JP2019168261A - Measuring device and measuring method - Google Patents

Abstract

An object component is accurately calibrated even when measurement data derived from a plurality of components are not statistically independent. An analysis processing unit generates R subsets from a plurality of optical spectra and performs an independent component analysis on each of the R subsets using N types of component amounts as independent components. In this way, the component amount series and the component calibration spectrum are specified for each component, and for each of the R subsets, an evaluation index corresponding to the non-Gaussian nature of the distribution of the component amounts in the component amount series is calculated and evaluated. The subset having the largest non-Gaussian property indicated by the index is selected as the optimal subset, and the target component calibration spectrum is identified from the N component calibration spectra calculated by the independent component analysis for the optimal subset, and the target component is identified. The calculated value of the inner product of each of a plurality of evaluation optical spectra measured for a sample whose component amount is known and the target component calibration spectrum, and the component amount of the target component contained in the sample To determine the calibration curve from the relationship. [Selection diagram] FIG.

Description

  The present invention relates to a technique for calibrating the amount of a target component contained in a subject.

  Conventionally, a technique for calibrating the amount of a target component by using independent component analysis (ICA) has been proposed. For example, in Patent Document 1, a spectrum derived from chlorophyll is estimated as an independent component by performing independent component analysis on a plurality of optical spectra measured from green vegetables, and unknown using the estimated spectrum. A calibration technique for calibrating the amount of chlorophyll in green vegetables is disclosed.

JP 2013-36973 A

  In order to realize highly accurate independent component analysis, it is necessary that a plurality of independent components to be estimated are statistically independent. However, there is a case where statistical independence is not sufficiently ensured among a plurality of components in actuality for measurement data to be subjected to independent component analysis.

  In order to solve the above problems, a calibration device according to a preferred embodiment of the present invention is a calibration device for calibrating the amount of a target component contained in a subject, and the target component calibration corresponding to the target component An analysis processing unit that generates a spectrum for analysis, a calibration curve representing a relationship between a calculated value of the inner product of the target component calibration spectrum and the optical spectrum of the subject, and a component amount of the target component; and the calculated value A calibration processing unit that specifies the amount of the target component corresponding to the standard curve from the calibration curve, and the analysis processing unit is a sample containing (a) N types (N is an integer of 1 or more). A process of generating R (R is an integer greater than or equal to 2) subsets from a plurality of optical spectra measured for (b) the N types of component amounts for each of the R subsets; Perform independent component analysis as independent components (C) processing for identifying a component amount series and a component calibration spectrum for each of a plurality of components; and (c) non-Gaussian distribution of component amounts in the component amount sequence for each of the R subsets. A process for calculating an evaluation index according to sex; (d) a process for selecting a subset having the maximum non-Gaussian property indicated by the evaluation index as an optimal subset; and (e) the independent component for the optimal subset. A process of specifying the target component calibration spectrum from a plurality of component calibration spectra calculated in the analysis; and (f) each of a plurality of evaluation optical spectra measured for a sample whose component amount of the target component is known. And a process for specifying the calibration curve from the relationship between the calculated value of the inner product of the target component calibration spectrum and the component amount of the target component contained in the sample. To.

  A calibration method according to a preferred embodiment of the present invention is a method in which a computer calibrates the amount of a target component contained in a subject, the target component calibration spectrum corresponding to the target component, and the target component calibration. Analysis processing for generating a calibration curve representing the relationship between the calculated value of the inner product of the spectrum for use and the optical spectrum of the subject and the component amount of the target component, and the component amount of the target component corresponding to the calculated value Calibration processing for specifying the calibration curve from the calibration curve, and the analysis processing includes: (a) R (from a plurality of optical spectra measured for a sample containing N types of components (N is an integer of 1 or more) ( (B is an integer greater than or equal to 2) generating a subset, and (b) performing independent component analysis for each of the R subsets using the N component amounts as independent components. Of multiple ingredients A process for identifying a component quantity series and a component calibration spectrum, and (c) for each of the R subsets, calculating an evaluation index according to the non-Gaussian distribution of the component quantity in the component quantity series (D) a process of selecting a subset having the maximum non-Gaussian property indicated by the evaluation index as an optimal subset; and (e) a plurality of components calculated by the independent component analysis for the optimal subset. A process of identifying the target component calibration spectrum from the calibration spectrum; (f) each of a plurality of evaluation optical spectra measured for a sample whose component amount of the target component is known; and the target component calibration spectrum; And a process for specifying the calibration curve from the relationship between the calculated value of the inner product and the component amount of the target component contained in the sample.

It is explanatory drawing of the principle of the method of calibrating the component quantity of the target component. It is explanatory drawing of an optical spectrum. It is a block diagram which illustrates the composition of a calibration device. It is a flowchart which illustrates the procedure of operation | movement of a calibration apparatus. It is explanatory drawing of a calibration process. It is explanatory drawing of an analysis process. It is a flowchart which illustrates the procedure of an analysis process. It is a graph which shows the relationship between an estimated kurtosis and separation accuracy. It is explanatory drawing of the process which specifies the spectrum for objective component calibration. It is a flowchart which illustrates the procedure of the analysis process in 2nd Embodiment.

