WO2022195935A1 - 学習用データ作成方法及び学習用データ作成装置 - Google Patents
学習用データ作成方法及び学習用データ作成装置 Download PDFInfo
- Publication number
- WO2022195935A1 WO2022195935A1 PCT/JP2021/036791 JP2021036791W WO2022195935A1 WO 2022195935 A1 WO2022195935 A1 WO 2022195935A1 JP 2021036791 W JP2021036791 W JP 2021036791W WO 2022195935 A1 WO2022195935 A1 WO 2022195935A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- peak
- data
- peaks
- measurement
- extracted
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 163
- 238000012549 training Methods 0.000 title claims abstract description 10
- 238000005259 measurement Methods 0.000 claims abstract description 122
- 239000013074 reference sample Substances 0.000 claims abstract description 82
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 29
- 238000000605 extraction Methods 0.000 claims description 49
- 238000003860 storage Methods 0.000 claims description 25
- 239000000284 extract Substances 0.000 claims description 17
- 230000007547 defect Effects 0.000 claims description 15
- 239000000523 sample Substances 0.000 description 25
- 238000012545 processing Methods 0.000 description 21
- 150000002500 ions Chemical class 0.000 description 14
- 238000011156 evaluation Methods 0.000 description 10
- 235000013305 food Nutrition 0.000 description 7
- 238000010801 machine learning Methods 0.000 description 7
- 239000000126 substance Substances 0.000 description 7
- 230000014759 maintenance of location Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 241000251468 Actinopterygii Species 0.000 description 2
- 239000012472 biological sample Substances 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000004949 mass spectrometry Methods 0.000 description 2
- 235000013372 meat Nutrition 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 238000000862 absorption spectrum Methods 0.000 description 1
- 150000001412 amines Chemical class 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000000035 biogenic effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000000491 multivariate analysis Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
- G01N30/02—Column chromatography
- G01N30/86—Signal analysis
- G01N30/8624—Detection of slopes or peaks; baseline correction
- G01N30/8631—Peaks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
- G01N30/02—Column chromatography
- G01N30/86—Signal analysis
- G01N30/8624—Detection of slopes or peaks; baseline correction
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
- G01N30/02—Column chromatography
- G01N30/62—Detectors specially adapted therefor
- G01N30/72—Mass spectrometers
- G01N30/7206—Mass spectrometers interfaced to gas chromatograph
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
- G01N30/02—Column chromatography
- G01N30/86—Signal analysis
- G01N30/8693—Models, e.g. prediction of retention times, method development and validation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/02—Food
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
- G01N30/02—Column chromatography
- G01N30/62—Detectors specially adapted therefor
- G01N30/72—Mass spectrometers
Definitions
- the present invention relates to a technique for creating learning data from measurement data obtained by measuring a sample using an analyzer such as a chromatograph mass spectrometer.
- Patent Document 1 describes evaluating the freshness of fish meat by quantitatively analyzing non-volatile biogenic amines generated when fish meat spoils.
- the discriminator created in this way is a feature amount common to the learning data with the same label information (for example, measured intensity peaks derived from multiple substances characteristic of reference samples with the same label information (same freshness)) position and intensity) to identify the freshness of the target sample. Therefore, erroneous determinations and oversights are less likely to occur than analysis based on only a single or specific type of indicator substance.
- a learning model is developed by machine learning using a large amount of learning data in which the feature values contained in the measurement data of the reference sample are correctly extracted. need to build. For example, when using measurement data (mass chromatogram) obtained by gas chromatograph mass spectrometry, the feature amount (retention time, mass-to-charge ratio , and measured intensity) are extracted to create learning data.
- the problem to be solved by the present invention is to suppress the loss of feature amounts when creating learning data by extracting peaks from measurement data obtained by analyzing a reference sample whose attributes are known. It is to provide a technique capable of creating learning data.
- the present invention which has been made to solve the above problems, is a method for creating learning data used to create a discriminator for discriminating a plurality of target samples having mutually different attributes, comprising: Acquiring measurement data including a plurality of measured intensity peaks for a predetermined parameter for each of a plurality of reference samples with known attributes, Extracting peak information about the plurality of peaks using a first method prepared in advance for each of the measurement data of the plurality of reference samples, Determining the presence or absence of peak defects by comparing the peak information extracted by the first method with the peak information of a reference sample having the same attribute prepared in advance, Peak information about the plurality of peaks using a second method with a different algorithm and/or parameters for extracting peaks from the first method for the measurement data of the reference sample determined to be missing the peak.
