measured spectrum added to the differential absorbance spectrum is referred to as “calibration spectrum”). And).

  The absorbance spectrum database DB thus obtained calculates a blood glucose level from a calibration curve for calculating a blood glucose level by the in-vivo component calculation unit 23 and a spectrum measured for calculating the blood glucose level. This is used for discriminant analysis performed by the discriminant analysis unit 34 in order to determine whether or not the spectrum is suitable for the above.

  Here, the creation of the absorbance spectrum database DB will be described in detail with an example.

(Example 1)
In this example, the absorbance spectrum as data is collected by actual measurement. For example, if the purpose is to estimate a blood glucose level and a calibration curve is created using the blood glucose level as an objective variable, a glucose tolerance experiment is performed. The absorbance spectrum when the blood glucose level is high or low is collected as a pair with the blood glucose level, and a database in which various blood glucose levels and absorbance spectra are paired is created.

  Further, the measured data of the absorbance spectrum group can be used as it is as the data of the absorbance spectrum database DB as in 1), or converted into a differential absorbance spectrum from the center spectrum as in 3).

  Furthermore, even if converted into a differential absorbance spectrum, the absorbance spectrum can be obtained by using the differential absorbance spectrum group data as the absorbance spectrum database DB as in 3-1) or by adding it to the calibration spectrum as in 3-2). Data in the database DB can be used.

  When collecting the data of the absorbance spectrum by actual measurement as in this example, it is more preferable to use data obtained by performing a plurality of glucose load experiments. In other words, since there are differences in various experimental environments such as the room temperature of the room to be collected, the measured position, and the state of the subject (living body 14), the creation of a calibration curve including many of these factors has many experimental factors. This is because the accuracy is improved.

   Next, an example is described about the collection method of the absorbance spectrum database DB in this example.

  First, a set of an absorbance spectrum of a subject's living body and a blood glucose level of a target body component of the living body is measured a plurality of times. In this measurement, the concentration of the target body component is measured at almost the same timing as the measurement of the absorbance spectrum.

  As described above, when measuring such a set of absorbance spectrum and blood glucose level, it is better to collect a large number of absorbance spectra when there is a difference in blood glucose level and when the measured environment is different. Therefore, during the period when the blood sugar level is normal, and after that, the glucose tolerance test is performed, the blood sugar level is increased, and the absorbance spectrum data is collected during the period until the blood glucose level decreases, and the measurement of the respective absorbance spectrum data At this time, it is desirable to collect a set of data for which blood glucose levels are also measured, a plurality of times, for example, by changing the day.

  Further, even when a calibration curve is created for a subject, the living body in which data necessary for creating the calibration curve is measured is not limited to that subject. For example, any person other than the subject may be used.

   Further, when measuring blood glucose level, it is desirable to measure the absorbance spectrum in the wavelength range of 1200 nm to 1880 nm because the absorption derived from glucose molecules is large and the influence of absorption derived from water molecules is relatively small. .

  In addition, collection of absorbance spectra for the absorbance spectrum database DB is not performed using a living body, but a scatterer suspended solution such as a fat emulsion (eg, “Intralipid” manufactured by Fresenius Mold Co., Ltd.) is used as a simulated sample. By using the absorbance spectrum data at that time as a substitute for the absorbance spectrum data of the living body, the absorbance spectrum database DB may be constructed.

  In this case, if a solution containing 2 to 5% of the fat emulsion “Intralipid” is brought in as a simulation sample, measurement and measurement results in a state close to a living body can be obtained. Then, change the mixing ratio of body components such as glucose, protein, lipid, and water to the fat emulsion and the solution temperature within a preset range to cover their daily fluctuations, and create multiple types of test solutions For each of the obtained test solutions, an absorbance spectrum is measured, and these may be used as a simulated absorbance spectrum of a living body.

(Example 2)
Although Example 1 was a method of obtaining data by actual measurement, in this example, the absorbance is reduced by simulation of the above 2) so that there is little variation when discriminant analysis is performed and more accurate discrimination can be performed. This is an example of collecting spectrum data. For example, the absorbance spectrum database DB is constructed by obtaining the absorbance spectrum data using a light propagation simulation such as the Monte Carlo method.

  The Monte Carlo method adopted here uses a function based on the probability distribution of the target event for a uniform random number in the range of 0 to 1 generated by a computer. In this method, when light propagation is reproduced, the light incident on the medium is regarded as a collection of photons, and the behavior of each photon in the medium is expressed as an optical characteristic value (μα: absorption coefficient, μS: scattering). Tracking is based on the coefficient, p (θ): scattering phase function, n: refractive index).

