JP5985709B2 - Discriminant filter design method, discriminant method, discriminant filter set, discriminator, and program - Google Patents

Description

  The present invention relates to a discrimination filter design method, a discrimination method, a discrimination filter set, a discrimination device, and a program.

  Conventionally, techniques for analyzing an object based on fluorescence characteristics have been developed.

  For example, in the method for discriminating flour described in Patent Document 1, in a fluorescence spectrophotometer or the like, the excitation wavelength irradiated to the evaluation object and the fluorescence wavelength observed from the evaluation object are changed stepwise (wavelength scanning). ) To obtain fluorescence fingerprints consisting of three axes of excitation wavelength, fluorescence wavelength, and fluorescence intensity, and analyze the obtained fluorescence fingerprints to determine the variety and type of flour. .

  Further, in the method described in Patent Document 2, a fluorescent fingerprint is acquired in units of pixels by measuring position information in addition to fluorescent fingerprint information, and component distribution is visualized by color mapping. In addition, in the method described in Patent Document 3, the influence of the solvent content (or a disturbance factor for the specific component to be visualized) between the evaluation object and the sample is further removed, and the specific component of the evaluation object is further clarified. Analyzing.

  Here, the “fluorescence fingerprint” described in Patent Documents 1 to 3 is also called a fluorescence fingerprint (Excitation-Emission Matrix: EEM), and the fluorescence spectrum is obtained while continuously changing the wavelength of the excitation light applied to the sample. This means three-dimensional data obtained by measurement. Since the shape of the fluorescence fingerprint is determined in a component-specific manner like a fingerprint, the measurer can detect subtle component differences that cannot be determined only by fluorescence measurement at a single single wavelength. In addition, the measurer uses fluorescent fingerprint imaging that measures the fluorescent fingerprint for each pixel or each pixel block with positional information (that is, spatial information indicating the position of each pixel in the image) in addition to the fluorescent fingerprint information. By using, the distribution of the specific component in the evaluation object can be visualized.

  Further, in the method described in Patent Document 4, an absorption spectrum is developed for imaging, and a filter having a first transmission band and a filter having a second transmission band including the first transmission band are used. Normalization is performed to enable discrimination based on a spectral image.

JP 2010-185719 A Japanese Patent No. 3706914 JP 2010-266380 A International Publication No. 2009/123068

  However, in the conventional method, when analyzing only from the peak fluorescence intensity as in the fluorescence analysis method, etc., it can be measured and analyzed in a short time, but it is low precision. On the other hand, broadband fluorescence intensity distribution such as fluorescence fingerprints, etc. However, there is a problem that it takes time to measure and analyze although it is highly accurate.

  For example, in the methods described in Patent Documents 1 to 3, it is necessary to scan a wide wavelength range while changing the wavelength condition in order to obtain a fluorescent fingerprint. Therefore, the mechanism is complicated and downsizing is difficult, and measurement time However, it had the problem of requiring a minimum of several minutes.

  Further, in the method of Patent Document 4 and the like, an optical filter is used for imaging an absorption spectrum, but the number of measurement wavelengths is limited in terms of mechanism, and a fluorescence fingerprint can provide a large number of fluorescence spectra. Therefore, it has a problem that it cannot be applied as it is.

  The present invention has been made in view of the above problems, and implements a discriminating filter design method capable of performing measurement and analysis with high accuracy in a short time in order to evaluate an evaluation object based on fluorescence characteristics. It is an object of the present invention to provide a discriminating filter set that is a group of optical filters, a discriminating apparatus and discriminating method using the discriminating filter, and a series of programs that enable them.

  In order to achieve such an object, a discriminating filter for determining a combination of an excitation wavelength band to be irradiated to the evaluation object and a fluorescence wavelength band to be observed, which is appropriate for evaluating the evaluation object by fluorescence characteristics. A design method, a wavelength band setting step for setting at least one combination of the excitation wavelength band and the fluorescence wavelength band, and a plurality of evaluation targets whose evaluation values are known, which are acquired in advance by fluorescence spectroscopy in a wide wavelength range. In addition, based on fluorescence fingerprint information composed of each excitation wavelength and fluorescence intensity at each fluorescence wavelength, the excitation wavelength band set in the wavelength band setting step and the fluorescence intensity obtained in the fluorescence wavelength band are calculated by simulation. Using the fluorescence intensity calculation step and the fluorescence intensity calculated in the fluorescence intensity calculation step and the evaluation value as variables The fluorescence intensity while re-setting the combination of the excitation wavelength band and the fluorescence wavelength band in the estimation formula acquisition step for acquiring the estimation formula for estimating the evaluation value from the fluorescence intensity and the wavelength band setting step By repeatedly executing the calculation step and the estimation formula acquisition step, an appropriate estimation formula for obtaining the evaluation value of the evaluation target is obtained, and the most appropriate from the combination of the excitation wavelength band and the fluorescence wavelength band is most appropriate. And an optimization step for determining a simple estimation formula.

  The discriminating filter design method of the present invention is the discriminant filter design method described above, wherein at least one of the combinations of the excitation light wavelength band and the fluorescence wavelength band is a half of a normal narrowband interference filter. It is characterized by having a broad wavelength band that is clearly wider than the value width.

  The discriminant filter design method of the present invention is characterized in that, in the discriminant filter design method described above, the evaluation value is a value for discriminating, detecting, or quantifying the evaluation object.

  Further, the discriminant filter design method of the present invention is the discriminant filter design method described above, wherein, in the estimation formula acquisition step, the fluorescence intensity is an explanatory variable and the evaluation value is an objective variable, and the estimation is performed by multivariate analysis. It is characterized by acquiring an expression.

  The discriminating filter design method of the present invention is the discriminant filter design method described above, wherein, in the fluorescence intensity calculation step, the fluorescence intensity at each excitation wavelength and each fluorescence wavelength is calculated with respect to the fluorescence fingerprint information. By integrating, the fluorescence intensity in the excitation wavelength band and the fluorescence wavelength band is calculated.

  The discriminant filter design method of the present invention is the discriminant filter design method described above, wherein, in the optimization step, among the plurality of estimation formulas acquired in the estimation formula acquisition step, the known evaluation What is obtained is the above-described appropriate estimation formula, which has a good balance between the two, having a small error with respect to the value and good estimation accuracy even for unknown samples.

  In addition, the discrimination method of the present invention is a discrimination method for evaluating an evaluation object based on fluorescence characteristics, and is an excitation wavelength side discrimination that transmits light having an excitation wavelength in the excitation wavelength band determined by the above-described discrimination filter design method. A discrimination filter setting step for setting a filter and a fluorescence wavelength side discrimination filter that transmits light having a fluorescence wavelength in a predetermined fluorescence wavelength band, and the evaluation object is irradiated with an excitation wavelength in the excitation wavelength band, and the evaluation object Execution of causing the filter setting step and the fluorescence measurement step to be executed at least once in a combination of at least one of the excitation wavelength band and the fluorescence wavelength band. The control step and the above-mentioned obtained in the fluorescence measurement step in a predetermined estimation formula with the fluorescence intensity and the evaluation value of the evaluation target as variables An estimation calculation step of calculating the evaluation value by substituting the light intensity, and the estimation formula includes the excitation wavelength band and the fluorescence wavelength band of the plurality of evaluation objects whose evaluation values are known. It is characterized in that it is preliminarily optimized as an appropriate estimation formula for obtaining the evaluation value of the evaluation object together with the fluorescence intensity information in the combination of the discrimination filters.

  Moreover, the discrimination method of the present invention is characterized in that, in the discrimination method described above, the evaluation value is a value for discriminating, detecting, or quantifying the evaluation object.

  Further, according to the discrimination method of the present invention, in the discrimination method described above, the estimation formula is acquired by optimization by multivariate analysis using the fluorescence intensity as an explanatory variable and the evaluation value as an objective variable. It is characterized by being.

  The discrimination filter set of the present invention is a set of optical filters for discriminating an evaluation object based on fluorescence characteristics, and has an optical characteristic approximated to the excitation wavelength band determined by the discrimination filter design method described above. It is at least one combination of an excitation side optical filter and a fluorescence side optical filter having optical characteristics approximated to the fluorescence wavelength band determined by the discriminating filter design method, and it is possible to discriminate, detect and quantify an evaluation target. It is used for the measurement for performing.

  In the discrimination filter set of the present invention, the excitation side optical filter is used with an excitation light source, and the fluorescence side optical filter is used with a fluorescence intensity detector. It is used to estimate.

  The discriminating apparatus of the present invention is a discriminating apparatus including a measuring unit and a control unit that are provided with the discriminating filter set described above in the light source and detector portions, which are suitable for discriminating an evaluation target based on fluorescence characteristics. The control unit includes a discrimination filter setting means for setting the excitation wavelength side discrimination filter and the fluorescence wavelength side discrimination filter in a predetermined combination in the discrimination filter set, and an excitation wavelength band for the evaluation target. Fluorescence measuring means for irradiating an excitation wavelength and measuring the fluorescence intensity of the fluorescence wavelength band from the evaluation object, and at least one combination of the excitation wavelength band and the fluorescence wavelength band, the filter setting means and the fluorescence measurement Execution control means for executing the processing by the means at least once, and the fluorescence intensity and the evaluation value of the evaluation object as variables The constant of the estimate equation by substituting the fluorescence intensity obtained by the fluorescent measurement step, characterized in that and an estimation calculation means for calculating the evaluation value.