- Another aspect of the present invention which has been made to solve the above problems, is a device for creating learning data used to create a discriminator for discriminating a plurality of target samples having mutually different attributes, , a measurement data acquisition unit that acquires mass chromatogram data by measurement using a chromatograph-mass spectrometer for each of a plurality of reference samples with known attributes; a reference sample information storage unit storing peak information included in a mass chromatogram of a reference sample having the same attribute as the reference sample; a method information storage unit storing information of a first method for extracting peaks from the mass chromatogram data and a second method having different algorithms and/or parameters for extracting peaks from the first method; , a first peak extraction unit that extracts peak information about the plurality of peaks using the first method for each of the plurality of mass chromatogram data of the reference samples; a first determination unit that determines whether or not there is a missing peak by comparing the peak information extracted by the first peak extraction unit with the peak information of the reference sample having the
- the reference sample is a food sample
- the above attributes are freshness and origin
- the reference sample is a biological sample, for example, the presence or absence of a specific disease.
- measurement data including a plurality of measured intensity peaks for a predetermined parameter is acquired for each of a plurality of samples whose attributes are known.
- mass chromatogram data is acquired as measurement data by measurement using a chromatograph mass spectrometer for each of a plurality of reference samples with known attributes. Acquisition of measurement data may be performed by actually measuring a sample with an analyzer, or may be performed by reading out measurement data obtained in a previous measurement.
- the peak of the measured intensity is extracted using the first method prepared in advance. Then, it is determined whether or not all the peaks have been extracted (whether or not there is a lack of peaks) by comparing the extracted peaks with the peak information of the reference sample of the same attribute prepared in advance.
- Information registered in a library for example, can be used as the peak information of the reference sample.
- the measurement data of a plurality of reference samples having the same attribute are integrated to create integrated measurement data, and peak information extracted from the integrated measurement data by the first method can be used as the peak information of the reference sample.
- the first method and the second method include combinations of algorithms for extracting peaks and parameters such as thresholds for peak heights. A conventionally used algorithm may be used as the algorithm for extracting the peak, or a characteristic algorithm that will be described in the examples below may be used.
- the second method For the measurement data of the sample determined to have a missing peak, the second method with a different algorithm or parameter for extracting the peak from the first method is used to extract the peak of the measured intensity. Then, the peak extracted by the second method is compared with peak information of a reference sample having the same attribute, which is prepared in advance, to determine whether or not there is a missing peak. Finally, from the measured data determined to have no missing peaks extracted by the first method or the second method, the feature values corresponding to each of the plurality of peaks of the measured data are acquired to create learning data. do.
- the learning data creation method and apparatus from the measurement data determined to have no missing peaks, the feature amount corresponding to the peak of the measured intensity of the measurement data is acquired to create the learning data. , it is possible to create learning data that suppresses missing features. Even if the first method cannot extract all the peaks from the measured data, the second method can be used to extract all the peaks from the measured data. is created, it is possible to suppress the decrease in the number of learning data.
- FIG. 1 is a configuration diagram of a principal part of a sample evaluation system including an embodiment of a learning data creation device according to the present invention
- 4 is a flow chart according to an embodiment of a learning data creation method according to the present invention
- FIG. 2 is a schematic diagram of mass chromatogram data, which is measurement data acquired in this example.
- the schematic diagram regarding the conventional peak extraction method An example of the results of extracting peaks from a mass chromatogram using a conventional peak extraction method. Another example of the result of extracting peaks from a mass chromatogram by a conventional peak extraction method. Yet another example of the results of extracting peaks from a mass chromatogram using conventional peak extraction methods.
- the sample evaluation system 1 of this embodiment is used to estimate the attributes of a target sample. Specifically, for example, it is used to estimate the freshness and origin of a target sample, which is food, and to determine whether or not a subject has a specific disease based on a biological sample.
- FIG. 1 shows the main configuration of a sample evaluation system 1 including the learning data creation device of this embodiment.
- a sample evaluation system 1 of this embodiment is roughly divided into a gas chromatograph-mass spectrometer (GC-MS) 2 and a control/processing unit 3 .
- GC-MS gas chromatograph-mass spectrometer
- the control/processing unit 3 includes a storage unit 31.
- the storage unit 31 is provided with a method information storage unit 311 and a reference sample information storage unit 312 .
- the method information storage unit 311 stores method information used for extracting peaks from mass chromatogram data obtained by sample measurement using the gas chromatograph mass spectrometer 2 .
- the method information includes information on a combination of parameters such as an algorithm for extracting peaks and a threshold for peak height. Algorithms for extracting peaks may include the conventionally used link point method, horizontal method, new baseline method, and a method using minimal measurement points (described later), which is characteristic of this embodiment.
- the reference sample information storage unit 312 stores mass chromatogram data of a plurality of reference samples with different attributes and peak information (retention times, mass-to-charge ratios, and measured intensities) on a plurality of peaks appearing in the mass chromatograms for various samples. ) is stored.
- the storage unit 312 also stores measurement conditions and the like for measuring various samples with the gas chromatograph mass spectrometer 2 .
- the control/processing unit 3 further includes, as functional blocks, a measurement data acquisition unit 32, a reference sample data generation unit 33, a first peak extraction unit 34, a second peak extraction unit 35, a determination unit 36, a measurement data display unit 36, A learning data creating unit 38 , a learning model creating unit 39 , and a discriminator creating unit 40 are provided.