<Principle of calibration>
Prior to the description of the calibration device 100 according to the preferred embodiment of the present invention, the principle of calibrating the component amount of the target component will be described. The target component is a biological component such as glucose, albumin, sodium chloride, hemoglobin, lipid or alcohol. The component amount is, for example, the concentration of the target component exemplified above. As described in detail below, independent component analysis is used for calibration of the component amounts.

Assume spectroscopic measurement of a sample containing N types (N is an integer of 1 or more). An optical spectrum matrix X representing the result of spectroscopic measurement for M samples is expressed as a product of a component specific spectrum matrix Y and a component quantity matrix W, as expressed by the following formula (1). FIG. 1 schematically shows Equation (1).

  The optical spectrum matrix X of Expression (1) is a matrix in which M optical spectra X1 to XM respectively measured from M samples are arranged in the horizontal direction. The optical spectrum Xm of one m-th (m = 1 to M) sample is a K-dimensional column vector in which K intensities xm (1) to xm (K) corresponding to K wavelengths are arranged. It is expressed by The optical spectrum matrix X may be a matrix in which M optical spectra X1 to XM measured at different time points by spectroscopic measurement for one sample are arranged in time series.

  As illustrated in FIG. 2, the optical spectrum Xm of the mth sample is generated from the optical spectrum Xm_d observed in the spectroscopic measurement for the sample. The optical spectrum Xm_d is, for example, an absorbance spectrum or a diffuse reflection spectrum. Although the M optical spectra X1_d to XM_d are sufficiently approximated, in FIG. 2, the difference between the M optical spectra X1_d to XM_d is emphasized for convenience.

  When M optical spectra X1_d to XM_d that are very close to each other as illustrated in FIG. 2 are applied to the independent component analysis, the analysis accuracy may be lowered. For example, the influence of the solvent on the optical spectrum Xm_d changes depending on the concentration of the solute (containing component), and the accuracy of independent component analysis may be reduced. In consideration of the above circumstances, the optical spectrum Xm is calculated by predetermined preprocessing for each optical spectrum Xm_d. Specifically, as illustrated in FIG. 2, each optical spectrum Xm is calculated by pre-processing for subtracting the average spectrum Xave of M optical spectra X1_d to XM_d from each optical spectrum Xm_d. Note that the average spectrum Xave applied to the preprocessing is held in a storage device, for example. According to the above method, even when the influence of the solvent on the optical spectrum Xm_d changes depending on the concentration of the solute, the influence is reduced, and highly accurate independent component analysis is realized. Note that the specific content of the preprocessing is not limited to the above example (subtraction of the average spectrum Xave). For example, a null space projection that generates the optical spectrum Xm by projecting the optical spectrum Xm_d onto the null space of the average spectrum Xave may be executed as preprocessing. Further, a process for converting the optical spectrum Xm_d into an average of 0 and a variance of 1 may be included in the preprocessing. Note that the pre-processing exemplified above can be omitted. That is, the optical spectrum Xm_d may be used as the optical spectrum Xm of the optical spectrum matrix X.

  The component inherent spectrum matrix Y of the formula (1) is a matrix in which N component inherent spectra Y1 to YN respectively corresponding to N types of components are arranged in the horizontal direction. The component intrinsic spectrum Yn of one component of the nth (n = 1 to N) component is a K-dimensional array in which K intensities yn (1) to yn (K) corresponding to K wavelengths are arranged. Expressed as a vector. The component intrinsic spectrum Yn of the nth component is a spectrum representing, for example, an extinction coefficient series (that is, an extinction coefficient series for each wavelength) unique to the component.

  The component quantity matrix W of the mathematical formula (1) is a matrix in which N component quantity series W1 to WN respectively corresponding to N types of components are arranged in the vertical direction. The component amount series Wn of the nth one component is a row vector in which the component amounts wn (m) of the nth component contained in each sample are arranged for M samples.

  As the independent component analysis using Equation (1), an independent component analysis in which unknown N component specific spectra Y1 to YN derived from N types of components are regarded as statistically independent components is assumed. However, the N component specific spectra Y1 to YN may not satisfy the condition that they are statistically independent from each other. On the other hand, even when the N component specific spectra Y1 to YN are not statistically independent, the condition that the component amounts (for example, concentrations) of the N types of components are independent of each other and statistically independent is satisfied. By executing the independent component analysis considering the component amounts of N types of components (component amount series Wn) as independent components, it is possible to estimate the N component amount series W1 to WN with high accuracy. is there. In addition to the component amount series Wn, N component specific spectra Y1 to YN can be accurately estimated.

  Against the background described above, N component quantity sequences are obtained by executing independent component analysis in which each of the N component quantity sequences W1 to WN in the component quantity matrix W of Equation (1) is regarded as an independent component. W1 to WN are estimated, and N component specific spectra Y1 to YN are also estimated simultaneously with each component amount series Wn. Even when each component amount series Wn is regarded as an independent component, an existing method can be used as the independent component analysis method itself. For example, the method disclosed in Japanese Patent Application Laid-Open No. 2013-160574 or Japanese Patent Application Laid-Open No. 2006-65803, which is the prior application of the applicant of the present application, or other known methods is used.