  Further, the discriminating apparatus of the present invention is characterized in that in the discriminating apparatus described above, the evaluation object is a value for discriminating, detecting or quantifying.

  In the discriminating apparatus of the present invention, an optimum estimation formula is obtained for estimating the evaluation value by multivariate analysis using the fluorescence intensity as an explanatory variable and the evaluation value as an objective variable. It is characterized by.

  Further, the program of the present invention stores a memory for determining a combination of an excitation wavelength band to be irradiated to the evaluation object and a fluorescence wavelength band to be observed, which is appropriate for evaluating the evaluation object by fluorescence characteristics. A storage unit is a program for causing a computer including a calculation unit and a calculation unit to execute each excitation wavelength over a wide wavelength range acquired in advance by fluorescence spectroscopy for a plurality of evaluation targets whose evaluation values are known. A wavelength band setting step for setting at least one combination of the excitation wavelength band and the fluorescence wavelength band in the calculation unit, and an evaluation value. Based on fluorescence fingerprint information of fluorescence intensity at each excitation wavelength and each fluorescence wavelength acquired in advance for a plurality of known evaluation targets, the wavelength band The fluorescence intensity calculation step for calculating the fluorescence intensity obtained in the excitation wavelength band and the fluorescence wavelength band set in the setting step, and the fluorescence intensity and the evaluation value calculated in the fluorescence intensity calculation step as variables. , An estimation formula acquisition step for acquiring an estimation formula for obtaining the evaluation value, and resetting the combination of the excitation wavelength band and the fluorescence wavelength band in the wavelength band setting step, while the fluorescence intensity calculation step and the above An optimization step for obtaining the optimal estimation formula for obtaining the evaluation value of the evaluation target by repeatedly executing the estimation formula acquisition step and determining the combination of the excitation wavelength band and the fluorescence wavelength band; , Is executed.

  In the program according to the present invention, the evaluation value is a value for discriminating, detecting, or quantifying the evaluation object.

  Further, the program of the present invention, in the program described above, acquires the estimation formula by multivariate analysis or the like using the fluorescence intensity as an explanatory variable and the evaluation value as an objective variable in the estimation formula acquisition step. It is characterized by.

  According to the present invention, at least one combination of the excitation wavelength band and the fluorescence wavelength band is set, and each excitation wavelength and each of the plurality of evaluation objects whose evaluation values are known are acquired in advance by fluorescence spectroscopy in a wide wavelength range. Based on the fluorescence fingerprint information consisting of the fluorescence intensity at the fluorescence wavelength, the fluorescence intensity obtained in the set excitation wavelength band and fluorescence wavelength band is calculated by simulation, and the evaluation value is calculated using the calculated fluorescence intensity and evaluation value as variables. While obtaining an estimation formula for obtaining and re-execution of the above processing while resetting the combination of the excitation wavelength band and the fluorescence wavelength band, an optimum estimation formula for obtaining the evaluation value of the evaluation target is obtained, A combination of excitation wavelength band and fluorescence wavelength band is determined. Thereby, in order to evaluate an evaluation object by a fluorescence characteristic, this invention has an effect that a highly accurate measurement and analysis can be performed in a short time. More specifically, by using at least one combination of filters that transmit the determined excitation wavelength band and the fluorescence wavelength band, scanning is performed by gradually changing the excitation wavelength on the irradiation side and the fluorescence wavelength on the observation side. Compared with the method of discriminating by the fluorescence fingerprint which needs to measure all combinations of wavelengths, it is possible to perform measurement including only necessary information in a short time.

  Further, according to the present invention, in the above, the evaluation value is a value for discriminating, detecting, or quantifying the evaluation object, and therefore, the evaluation object is discriminated, detected, or quantified. The estimation formula can be obtained, and the evaluation object can be evaluated with the fluorescence characteristics with high accuracy in a short time.

  Further, according to the present invention, when calculating the fluorescence intensity in the above, by integrating the fluorescence intensity at each excitation wavelength and each fluorescence wavelength with respect to the fluorescence fingerprint information, the excitation wavelength band and the fluorescence wavelength band are integrated. Calculating fluorescence intensity is more accurate than the conventional fluorescence intensity in a narrow band (before integration) because a large signal including necessary information can be observed and the S / N ratio can be greatly improved. There is an effect that measurement and analysis can be performed.

  Further, according to the present invention, in the above, at the time of optimization, among the plurality of obtained estimation formulas, there is little error from a known evaluation value, and an unknown sample (an evaluation target that has not been used for creating an estimation formula) ), Which is well-estimated and is well balanced, is obtained as an optimal estimation formula for obtaining the evaluation value of the evaluation target, so that accurate measurement and estimation are possible even for unknown samples There is an effect that can be performed.

FIG. 1 is a diagram schematically showing an overview of a discrimination filter design method in the present embodiment. FIG. 2 is a flowchart showing an example of a discrimination filter design method in the present embodiment. FIG. 3 is a block diagram illustrating an example of the configuration of the simulation apparatus 111 according to the present embodiment. FIG. 4 is a block diagram illustrating an example of the configuration of the determination apparatus 100 according to the present embodiment. FIG. 5 is a block diagram illustrating an example of the configuration of the measurement unit 110. FIG. 6 is a flowchart showing an example of a discrimination filter design method executed by the simulation apparatus 111 of the present embodiment. FIG. 7 is a diagram showing the filter characteristics of one discrimination filter set on a fluorescent fingerprint. FIG. 8 is a diagram schematically showing a method for converting a sample spectrum into a sensor output based on filter characteristics. FIG. 9 is a diagram showing the coefficients and the positions and sizes of the measurement windows obtained on the fluorescent fingerprint when the number of measurement windows is set to two (estimation formula: Y = aW ₁ + bW ₂ ). FIG. 10 is a diagram showing the coefficients and the positions and sizes of the measurement windows obtained on the fluorescent fingerprint when the number of measurement windows is set to 3 (estimation formula: Y = aW ₁ + bW ₂ + cW ₃ ). is there. FIG. 11 is a diagram showing the coefficient, the position and the size of the measurement window on the fluorescent fingerprint obtained when the number of measurement windows is set to 4 (estimation formula: Y = aW ₁ + bW ₂ + cW ₃ + dW ₄ ). It is. FIG. 12 is a graph in which the vertical axis represents the predicted value based on the estimation formula obtained when the number of measurement windows is set to two, and the horizontal axis represents the actual value measured by the AOAC official method. FIG. 13 is a graph in which the vertical axis represents the predicted value based on the estimation formula obtained when the number of measurement windows is set to three, and the horizontal axis represents the actual value measured by the AOAC official method. FIG. 14 is a graph in which the vertical axis represents the predicted value based on the estimation formula obtained when the number of measurement windows is set to four, and the horizontal axis represents the actual value measured by the AOAC official method. FIG. 15 is a flowchart illustrating an example of the operation of the determination apparatus 100 according to the present embodiment. FIG. 16 is a diagram illustrating an example (one-point measurement) of the measurement unit 110 that is set by the discrimination filter setting unit 102e. FIG. 17 is a diagram illustrating another example (imaging measurement) of the measurement unit 110 that is set by the discrimination filter setting unit 102e. FIG. 18 is a diagram illustrating filter characteristics of an ideal filter created in the present embodiment. FIG. 19 is a diagram showing the actual filter characteristics of the two discriminating filters (optical filters) created in this example. FIG. 20 shows a color image obtained by photographing black asphalt with an ice / water film (left figure) and a filter manufactured by a near infrared camera XSVA XS FPA-1.7-320 (trade name) manufactured by Xenics. It is a figure which shows the ice / water discrimination | determination result image (right figure) measured by mounting | wearing.

  Hereinafter, examples of preferred embodiments of a discrimination filter design method, a discrimination method, a discrimination filter set, a discrimination device, and a program according to an embodiment of the present invention will be described in detail with reference to FIGS. To do. Note that the present invention is not limited to the embodiments. Further, in the present embodiment, various conditions such as the evaluation object, the fluorescence intensity measurement condition, and the statistical analysis method can be selected as appropriate without departing from the object of the present invention.

[I. Discriminant filter design method]
First, an overview of the discrimination filter design method in the present embodiment will be described with reference to FIGS. 1 and 2. Here, FIG. 1 is a diagram schematically showing an outline of the discrimination filter design method in the present embodiment.

  In the present embodiment, the “fluorescence fingerprint” is also called an excitation-emission matrix (EEM), and as shown in FIG. 1, the excitation wavelength (nm) for irradiating the evaluation object and the evaluation object 3 is a contour graph of three-dimensional data composed of three axes of an excitation wavelength, a fluorescence wavelength, and a fluorescence intensity, representing a fluorescence intensity of an evaluation object at a fluorescence wavelength (nm) of light emitted from the object.

  As shown in FIG. 1, “fluorescence fingerprint” can be represented as a plane by plotting the fluorescence intensity at each point as a contour plot, with the vertical axis representing the fluorescence wavelength and the horizontal axis representing the excitation wavelength, for example. Fluorescent fingerprints have the advantages of being able to characterize the evaluation object without pretreatment such as fluorescent staining, being easy to operate and measuring in a short time, and having higher sensitivity than absorbance. Therefore, it is a widely used technique for analysis of the internal structure of food, etc., measurement of suspended matters in the atmosphere, and identification of dye raw materials. Thus, since the fluorescence fingerprint is the fluorescence information specific to the component having a large amount of information in three dimensions, the measurer can identify the component by using the fluorescence fingerprint and is nondestructive. Can be measured. Fluorescence fingerprint (excitation fluorescence matrix) is a measurement technology that has been put to practical use because computers can handle a large amount of information. In addition to detection and quantification of mold poisons (DON (deoxynivalenol), aflatoxin, etc.) In addition, such as estimation of the mixing ratio of buckwheat flour in dry noodles, discrimination of mango production areas, etc. have been applied to discrimination and quantification that could not be realized so far.