- the substance of the control/processing unit 3 is a computer called a personal computer or a higher-performance work station, and each functional block described above is executed by executing a sample evaluation system program pre-installed in the computer by the processor of the computer. is embodied.
- An input unit 4 such as a keyboard and a mouse and a display unit 5 such as a liquid crystal display are connected to the control/processing unit 3 .
- a method information storage unit 311, a reference sample information storage unit 312, a measurement data acquisition unit 32, a reference sample data generation unit 33, a first peak extraction unit 34, a second peak extraction unit 35, a determination unit 36, a measurement data display unit 36, and a learning data creation unit 38 constitute the learning data creation device 10 of this embodiment.
- the learning data creation device 10 is incorporated as part of the control/processing unit 3, but the learning data creation device 10 may be configured as an independent device from the control/processing unit 3.
- a discriminator is created for estimating the attributes of the target sample.
- the user sets a reference sample in an autosampler (not shown) connected to the gas chromatograph mass spectrometer 2 prior to creating the discriminator.
- the reference sample is of the same type as the target sample and has known attributes. Also, a plurality of reference samples are set for each attribute.
- the measurement data acquisition unit 32 introduces the reference samples set in the autosampler into the gas chromatograph mass spectrometer 2 in a predetermined order.
- each component is ionized by an ionization source such as an electron ionization source after being introduced into the mass spectrometer, and ion detection is performed after mass separation. detected by the instrument.
- Output signals from the ion detector are sequentially sent to the control/processing unit 3 , and measurement data (mass chromatogram data) of each reference sample are stored in the storage unit 31 . Mass chromatogram data is thus obtained for all reference samples (step 1).
- the mass chromatogram data represents the measured intensity of ions with respect to two parameters, time and mass-to-charge ratio, as schematically shown in FIG. At time (retention time centered at t1), the mass-to-charge ratio of ions generated from the component has a peak.
- the reference sample data creation unit 33 groups the reference samples by attribute. Then, mass chromatogram data of reference samples having the same attribute are integrated to create integrated mass chromatogram data (step 2). The reference sample data creation unit 33 subsequently activates the first peak extraction unit 34 to create an extracted ion current chromatogram (a chromatogram of the measured intensity of ions with a specific mass-to-charge ratio) from the integrated mass chromatogram data. Then, peaks are extracted using the algorithm and parameters stored in the method information storage unit 311 (step 3).
- FIG. 4 As a method for extracting peaks from an extracted ion current chromatogram or a total ion current chromatogram showing the intensity of all ions, conventionally, for example, a link point method, a horizontal method, and the like are used (FIG. 4).
- a link point method the peaks extracted by each method are hatched.
- the peak start point S is the point at which the slope of the chromatogram waveform exceeds a predetermined value
- the peak end point E is the point at which the slope of the waveform thereafter falls below the predetermined value.
- a baseline is defined by connecting the peak start point S and the peak end point E, and the peak is extracted.
- the horizontal method after defining the peak start point S and the peak end point E in the same manner as above, a horizontal line is drawn through the point with the smaller measured intensity of these two points, and the intersection with the perpendicular line from the other point is defined as the baseline and the peak is extracted.
- a new baseline method is used.
- the peak start point S and the peak end point E are defined in the same manner as described above, and the peaks are separated with the minimum point located between the two peaks as the peak separation point M.
- the baseline may increase with time in the latter half of the chromatogram due to a phenomenon called column bleed.
- peaks are automatically extracted from chromatogram data using preset algorithms such as the link point method, horizontal method, and new baseline method. As shown in FIG. 5, the rise of the baseline due to the drift in the latter half of the chromatogram is extracted as a peak.
- this parameter is set to a value of 50, but even with this parameter value, the rise of the baseline due to drift is still extracted as a peak.
- FIG. 7 shows this parameter set to a value of 100. In FIG. 7, the drift is not extracted as a peak, and it can be said that this point is appropriate. However, the slope of the baseline is too large in the first half of the chromatogram, making it difficult to obtain correct peak heights and areas.
- the peaks are extracted by connecting the minimum points of the chromatogram waveform to define the baseline.
- the chromatogram is obtained by connecting many measurement points.
- the algorithm of the present embodiment among the measurement points forming the chromatogram, the one with the smaller measured intensity than any of the two adjacent measurement points is extracted as the minimal measurement point.
- a baseline as shown in the lower part of FIG. 8 is defined.
- the method of defining the baseline from the minimum measurement points is not limited to linear interpolation, and may be an approximation curve or the like connecting these minimum measurement points.
- peaks whose height from the baseline exceeds a predetermined threshold are extracted.
- peaks do not overlap unless both of the two parameters, retention time and mass-to-charge ratio, are common, so a baseline can be defined by linear interpolation, but the parameters for measured intensity are In one chromatogram (such as a total ion current chromatogram), overlapping peaks may appear, in which case a minima may appear between the two peaks.