数式(2)の光学スペクトルＸSは、被検体に対する分光測定で観測されたスペクトルであり、相異なる波長に対応するＫ個の強度ｘ(1)〜ｘ(K)を配列したＫ次元の列ベクトルで表現される。また、数式(2)の成分検量用スペクトルＺnは、前述の独立成分分析により推定された成分固有スペクトル行列Ｙの一般化逆行列Ｙ⁺における第ｎ行に相当するＫ次元の行ベクトルである。 When the component eigenspectral matrix Y is estimated by the independent component analysis described above, the calculation value α corresponding to the unknown component amount of the nth component (target component) contained in the subject can be calculated. Is possible. Specifically, the calculated value α is calculated as the product of the component calibration spectrum Zn of the nth component and the optical spectrum XS of the subject, as expressed by the following formula (2).

The optical spectrum XS in Equation (2) is a spectrum observed by spectroscopic measurement on the subject, and is a K-dimensional column vector in which K intensities x (1) to x (K) corresponding to different wavelengths are arranged. It is expressed by Further, the component calibration spectrum Zn in Equation (2) is a K-dimensional row vector corresponding to the n-th row in the generalized inverse matrix Y ⁺ of the component eigenspectral matrix Y estimated by the independent component analysis described above.

  As understood from the equation (2), an arithmetic value α corresponding to the component amount of the component contained in the subject is obtained by the inner product of the component calibration spectrum Zn of the nth component and the optical spectrum XS of the subject. Is calculated. However, the numerical value of each element of the component inherent spectrum matrix Y estimated by independent component analysis is meaningless (so-called scaling indefiniteness), and the shape of each component inherent spectrum Yn of the component inherent spectrum matrix Y is true It is in proportion to the spectrum. Therefore, the calculated value α calculated by the formula (2) is a numerical value proportional to the true value of the component amount. The actual component amount of the n-th component is obtained by applying the calculated value α calculated by the inner product of Equation (2) to a calibration curve representing the relationship between the calculated value α and the component amount (true value) C. Calculated. The component calibration spectrum Zn of the target component to be calibrated among the N types of components will be referred to as “target component calibration spectrum Z” in the following description.

  As described above, according to the independent component analysis in which the component amount is regarded as an independent component, each component for each optical spectrum Xm is obtained even when each component intrinsic spectrum Yn derived from N types of components is not statistically independent. If the variation in quantity is statistically independent, the component quantity matrix W (component quantity series Wn) and the component intrinsic spectrum matrix Y (component intrinsic spectrum Yn and component calibration spectrum Zn) can be estimated with high accuracy. is there.

<First Embodiment>
The calibration apparatus 100 according to the first embodiment using the independent component analysis described above will be described. The calibration device 100 is a measurement device that calibrates the amount of a target component contained in a subject from an optical spectrum observed by spectroscopic measurement of the subject (for example, a living body).

  FIG. 3 is a block diagram illustrating the configuration of the calibration apparatus 100. As illustrated in FIG. 3, a measuring instrument 200 for measuring the optical spectrum XS of the subject is connected to the calibration apparatus 100. As the measuring device 200, a spectroscopic measuring device that outputs a measurement signal corresponding to the optical characteristics of the subject by spectroscopic measurement of the subject is preferably used as the measuring device 200. Note that the measuring instrument 200 may be built in the calibration apparatus 100.

As illustrated in FIG. 3, the calibration device 100 includes a control device 11, a storage device 12, and a display device 13. The control device 11 is, for example, a CPU (Central Processing Unit) or an FPGA (Field
It is configured to include a processing circuit such as a Programmable Gate Array) and executes various processes for calculating the component amount of a target component (for example, glucose) contained in the subject. The storage device 12 stores a program executed by the control device 11 and various data used by the control device 11. For example, a semiconductor memory is suitable as the storage device 12. The display device 13 is composed of, for example, a liquid crystal display panel, and displays various types of information including the component amounts calculated by the control device 11.

  As illustrated in FIG. 3, the control device 11 (exemplary computer) executes a program stored in the storage device 12 to thereby calculate a plurality of elements (analysis processing unit 21) for calculating the component amount of the target component. And function as a calibration processing unit 22). In addition, you may implement | achieve the function of the control apparatus 11 with the some apparatus comprised mutually separately.

  The analysis processing unit 21 generates the aforementioned target component calibration spectrum Z corresponding to the target component, and a calibration curve Cc representing the relationship between the target component calculation value α and the component amount C (true value). The target component calibration spectrum Z and the calibration curve Cc generated by the analysis processing unit 21 are stored in the storage device 12. The calibration processing unit 22 uses the optical spectrum XS of the subject measured by the measuring instrument 200 and the target component calibration spectrum Z and the calibration curve Cc generated by the analysis processing unit 21, and the objective contained in the subject. Specify the component amount of the component.

  FIG. 4 is a flowchart illustrating a specific procedure of processing in which the calibration apparatus 100 calibrates the component amount of the target component. For example, the process in FIG. 4 is started in response to an instruction from the user. When the process of FIG. 4 is started, the analysis processing unit 21 specifies the target component calibration spectrum Z and the calibration curve Cc by executing the analysis process Sa. The specific procedure of the analysis process Sa will be described later.