  Conventionally, fluorescence analysis is used as another measurement method. In the fluorescence analysis method, a method of measuring the concentration of an evaluation target by irradiating an evaluation target with excitation light of a specific wavelength and measuring the fluorescence intensity of the specific wavelength that is the response peak is generally used. is there. 1 indicates an example of a combination of an excitation wavelength and a fluorescence wavelength used in the fluorescence analysis method. As described above, the conventional fluorescence analysis method can measure and analyze in a short time with a small number of conditions. However, since only the peak information is used, the contour line shape information is not included, and fluctuation (disturbance) etc. In addition, it is difficult to quantify without such a prominent peak. On the other hand, since the method using the fluorescent fingerprint described above uses all wavelength conditions, all the information of the contour line shape can be used and the influence of fluctuations and the like can be removed. There is a problem that it takes a long time to measure and analyze because of the enormous volume.

  In view of such a conventional method, the present inventors can simplify a conventional measurement / analysis method and perform high-precision measurement and analysis in a short time in order to evaluate an evaluation object by fluorescence characteristics. A method was developed. A broken-line rectangle in FIG. 1 schematically represents the feature according to the present embodiment on a fluorescence fingerprint, and specifically, a specific excitation wavelength band and a specific fluorescence wavelength band used in the present embodiment. Represents a combination of. In the following description, the rectangular region composed of the excitation wavelength band and the fluorescence wavelength band may be referred to as a “measurement window”.

  In the present embodiment, unlike the fluorescence analysis method, it is not used under one condition of the excitation wavelength and the fluorescence wavelength, and it is not necessary to scan all the wavelength conditions as in the fluorescence fingerprint method. That is, as shown by the broken-line rectangle (measurement window) in FIG. 1, since a combination of a specific excitation wavelength band and a specific fluorescence wavelength band is used, measurement and analysis can be performed in a short time with a small number of conditions. In addition, since the contour shape information is included, a highly accurate analysis result can be obtained. In other words, the present embodiment has improved the conventional discrimination method that conventionally had to measure fluorescent fingerprints by scanning all wavelengths or through a huge number of narrowband filters, The broadband optical filter that covers the part where the target information can be obtained dramatically simplifies the necessary measurement, and also enables reduction in measurement time and easy application to imaging.

  In the present embodiment, it is not necessary to obtain the fluorescence intensity distribution inside the measurement window by scanning, and it is only necessary to obtain the fluorescence intensity over the entire measurement window. For example, by using at least one pair of optical filters corresponding to the measurement window (combination of excitation wavelength side discrimination filter and fluorescence wavelength side discrimination filter), the fluorescence intensity on the entire measurement window is obtained with a simple mechanism instead of digital. can do. The position (center wavelength) and size (bandwidth) of the measurement window are not arbitrary, and it is necessary to appropriately set a range in which target information can be obtained as shown as an example below. Here, FIG. 2 is a flowchart showing an example of a discrimination filter design method in the present embodiment.

  As shown in FIG. 2, first, at least one combination of an excitation wavelength band and a fluorescence wavelength band is set (step SA-1). Note that the number and band of combinations may be arbitrarily set, for example, may be set randomly based on a random number table, may be set by the user, or a predetermined initial value may be used. Good. Here, at least one or more of the combination of the excitation light wavelength band and the fluorescence wavelength band may be a broad wavelength band clearly wider than the half-value width (about 10 nm) of a normal narrow band interference filter.

  Then, based on the fluorescence fingerprint information measured in advance by fluorescence spectroscopy using a fluorescence spectrophotometer or the like, the fluorescence intensity obtained in the excitation wavelength band and fluorescence wavelength band set in Step SA-1 is calculated (Step S1). SA-2). Here, the fluorescence fingerprint information is information relating to the above-described fluorescence fingerprint referred to in FIG. 1, and the fluorescence intensity distribution at each excitation wavelength and each fluorescence wavelength acquired in advance for a plurality of evaluation objects with known evaluation values. It is information which shows. As an example, the fluorescence fingerprint information includes, for example, a predetermined excitation wavelength range such as a wide wavelength range of 200 nm to 900 nm and a predetermined fluorescence wavelength range. The spectrum information (fluorescence spectrum) of the fluorescence intensity generated from the evaluation object is measured by changing continuously. A fluorescence fingerprint can be acquired by sequentially changing the excitation wavelength, acquiring a fluorescence spectrum in the same manner, and measuring the fluorescence spectrum over the entire excitation wavelength range.

  Here, the evaluation value is a value for discriminating, detecting, or quantifying the evaluation target. For example, a discrimination value of “0” or “1” in the two-group discrimination, a quantitative value such as a concentration, or the like. It may be. Further, as an example of a method for calculating fluorescence intensity, fluorescence intensity in each excitation wavelength band and each fluorescence wavelength band may be calculated by integrating fluorescence intensity in each excitation wavelength and each fluorescence wavelength with respect to fluorescence fingerprint information. . The evaluation object is an arbitrary sample that is an object to be evaluated by fluorescence specification, and may be a food, a material, a road surface, or the like as an example.

  And the candidate of the estimation formula for calculating | requiring the said evaluation value is acquired using the fluorescence intensity and evaluation value which were calculated in step SA-2 as a variable (step SA-3). Here, multivariate analysis (multiple regression analysis, PLS regression analysis, discriminant analysis, principal component analysis, factor analysis, cluster analysis, etc.) is performed using the fluorescence intensity as an explanatory variable and the evaluation value as an objective variable. The estimation formula may be acquired by For example, a calibration curve of evaluation values with respect to fluorescence intensity may be created using a least square method or the like.

  Then, it is determined whether or not the end condition is satisfied (step SA-4). For example, the termination condition may be a case where the above-described processing is repeated a predetermined number of times, or may be a case where an estimation formula having an error within a predetermined range is obtained, and the termination is designated by the user. It may be the case.

  When the termination condition is not satisfied (step SA-4, No), the process returns to step SA-1, and the above-described steps SA-1 to SA-4 are performed while resetting the combination of the excitation wavelength band and the fluorescence wavelength band. Repeat the process. Note that the number and band of combinations may be reset arbitrarily as in step SA-1, but are reset using a known optimization method such as a genetic algorithm or a simulated annealing method in the iterative process. May be performed.

  On the other hand, when the termination condition is satisfied (step SA-4, Yes), an optimal estimation formula for obtaining the evaluation value of the evaluation target is obtained from the obtained candidate estimation formulas, and the optimal estimation formula is A combination of an excitation wavelength band and a fluorescence wavelength band under the obtained conditions is determined (step SA-5). For example, among a plurality of candidate estimation formulas, a formula having a small error between an evaluation value (estimation value) obtained by the estimation formula and a known evaluation value (actual measurement value) may be acquired as an optimal estimation formula.

  The above is an example of the discrimination filter design method in the present embodiment. A more specific example of the discrimination filter design method will be described in detail below.

[II. Simulation device and discrimination device]
Next, an example of the configuration of the simulation apparatus and the determination apparatus in the present embodiment will be described below with reference to FIGS. The simulation apparatus in the present embodiment is an apparatus such as a computer that executes the above-described discriminating filter design method, and the discriminating apparatus in the present embodiment includes an excitation wavelength band determined by the above-described discriminating filter design method and A discriminating filter set (optical filter) that transmits the fluorescence wavelength band is mounted on the measurement unit, and the target for evaluation is discriminated, detected, or quantified with high accuracy in a short time from the fluorescence intensity measured from each wavelength condition. Device. Here, FIG. 3 is a block diagram showing an example of the configuration of the simulation apparatus 111 in the present embodiment, and conceptually shows a part related to the present invention in the configuration.

  As shown in FIG. 3, the simulation apparatus 111 schematically includes a storage unit 116 and an operation unit 112 such as a CPU that controls the entire system in an overall manner.

  Various databases and tables (fluorescent fingerprint file 116a and candidate estimation expression file 116b) stored in the storage unit 116 are storage means such as a fixed disk device. For example, the storage unit 116 stores various programs, tables, files, databases, and the like used for various processes.

  Among these components of the storage unit 116, the fluorescence fingerprint file 116a stores fluorescence fingerprint information of fluorescence intensity at each excitation wavelength and each fluorescence wavelength for a plurality of evaluation objects whose evaluation values are known. Means. Here, the fluorescence fingerprint information stored in the fluorescence fingerprint file 116a may be information measured by an external fluorescence spectrophotometer, and is connected to an external device (remotely) via an external storage device such as a USB memory or the Internet. Fluorescent fingerprint information may be obtained from a fluorescent spectrophotometer on the ground.