- linear interpolation of minimum measurement points including minimum points between peaks, fails to define the baseline correctly.
- finding an approximate curve for all minimum measurement points reduces the influence of outliers such as local minima in superimposed peaks and correctly defines the baseline. do it.
- the analyst may eliminate outliers, such as minima between peaks, and linearly interpolate other minima to define the baseline.
- FIG. 5 shows the result of having extracted the peak (circle in the figure shows a peak top) higher than the determined threshold value.
- the method of this example can be used to correctly extract peaks.
- the position information of a plurality of peaks extracted from integrated mass chromatogram data is used as the peak information of the reference sample.
- the reference sample data creation unit 33 stores the peak information of the reference sample in the reference sample information storage unit 312 for each attribute.
- the first peak extraction unit 34 extracts peaks from the mass chromatogram data of each reference sample using the algorithm and parameters stored in the method information storage unit 311 (Step 5).
- the determination unit 36 reads from the reference sample information storage unit 312 the peak information of the reference sample having the same attribute as that of the reference sample. Then, the peak information of the reference sample and the peak information of the standard sample are compared (step 6).
- the determination unit 36 determines that the peak information of the reference sample matches the peak information of the reference sample (that is, there is no missing peak) (NO in step 7)
- the mass chromatogram data of the reference sample is extracted.
- a feature amount corresponding to the peak is acquired to create learning data (step 11).
- the feature quantity extracted here includes a combination of retention time and mass-to-charge ratio corresponding to the peak top of each peak.
- the height or area value of the peak may be included in the feature amount. In other words, as training data, something equivalent to a peak list is created.
- the reference sample Display a screen notifying that a peak is missing for the sample.
- the measurement data display unit 37 also creates an extracted ion current chromatogram corresponding to the mass-to-charge ratio of the missing peak from the mass chromatogram data of the reference sample and displays it on the screen of the display unit 5 . In the mass chromatogram displayed on the screen, the positions where the peaks were not extracted by the first peak extractor 34 although they should be extracted (that is, the positions included in the peak information of the reference sample).
- the mark is superimposed on the retention time and mass-to-charge ratio position of the peak that was not included in the peak information of the reference sample.
- the user confirms the waveform of the extracted ion current chromatogram at the position where the mark is displayed, and determines whether the second peak extraction unit 35 can extract the peak by changing the algorithm or parameters. can do.
- the determination unit 36 also determines whether or not the number of times the peak extraction process has been performed on the reference sample has reached a predetermined number (predetermined number of times). This number is set to, for example, 5 times. At this stage, the first peak extraction unit 34 has only extracted the peak (the number of times of peak extraction processing is 1), so it is determined that the predetermined number of times has not been reached (NO in step 8). be.
- the second peak extraction unit 35 changes the algorithm and/or parameters used in the previous peak extraction (step 9).
- the algorithm is not changed, but the threshold (baseline height) for determining the peak is lowered.
- the threshold (parameter) for peak extraction is changed to 90% of the threshold for peak extraction by the first peak extraction unit 34 (the threshold is lowered by 10%).
- the second peak extraction unit 35 extracts peaks again from the mass chromatogram data of the reference sample using the changed threshold (step 5). Then, it is compared with the peak information of the reference sample again (step 6), and if it is determined that there is no missing peak (NO in step 7), the feature amount corresponding to each of the plurality of extracted peaks is obtained and learned. data is created (step 11). After creating the learning data, it is determined whether or not the mass chromatogram data of all the reference samples have been processed (step 12). Then, if there is unprocessed data, the mass chromatogram data of the next reference sample is processed in the same procedure as above from step 5 onwards.
- step 7 it is determined whether the number of times of peak extraction processing has reached a predetermined number (step 8). again change the algorithm and/or parameters (step 9).
- the threshold is set to 100 during peak extraction processing by the first peak extraction unit 34, and the threshold is lowered by 10. Alternatively, the threshold may be lowered by 10% with reference to the time of the previous peak extraction processing.
- step 8 If the peak loss is not resolved even after performing the peak extraction process a predetermined number of times (YES in step 8), the process for the mass chromatogram data of the reference sample is terminated, and the mass chromatogram of the next reference sample is processed. For the data, the processing after step 5 is performed in the same procedure as above.
- the learning model building unit 39 builds a learning model by machine learning using the learning data created in the above procedure (step 13).
- a machine learning method for example, what is called supervised learning can be used.
- multivariate analysis methods such as logistic regression, orthogonal partial least squares, and k-nearest neighbors. can be done.
- the discriminator creation unit 40 creates a discriminator based on the learning model and stores it in the storage unit 31 (step 14).
- the second peak extraction unit 35 extracts the peak a predetermined number of times. try. Therefore, reduction in the number of learning data can be suppressed.
- the learning data is created after confirming that all peaks have been extracted by comparison with the peak information of the reference sample, it is possible to create learning data with no missing feature values.