  The calibration processing unit 22 calibrates the component quantity of the target component contained in the subject by the calibration process Sb (Sb1 to Sb4) using the target component calibration spectrum Z and the calibration curve Cc specified in the analysis process Sa. To do. FIG. 5 is an explanatory diagram of the calibration process Sb.

  The calibration processing unit 22 acquires the optical spectrum XS of the subject by analyzing the measurement signal supplied from the measuring instrument 200 (Sb1). Specifically, the calibration processing unit 22 calculates the optical spectrum XS by executing the above-described preprocessing on the optical spectrum X_d specified from the measurement signal. For the preprocessing for the optical spectrum X_d, for example, the above-described average spectrum Xave applied to the preprocessing (FIG. 2) of the optical spectrum Xm_d to be subjected to independent component analysis is used. For example, the average spectrum Xave generated in the analysis process Sa is held in the storage device 12 and used for the preprocessing in the calibration process Sb.

  As illustrated in FIG. 5, the calibration processing unit 22 calculates an arithmetic value α of the inner product of the optical spectrum XS of the subject and the target component calibration spectrum Z generated in the analysis process Sa (Sb2). The calibration processing unit 22 specifies the component amount C of the target component from the calculated value α using the calibration curve Cc generated in the analysis processing Sa (Sb3). Specifically, the calibration processing unit 22 specifies the component amount C corresponding to the calculated value α in the calibration curve Cc, as illustrated in FIG. The calibration processing unit 22 causes the display device 13 to display the component amount C of the target component specified by the calibration processing unit 22 as a measurement result (Sb4).

<Analysis processing Sa>
FIG. 6 is an explanatory diagram of the analysis process Sa. As illustrated in FIG. 6, the storage device 12 stores a plurality of optical spectra X0 and a plurality of evaluation data D in advance. The plurality of optical spectra X0 are spectra measured by spectroscopic measurement of a plurality of first samples containing N types of components including a target component (for example, glucose). The first sample is composed of the same components as the subject. Each optical spectrum X0 is used as sample data for learning for determining each component calibration spectrum Zn by independent component analysis. Each of the plurality of evaluation data D includes an optical spectrum (hereinafter referred to as “evaluation optical spectrum”) XE measured by spectroscopic measurement on the second sample, and a known component amount Ct of the target component in the second sample (ie, True value). In addition, about the some 1st sample, the component quantity of the target component does not need to be known. The optical spectrum X0 and the evaluation optical spectrum XE are spectra generated by performing the above-described preprocessing using the average spectrum Xave on the optical spectrum observed by the spectroscopic measurement of the second sample. A part of the first sample may be used as the second sample.

  FIG. 7 is a flowchart illustrating a specific procedure of the analysis process Sa. When the analysis process Sa is started, the analysis processing unit 21 generates R subsets P1 to PR (R is an integer of 2 or more) from a set of a plurality of optical spectra X0 (Sa1). Each of the R subsets P1 to PR is a set including M optical spectra X0 (X1 to XM). The number M of optical spectra X0 constituting each subset Pr (r = 1 to R) is set to be equal to or greater than the total number N of component specific spectra Yn specified by independent component analysis. The total number M of optical spectra X0 constituting each subset Pr may be different for each subset Pr, or may be common across R subsets P1 to PR. For example, the analysis processing unit 21 randomly generates optical spectra X1 to XM of each subset Pr from a set of a plurality of optical spectra X0 using random numbers. Step Sa1 is an example of the process (a). The optical spectrum X0 and the evaluation optical spectrum XE may be generated by preprocessing using the average spectrum Xave calculated for each subset Pr.

  The analysis processing unit 21 performs the above-described independent component analysis, assuming that the component amount is an independent component, for each of the R subsets P1 to PR, so that N component specific spectra Y1 (r) are obtained. ~ YN (r) and N component quantity series W1 (r) to WN (r) are calculated (Sa2). Specifically, N component specific spectra Y1 (r) to YN (r) and N components are obtained by independent component analysis using a sequence of M optical spectra X1 to XM included in the subset Pr as an optical spectrum matrix X. The component amount series W1 (r) to WN (r) are calculated. Further, the analysis processing unit 21 calculates a component calibration spectrum Zn (r) corresponding to each component specific spectrum Yn (r) (Sa3). That is, for each of the R subsets P1 to PR, N component quantity sequences W1 (r) to WN (r) and N component calibration spectra Z1 (r) to ZN (r) are specified. The Steps Sa2 and Sa3 are examples of the process (b). Note that the component amount series Wn (r) and the component calibration spectrum Zn (r) may be specified for some of the N types of components.

  The analysis processing unit 21 calculates an evaluation index Er (E1 to ER) for each of the R subsets P1 to PR (Sa4). The evaluation index Er of the subset Pr is an index of the degree to which the subset Pr is appropriate for independent component analysis. The order of calculation of the component calibration spectrum Zn (r) (Sa3) and calculation of the evaluation index Er (Sa4) may be reversed. Step Sa4 is an example of the process (c).