  The storage unit 116 is configured to emit fluorescence at a preset excitation wavelength range (200 nm to 400 nm in the example of FIG. 1), a fluorescence wavelength range (200 nm to 800 nm in the example of FIG. 1), and a wavelength pitch (for example, 20 nm intervals). A plurality of data files of fluorescent fingerprints to be evaluated, which are measured by a spectrophotometer or the like and whose evaluation values are known, are accumulated. In other words, in the predetermined excitation wavelength range and the predetermined fluorescence wavelength range, the fluorescence intensity is measured for all combinations of wavelength conditions in the evaluation object while gradually changing the excitation wavelength and the observed fluorescence wavelength. To obtain fluorescent fingerprint information. That is, with an existing spectrofluorometer, excitation wavelengths (for example, m wavelength conditions obtained by dividing the measurement wavelength range by the data acquisition interval) and fluorescence wavelengths (for example, n by dividing the measurement wavelength range by the data acquisition interval) The fluorescence intensity of the evaluation object is acquired under a total of m × n wavelength conditions while changing the combination of (wavelength conditions). In addition, according to the setting example enumerated above with reference to FIG. Here, the measurer may adjust the number of times of measurement (for example, three times for each sample).

  Thus, the fluorescence intensity information measured while changing the wavelength condition is stored in the fluorescence fingerprint file 116a in association with the wavelength condition. Here, data preprocessing for extracting necessary information by removing noise (for example, scattered light of the excitation light, its secondary light, tertiary light, etc.) may be performed on the obtained fluorescent fingerprint information. For example, data in a range where the fluorescence wavelength is shorter than the excitation wavelength may be removed. This is a process for analyzing only the fluorescence intensity data of the fluorescence wavelength longer than the excitation wavelength because the fluorescence wavelength emitted by the evaluation object is longer than the excitation wavelength. Note that data in a region where the luminance value of the excitation light source is low (for example, an excitation wavelength of 240 nm or less) and data in a region where the sensitivity of fluorescence and detector is low (for example, a fluorescence wavelength of 800 nm or more) may be deleted. The above is an example of a method for acquiring fluorescent fingerprint information.

  The candidate estimation formula file 116b is candidate estimation formula storage means for storing the candidate estimation formula (discriminant formula, calibration curve, etc.) acquired by the estimation formula acquisition unit 112c. Here, the candidate estimation formula file 116b may store the excitation wavelength band and the fluorescence wavelength band set by the wavelength band setting unit 112a in association with the candidate estimation formula.

  In FIG. 3, the calculation unit 112 (wavelength band setting unit 112 a to optimization unit 112 d) stores a control program such as an OS (Operating System), a program that defines various processing procedures, and necessary data. An internal memory for And the calculating part 112 performs the information processing for performing various processes with these programs. The calculation unit 112 includes a wavelength band setting unit 112a, a fluorescence intensity calculation unit 112b, an estimation formula acquisition unit 112c, and an optimization unit 112d in terms of functional concept.

  Among these, the wavelength band setting unit 112a is a wavelength band setting unit that sets at least one combination of the excitation wavelength band and the fluorescence wavelength band. As the number of combinations of excitation wavelength bands and fluorescence wavelength bands set by the wavelength band setting unit 112a increases, the number of variables increases, the estimation formula becomes complicated, the processing time becomes longer, and the estimation accuracy for an unknown sample decreases. Therefore, it may be limited to a predetermined number (such as an upper limit number). In the repetition process, the wavelength band setting unit 112a may set the excitation wavelength band and the fluorescence wavelength band by a brute force formula, or set by a known optimization method such as a genetic algorithm or a simulated annealing method. Also good. Here, in the wavelength band setting unit 112a, at least one of the combination of the excitation light wavelength band and the fluorescence wavelength band to be set is a wide band that is clearly wider than the half width (about 10 nm) of a normal narrow band interference filter. May be the wavelength band.

  Further, the fluorescence intensity calculation unit 112b calculates the fluorescence intensity obtained in the excitation wavelength band and the fluorescence wavelength band set by the wavelength band setting unit 112a by simulation based on the fluorescence fingerprint information stored in the fluorescence fingerprint file 116a. It is a fluorescence intensity calculation means. For example, the fluorescence intensity calculation unit 112b calculates the fluorescence intensity in the excitation wavelength band and the fluorescence wavelength band by integrating the fluorescence intensity at each excitation wavelength and each fluorescence wavelength with respect to the fluorescence fingerprint information. As an example, the fluorescence intensity calculation unit 112b integrates the fluorescence intensities of all the wavelength conditions within the set excitation wavelength band and the fluorescence wavelength band based on the fluorescence fingerprint information, thereby obtaining the excitation wavelength band and the fluorescence wavelength band. You may acquire the fluorescence intensity obtained by.

  In addition, the estimation formula acquisition unit 112c uses the fluorescence intensity calculated by the fluorescence intensity calculation unit 112b and the known evaluation value of the evaluation target for measuring the fluorescence intensity as variables, and uses the estimation formula for obtaining the evaluation value. It is an estimation formula acquisition means to acquire. As an example, the estimation formula acquisition unit 112c may acquire the estimation formula by multivariate analysis or the like using the fluorescence intensity as an explanatory variable and the evaluation value as a target variable. As an example, the multivariate analysis may be multiple regression analysis, PLS regression analysis, multiple regression analysis, discriminant analysis, principal component analysis, factor analysis, cluster analysis, or the like. For example, a calibration curve of evaluation values with respect to fluorescence intensity may be created using a least square method or the like.

  Further, the optimization unit 112d performs evaluation by repeatedly executing the processing by the fluorescence intensity calculation unit 112b and the estimation formula acquisition unit 112c while the wavelength band setting unit 112a resets the combination of the excitation wavelength band and the fluorescence wavelength band. This is an optimization means for obtaining an appropriate estimation formula for obtaining an evaluation value of an object and determining a combination of an excitation wavelength band and a fluorescence wavelength band. For example, the optimization unit 112d may acquire, as an optimal estimation formula, the one with the smallest error from a known evaluation value among the plurality of candidate estimation formulas acquired by the estimation formula acquisition unit 112c. The optimization unit 112d is not limited to this, and the optimization unit 112d uses an estimation formula evaluation value such as a misclassification rate, selects an appropriate estimation formula from candidate estimation formulas in consideration of estimation accuracy for an unknown sample, or these The determination may be made in consideration of a plurality of criteria. Here, the iterative process by the wavelength band setting unit 112a, the fluorescence intensity calculation unit 112b, and the estimation formula acquisition unit 112c controlled by the optimization unit 112d may be a predetermined number of times and satisfies a predetermined end condition. In some cases (for example, when an estimation formula having an error within a predetermined range is obtained), the processing may be terminated.

  The above is the function of each unit of the simulation apparatus 111 that executes the discrimination filter design method of the present embodiment. Below, each part of the discrimination device 100 connected with the measurement part 110 of this Embodiment is demonstrated. Here, FIG. 4 is a block diagram showing an example of the configuration of the determination apparatus 100 of the present embodiment, and conceptually shows a part related to the present invention in the configuration.

  As shown in FIG. 4, the discriminating apparatus 100 roughly includes a control unit 102 such as a CPU that controls the entire system, a storage unit 106, a measurement unit 110, and an input device 113 in the present embodiment. And an output device 114. The input / output control interface unit 108 is an interface connected to the measurement unit 110, the input device 113, and the output device 114.

  In FIG. 4, the input / output control interface unit 108 controls the measurement unit 110, the input device 113, and the output device 114. Here, as the output device 114, in addition to a monitor (including a home television), a speaker can be used (hereinafter, the output device 114 may be described as a monitor). As the input device 113, a keyboard, a mouse, a microphone, and the like for inputting measurement conditions and the like can be used. The measurement unit 110 is a device that measures the fluorescence intensity to be evaluated, and includes a light source and a detector. The spectroscope has the transmission characteristics of the excitation wavelength band and the fluorescence wavelength band obtained by the discrimination filter design method. Has a discriminating filter. Here, FIG. 5 is a block diagram illustrating an example of the configuration of the measurement unit 110.

  As shown in FIG. 5, the measurement unit 110 includes a spectral illumination unit 11 and a spectral detection unit 12. The spectral illumination unit 11 is a device that irradiates the evaluation target 13 with excitation light in a predetermined excitation wavelength band to generate fluorescence from the components of the evaluation target 13. The spectral illumination unit 11 may include excitation wavelength band variable means for changing the excitation wavelength band to be irradiated. As illustrated, the spectroscopic illumination unit 11 includes a light source 11a and a spectroscope 11b.

  As the light source 11a of the spectral illumination unit 11, a xenon lamp, a tungsten lamp, a wavelength tunable laser, or the like that emits white light can be used. The light emitted from the light source 11a has a wide wavelength band from, for example, the ultraviolet to the visible region, and is split into excitation light having a predetermined wavelength band in the spectroscope 11b. The measurer can set the excitation light to an arbitrary wavelength band by adjusting the excitation wavelength band variable means. When a wavelength tunable laser is used as the light source 11a, the spectroscope 11b is unnecessary because excitation light having a desired wavelength band can be obtained directly from the light source 11a.

  As the spectroscope 11b, for example, a band pass filter (BPF, interference filter), an AOTF (Acoustic Optical Tunable Filter), a liquid crystal tunable filter, a diffraction grating, or the like that transmits a predetermined excitation wavelength band can be used. Moreover, you may use the LED light source, DLP light source, etc. which integrated the light source 11a and the spectrometer 11b. In addition, the combination which can select a wavelength band widely is a band pass filter for a xenon light source, and is optimal for the spectroscopic purpose of the present invention, but is not limited thereto.