- the user may instruct the second peak extractor 35 to extract a peak.
- the mass chromatogram data obtained by measuring the reference sample with the gas chromatograph mass spectrometer 2 was processed.
- a configuration similar to that of the embodiment can be applied.
- the peak here may also include downward peaks appearing in the absorption spectrum or the like.
- One aspect of the present invention is a method for creating learning data used to create a discriminator for identifying a plurality of target samples with different attributes, wherein for each of a plurality of reference samples with known attributes, obtaining measurement data comprising a plurality of measured intensity peaks for a given parameter; Extracting peak information about the plurality of peaks using a first method prepared in advance for each of the measurement data of the plurality of reference samples, Determining the presence or absence of peak defects by comparing the peak information extracted by the first method with the peak information of a reference sample having the same attribute prepared in advance, Peak information about the plurality of peaks using a second method with a different algorithm and/or parameters for extracting peaks from the first method for the measurement data of the reference sample determined to be missing the peak.
- FIG. 8 Another aspect of the present invention is an apparatus for creating a learning model used in machine learning for creating a learning model constituting a classifier that identifies a plurality of target samples having different attributes, a measurement data acquisition unit that acquires mass chromatogram data by measurement using a chromatograph-mass spectrometer for each of a plurality of reference samples with known attributes; a reference sample information storage unit storing peak information included in a mass chromatogram of a reference sample having the same attribute as the reference sample; a method information storage unit storing information of a first method for extracting peaks from the mass chromatogram data and a second method having different algorithms and/or parameters for extracting peaks from the first method; , a first peak extraction unit that extracts peak information about the plurality of peaks using the first method for each of the plurality of mass chromatogram data of the reference samples; a first determination unit that determines whether or not there is a missing peak by comparing the peak information extracted by the first peak extraction unit with the peak information of the reference sample
- the learning data creation method of the first item first, for each of a plurality of samples whose attributes are known, measurement data including a plurality of measured intensity peaks for a predetermined parameter is acquired.
- mass chromatogram data is acquired as measurement data by measurement using a chromatograph mass spectrometer. Acquisition of measurement data may be performed by actually measuring a sample with an analyzer, or may be performed by reading out measurement data obtained in a previous measurement.
- the peak of the measured intensity is extracted using the first method prepared in advance. Then, it is determined whether or not all the peaks have been extracted (whether or not there is a lack of peaks) by comparing the extracted peaks with the peak information of the reference sample of the same attribute prepared in advance.
- Information registered in a library for example, can be used as the peak information of the reference sample.
- the second method For the measurement data of the sample determined to have a missing peak, the second method with a different algorithm or parameter for extracting the peak from the first method is used to extract the peak of the measured intensity. Then, the peak extracted by the second method is compared with peak information of a reference sample having the same attribute, which is prepared in advance, to determine whether or not there is a missing peak. Finally, from the measured data determined to have no missing peaks extracted by the first method or the second method, the feature values corresponding to each of the plurality of peaks of the measured data are acquired to create learning data. do.
- a feature quantity corresponding to the peak of the measured intensity of the measured data is obtained from the measured data determined to have no missing peaks. Since the learning data is created by using the method, it is possible to create the learning data in which the loss of the feature amount is suppressed. Even if the first method cannot extract all the peaks from the measured data, the second method can be used to extract all the peaks from the measured data. , it is possible to suppress the decrease in the number of learning data.
- the user confirms on the screen the measurement data determined to have a missing peak, thereby confirming whether or not there is an extractable intensity peak. can do.
- the user designates only the measurement data in which the peak of the intensity that can be extracted exists, thereby reducing the processing load related to the extraction of peak information by the second method. can be mitigated.
- the peak information of reference samples is obtained even for samples for which measurement data has not been recorded in the database, such as samples for which a sufficient number of measurements have not been performed in the past or unknown samples. can be used.
- the first method and/or the second method selects, among a plurality of measurement points constituting the measurement data, a minimum measurement point for which the measured intensity of the measurement point is lower than the measured intensities of both adjacent measurement points. and extract as determining a baseline at the plurality of measurement points using the minimum measurement points; At each of the plurality of measurement points, a peak is extracted based on whether a value obtained by subtracting the baseline from the measured intensity at the measurement point exceeds a predetermined threshold.
- the magnitude of the baseline also changes over time. Since the baseline is determined using the minimal measurement points in the learning data creation method of Section 6, it is suitable for extracting peaks from measurement data in which not only the measured intensity but also the baseline changes with respect to the parameter. can be used. Also, when extracting a peak from such measurement data, only the threshold used to determine whether it is a peak is changed from the first method, as in the learning data creation method described in Section 7. can be used as the second method.