  In independent component analysis that separates statistically independent components, the higher the non-Gaussianness that is an index of statistical independence, the easier it is to separate each component. Therefore, the subset Pr having a high non-Gaussian property among the R subsets P1 to PR is suitable for independent component analysis. However, since the component amount of each component is unknown for the subset Pr, the non-Gaussian property cannot be directly evaluated for the subset Pr. Against the background described above, in the first embodiment, non-Gaussianity is evaluated for N component quantity sequences W1 (r) to WN (r) estimated from each subset Pr. Specifically, the kurtosis of the distribution of M component amounts wn (1) to wn (M) in the component amount series Wn (r) of each component is used as a non-Gaussian index.

Specifically, the analysis processing unit 21 calculates the evaluation index Er from the kurtosis κn (κ1 to κN) corresponding to each of the N component amount series W1 (r) to WN (r). The kurtosis κn is a high-order statistic representing the sharpness of the distribution of M component amounts wn (1) to wn (M) in the component amount series Wn (r), and is calculated by the following equation (3). Is done.

  In the equation (3), symbol σ is a standard deviation of M component amounts wn (1) to wn (M), and symbol μ is an average of M component amounts wn (1) to wn (M). is there. In addition, since the kurtosis κn is calculated using the estimation result of the component amount, not the true value of the component amount, the kurtosis κn is also expressed as the estimated kurtosis.

  The analysis processing unit 21 calculates the representative values of the N kurtosis values κ1 to κN calculated by the calculation of Expression (3) as the evaluation index Er of the subset Pr. For example, the representative value is, for example, an average value, median value, maximum value, or minimum value of N kurtosis κ1 to κN. Of the N kurtosis κ1 to κN, the kurtosis κn corresponding to the component amount series Wn (r) of the target component (that is, the representative value of the N kurtosis κ1 to κN) may be selected as the evaluation index Er. . As understood from the above description, the evaluation index Er becomes a larger numerical value as the non-Gaussianity of the component amount series Wn (r) is higher. Therefore, the subset Pr having a larger evaluation index Er among the R subsets P1 to PR can be evaluated as being suitable for independent component analysis. FIG. 8 is a graph showing the relationship between the estimated kurtosis (representative values of N kurtosis κ1 to κN) and the separation accuracy by independent component analysis. It can be confirmed from FIG. 8 that the higher the estimated kurtosis (that is, the higher the non-Gaussian property), the higher the separation accuracy by independent component analysis.

  When the evaluation index Er for each subset Pr is calculated by the processing illustrated above, the analysis processing unit 21 selects one subset Pr (hereinafter referred to as “the evaluation index Er” among the R subsets P1 to PR). “Optimum subset P”) is selected (Sa5). That is, the analysis processing unit 21 selects, as the optimum subset P, a subset Pr that has a high possibility that N components are separated with high accuracy by independent component analysis. Step Sa5 is an example of the process (d).

  The analysis processing unit 21 specifies the target component calibration spectrum Z from the N component calibration spectra Z1 (r) to ZN (r) corresponding to the optimum subset P as described in detail below (Sa6, Sa7). ). Steps Sa6 and Sa7 are examples of the process (e). FIG. 9 is an explanatory diagram of the processing (Sa6, Sa7) for specifying the target component calibration spectrum Z.

  The analysis processing unit 21 calculates an evaluation index Vn (r) for each of the N component calibration spectra Z1 (r) to ZN (r) calculated from the optimal subset P (Sa6). The evaluation index Vn (r) is an index of the accuracy with which the component calibration spectrum Zn (r) corresponds to the target component. Specifically, as illustrated in FIG. 9, the analysis processing unit 21 calculates the component calibration spectrum Zn (r) and the evaluation optical spectrum XE of the evaluation data D for each of the plurality of evaluation data D. The calculated value α of the inner product of is calculated. That is, for one component calibration spectrum Zn (r), a plurality of calculated values α corresponding to different evaluation optical spectra XE are calculated.

  FIG. 9 shows a plurality of calculated values α calculated between the component calibration spectrum Zn (r) and each evaluation optical spectrum XE, and a plurality of component amounts Ct (that is, true values) indicated by each evaluation data D. ) Is illustrated. When the component calibration spectrum Zn (r) corresponds to the target component of the N types of components, a strong correlation is observed between the calculated value α and the component amount Ct, as illustrated in FIG. Against the background described above, the analysis processing unit 21 uses the evaluation index Vn () as an index of the degree of correlation between the calculated value α calculated for the component calibration spectrum Zn (r) and the component amount Ct indicated by the evaluation data D. Calculate as r). For example, a correlation coefficient (correlation degree) between the calculated value α and the component amount Ct is suitable as the evaluation index Vn (r). As understood from the above description, the evaluation index Vn (r) becomes a larger numerical value as the possibility that the component calibration spectrum Zn (r) corresponds to the target component is higher.

  The analysis processing unit 21 selects one component calibration spectrum having the maximum evaluation index Vn (r) among the N component calibration spectra Z1 (r) to ZN (r) calculated from the optimum subset P. Zn (r) is selected as the target component calibration spectrum Z (Sa7). As understood from the above description, the calculated value α corresponding to the component amount of the target component contained in the subject is calculated from the inner product of the target component calibration spectrum Z and the optical spectrum XS of the subject.