  The spectroscopic detection unit 12 is a device that acquires the fluorescence intensity of the evaluation target 13 in a predetermined fluorescence wavelength band and transmits the fluorescence intensity information to the determination device 100. The spectroscopic detection unit 12 selectively captures a specific fluorescence wavelength band from the fluorescence emitted from the evaluation object 13 via the spectroscope, and measures the fluorescence intensity. The spectroscopic detection unit 12 includes fluorescence wavelength band variable means for changing the observed fluorescence wavelength band. The spectroscopic detection unit 12 includes a spectroscope 12a and a detector 12b.

  By irradiating the excitation light, the fluorescence emitted from the measurement object 13 is split into light having a predetermined wavelength by the spectroscope 12a. The spectroscope 12a has means for setting the fluorescence wavelength band to be observed. As the spectroscope 12a, for example, a band-pass filter (BPF, interference filter), an AOTF (Acoustic Optical Tunable Filter), a liquid crystal tunable filter, a PGP (prism, grating, prism), diffraction, which transmits a predetermined fluorescence wavelength band. A lattice or the like can be used. A spectroscope capable of selecting a wide wavelength band at low cost is a bandpass filter, which is optimal for the spectroscopic purpose of the present invention, but is not limited thereto.

  By passing through the spectroscope 12a, the intensity of light having a predetermined wavelength band in the fluorescence emitted from the measurement object 13 is detected by the detector 12b. The excitation wavelength band to be irradiated and the fluorescence wavelength band to be observed are a combination calculated by the discriminating filter design method, and these can be arbitrarily set manually or automatically by the measurer. Preferably, the measurement unit 110 includes means for automatically changing the excitation wavelength band and the fluorescence wavelength band.

  The spectroscopes 11b and 12a use a liquid crystal tunable filter, a diffraction grating, or the like that transmits light in a narrow band when measuring a fluorescent fingerprint, and on the other hand, according to the wavelength band determined by the discriminating filter design method. When performing measurement and evaluation according to the embodiment, an interference filter (optical filter) or the like that transmits broadband light may be used.

  In FIG. 4, the control unit 102 has an internal memory for storing a control program such as an OS (Operating System), a program that defines various processing procedures, and necessary data. And the control part 102 performs the information processing for performing various processes by these programs. The control unit 102 includes a discrimination filter setting unit 102e, a fluorescence measurement unit 102f, an execution control unit 102g, and an estimation calculation unit 102h in terms of functional concept.

  The discrimination filter setting unit 102e includes an excitation wavelength side discrimination filter that transmits light of a predetermined excitation wavelength band determined by the discrimination filter design method, and a fluorescence wavelength side discrimination filter that transmits light of a predetermined fluorescence wavelength band. It is a discrimination filter setting means for setting. When there are a plurality of combinations of excitation wavelength bands and fluorescence wavelength bands determined by the discriminating filter design method, the discriminating filter setting unit 102e performs setting for each set.

  The fluorescence measurement unit 102f irradiates the evaluation target with light in the excitation wavelength band via the measurement unit 110 with the combination of filters set by the discrimination filter setting unit 102e, and measures the fluorescence measured from the evaluation target. This is a fluorescence measuring means for acquiring the fluorescence intensity in the wavelength band.

  The execution control unit 102g is an execution control unit that executes the processes of the discrimination filter setting unit 102e and the fluorescence measurement unit 102f once for each combination of the excitation wavelength band and the fluorescence wavelength band.

  In addition, the estimation calculation unit 102h includes an estimation formula calculation unit that calculates an estimation formula for obtaining the evaluation value using the fluorescence intensity measured by the fluorescence measurement unit 102f and the evaluation value of the evaluation target as variables, and the estimation formula, This is an evaluation value calculation means for calculating an evaluation value by substituting the fluorescence intensity obtained by the fluorescence measurement unit 102f. Further, the estimation calculation unit 102h may store the evaluation result (evaluation result of the evaluation value, etc.) in the storage unit 106, display it on the output device 114 such as a monitor, or output it from a printer or the like. It may be printed out on the device 114.

  The above is the description of each unit of the determination apparatus 100.

[III. Discriminant filter design method executed in simulation apparatus 111]
Next, an example of the discrimination filter design method executed in the simulation apparatus 111 in the present embodiment will be described below with reference to FIGS. Here, FIG. 6 is a flowchart illustrating an example of a discrimination filter design method executed in the simulation apparatus 111 of the present embodiment.

  As shown in FIG. 6, first, the wavelength band setting unit 112a sets the filter characteristics of at least one discrimination filter set (step SB-1). Here, FIG. 7 is a diagram showing the filter characteristics of one discrimination filter set on a fluorescent fingerprint. As shown in FIG. 7, the filter characteristics shown here are the position and size of the measurement window configured from the excitation wavelength band and the fluorescence wavelength band. The wavelength band setting unit 112a may set a preset value as an initial value as a filter characteristic, may be set arbitrarily based on a random number table or the like, or may provide an input device (not shown) to the user. May be set via Further, the wavelength band setting unit 112a may allow the user to specify the number of combinations of the excitation wavelength band and the fluorescence wavelength band via the input device. Thereby, the wavelength band setting unit 112a can reduce the processing time by setting the filter characteristics with the designated number of combinations. In practice, in order to avoid bias in results due to differences in initial values, there is an algorithm that automatically and comprehensively changes the wavelength band settings comprehensively with the target wavelength range and the number of measurement windows as the initial values. Often assembled.

  Then, the fluorescence intensity calculation unit 112b converts the fluorescence fingerprint spectrum stored in the fluorescence fingerprint file 116a into a sensor output having the filter characteristics set by the wavelength band setting unit 112a (step SB-2). Here, FIG. 8 is a diagram schematically showing a method of converting the sample spectrum into the sensor output based on the filter characteristics. Note that λ represents a wavelength.

As shown in FIG. 8a, the fluorescence intensity calculation unit 112b uses the wavelength band as shown in FIG. 8b in the sample spectrum S (λ) of the fluorescence intensity obtained for each wavelength condition stored in the fluorescence fingerprint file 116a. Based on the transmission characteristic T _i (λ) of the optical filter set by the setting unit 112a, the position (center wavelength) and size (bandwidth) of the measurement window are matched. In this example, the transmittance is 100% in the transmission band and 0% in the non-transmission band, but may be arbitrarily set in accordance with the characteristics of the optical filter. Then, as shown in FIG. 8c, the fluorescence intensity calculation unit 112b calculates the integrated value (area of the hatched area in the drawing) of the fluorescence intensity in the range of the measurement window as the sensor output O _i . That is, the fluorescence intensity calculation unit 112b virtually integrates the fluorescence intensity within the wavelength band according to the set filter characteristics from the sample spectrum S (λ), thereby virtually transmitting the sensor output O transmitted through the wavelength band filter. Convert to _i .

  Returning to FIG. 6 again, the estimation formula acquisition unit 112c creates an estimation formula for estimating the evaluation value from the sensor output of the fluorescence intensity calculated by the fluorescence intensity calculation unit 112b (step SB-3). Since the fluorescent fingerprint file 116a stores evaluation values for each fluorescent fingerprint, the estimation formula acquisition unit 112c creates an estimation formula for obtaining the evaluation value from the sensor output using the known evaluation value and the sensor output as variables. To do. For example, the estimation formula acquisition unit 112c may create a relational expression such as a calibration curve for obtaining an evaluation value for the sensor output of the fluorescence intensity by multiple regression analysis or the like. In addition, the estimation formula acquisition unit 112c may acquire the estimation formula by multivariate analysis using the fluorescence intensity as an explanatory variable and the evaluation value as a target variable. Note that the estimation formula acquisition unit 112c may store the created estimation formulas based on various wavelength bands (that is, measurement windows) in the candidate estimation formula file 116b in association with the filter characteristics.

  Then, the optimization unit 112d evaluates the obtained estimation formula created by the estimation formula acquisition unit 112c (step SB-4). For example, the optimization unit 112d performs evaluation by acquiring an error between an evaluation value (estimated value) calculated by substituting the sensor output of the fluorescence intensity into the candidate estimation formula and a known evaluation value (actually measured value). You may go. Further, in addition to estimation accuracy with respect to an unknown sample (an evaluation object that has not been used for creating an estimation formula), the evaluation may be performed because both are balanced. In addition, the optimization unit 112d may evaluate the candidate estimation formula using a known estimation formula evaluation method such as a misclassification rate.

  Then, the optimization unit 112d determines whether or not an end condition is satisfied (step SB-5). For example, the optimization unit 112d may determine that the end condition is satisfied when the evaluation result of the candidate estimation formula is a predetermined condition. As an example, the optimization unit 112d satisfies the termination condition when the error or the estimation accuracy of the unknown sample is within a predetermined range, or when it is determined that it has converged to the optimal solution in an iterative process using a known optimization method. May be determined.

  When it is determined that the termination condition is not satisfied (step SB-5, No), the optimization unit 112d returns the process to step SB-1 and resets the filter characteristics, and then performs steps SB-1 to SB- described above. Step 5 is repeatedly executed. Here, the wavelength band setting unit 112a may reset the filter characteristics using a known optimization method such as a genetic algorithm or a simulated annealing method in an iterative process.

  On the other hand, when it is determined that the termination condition is satisfied (step SB-5, Yes), the optimization unit 112d selects the filter characteristic from which the optimum estimation formula is obtained (step SB-6). If the end condition is that the error or misjudgment rate is within a predetermined range, or if it is determined that it has converged to an optimal solution in an iterative process using a known optimization method, the optimal estimation formula is obtained last. This is a candidate estimation formula. When the termination condition is the number of repetitions or the like, the optimization unit 112d may select an optimal estimation formula from the candidate estimation formulas based on estimation formula evaluation values such as errors and misjudgment rates.