Landscapes
- Chemical & Material Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Engineering & Computer Science (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
Abstract
Description
属性が既知である複数の参照試料のそれぞれについて、所定のパラメータに対する複数の測定強度のピークを含む測定データを取得し、
前記複数の参照試料の測定データのそれぞれについて、予め用意された第1メソッドを用いて前記複数のピークに関するピーク情報を抽出し、
前記第1メソッドにより抽出されたピーク情報を、予め用意された、同じ属性の基準試料のピーク情報と照合することによりピークの欠損の有無を判定し、
前記ピークが欠損していると判定された参照試料の測定データについて、前記第1メソッドとはピークを抽出するためのアルゴリズム又は/及びパラメータが異なる第2メソッドを用いて前記複数のピークに関するピーク情報を抽出し、
前記第2メソッドにより抽出されたピーク情報を、前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定し、
前記第1メソッド又は前記第2メソッドにより抽出されたピークに欠損がないと判定された測定データから、前記複数のピークのそれぞれに対応する特徴量を取得して学習用データを作成する
ものである。
属性が既知である複数の参照試料のそれぞれについて、クロマトグラフ質量分析装置を用いた測定によりマスクロマトグラムデータを取得する測定データ取得部と、
前記参照試料と同じ属性の基準試料のマスクロマトグラムに含まれるピーク情報が保存された基準試料情報記憶部と、
前記マスクロマトグラムデータからピークを抽出するための第1メソッド、及び該第1メソッドとはピークを抽出するためのアルゴリズム又は/及びパラメータが異なる第2メソッドの情報が保存されたメソッド情報記憶部と、
前記複数の参照試料のマスクロマトグラムデータのそれぞれについて、前記第1メソッドを用いて前記複数のピークに関するピーク情報を抽出する第1ピーク抽出部と、
前記第1ピーク抽出部により抽出されたピーク情報を、同じ属性の前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定する第1判定部と、
前記第1判定部によりピークが欠損していると判定された参照試料のマスクロマトグラムデータについて、前記第2メソッドを用いて前記複数のピークに関するピーク情報を抽出する第2ピーク抽出部と、
前記第2メソッドにより抽出されたピーク情報を、前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定する第2判定部と、
前記第1判定部又は前記第2判定部によりピークに欠損がないと判定されたマスクロマトグラムデータから、前記複数のピークのそれぞれに対応する特徴量を取得して学習用データを作成する学習用データ作成部と
を備える。
上述した複数の例示的な実施形態は、以下の態様の具体例であることが当業者により理解される。
本発明の一態様は、互いに属性が異なる複数の目的試料を識別する識別器を作成するために使用する学習用データを作成する方法であって
属性が既知である複数の参照試料のそれぞれについて、所定のパラメータに対する複数の測定強度のピークを含む測定データを取得し、
前記複数の参照試料の測定データのそれぞれについて、予め用意された第1メソッドを用いて前記複数のピークに関するピーク情報を抽出し、
前記第1メソッドにより抽出されたピーク情報を、予め用意された、同じ属性の基準試料のピーク情報と照合することによりピークの欠損の有無を判定し、
前記ピークが欠損していると判定された参照試料の測定データについて、前記第1メソッドとはピークを抽出するためのアルゴリズム又は/及びパラメータが異なる第2メソッドを用いて前記複数のピークに関するピーク情報を抽出し、
前記第2メソッドにより抽出されたピーク情報を、前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定し、
前記第1メソッド又は前記第2メソッドにより抽出されたピークに欠損がないと判定された測定データから、前記複数のピークのそれぞれに対応する特徴量を取得して学習用データを作成する
ものである。
本発明の別の一態様は、互いに属性が異なる複数の目的試料を識別する識別器を構成する学習モデルを作成するための機械学習において使用する学習モデルを作成する装置であって、
属性が既知である複数の参照試料のそれぞれについて、クロマトグラフ質量分析装置を用いた測定によりマスクロマトグラムデータを取得する測定データ取得部と、
前記参照試料と同じ属性の基準試料のマスクロマトグラムに含まれるピーク情報が保存された基準試料情報記憶部と、
前記マスクロマトグラムデータからピークを抽出するための第1メソッド、及び該第1メソッドとはピークを抽出するためのアルゴリズム又は/及びパラメータが異なる第2メソッドの情報が保存されたメソッド情報記憶部と、
前記複数の参照試料のマスクロマトグラムデータのそれぞれについて、前記第1メソッドを用いて前記複数のピークに関するピーク情報を抽出する第1ピーク抽出部と、
前記第1ピーク抽出部により抽出されたピーク情報を、同じ属性の前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定する第1判定部と、
前記第1判定部によりピークが欠損していると判定された参照試料のマスクロマトグラムデータについて、前記第2メソッドを用いて前記複数のピークに関するピーク情報を抽出する第2ピーク抽出部と、
前記第2メソッドにより抽出されたピーク情報を、前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定する第2判定部と、
前記第1判定部又は前記第2判定部によりピークに欠損がないと判定されたマスクロマトグラムデータから、前記複数のピークのそれぞれに対応する特徴量を取得して学習用データを作成する学習用データ作成部と