  As a result, light propagation can be statistically reproduced from the behavior of all photons. Note that θ of the scattering phase function is an angle change in the traveling direction of light caused by one scattering when symmetry is assumed with respect to the azimuth angle.

  An effective description of the scattering phase function is expressed by the Henry-Greenstein function as shown in the following equation (1), and is often used to express scattering of biological tissue including red blood cells.

  In the equation (1), “g” is an anisotropic scattering parameter, and the scattering anisotropy expressed by p (θ) can be characterized more easily. g takes a value from 1 to −1, and when g = 1, 0, −1, the scattering characteristic is represented by perfect forward scattering, isotropic scattering, and backscattering corresponding to each. Scattering characteristics in a living body are expressed by very strong forward scattering, and scattering characteristics of a medium having weak absorption can be expressed almost by isotropic scattering.

  Also, based on the above approximation, the scattering coefficient μS is provided by the equivalent scattering coefficient μS = (1−g) μS. The optical path length L between two consecutive interactions (absorption, scattering) in the propagation of light is expressed by the following equation (2). Changes in the zenith angle and azimuth angle of refraction during interaction (θ, φ) are expressed by the following equations (3) and (4), respectively.

In the above formula, each of R ₁ , R ₂ , R ₃ is a uniform random number between 0 and 1, and ∫ (θ) is the cumulative probability of the scattering phase function. In the first interaction, photon energy is absorbed based on the light absorption characteristics of the medium, and is expressed by the following equation (5).

  By the way, the skin of a living body includes a stratum corneum, a granular layer, a spiny layer, and a basal layer. The basal layer consists of a single layer of cells, and new cells are created at a certain level by cell division. New cells move upward toward the skin surface and change in the order of spinous cells → granule cells → keratinocytes, resulting in the formation of a stratum corneum. In the case of normal keratinocytes, one layer of keratinocytes peels off every day. The epidermis layer is reborn about every two weeks after the basal cells are newly born to become horny cells, about two weeks until the horny cells reach the skin surface and peel off. This means that the epidermal layer is composed of more than 20 cell layers. In the dermis layer, there are sweat glands, sebaceous glands and hair roots in addition to blood vessels, lymph vessels and nerves. Also, due to active physiological activity in the dermis layer, nutrients supplemented from blood are used to create cells in the basal layer. Furthermore, the dermis layer has a fibrous tissue that provides skin elasticity while retaining the skin. Subcutaneous tissue is thought to be formed mainly by subcutaneous fat. Therefore, it can be considered that glucose is mainly present in the dermis layer and almost no glucose is present in the epidermis layer or subcutaneous tissue.

  FIG. 3 shows a simulation model of the skin. In the figure, numbers 30, 31, 32 and 33 are the epidermis layer, dermis layer, subcutaneous tissue, muscle and bone, respectively. For example, near-infrared light irradiated on the skin from the light-emitting side optical fiber 6 propagates along the optical path X and is received by the light-receiving side optical fiber 10.

  Here, when the light propagation simulation is performed based on the Monte Carlo method, the optical parameters such as the absorption coefficient μα, the scattering coefficient μS, the refractive index n, and the anisotropic scattering parameter g of each of the epidermis layer 30, the dermis layer 31, and the subcutaneous tissue 32 are used. A characteristic value is required. In general, the “g” of living tissue is a very strong forward scatter of 0.8-0.95. Therefore, “g” used in this example is set to be constant 0.9. In addition, the refractive index n of these layers is also constant 1.34.

  Then, in vivo parameters such as glucose, protein, lipid, moisture and temperature, and scattering coefficient are changed within a preset range so as to cover the daily fluctuation of the in vivo parameters, and the creation of the absorbance spectrum by simulation is repeated. In this way, a plurality of simulated absorbance spectra are obtained. As a result, instead of measuring data from a living body, a data table, that is, an absorbance spectrum database DB can be obtained by simulation.

  When the data of the absorbance spectrum database DB is created using simulation as in this example, there are the following advantages.

  That is, in the simulation, a daily fluctuation range of the concentration in the target body component such as glucose is determined in advance, and a difference value is simulated at equal intervals around the origin “ORG” in the fluctuation range. When measuring data necessary for creating a calibration curve from a living body, it is difficult to obtain differential values at equal intervals within a short time.