  The above is an example of a simulation program (required number) that realizes the discrimination filter design method according to the present embodiment.

[IV. Example 1]
Here, Example 1 to which the discrimination filter design method (simulation program) of the above-described embodiment is applied will be described below with reference to FIGS.

The purpose of Example 1 was to obtain an appropriate estimation formula and discriminating filter characteristics for estimating the riboflavin content using yogurt as an evaluation object. For the evaluation object, 42 samples were used to obtain each fluorescence fingerprint data (225 wavelength condition = excitation side 15 wavelength condition × fluorescence side 15 wavelength condition) in advance. Moreover, riboflavin quantification was performed based on an AOAC (Association of Official Analytical Chemistry) official method as a known evaluation value. Here, the measurement conditions of the fluorescence fingerprint are as follows.
Measurement wavelength range: excitation wavelength 250 to 550 nm / fluorescence wavelength 300 to 600 nm
・ Data acquisition interval: 20nm
・ Number of measurements: 3 times for each sample

Using the fluorescent fingerprint data with a known riboflavin content measured under the above conditions, the discrimination filter of the above-described embodiment was designed. Note that, using multiple regression analysis as a method for obtaining candidate estimation formulas, the optimal estimation formula and filter characteristics were determined. FIG. 9 is a diagram showing the coefficients and the positions and sizes of the measurement windows obtained on the fluorescent fingerprint when the number of measurement windows is set to two (estimation formula: Y = aW ₁ + bW ₂ ). Further, FIG. 10 shows the coefficient, the position of the measurement window, and the size and the size obtained on the fluorescent fingerprint when the number of measurement windows is set to 3 (estimation formula: Y = aW ₁ + bW ₂ + cW ₃ ). FIG. Further, FIG. 11 shows the coefficient, the position and the size of the measurement window obtained on the fluorescent fingerprint when the number of measurement windows is set to 4 (estimation formula: Y = aW ₁ + bW ₂ + cW ₃ + dW ₄ ). It is a figure. One measurement window corresponds to a combination of one excitation side filter and one fluorescence side filter. Here, in the estimation formula, Y represents the riboflavin content, W _{1 to 4} represent the fluorescence intensity obtained in the measurement windows 1 to 4, and a to d are coefficients thereof. The coefficients ₁ , ₂ , ₃ , and ₄ are the coefficients of W ₁ , W ₂ , W ₃ , and W ₄ , and indicate a, b, c, and d, respectively.

  As shown in FIGS. 9 to 11, it can be seen that the measurement window is set in a characteristic portion of the fluorescent fingerprint. Here, the measurement window is also set at a position where the fluorescence intensity is low, but the coefficient of such a measurement window is a negative number (such as coefficient 1 in FIG. 9), so the background is removed. It seems to be effective. In FIG. 10, the ranges of the measurement window 1 and the measurement window 2 overlap, and in FIG. 11, the ranges of the measurement window 3 and the measurement window 4 further overlap. Thus, the reason why the measurement windows are nested is that the outer measurement window has a role of correcting the inner measurement window. For example, in FIG. 10, the coefficient 1 of the outer measurement window 1 is a negative number, the coefficient 2 of the inner measurement window 2 is a positive number, and the fluorescence intensity of the inner measurement window is corrected. . In other words, as a characteristic of the discrimination filter designed by simulation, the transmission band of one discrimination filter may be included in the transmission band of the other discrimination filter. By using such a set of discriminating filters, the sample can be discriminated appropriately because the discriminating filter with the wider transmission band functions as a normalization filter to eliminate the influence of disturbance factors, and the transmission band This is because the narrower discrimination filter functions as an identification filter that passes the wavelength from which the identification target is accurately separated.

Subsequently, the estimation results of the estimation equations obtained in each case of FIGS. 9 to 11 were examined. FIG. 12 is a graph in which the vertical axis represents the predicted value based on the estimation formula obtained when the number of measurement windows is set to two, and the horizontal axis represents the actual value measured by the AOAC official method. FIG. 13 is a graph in which the vertical axis represents the predicted value based on the estimation formula obtained when the number of measurement windows is set to three, and the horizontal axis represents the actual value measured by the AOAC official method. FIG. 14 is a graph in which the vertical axis represents the predicted value based on the estimation formula obtained when the number of measurement windows is set to four, and the horizontal axis represents the actual value measured by the AOAC official method. Here, the points in the graph represent one sample of the evaluation object, and 42 samples are shown in total. R ² is the square of the correlation coefficient and represents the accuracy (contribution rate) of the regression analysis. RMSECV is a root mean square error (Root-Mean-Square Error of Cross-Validation) of the cross-validation method. For reference, in the estimation formula obtained as a result of PLS regression analysis using all wavelength conditions of fluorescent fingerprint data, R ² = 0.97 and RMSECV = 0.092.

As shown in FIGS. 12 to 14, an estimation equation with an accuracy equivalent to or higher than the analysis by the conventional method using all wavelength conditions was obtained regardless of the number of measurement windows. In particular, in the case of the measurement window 3 or more, R ² is higher than the estimation formula of the conventional method, indicating that the estimation formula well represents the actual measurement value. Furthermore, in the case of three or more measurement windows, it was shown that there are fewer errors and better accuracy than the estimation formula of the conventional method.

  As described above, the first embodiment shows that when the estimation is performed based on the fluorescence intensity in the wavelength band of the measurement window, the accuracy equal to or higher than that of the conventional analysis method using all the wavelength conditions of the fluorescence fingerprint can be obtained. It was done.

[V. Example of discrimination device]
Next, an example of the determination device 100 according to the present embodiment will be described below with reference to FIGS. Here, FIG. 15 is a flowchart showing an example of the operation of the determination apparatus 100 in the present embodiment.

  As shown in FIG. 15, first, the discrimination filter setting unit 102e sets an excitation wavelength side discrimination filter that transmits light in a predetermined excitation wavelength band in the spectroscope 11b of the measurement unit 110, and light in a predetermined fluorescence wavelength band. Is set in the spectroscope 12a of the measurement unit 110 (step SC-1). Here, the filter to be set has a filter characteristic determined according to the discriminating filter design method of the above-described embodiment. When there are a plurality of measurement windows (when there are a plurality of combinations of excitation wavelength bands and fluorescence wavelength bands), the discrimination filter setting unit 102e performs setting for each set. Here, FIG. 16 is a diagram illustrating an example of the measurement unit 110 in which the filter setting is performed by the discrimination filter setting unit 102e.

  As shown in FIG. 16, in this example, the measurement unit 110-1 has a small light source 11 a-1 as a light source 11 a of the spectral illumination unit 11 and a filter having excitation wavelength variable means for measuring one point of the sample 13-1. A switch 11b-1, a coaxial optical fiber 10-1 that irradiates the sample 13-1 with excitation light and makes the fluorescence from the sample 13-1 enter, a filter switch 12a-1 having a fluorescence wavelength varying means, and fluorescence A detector 12b-1 for detecting the intensity is provided. The filter switchers 11b-1 and 12a-1 are provided with a filter having the above filter characteristics, and are configured to be switchable under the control of the discrimination filter setting unit 102e.

  With this configuration, excitation light in a predetermined excitation wavelength band is sent to the excitation light irradiation unit via the coaxial optical fiber 10-1 in the spectroscope 11b-1, and the measurement object is transmitted from the excitation light irradiation unit. A certain sample 13-1 is irradiated. Therefore, the sample 13-1 emits fluorescence with a component-specific pattern. This eliminates the need for wavelength scanning for measuring fluorescent fingerprints using a monochromator or the like, thereby making it possible to simplify and downsize the measurement unit 110 and to speed up the measurement. Here, FIG. 17 is a diagram illustrating another example (imaging measurement) of the measurement unit 110 in which the filter is set by the discrimination filter setting unit 102e.

  In this example, for imaging measurement of the sample 13-2, the measurement unit 110-2 includes an ultraviolet light source 11a-2, a bandpass filter 11b-2 including excitation wavelength variable means, and a sample 13- irradiated with excitation light. 2 includes an objective lens 10-2 for forming an image of fluorescence from 2, a bandpass filter 12a-2 having a fluorescence wavelength varying means, and a monochrome CCD camera 12b-2 for imaging fluorescence intensity associated with a pixel. .

  With this configuration, excitation light in a predetermined excitation wavelength band is sent to the excitation light irradiation unit in the spectroscope 11b-2, and the sample 13-2 that is the measurement object is irradiated from the excitation light irradiation unit. . Therefore, the sample 13-2 emits fluorescence with a component-specific pattern. The objective lens 10-2 converts the image of the sample 13-2 into a size suitable for photographing, and only the fluorescence in a predetermined fluorescence wavelength band is captured by the monochrome CCD camera 12-2, and each pixel value representing the fluorescence intensity is converted into each pixel value. Converted to a monochrome image based on it. That is, the fluorescence intensity at each measurement location on the sample 13-2 is determined from the brightness of each pixel of the monochrome image. Therefore, the fluorescence intensity at a specific excitation wavelength and fluorescence wavelength can be measured at a time at all points (pixels) on the sample 13-2. When acquiring the fluorescence intensity of the n wavelength condition at each point on the sample 13-2, n fluorescence images are taken while changing the combination of the excitation wavelength band and the fluorescence wavelength band. The bandpass filters 11b-2 and 12a-2 have the predetermined filter characteristics, and each wavelength variable unit can switch the filter according to the control from the discrimination filter setting unit 102e. By using a bandpass filter having a predetermined filter characteristic, the number of optical filters can be reduced and the accuracy can be improved.