を備える。
第1項に記載の学習用データ作成方法において、
前記第1メソッド又は前記第2メソッドにより抽出されたピーク情報にピークの欠損があると判定された測定データを画面表示する。
第2項に記載の学習用データ作成方法において、
前記画面表示された測定データのうち、使用者により指定されたものについて前記第2メソッドによりピーク情報を抽出する。
第1項から第3項のいずれかに記載の学習用データ作成方法において、
属性が同一である複数の目的試料のそれぞれについて前記測定データを取得し、
前記属性が同一である複数の参照試料の測定データを積算して積算測定データを作成し、
前記基準試料のピーク情報として、前記第1メソッドを用いて前記積算測定データに含まれる複数のピークに関するピーク情報を抽出したものを使用する。
第1項から第4項のいずれかに記載の学習用データ作成方法において、
アルゴリズム又は/及びパラメータが異なるメソッドを用いて前記複数のピークに関するピーク情報を抽出し、前記ピークの欠損の有無を判定する処理を、該ピークの欠損がないと判定されるまで所定回数繰り返す。
第1項から第5項に記載の学習用データ作成方法において、
前記第1メソッド及び/又は前記第2メソッドが、前記測定データを構成する複数の測定点のうち、当該測定点の測定強度が両隣の測定点の測定強度のいずれよりも低いものを極小測定点として抽出し、
前記極小測定点を用いて前記複数の測定点におけるベースラインを決定し、
前記複数の測定点のそれぞれにおいて、該測定点の測定強度から前記ベースラインを差し引いた値が予め決められた閾値を超えていることに基づいてピークを抽出する。
第6項に記載の学習用データ作成方法において、
前記第1メソッドと前記第2メソッドがいずれも前記極小測定点を用いてベースラインを決定してピークを抽出するものであり、前記第1メソッドと前記第2メソッドにおいて前記閾値が異なる。
10…学習用データ作成装置
2…ガスクロマトグラフ質量分析装置
3…制御・処理部
31…記憶部
311…メソッド情報記憶部
312…基準試料情報記憶部
312…記憶部
32…測定データ取得部
33…基準試料データ作成部
34…第1ピーク抽出部
35…第2ピーク抽出部
36…判定部
37…測定データ表示部
38…学習用データ作成部
39…学習モデル構築部
39…識別器作成部
4…入力部
5…表示部
Claims (8)
- 互いに属性が異なる複数の目的試料を識別する識別器を作成するために使用する学習用データを作成する方法であって
属性が既知である複数の参照試料のそれぞれについて、所定のパラメータに対する複数の測定強度のピークを含む測定データを取得し、
前記複数の参照試料の測定データのそれぞれについて、予め用意された第1メソッドを用いて前記複数のピークに関するピーク情報を抽出し、
前記第1メソッドにより抽出されたピーク情報を、予め用意された、同じ属性の基準試料のピーク情報と照合することによりピークの欠損の有無を判定し、
前記ピークが欠損していると判定された参照試料の測定データについて、前記第1メソッドとはピークを抽出するためのアルゴリズム又は/及びパラメータが異なる第2メソッドを用いて前記複数のピークに関するピーク情報を抽出し、
前記第2メソッドにより抽出されたピーク情報を、前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定し、
前記第1メソッド又は前記第2メソッドにより抽出されたピークに欠損がないと判定された測定データから、前記複数のピークのそれぞれに対応する特徴量を取得して学習用データを作成する
ものである、学習用データ作成方法。 - 前記第1メソッド又は前記第2メソッドにより抽出されたピーク情報にピークの欠損があると判定された測定データを画面表示する、請求項1に記載の学習用データ作成方法。
- 前記画面表示された測定データのうち、使用者により指定されたものについて前記第2メソッドによりピーク情報を抽出する、請求項2に記載の学習用データ作成方法。
- 属性が同一である複数の目的試料のそれぞれについて前記測定データを取得し、
前記属性が同一である複数の参照試料の測定データを積算して積算測定データを作成し、
前記基準試料のピーク情報として、前記第1メソッドを用いて前記積算測定データに含まれる複数のピークに関するピーク情報を抽出したものを使用する
ものである、請求項1に記載の学習用データ作成方法。 - アルゴリズム又は/及びパラメータが異なるメソッドを用いて前記複数のピークに関するピーク情報を抽出し、前記ピークの欠損の有無を判定する処理を、該ピークの欠損がないと判定されるまで所定回数繰り返す、請求項1に記載の学習用データ作成方法。
- 前記第1メソッド及び/又は前記第2メソッドが、前記測定データを構成する複数の測定点のうち、当該測定点の測定強度が両隣の測定点の測定強度のいずれよりも低いものを極小測定点として抽出し、
前記極小測定点を用いて前記複数の測定点におけるベースラインを決定し、
前記複数の測定点のそれぞれにおいて、該測定点の測定強度から前記ベースラインを差し引いた値が予め決められた閾値を超えていることに基づいてピークを抽出する
ものである、請求項1に記載の学習用データ作成方法。 - 前記第1メソッドと前記第2メソッドがいずれも前記極小測定点を用いてベースラインを決定してピークを抽出するものであり、前記第1メソッドと前記第2メソッドにおいて前記閾値が異なる、請求項6に記載の学習用データ作成方法。
- 互いに属性が異なる複数の目的試料を識別する識別器を作成するために使用する学習用データを作成する装置であって、
属性が既知である複数の参照試料のそれぞれについて、クロマトグラフ質量分析装置を用いた測定によりマスクロマトグラムデータを取得する測定データ取得部と、
前記参照試料と同じ属性の基準試料のマスクロマトグラムに含まれるピーク情報が保存された基準試料情報記憶部と、
前記マスクロマトグラムデータからピークを抽出するための第1メソッド、及び該第1メソッドとはピークを抽出するためのアルゴリズム又は/及びパラメータが異なる第2メソッドの情報が保存されたメソッド情報記憶部と、