  However, according to the simulation, it is possible to quickly obtain an optimum difference value for creating a calibration curve having a stable and reliable estimation accuracy without collecting a lot of data.

Further, it is very difficult to determine whether the actually measured data includes a weak signal of glucose. However, according to the simulation, such a weak signal of glucose can be reliably reproduced. Furthermore, it is possible to prevent the influence of an error occurring between devices for measuring an absorbance spectrum from a living body and a noise component generated during the measurement on a weak glucose signal. Therefore, there is no accidental correlation in the simulation. The great advantage of using the absorbance spectrum data from another simulation is data that does not depend on the environment for each measurement. It does not affect the selection criteria. When the absorbance spectrum database DB is a data group collected by actual measurement, noise due to the actual measurement environment that created the absorbance spectrum database DB is mixed in the data.

  In this case, the discriminant analysis criteria described later and the absorbance spectrum database DB used for the creation of the calibration curve become data biased toward a certain actual measurement environment, and the data selection criteria become ambiguous. On the other hand, by using the data obtained by the simulation for the data group used as the criterion, it is possible to select the data based on the consistent criterion without depending on the actual measurement environment.

  Moreover, after the data by the said simulation are initially stored in the memory | storage device 19a as a differential absorption spectrum as shown below, and the calibration spectrum mentioned above is measured, this difference is used as a fluctuation | variation amount of the component in the living body from there. A value obtained by adding an absorbance spectrum to the previously defined calibration spectrum may be used as an absorbance spectrum database DB for discriminant analysis and creation of a calibration curve described later.

  By doing so, the absorbance spectrum database DB becomes a spectrum group that better compensates for changes in the absorbance spectrum of the subject by compensating for changes from the actually measured calibration spectrum as a reference.

  The differential absorbance spectrum may be created by actually measuring a living body and collecting a plurality of absorbance spectra, or may be created based on a simulation absorbance spectrum group as described above.

  For example, the absorbance spectrum group of the simulation is created in consideration of various fluctuation factors, for example, in this case, particularly the concentration and volume of the component in the living body. The absorbance spectrum at an intermediate value of concentration or volume is defined as the center spectrum. Using this center spectrum as a reference value, the difference between each component and each absorbance spectrum that varies depending on the concentration distribution among the components and the center spectrum is taken, and discriminant analysis and a calibration curve are created using this value as the difference spectrum. Therefore, the absorbance spectrum database DB is necessary. As the center spectrum, an average spectrum of a spectrum group created by simulation may be used.

  As described above, the method of using the differential absorbance spectrum group of the simulation as the absorbance spectrum database DB is not limited to creating a spectrum by adding the calibration spectrum. A method of using the spectrum obtained by subtracting the center spectrum from the absorbance spectrum actually measured for measuring the blood glucose level as the absorbance spectrum database DB as the differential absorbance spectrum may be used.

  Furthermore, in creating the absorbance spectrum data of the simulation, it is preferable to consider as a parameter the main factors that are likely to vary and affect the measurement. For example, when glucose is the measurement target, parameters such as in vivo protein, lipid, temperature, water content, and scattering coefficient are selected in addition to glucose, and calculated so as to have a change amount that takes into account the biotransformation. The data group obtained from is used as the absorbance spectrum database DB.

  The calibration curve at this time is created by multivariate analysis of the absorbance spectrum database DB composed of the above-described differential absorbance spectrum group. The calibration curve will be described later.

  Similarly, the discriminant analysis is performed by subtracting the center spectrum from the measured spectrum to form a differential absorbance spectrum, and comparing with the absorbance spectrum database DB configured by the above-described differential absorbance spectrum group.

  The discriminant analysis performed in the discriminant analysis unit 24 is performed using a method of analyzing and classifying data similarity such as principal component analysis, cluster analysis, and pattern classification. In this embodiment, principal component analysis is adopted. Yes.

  Here, the principal component analysis employed in the present embodiment is one of multivariate analyses, and is very effective in reducing data. The main vector of the entire sample is calculated as the distribution structure of the samples represented by multidimensional vectors, and the degree of similarity of each sample with respect to the vector is calculated. Specifically, when singular value decomposition is performed on the data and the data is viewed as a matrix, the data is decomposed into eigenvectors normalized in the column direction and eigenvectors normalized in the row direction. Data projection is performed using these values. The direction coordinate of the axis that captures the most sample variance is called first loading, and the projection of each sample onto that axis is called the first score. Subsequently, the second loading and the second score are called in descending order of contribution.