  When the target component calibration spectrum Z is specified, the analysis processing unit 21 specifies the calibration curve Cc (Sa8). Specifically, a straight line expressed by a single regression equation (Ct = aα + b) that approximates the relationship between a plurality of calculated values α and a plurality of component amounts Ct is specified as a calibration curve Cc. The target component calibration spectrum Z and the calibration curve Cc specified in the analysis process Sa (Sa1 to Sa8) exemplified above are applied to the calibration process Sb described above with reference to FIG. Step Sa8 is an example of the process (f).

  As described above, in the first embodiment, the evaluation index Er corresponding to the non-Gaussian distribution of the M component amounts wn (1) to wn (M) in the component amount series Wn (r) is used. An optimal subset P is selected. That is, a subset P suitable for independent component analysis is selected. Therefore, there is an advantage that the component amount of the target component can be estimated with high accuracy.

Second Embodiment
A second embodiment of the present invention will be described. In the following examples, elements having the same functions as those of the first embodiment are diverted using the same reference numerals used in the description of the first embodiment, and detailed descriptions thereof are appropriately omitted.

  In the second embodiment, the analysis process Sa illustrated in FIG. 7 is replaced with the analysis process Sc of FIG. When the analysis process Sc is started, the analysis processing unit 21 generates R subsets P1 to PR from the plurality of optical spectra X0 stored in the storage device 12, as in step Sa1 of the first embodiment (Sc1). ). Step Sc1 is an example of the process (a).

  The analysis processing unit 21 selects a Q subset (Q is an integer less than or equal to 1 and less than R) suitable for independent component analysis from the R subsets P1 to PR (hereinafter referred to as “selection process”). ) Is executed (Sa2 to Sa6).

  When the selection process is started, the analysis processing unit 21 performs N component analysis for each of the R subsets P1 to PR by performing independent component analysis considering the component amount as an independent component, as in step Sa2 of the first embodiment. The component intrinsic spectra Y1 (r) to YN (r) and N component quantity sequences W1 (r) to WN (r) are calculated (Sc2). Similarly to step Sa3 of the first embodiment, the analysis processing unit 21 uses component calibration spectra Z1 (r) to ZN (r) corresponding to the N component specific spectra Y1 (r) to YN (r), respectively. ) Is specified for each subset Pr (Sc3). Steps Sc2 and Sc3 are examples of the process (b).

  The analysis processing unit 21 calculates an evaluation index Vn (r) for each of the N component calibration spectra Z1 (r) to ZN (r) of each subset Pr (Sc4). For the calculation of the evaluation index Vn (r), the same method as in the first embodiment is employed. That is, for example, the correlation coefficient (degree of correlation) between the calculated value α of the inner product of the component calibration spectrum Zn (r) and each evaluation optical spectrum XE and the component amount Ct indicated by the evaluation data D is the evaluation index Vn. Calculated as (r).

  The analysis processing unit 21 calculates the evaluation index Er (E1 to ER) for each of the R subsets P1 to PR (Sc5). For the calculation of the evaluation index Er, the same method as in the first embodiment is adopted. That is, for example, the representative value of the kurtosis κn of the distribution of the M component amounts wn (1) to wn (M) in each component amount series Wn (r) is calculated as the evaluation index Er of the subset Pr. The order of calculation of the evaluation index Vn (r) of each component calibration spectrum Zn (r) (Sc3, Sc4) and calculation of the evaluation index Er of each subset Pr (Sc5) may be reversed. Step Sc5 is an example of the process (c).

  The analysis processing unit 21 selects Q subsets P from the R subsets P1 to PR according to the evaluation index Er (Sc6). Specifically, the analysis processing unit 21 selects Q subsets P positioned in the descending order of the evaluation index Er among the R subsets P1 to PR (elite selection). That is, Q subsets P suitable for independent component analysis are selected from the R subsets P1 to PR. Step Sc6 is an example of process (g). For selecting Q subsets P, various selection algorithms such as roulette selection or tournament selection are used.

  After selecting Q subsets P, the analysis processing unit 21 determines whether a predetermined condition (hereinafter referred to as “end condition”) for ending the repetition of the selection process is satisfied (Sc7). The termination condition is, for example, that the maximum value of the evaluation index Er exceeds a predetermined threshold, or that the number of times the selection process has been repeated exceeds a predetermined threshold. That is, the selection process (Sc2 to Sc6) is repeated until a subset P suitable for independent component analysis is selected. Note that the success or failure of the end condition may be determined between the calculation of the evaluation index Er (Sc5) and the selection of the Q subsets P (Sc6).