  Returning to FIG. 15 again, the fluorescence intensity of the fluorescence from the evaluation object is measured via the measurement unit 110 with the filter set set by the discrimination filter setting unit 102e (step SC-2). For example, image information of n fluorescent images taken by the measurement unit 110 is transferred to the determination device 100.

  And the execution control part 102g determines whether a measurement is complete | finished (step SC-3). For example, when there are a plurality of combinations of excitation wavelength bands and fluorescence wavelength bands, the execution control unit 102g determines whether all the discrimination filter sets have been set.

  If it is determined not to end the measurement (No at Step SC-3), the execution control unit 102g returns the process to Step SC-1, causes the remaining discrimination filter set to be set, and the above-described Steps SC-1 to SC-. Step 3 is repeatedly executed.

  On the other hand, when it is determined that the measurement is to be ended (step SC-3, Yes), the estimation calculation unit 102h calculates an evaluation value by substituting the fluorescence intensity obtained by the fluorescence measurement unit 102f into a predetermined estimation formula. (Step SC-4). The predetermined estimation formula is an optimum estimation formula obtained together with the evaluation target whose evaluation value is known together with the filter characteristics described above by the estimation calculation unit 102h. Further, the estimation calculation unit 102h may store the evaluation result (evaluation result of the evaluation value, etc.) in the storage unit 106, display it on the output device 114 such as a monitor, or output it from a printer or the like. It may be printed out on the device 114. When an evaluation value is calculated for each pixel based on n fluorescent images photographed by the measurement unit 110, an image having a color and brightness corresponding to the evaluation value can be created.

  This is the end of the description of the example of the determination device 100 according to the present embodiment.

[VI. Discriminant filter creation method]
Next, a discrimination filter creation method and an embodiment thereof (embodiment 2) will be described below with reference to FIGS. That is, an example of a method for creating a discrimination filter having filter characteristics (combination of excitation wavelength band and fluorescence wavelength band) determined together with an appropriate estimation formula in the above embodiment will be described.

  A bandpass spectral filter that transmits a specific wavelength band can be realized by a multilayer optical filter having a multilayer film made of a thin film material on a transparent substrate. As the transparent substrate, for example, inorganic glass such as soda lime glass, borosilicate glass, and crown glass, and organic glass such as acrylic, polycarbonate, polysulfone, polyester, and cellulose can be used. .

  A combination of thin film materials for forming the multilayer film is not particularly limited as long as it has a refractive index capable of reducing the transmittance when the multilayer film is formed. For example, inorganic oxides (oxides such as titanium, zinc, silicon, and iron, or composite oxides including a plurality of them), inorganic halides (halides such as magnesium, sodium, and aluminum), single metals (silicon, gold, silver) , Copper, nickel, etc. or a mixed zone containing a plurality of them can be used. Depending on the target filter characteristics, thin films of these materials are laminated.

  As a method for forming the multilayer film, a general-purpose thin film forming technique can be used. For example, the multilayer film can be formed by a vacuum deposition method, an ion assist deposition method, a sputtering method, or the like. Here, FIG. 18 is a diagram showing the filter characteristics of the ideal filter created in the present embodiment. The horizontal axis represents the wavelength, and the vertical axis represents the transmittance.

  The theoretically designed band pass filter has transmission characteristics as shown in FIG. The wavelength band of one filter Filter1 is nested, including the wavelength band of the other filter Filter2. In order to realize these filter characteristics, a multilayer film was formed by using alternating layers of SiO2 and TiO2 on the white plate glass B270 to create a spectral filter. As for the number of layers, the high-pass filter (Filter 1) has 35 single-sided coating layers, and the band-pass filter (Filter 2) has 35 double-sided coating layers and 27 layers. An ion assist method was used for forming the thin film. FIG. 19 is a diagram showing the filter characteristics of the two filters created in this example. The filter characteristic is a transmission characteristic measured with a spectrophotometer. The Theoretical Filter indicates the ideal filter characteristic to be designed, and the Optical Filter indicates the filter characteristic of the actually created filter.

  As shown in FIG. 19, the two filters achieve almost ideal filter characteristics. In addition, although the transmittance | permeability higher than an ideal filter characteristic is seen by the high wavelength side, this can be set so that it may not detect by utilizing the property of a camera. That is, in this example, Filter 1 uses the sensitivity limitation of the camera, does not require a filter as designed, and can be replaced with a high-pass filter.

  That is, since the sensitivity of the near-infrared camera to which the filter is to be mounted is up to 1700 nm, transmitted light of 1700 nm or more is not sensed and does not affect the accuracy. Therefore, Filter 1 is manufactured as a high-pass filter. Here, FIG. 20 shows a color image (left figure) obtained by photographing black asphalt on which an ice / water film is formed and a near infrared camera XSVA XS FPA-1.7-320 (trade name) manufactured by Xenics. It is a figure which shows the ice / water discrimination | determination result image (right figure) which mounted | wore and measured with the produced filter. The upper part is a frozen sample, and the lower part is a wet sample. Pixels that are determined to be in a frozen state are represented in gray, and pixels that are determined to be in a wet state are represented in black. The correct rate of discrimination (correct rate) was 90.28% calculated from the number of discrimination pixels of each asphalt sample.

  As shown in FIG. 20, it was possible to determine whether the asphalt was wet or frozen using the created filter. In this experimental example, the irradiation is white light and filtering on the irradiation side is not required. In other words, this is an example in which the discrimination is not made by the fluorescence fingerprint but by the absorption spectrum using the data of only the fluorescence wavelength axis. However, according to this embodiment, it was confirmed that a filter having an ideal filter characteristic can be manufactured.

[VII. Other Embodiments]
Although the embodiments of the present invention have been described so far, the present invention is not limited to the above-described embodiments, but can be applied to various different embodiments within the scope of the technical idea described in the claims. It may be implemented.

  For example, in the present embodiment, multiple regression analysis is given as an example of the estimation formula acquisition method. However, the present invention is not limited thereto, and any statistical analysis process may be performed. For example, the statistical analysis process may be performed by multivariate analysis or data mining. Further, multivariate analysis or data mining may be performed using at least one technique among data structure analysis, discriminant analysis, pattern classification, multidimensional data analysis, regression analysis, and learning machine. Further, the data structure analysis may be performed using at least one technique among principal component analysis, factor analysis, correspondence analysis, and independent component analysis. Also, the discriminant analysis may be performed using a linear discriminant analysis method or a nonlinear discriminant analysis method. Pattern classification may also be performed by cluster analysis or multidimensional scaling. Further, the learning machine may use at least one technique among a neural network, a support vector machine, a self-organizing map, group learning, and a genetic algorithm.

  In addition, among the processes described in the embodiment, all or part of the processes described as being automatically performed can be performed manually, or the processes described as being performed manually can be performed. All or a part can be automatically performed by a known method.

  In addition, unless otherwise specified, the processing procedures, control procedures, specific names, information including registration data for each processing, parameters such as search conditions, screen examples, and database configurations shown in the above documents and drawings Can be changed arbitrarily.

  Moreover, regarding the determination apparatus 100, each illustrated component is functionally conceptual and does not necessarily need to be physically configured as illustrated.

  For example, the processing functions provided in each device of the simulation device 111 and the discriminating device 100, in particular, the processing functions performed by the calculation unit 112 and the control unit 102, are all or any part of the processing functions are CPU (Central Processing Unit). Alternatively, it may be realized by a program interpreted and executed by the CPU, or may be realized as hardware by wired logic. Note that the program is recorded on a non-transitory computer-readable recording medium including a programmed instruction for causing the computer to execute the method according to the present invention, which will be described later, and the determination device 100 as necessary. Read mechanically. That is, the storage unit 116 such as ROM or HD, the storage unit 106, and the like store computer programs for giving instructions to the CPU and performing various processes in cooperation as an OS (Operating System). The computer program is executed by being loaded into the RAM, and constitutes a calculation unit and a control unit in cooperation with the CPU.

  The computer program may be stored in an application program server connected to the determination apparatus 100 via an arbitrary network, and may be downloaded in whole or in part as necessary. .

  In addition, the program according to the present invention may be stored in a computer-readable recording medium, and may be configured as a program product. Here, the “recording medium” is any memory card, USB memory, SD card, flexible disk, magneto-optical disk, ROM, EPROM, EEPROM, CD-ROM, MO, DVD, Blu-ray Disc, etc. Of “portable physical media”.

  The “program” is a data processing method described in an arbitrary language or description method, and may be in any format such as source code or binary code. The “program” is not necessarily limited to a single configuration, but is distributed in the form of a plurality of modules and libraries, or in cooperation with a separate program represented by an OS (Operating System). Including those that achieve the function. Note that a well-known configuration and procedure can be used for a specific configuration for reading a recording medium, a reading procedure, an installation procedure after reading, and the like in each device described in the embodiment.

  Various databases and the like (fluorescent fingerprint file 116a and candidate estimation formula file 116b etc.) stored in the storage unit 116 and the storage unit 106 are a memory device such as RAM and ROM, a fixed disk device such as a hard disk, a flexible disk, an optical disk, etc. The storage means stores various programs, tables, databases, web page files, etc. used for various processes and website provision.