  On the other hand, the change in glucose in the body, which is the measurement target of the in-vivo component measurement device of the present invention, is very small. Therefore, even if it considers from the surface of the contribution to an absorbance spectrum, an influence will become very small. That is, from the viewpoint of the main component described above, the influence of the blood glucose level hardly appears in the loading with the very high contribution such as the first and second. A result that shows an influence on the loading of a slightly higher number of components is obtained, and a result that an influence on glucose appears around the seventh and eighth loadings.

  Here, in this embodiment, the discriminant analysis unit 24 performs principal component analysis using the measured absorbance spectrum A and the absorbance spectrum database DB as one data group, and uses the obtained loading and score to perform discrimination. The score of the number of components defined to be used for viewing is stored in the storage device 19a, and the score range is set using only the score of the absorbance spectrum database DB among the scores.

  Specifically, the discriminant analysis unit 24 calculates the maximum value and the minimum value of the score values in the absorbance spectrum database DB, and the measured absorbance spectrum score value exists in the range of the maximum value and the minimum value. If it is within the range, it is determined that the absorbance spectrum is suitable, and if it is outside the range, it is determined to be inappropriate.

  For example, in the case of the result shown in FIG. 4A, the measured spectrum (one point at the center of the circle in the figure) is between the maximum value and the minimum value of the score values obtained from the absorbance spectrum database DB. Since the score value exists, it is determined to be suitable.

  Here, if it is determined that the spectrum measured by the discriminant analysis unit 24 is appropriate, the blood glucose level is calculated using the measured spectrum and the already stored calibration curve.

  If it is determined that the spectrum measured by the one-discriminant analyzer 24 is not suitable, the measured spectrum is discarded from the storage device 19a.

  In the example shown in FIG. 4A, the number of principal components used as an example is tested as one, but the number of principal components used in the test is not limited to one. Using two or more principal component numbers, data may be discriminated from similarity in a plane or multi-dimension instead of one axis (an example is shown in FIG. 4B).

  Here, an example of the result of selecting data by principal component analysis is shown in FIG. 5, FIG. 6, and FIG. The principal component analysis here was performed using an absorbance spectrum database DB and an absorbance spectrum obtained by measuring a living body in order to estimate blood glucose. Further, as the absorbance spectrum database DB, a differential absorbance spectrum created by simulation was used.

  First, FIG. 5 shows a blood glucose level (A) obtained from a blood collection type blood glucose meter, and a blood glucose level (B) measured by the in-vivo component measuring apparatus of the present invention shown in FIG. 1 at the same time. 6 (a) to (c) and FIGS. 7 (a) to (c), after a series of measurements, the one with good estimation accuracy and the one with poor estimation were randomly selected for each of the three measurements. The principal component analysis was performed for each using the measured absorbance spectrum. In FIG. 5, the data determined to have good estimation accuracy are data C, D, and F, and the data determined to have poor estimation accuracy are data A, B, and E. FIGS. 6 (a) to 6 (c). Corresponds to data C, D, F, and FIGS. 7A to 7C correspond to data A, B, E. In FIGS. 6 and 7, the number of principal components on the x axis is 7 and the number of principal components on the y axis is 8.

  In addition, in the creation of absorbance spectrum data for simulation, six parameters of protein, lipid, temperature, water content, scattering coefficient, and glucose are calculated so as to have a variation range that takes into account biological changes, and obtained from there. These data groups were used as the absorbance spectrum database DB.

  As a result, the three data A, B, and E having poor estimation accuracy are plotted at positions outside the absorbance spectrum database DB in the result of the principal component analysis, and the three data C, D, and F having relatively good estimation accuracy are plotted. Is very close to the absorbance spectrum database DB.

 From this, it can be said that only data with good estimation accuracy can be selected by selecting data using principal component analysis.

  Next, a calibration curve used for the in-vivo component calculation unit 23 will be described.

 The calibration curve used in the in-vivo component calculation unit 23 is created by performing multivariate analysis such as PLS regression analysis. In the case of this embodiment, using PLS regression analysis, a plurality of absorbance spectra (referred to as explanatory variables) and values of components (referred to as objective variables) that are measured in pairs with each other and are later quantified are used. Created using data groups. In this embodiment, the explanatory variable corresponds to the absorbance spectrum database DB, and the objective variable corresponds to the blood glucose level.

   After the calibration curve is created, the absorbance spectrum is measured, and the in-vivo component calculation unit 23 estimates (calculates) the blood glucose level from the measured and acquired absorbance spectrum by using the calibration curve. By using this calibration curve, the concentration value of the target in-vivo component of the subject can be estimated with good accuracy from the near-infrared absorbance spectrum measured noninvasively from the subject.