  When the termination condition is not satisfied (Sc7: NO), the analysis processing unit 21 partially changes the subset P for each of the Q subsets P, so that the R subsets P1 to PR are changed. Is generated (Sc8). Specifically, the analysis processing unit 21 uses one or more optical spectra Xm among the M optical spectra X1 to XM included in each subset P as another optical spectrum X0 stored in the storage device 12. To generate a new subset P. For example, the analysis processing unit 21 determines the presence / absence of replacement for each of the M optical spectra X1 to XM included in the subset P with a predetermined probability. Then, the analysis processing unit 21 replaces each optical spectrum Xm determined to be a replacement target with an optical spectrum X0 randomly selected by a random number (for example, a uniform random number). By partially changing each of the Q subsets P, a total of R subsets P1 to PR are generated. As understood from the above description, the process of partially changing the subset P corresponds to a mutation in the genetic algorithm. Step Sc8 is an example of the process (h). Further, the number R of subsets P generated from each subset P may be different for each subset P. For example, it is assumed that the subset P having a larger evaluation index Er increases the number R of subsets P generated from the subset P.

  When the above processing is completed, the sorting processing is executed for the updated R subsets P1 to PR. As will be understood from the above description, the selection processing for selecting Q subsets P from the R subsets P1 to PR according to the evaluation index Er is performed until the end condition is satisfied. Iterate while updating ~ PR sequentially. Therefore, Q subsets P suitable for independent component analysis are selected. When the maximum value of the evaluation index Er calculated for the R subsets P selected first exceeds the threshold value, the selection process is executed only once. Further, the number R of the subsets P to be selected may be different for each selection process.

  When the end condition is satisfied (Sc7: YES), the analysis processing unit 21 selects the one subset Pr having the maximum evaluation index Er among the Q subsets P selected by the repetition of the selection process as the optimum subset. Select as P (Sc9). That is, as in the first embodiment, the subset P that is highly likely to have been separated with high accuracy by the independent component analysis is selected as the optimum subset P. Step Sc9 is an example of process (d).

  The analysis processing unit 21 specifies the target component calibration spectrum Z from the N component calibration spectra Z1 (r) to ZN (r) calculated for the optimal subset P (Sc10). Specifically, the analysis processing unit 21 has the largest evaluation index Vn (r) calculated in step Sc4 among the N component calibration spectra Z1 (r) to ZN (r) corresponding to the optimum subset P. One component calibration spectrum Zn (r) is selected as the target component calibration spectrum Z. Steps Sc4 and Sc10 are examples of the process (e).

  When the target component calibration spectrum Z is specified, the analysis processing unit 21 specifies the calibration curve Cc (Sc11). Specifically, as in the first embodiment, a straight line expressed by a single regression equation (Ct = aα + b) that approximates the relationship between a plurality of calculated values α and a plurality of component amounts Ct is specified as a calibration curve Cc. The The target component calibration spectrum Z and the calibration curve Cc specified in the analysis process Sc exemplified above are applied to the calibration process Sb described above with reference to FIG. Step Sc11 is an example of the process (f).

  As described above, in the second embodiment, the evaluation index Er corresponding to the non-Gaussian distribution of the M component amounts wn (1) to wn (M) in the component amount series Wn (r) is used. An optimal subset P is selected. That is, a subset P suitable for independent component analysis is selected. Therefore, similarly to the first embodiment, there is an advantage that the component amount of the target component can be estimated with high accuracy. In the second embodiment, the selection process for selecting Q subsets P from the R subsets P1 to PR according to the evaluation index Er is repeated while updating each subset Pr. A subset P suitable for component analysis can be selected with high accuracy. Therefore, the effect that the component amount of the target component can be estimated with high accuracy is particularly remarkable.

<Modification>
Each form illustrated above can be variously modified. Specific modes of modifications that can be applied to the above-described embodiments are exemplified below. Note that two or more aspects arbitrarily selected from the following examples can be appropriately combined as long as they do not contradict each other.

(1) In each of the above embodiments, the kurtosis of the distribution of the M component amounts wn (1) to wn (M) in each component amount series Wn (r) is exemplified as the evaluation index Er of the subset Pr. The non-Gaussian index relating to the distribution of the M component amounts wn (1) to wn (M) is not limited to kurtosis. For example, various indexes such as KL (Kullback-Leibler) divergence, negentropy, or mutual information amount are used as the evaluation index Er. The evaluation index Er may be calculated according to two or more types of indexes exemplified above.

(2) In each of the above-described embodiments, the non-Gaussian index relating to the distribution of the M component amounts wn (1) to wn (M) is used as the evaluation index Er, but the method for calculating the evaluation index Er is as described above. It is not limited to illustration. For example, the evaluation index Er may be calculated according to a non-Gaussian index and other indices. For example, the evaluation index Er of the subset Pr may be calculated from the representative values of N kurtosis κ1 to κN and N evaluation indexes V1 (r) to VN (r). For example, a multiplication value or a weighted sum of the representative values of N kurtosis κ1 to κN and the representative values of N evaluation indexes V1 (r) to VN (r) is suitable as the evaluation index Er. In addition, a standard optical spectrum corresponding to the target component (hereinafter referred to as “reference optical spectrum”) is stored in the storage device 12 in advance, and the similarity between each component specific spectrum Yn (r) and the reference optical spectrum is set to The evaluation index Er may be considered together with the Gaussian index.