  The simulation apparatus 111 and the determination apparatus 100 may be configured as an information processing apparatus such as a known personal computer or workstation, or may be configured by connecting any peripheral device to the information processing apparatus. . Further, the determination apparatus 100 may be realized by installing software (including programs, data, and the like) that causes the information processing apparatus to realize the method of the present invention.

  Furthermore, the specific form of distribution / integration of the devices is not limited to that shown in the figure, and all or a part of them may be functional or physical in arbitrary units according to various additions or according to functional loads. Can be distributed and integrated. That is, the above-described embodiments may be arbitrarily combined and may be selectively implemented.

  As described above in detail, according to the present invention, a discriminating filter design method, a discriminating method, and a discriminating method capable of performing highly accurate measurement and analysis in a short time in order to evaluate an evaluation object by fluorescence characteristics. Since a filter set, a discriminating apparatus, and a program can be provided, it is extremely useful in various fields such as optics, food processing, product inspection, field, medicine, pharmaceuticals, drug discovery, basic research, and clinical examinations.

  In particular, since the present invention can accurately evaluate the evaluation object regardless of non-destructive and non-contact measurement, it can be widely used in various fields such as basic research areas and inspections at food manufacturing sites. It is a very versatile technique that can be performed. In addition, the present invention can be applied to 100% inspection and real-time monitoring in factories and the like by reducing the measurement time by narrowing down the measurement conditions. Further, the present invention can be applied to an imaging apparatus and can be simplified, and therefore can be applied to practical application of visualization apparatuses having a wide variety of component distributions.

DESCRIPTION OF SYMBOLS 11 Spectral illumination part 11a Light source 11b Spectroscope 12 Spectroscopic detection part 12a Spectroscope 12b Detector 13 Evaluation object 111 Simulation apparatus 112 Arithmetic part 112a Wavelength band setting part 112b Fluorescence intensity calculation part 112c Estimation formula acquisition part 112d Optimization part 100 Discrimination Device 102 Control unit 102e Discriminant filter setting unit 102f Fluorescence measurement unit 102g Execution control unit 102h Estimation calculation unit 106 Storage unit 116 Storage unit 116a Fluorescence fingerprint file 116b Candidate estimation formula file 108 Input / output control interface unit 110 Measurement unit 113 Input device 114 Output apparatus

Claims (11)

A method for designing a discriminating filter for determining a combination of an excitation wavelength band to be irradiated to the evaluation object and a fluorescence wavelength band to be observed, which is appropriate for evaluating the evaluation object by fluorescence characteristics,
A wavelength band setting step for setting at least one combination of the excitation wavelength band and the fluorescence wavelength band;
Based on fluorescence fingerprint information consisting of each excitation wavelength and fluorescence intensity at each fluorescence wavelength, obtained in advance by fluorescence spectroscopy in which the wavelength is continuously changed in a wide wavelength range for a plurality of evaluation targets whose evaluation values are known, A fluorescence intensity calculation step of calculating, by simulation, the fluorescence intensity obtained in the excitation wavelength band and the fluorescence wavelength band set in the wavelength band setting step;
Using the fluorescence intensity calculated in the fluorescence intensity calculation step and the evaluation value as variables, an estimation formula acquisition step for acquiring an estimation formula for estimating the evaluation value from the fluorescence intensity;
While repeatedly setting the combination of the excitation wavelength band and the fluorescence wavelength band in the wavelength band setting step, the evaluation value of the evaluation target is calculated by repeatedly executing the fluorescence intensity calculation step and the estimation formula acquisition step. Obtaining an appropriate estimation equation for obtaining, and obtaining the most appropriate evaluation value from a plurality of the combinations of the excitation wavelength band and the fluorescence wavelength band, and the combination and the estimation of the excitation wavelength band and the fluorescence wavelength band. An optimization process to determine the formula;
Including
The discriminant filter design method, wherein the evaluation value is a value for quantifying the evaluation object.   2. The discriminant filter design method according to claim 1, wherein at least one of the combination of the excitation light wavelength band and the fluorescence wavelength band has a broad wavelength band wider than 10 nm. In the estimation formula acquisition step,
3. The discriminant filter design method according to claim 1, wherein the estimation formula is acquired by multivariate analysis using the fluorescence intensity as an explanatory variable and the evaluation value as an objective variable. In the fluorescence intensity calculation step,
The fluorescence intensity in the excitation wavelength band and the fluorescence wavelength band is calculated by integrating the fluorescence intensity at each excitation wavelength and each fluorescence wavelength with respect to the fluorescence fingerprint information. Item 4. The discrimination filter design method according to any one of Items 1 to 3. In the above optimization process,
Among the plurality of estimation formulas acquired in the estimation formula acquisition step, there is little error from the known evaluation value, and the estimation accuracy is good even for unknown samples, and both are balanced. Is obtained as the appropriate estimation formula, the discriminant filter design method according to any one of claims 1 to 4. It is a discriminating method for evaluating an evaluation object by fluorescence characteristics,
An excitation wavelength-side discrimination filter that transmits light having an excitation wavelength in the excitation wavelength band, which is determined by the discrimination filter design method according to any one of claims 1 to 5, and light having a fluorescence wavelength in a predetermined fluorescence wavelength band A discrimination filter setting step for setting a fluorescence wavelength side discrimination filter for transmitting
Irradiating the evaluation object with an excitation wavelength in the excitation wavelength band, and measuring a fluorescence intensity in the fluorescence wavelength band from the evaluation object;
An execution control step of executing the discrimination filter setting step and the fluorescence measurement step at least once with at least one combination of the excitation wavelength band and the fluorescence wavelength band;
An estimation operation for calculating the evaluation value by substituting the fluorescence intensity obtained in the fluorescence measurement step into a predetermined estimation formula using the evaluation value for quantifying the fluorescence intensity and the evaluation target as variables. Process,
Including
The above estimation formula is
As an appropriate estimation formula for obtaining the evaluation value of the evaluation object, together with the fluorescence intensity information in the combination of the excitation wavelength band and the discrimination filter of the fluorescence wavelength band of a plurality of evaluation objects whose evaluation values are known Pre-optimized,
A distinction method characterized by. The above estimation formula is
The determination method according to claim 6, wherein the fluorescence intensity is used as an explanatory variable, and the evaluation value is used as an objective variable and is optimized and acquired by multivariate analysis. A measurement unit having a discrimination filter set determined by the discrimination filter design method according to any one of claims 1 to 5 suitable for discriminating an evaluation object based on fluorescence characteristics in a light source and a detector. A discrimination device including a control unit,
The control unit
Discrimination filter setting means for setting an excitation wavelength side discrimination filter and a fluorescence wavelength side discrimination filter in a predetermined combination in the discrimination filter set;
Irradiating the evaluation object with an excitation wavelength in the excitation wavelength band, and measuring fluorescence intensity in the fluorescence wavelength band from the evaluation object;
Execution control means for executing the processing by the discrimination filter setting means and the fluorescence measurement means at least once in a combination of at least one of the excitation wavelength band and the fluorescence wavelength band;
An estimation operation for calculating the evaluation value by substituting the fluorescence intensity obtained by the fluorescence measurement means into a predetermined estimation formula using the evaluation value for quantifying the fluorescence intensity and the evaluation target as variables. Means,
A discrimination apparatus comprising: The discrimination device is
The discriminating apparatus according to claim 8, wherein an optimum estimation formula for estimating the evaluation value is obtained by multivariate analysis using the fluorescence intensity as an explanatory variable and the evaluation value as an objective variable. . A computer having a storage unit and a calculation unit for determining a combination of an excitation wavelength band to be irradiated to the evaluation object and a fluorescence wavelength band to be observed, which is appropriate for evaluating the evaluation object by fluorescence characteristics A program for executing
The storage unit
A storage region for fluorescence fingerprint information of each excitation wavelength and fluorescence intensity at each fluorescence wavelength over a wide wavelength range, obtained by fluorescence spectroscopy in which the wavelength is continuously changed for a plurality of evaluation targets whose evaluation values are known,
With
In the above calculation unit,
A wavelength band setting step for setting at least one combination of the excitation wavelength band and the fluorescence wavelength band;
The excitation wavelength band and the fluorescence set in the wavelength band setting step based on the fluorescence fingerprint information of the fluorescence intensity at each excitation wavelength and each fluorescence wavelength acquired in advance for a plurality of evaluation targets whose evaluation values are known. A fluorescence intensity calculating step for calculating the fluorescence intensity obtained in the wavelength band;
With the fluorescence intensity calculated in the fluorescence intensity calculation step and the evaluation value as variables, an estimation formula acquisition step for acquiring an estimation formula for obtaining the evaluation value;
While repeatedly setting the combination of the excitation wavelength band and the fluorescence wavelength band in the wavelength band setting step, the evaluation value of the evaluation target is calculated by repeatedly executing the fluorescence intensity calculation step and the estimation formula acquisition step. Obtaining the optimum estimation formula for obtaining, and obtaining the most appropriate evaluation value from the plurality of combinations of the excitation wavelength band and the fluorescence wavelength band, and the combination and the estimation of the excitation wavelength band and the fluorescence wavelength band. An optimization process to determine the formula;
And execute
The evaluation value is a value for quantifying the evaluation object. In the estimation formula acquisition step,
11. The program according to claim 10, wherein the estimation formula is acquired by multivariate analysis or the like using the fluorescence intensity as an explanatory variable and the evaluation value as an objective variable.