  Thus, in the in-vivo component measuring apparatus of the present embodiment, the in-vivo component calculating unit 23 creates a calibration curve by performing multivariate analysis using the absorbance spectrum database DB, and then obtains it by measuring with this absorbance spectrum database DB. The discriminant analysis unit 24 discriminates and analyzes the absorbance spectrum thus obtained, and determines whether the spectrum is appropriate or not. If it is determined that the spectrum is appropriate, the endogenous component calculation unit 23 first determines based on the determination result. Using the prepared calibration curve, the blood glucose level is estimated and calculated, and when it is determined to be inappropriate, the absorbance spectrum obtained by measurement is discarded.

   At this time, whether or not the measured spectrum is appropriate based on the absorbance spectrum database DB is judged because the spectrum having a high similarity to the spectrum group used for the calibration curve is more than the spectrum having a low similarity. This is because it is easy to predict accurately.

Next, the operation of the in-vivo component measuring device of the present embodiment will be described with reference to examples.
Example 1
In this example, the absorbance spectrum database DB is prepared using the above-described Monte Carlo method simulation and the differential absorbance spectrum <Example of 3-2) described above>.

  First, before actually measuring the blood glucose level, in the arithmetic unit 19, the CPU 19b executes the database creation program to cause the database creation unit 26 to function, and this database creation unit 26 causes the database creation unit 26 to function as shown in FIG. As described above, the absorbance spectrum database by simulation is created by simulation using the Monte Carlo method (step S1). Further, from this created absorbance spectrum database, the center spectrum (= the intermediate value of the concentration and volume of each component) A series of operations for creating a difference spectrum group by taking the difference from the (absorbance spectrum) and storing the difference spectrum group in the storage unit 19b is performed, and the previous stage of database creation is completed.

  Thereafter, before actually measuring the blood glucose level, the absorbance spectrum database DB for finally creating the absorbance spectrum database DB using the in-vivo component measuring apparatus of the present embodiment is measured once using the ecology 4.

  Next, the actual measurement process will be briefly described.

  First, the above-described database creation program is executed by the CPU 19b of the arithmetic unit 19 to cause the database creation unit 26 to function, and then the distal end surface of the measurement probe 8 is applied to the surface of the living body 14 such as a human arm with a predetermined pressure. The light source 1 is caused to emit light while being in contact with. Thereby, the near infrared light emitted from the light source 1 is applied to the surface of the living body 14 from the tip of the measurement probe 8 via the incident body 5 and the side optical fiber 6, and the irradiated near infrared light is further irradiated to the living body 14. After being reflected or diffused in the light, it becomes reflected light and is received by the light receiving side optical fiber 10 from the tip of the measurement probe 8.

  The received reflected light is emitted from the light emitting body 12 through the light receiving side optical fiber 10, is incident on the diffraction grating 16 through the shutter 21 a, the lens 15 a, and the reflecting mirror 22, and is then spectrally detected. The The light receiving signal output from the light receiving element 17 is AD converted by the A / D converter 18 and then input to the arithmetic unit 19.

  Here, the database creation unit 26 of the CPU unit 19b of the arithmetic unit 19 uses the function of the absorbance spectrum acquisition unit 25 to receive the light reception signal (reference signal) by the reflected light for reference and the light reception by the reflected light from the living body 14. An absorbance spectrum is acquired from the signal (biological signal). Here, the absorbance Abs is expressed as Abs = log 10 (Ref / Sig), where Ref is a reference signal and Sig is a biological signal. The database creation unit 26 adds the acquired absorbance spectrum to each of the difference spectrum groups previously stored in the storage unit 19a, creates an absorbance spectrum database DB, stores it in the storage unit 19a, and ends the database creation. (Step S3).

  This completes the preparation for measuring the actual blood glucose level of the subject.

  Next, when measuring the actual blood glucose level of the examiner, the CPU 19b of the arithmetic device 19 executes the program for actual measurement operation from the storage unit 19a to set the actual measurement operation state.

  In such a state, the biological signal Sig and the reference signal Ref of the subject are measured as described above, that is, spectrum measurement is performed (step S4). Then, from the spectrum obtained by this measurement, an absorbance spectrum is obtained by the absorbance spectrum acquisition unit 23 of the CPU 19b (step S5).