  A configuration using only the correlation coefficient (correlation) between the calculated value α calculated from the component calibration spectrum Zn (r) and the known component amount Ct, or the similarity to the reference optical spectrum, as the evaluation index Er. Then, a tendency is observed that the evaluation index Er tends to be a numerical value that approximates a binary value of 0 or 1. According to the configuration in which the evaluation index Er is calculated according to the non-Gaussian index, the numerical value of the evaluation index Er is appropriately dispersed. Therefore, it is appropriately evaluated whether each subset Pr is suitable for independent component analysis. There is also an advantage of being able to do it.

(3) The optimal subset P may be selected using both the evaluation index Er corresponding to non-Gaussianity and other indices. For the process of selecting the optimal subset P using a plurality of indexes including the evaluation index Er, for example, a multipurpose genetic algorithm is preferably used.

(4) In the second embodiment, the calculation of each component calibration spectrum Zn (r) (Sc3) and the calculation of the evaluation index Vn (r) (Sc4) are performed in the selection process. After the selection (Sc9), the evaluation index Vn (r) of each component calibration spectrum Zn (r) may be calculated for the optimum subset P. That is, for each subset P other than the optimal subset P, calculation of the component calibration spectrum Zn (r) and the evaluation index Vn (r) is omitted.

(5) The calibration device 100 is also specified as an analysis device that generates the target component calibration spectrum Z and the calibration curve Cc. The analysis apparatus includes the analysis processing unit 21 exemplified in each of the above embodiments. In the analyzer, the calibration processing unit 22 can be omitted. An apparatus for generating the target component calibration spectrum Z and the calibration curve Cc (an apparatus having the analysis processing unit 21) and an apparatus for executing the calibration process Sb (an apparatus having the calibration processing unit 22) are mutually connected. It may be realized as a separate device.

DESCRIPTION OF SYMBOLS 100 ... Calibration apparatus, 200 ... Measuring device, 11 ... Control apparatus, 12 ... Memory | storage device, 13 ... Display apparatus, 21 ... Analysis process part, 22 ... Calibration process part.

Claims (4)

A calibration device for calibrating the amount of a target component contained in a subject,
Generating a target component calibration spectrum corresponding to the target component, and a calibration curve representing a relationship between the calculated value of the inner product of the target component calibration spectrum and the optical spectrum of the subject and the component amount of the target component An analysis processing unit to
A calibration processing unit that identifies the component amount of the target component corresponding to the calculated value from the calibration curve;
The analysis processing unit
(A) a process of generating R (R is an integer of 2 or more) subsets from a plurality of optical spectra measured for a sample containing N types (N is an integer of 1 or more);
(B) For each of the R subsets, by performing independent component analysis using the N types of component amounts as independent components, a component amount series and a component calibration spectrum for each of the plurality of components Processing to identify
(C) For each of the R subsets, processing for calculating an evaluation index according to non-Gaussianity of the component amount distribution in the component amount series;
(D) a process of selecting a subset having the maximum non-Gaussian property indicated by the evaluation index as an optimal subset;
(E) processing for identifying the target component calibration spectrum from a plurality of component calibration spectra calculated by the independent component analysis for the optimal subset;
(F) a calculated value of an inner product of each of a plurality of optical spectra for evaluation measured for a sample whose component amount of the target component is known and the spectrum for target component calibration, and the target component contained in the sample A calibration device that executes the process of specifying the calibration curve from the relationship between the component amount and The analysis processing unit
The process (a) to the process (c);
(G) After executing the process (c), a process of selecting Q (Q is an integer less than or equal to 1 and less than R) subsets from the R subsets according to the evaluation index;
(H) by repeating each of the Q subsets to generate a subset of R to be processed (b) by repeating a selection process including:
The calibration device according to claim 1, wherein in the process (d), the optimum subset is selected according to the evaluation index from the Q subsets after the selection process is repeated. The said analysis process part calculates the said evaluation parameter | index according to the kurtosis of the distribution of the component amount in the said component amount series about each of the said R subsets in the said process (c). 2. Calibration device. A method in which a computer calibrates the amount of a target component contained in a subject,
Generating a target component calibration spectrum corresponding to the target component, and a calibration curve representing a relationship between the calculated value of the inner product of the target component calibration spectrum and the optical spectrum of the subject and the component amount of the target component Analysis processing to
A calibration process for specifying the component amount of the target component corresponding to the calculated value from the calibration curve,
The analysis process is
(A) a process of generating R (R is an integer of 2 or more) subsets from a plurality of optical spectra measured for a sample containing N types (N is an integer of 1 or more);
(B) For each of the R subsets, by performing independent component analysis using the N types of component amounts as independent components, a component amount series and a component calibration spectrum for each of the plurality of components Processing to identify
(C) For each of the R subsets, processing for calculating an evaluation index according to non-Gaussianity of the component amount distribution in the component amount series;
(D) a process of selecting a subset having the maximum non-Gaussian property indicated by the evaluation index as an optimal subset;
(E) processing for identifying the target component calibration spectrum from a plurality of component calibration spectra calculated by the independent component analysis for the optimal subset;
(F) a calculated value of an inner product of each of a plurality of optical spectra for evaluation measured for a sample whose component amount of the target component is known and the spectrum for target component calibration, and the target component contained in the sample And a process of identifying the calibration curve from the relationship between the component amount and the calibration method.

Images