Then, the discriminant analysis unit 24 of the CPU 19b performs the above-described principal component analysis using the acquired absorbance spectrum and the data of the absorbance spectrum database DB (step S6), and further determines the existence range of the score values in the absorbance spectrum database DB. Calculate (step S7). In the data shown in FIG. 4A, the existence range is from −15.8 × 10 ⁻⁴ to + 4.9 × 10 ⁻⁴ .

  It is determined whether or not the obtained absorbance spectrum score value exists within the calculated existence range (step S8). If it is out of the range, the obtained absorbance spectrum is discarded and the absorbance spectrum is measured again (step S4). Return to).

  If it is within the range, it is determined that the acquired absorbance spectrum is suitable for blood sugar level estimation calculation, and based on this determination, the in-vivo component measurement unit 23 performs multivariate analysis based on the absorbance spectrum and the PLS regression analysis as described above. The blood glucose level is estimated and calculated using a calibration curve created in advance using analysis and stored in the storage unit 19a, and the result is output.

In the case of the data shown in FIG. 4A, the score value of the acquired absorbance spectrum is −17.9 × 10 ⁻⁴ , and thus is within the existing range shown above, and this absorbance spectrum is suitable. It is judged. Actually, the blood glucose level calculated from the calibration curve and the obtained absorbance spectrum of FIG. 4A is 119 mg / dl, and the blood glucose level measured using a blood sampling blood glucose meter for the same subject immediately before the actual measurement. Was 117 mg / dl, demonstrating that very good estimation accuracy was obtained.
(Example 2)
In Example 1 described above, the absorbance spectrum database DB is created by simulation, but in this example, a data group of absorbance spectra obtained by actual measurement is used. Since the apparatus configuration, discriminant analysis, etc. other than the creation of the absorbance spectrum database DB are the same as those in the first embodiment, only database creation will be described.

In this example, the subject for creating the database does not eat meals and beverages for about 4 hours after meals, measures the absorbance spectrum of the living body 14 in the fasted state using the apparatus of this embodiment, and further calculates the absorbance spectrum. Immediately after the measurement, the blood glucose level at that time is measured with a blood-collecting blood glucose level.
Then, the absorbance spectrum data and blood glucose level data obtained by these measurements are stored in the storage unit 19a as one set. This process of measuring and storing is performed every 5 minutes until 2 hours have elapsed from the start of measurement.

  Further, after 2 hours, sugar loading is performed. In this case, a toleran (oral glucose tolerance test drink, manufactured by Shimizu Pharmaceutical Co., Ltd.) is taken to promote an increase in blood sugar level. Confirm that the blood glucose level has risen from the fasting blood glucose level and then decreased to the blood glucose level at the start of the experiment, or continue the above measurement / memory process until 4 hours have elapsed after the glucose load. Do it.

  The above process is further performed for 5 days by changing the measurement date, and about 300 sets of absorbance spectrum data and blood glucose level data are collected, and an absorbance spectrum database DB is created and stored in the storage unit 19a.

  Then, using the obtained absorbance spectrum database DB as an explanatory variable and blood glucose level as an objective variable, a calibration curve is created and stored in the storage unit 19a by PLS regression analysis which is one of multivariate analyses.

After that, the blood glucose level of the actual subject is measured using the absorbance spectrum database DB and the calibration curve. The discriminant analysis and blood glucose level estimation calculation in this blood glucose level measurement are the same as those in the first embodiment. Omitted.
(Embodiment 2)
By the way, when measuring the absorbance spectrum of the living body 14, it is generally important to bring the measuring probe 8 into contact with the measurement site at a constant pressure, but it is difficult to contact the skin with a constant pressure. As a result, even if the skin can be brought into contact with the skin at a constant pressure, the skin condition changes from moment to moment due to the measurement probe 8 coming into contact with the skin many times and other factors. Therefore, the absorbance spectrum often changes greatly by pressing at a constant pressure.

  Therefore, the present embodiment is characterized in that a probe adjusting unit 40 for controlling the contact pressure of the measurement probe 8 to the living body 14 to be constant is provided as shown in FIG.

  The probe adjustment unit 40 includes a stepping motor 41 that moves the measurement probe 8 up and down, and a controller that controls the stepping motor 41 based on the analysis result of the discrimination analysis unit 24 by the CPU unit 19 b of the arithmetic unit 19. 42.

  That is, when an absorbance spectrum obtained from the living body 14 to be measured is included in the area of the score of the absorbance spectrum database DB obtained by the discriminant analysis unit 24, calculation of in-vivo components is performed using the absorbance spectrum and the calibration curve. Although it is the same as in the case of Embodiment 1 in that the value (blood glucose level) is output, the controller 42 determines the contact pressure of the measurement probe 8 to the living body 14 when it is not included in the score area of the spectrum database DB. In order to control, the movement of the stepping motor 41 is controlled, and the absorbance spectrum is remeasured.

  Thus, in the in-vivo component measuring apparatus of this embodiment, when the measurement is started as shown in FIG. 10, the controller 42 of the probe adjusting unit 40 gives a control signal to the stepping motor 41 under the control of the arithmetic unit 19. (Step S10), the stepping motor 41 is advanced (Step S11). As a result, the measurement probe 8 attached to the stepping motor 41 advances toward the skin of the living body 14 (step S12), and the operation of the stepping motor 41 is stopped after a predetermined time has elapsed from the output of the control signal (step S13). It is assumed that the measurement probe 8 is held by a holder provided on a table on which the arm of the living body 14 is placed.

  After the stop, spectrum measurement is started in the same manner as in the actual measurement of the first embodiment (step S14), and in step S15, the arithmetic unit 9 acquires the absorbance spectrum of the living body 14 from the biological signal Sig and the reference signal Reg, and stores the storage unit 19a. To remember.

  The discriminant analysis unit 24 performs principal component analysis using the acquired absorbance spectrum and the absorbance spectrum database DB created in advance as in the first embodiment (step S16), and the score value of the absorbance spectrum database B in step S17. In Step S18, it is checked whether or not it exists in the existing range of the score value of the absorbance spectrum database DB. If the score value of the absorbance spectrum acquired from the living body 14 is included, the acquired absorbance spectrum is included. Determines that it is suitable for estimating and calculating the blood sugar level, and passes the determination result to the in-vivo component calculation unit. Based on the determination result, the in-vivo component calculation unit 23 estimates and calculates the blood glucose level using the calibration curve as in the first embodiment (step S19).

  On the other hand, if it is not within the existing range of the score values in the absorbance spectrum database DB, the discriminant analysis unit 24 discards the absorbance spectrum acquired from the living body 14 and controls the controller 42 of the probe adjustment unit 40 to the stepping motor 41. Instruct to output a signal. Thus, the control of steps S10 to S13 is performed again by the probe adjusting unit 40. Thus, the control of the stepping motor 41 is repeated until the absorbance spectrum acquired from the living body 14 falls within the range of the score values in the absorbance spectrum database DB.

  The absorbance spectrum database DB may be prepared in advance by any of the methods described in the first embodiment.

  In the second embodiment, the measurement probe 8 is advanced in the direction of the living body 14 to change the contact pressure. However, the spectrum measurement may be performed by changing the position of contact with the living body 14. Moreover, you may carry out combining the position change and contact pressure.

  Further, the measurement may be performed by changing the contact pressure by controlling the measurement probe 8 so as to be retracted from the state in which the measurement probe 8 is in contact with the skin of the living body 14.

  It is also possible to set a maximum load value (maximum contact pressure value) and control the stepping motor 41 to limit the degree of pressing the measurement probe 8 against the skin of the living body 14.

  Furthermore, a fluctuation range of the overload value (contact pressure value) is provided, and the measurement probe 8 is moved in a direction in which the measurement probe 8 is pressed against or pulled away from the skin of the living body 14 within the set load value range. Alternatively, the stepping motor 41 may be controlled.

  In addition, as one of the discriminant analysis methods in the discriminant analysis unit 24, focusing on a part of the calibration spectrum defined above, a specified range is provided for the fluctuation of the value, and the newly measured absorbance spectrum is A method of determining similarity to the calibration spectrum, that is, whether the measurement is suitable or inappropriate in the measurement under this condition, may be employed depending on whether the measurement is performed with a change within a specified range.

  An example in which a prescribed value is provided in a part of this calibration spectrum is shown in FIG. When the calibration spectrum is measured in the wavelength band from 1200 nm to 1900 nm as shown in FIG. Β, if the spectrum is shown in the figure, the absorbance value near 1655 nm is assumed to be used for discriminant analysis, and this absorbance value is For example, if ± 2000 μAU is set as an allowable error range = specified value, whether the absorbance value at 1655 nm of the absorbance spectrum measured thereafter is within the specified value compared with the absorbance value at 1655 nm of the calibration spectrum measured earlier No, it is determined whether the measured spectrum is appropriate or inappropriate.

  Further, in the above-described method for determining a calibration spectrum value by providing a specified value, the point used for discriminant analysis is not limited to the use of an absorbance value of a certain wavelength, for example, an absorbance spectrum near 1450 nm shown in FIG. By subtracting the low power value of the absorbance spectrum (shown as index 1 in FIG. 11) near 1655 nm from the power peak value (shown as index 2 in FIG. 11), the measurement probe 8 emits near-infrared light. / By greatly affecting the amount of water present during light reception, or by subtracting the value near 1655 nm from the value near 1725 nm (described as index 3 in FIG. 11), The amount of fat present during light reception can be greatly influenced.

  Therefore, instead of using the absorbance value depending on one wavelength of 1655 nm, the specified value may be set using the above value.

(A) is a block diagram of Embodiment 1. FIG. (B) is a block diagram showing the functional configuration of the arithmetic device used in the first embodiment, and (d) is an enlarged front view of the distal end surface of the measurement probe used in the first embodiment. It is a block diagram of living body skin. It is a model figure of the skin of the biological body used for simulation. It is explanatory drawing of the principal component analysis used for the discriminant analysis part of Embodiment 1. FIG. It is an example of a blood glucose level measurement example for description of the principal component analysis of the discriminant analysis part of Embodiment 1. FIG. 6 is an explanatory diagram illustrating an example of principal component analysis of a discriminant analysis unit according to Embodiment 1. FIG. 6 is an explanatory diagram illustrating an example of principal component analysis of a discriminant analysis unit according to Embodiment 1. FIG. 6 is a flowchart for explaining the operation of the first embodiment of the first embodiment. FIG. 6 is a configuration diagram of a main part of a second embodiment. 6 is a flowchart for explaining operations of the second embodiment. It is explanatory drawing in the case of using a calibration spectrum for the discriminant analysis method of the discriminant analysis part in Embodiment 2. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Light source 2 Diffuser 3 Pinhole 4 Lens 5 Light incident body 6 Light emission side optical fiber 7 Reference light emission side optical fiber 8 Measurement probe 9 Reference probe 10 Light reception side optical fiber 11 Reference light reception side optical fiber 12 Light emission body 13 Light emission body 14 Living body 15a, 15b Lens 16 Diffraction grating 17 Light receiving element 18 A / D conversion circuit 19 Arithmetic device 19a Storage unit 19b CUP unit 20 Reference plate 21a, 21b Shutter 22 Reflector 23 Internal component calculation unit 24 Discriminant analysis unit 25 Absorbance spectrum acquisition Part 26 Database creation part

Claims (3)

A measurement probe for irradiating a living body to be measured with infrared light, an absorbance spectrum acquisition unit for acquiring an absorbance spectrum from reflected light or transmitted light from the living body, and absorbance of the living body under a plurality of conditions obtained in advance An absorbance spectrum database storing spectrum data; an in-vivo component calculation unit that calculates in-vivo components based on the absorbance spectrum acquired by the absorbance spectrum acquisition unit and the absorbance spectrum stored in the absorbance spectrum database; A discriminant analysis unit that performs discriminant analysis using the absorbance spectrum acquired by the absorbance spectrum acquisition unit and the absorbance spectrum stored in the absorbance spectrum database, and the in-vivo component calculation unit includes the discriminant analysis unit In the group of main analysis results obtained by discriminant analysis of Chemical Composition measurement apparatus and outputs the calculated value of the body component when it contains absorbance spectrum obtained by the light intensity spectrum acquisition unit. 2. The in-vivo component measurement apparatus according to claim 1, wherein the absorbance spectrum data stored in the absorbance spectrum database is absorbance spectrum data obtained by simulation. A measurement probe that irradiates a living body to be measured with infrared light, an absorbance spectrum acquisition unit that acquires an absorbance spectrum from reflected or transmitted light from the living body, and the absorbance of the living body under a plurality of conditions obtained in advance An absorbance spectrum database storing spectrum data; an in-vivo component calculation unit that calculates in-vivo components based on the absorbance spectrum acquired by the absorbance spectrum acquisition unit and the absorbance spectrum stored in the absorbance spectrum database; A discriminant analysis unit that performs discriminant analysis using the absorbance spectrum acquired by the absorbance spectrum acquisition unit and the absorbance spectrum stored in the absorbance spectrum database, and the discriminant analysis result of the discriminant analysis unit Based on the contact pressure of the measurement probe to the living body Chemical Composition measuring apparatus characterized by comprising a probe adjusting unit to adjust.

Images