前記複数の参照試料のマスクロマトグラムデータのそれぞれについて、前記第1メソッドを用いて前記複数のピークに関するピーク情報を抽出する第1ピーク抽出部と、
前記第1ピーク抽出部により抽出されたピーク情報を、同じ属性の前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定する第1判定部と、
前記第1判定部によりピークが欠損していると判定された参照試料のマスクロマトグラムデータについて、前記第2メソッドを用いて前記複数のピークに関するピーク情報を抽出する第2ピーク抽出部と、
前記第2メソッドにより抽出されたピーク情報を、前記基準試料のピーク情報と照合することによりピークの欠損の有無を判定する第2判定部と、
前記第1判定部又は前記第2判定部によりピークに欠損がないと判定されたマスクロマトグラムデータから、前記複数のピークのそれぞれに対応する特徴量を取得して学習用データを作成する学習用データ作成部と
を備える、学習用データ作成装置。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2023506724A JPWO2022195935A1 (ja) | 2021-03-19 | 2021-10-05 | |
CN202180094903.9A CN117083523A (zh) | 2021-03-19 | 2021-10-05 | 学习用数据制作方法以及学习用数据制作装置 |
EP21931678.3A EP4310494A1 (en) | 2021-03-19 | 2021-10-05 | Training data creation method and training data creation device |
US18/282,755 US20240167988A1 (en) | 2021-03-19 | 2021-10-05 | Learning Data Creation Method and Learning Data Creation Device |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US202163163205P | 2021-03-19 | 2021-03-19 | |
US63/163,205 | 2021-03-19 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022195935A1 true WO2022195935A1 (ja) | 2022-09-22 |
Family
ID=83319963
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2021/036791 WO2022195935A1 (ja) | 2021-03-19 | 2021-10-05 | 学習用データ作成方法及び学習用データ作成装置 |
Country Status (5)
Country | Link |
---|---|
US (1) | US20240167988A1 (ja) |
EP (1) | EP4310494A1 (ja) |
JP (1) | JPWO2022195935A1 (ja) |
CN (1) | CN117083523A (ja) |
WO (1) | WO2022195935A1 (ja) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009204397A (ja) * | 2008-02-27 | 2009-09-10 | Shimadzu Corp | クロマトグラフ用データ処理装置 |
JP2017122677A (ja) | 2016-01-08 | 2017-07-13 | 東洋製罐グループホールディングス株式会社 | イオン感応膜、イオン選択性電極、及びイオンセンサ |
JP2019056664A (ja) * | 2017-09-22 | 2019-04-11 | 東海電子株式会社 | におい判定システム、および、におい判定プログラム |
JP2019086475A (ja) * | 2017-11-09 | 2019-06-06 | 富士通株式会社 | 学習プログラム、検出プログラム、学習方法、検出方法、学習装置および検出装置 |
WO2020070786A1 (ja) * | 2018-10-02 | 2020-04-09 | 株式会社島津製作所 | 識別器の生成方法 |
JP2020165847A (ja) | 2019-03-29 | 2020-10-08 | 株式会社島津製作所 | 食品の品質判定方法、及び、食品品質判定装置 |
-
2021
- 2021-10-05 US US18/282,755 patent/US20240167988A1/en active Pending
- 2021-10-05 EP EP21931678.3A patent/EP4310494A1/en active Pending
- 2021-10-05 JP JP2023506724A patent/JPWO2022195935A1/ja active Pending
- 2021-10-05 CN CN202180094903.9A patent/CN117083523A/zh active Pending
- 2021-10-05 WO PCT/JP2021/036791 patent/WO2022195935A1/ja active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009204397A (ja) * | 2008-02-27 | 2009-09-10 | Shimadzu Corp | クロマトグラフ用データ処理装置 |
JP2017122677A (ja) | 2016-01-08 | 2017-07-13 | 東洋製罐グループホールディングス株式会社 | イオン感応膜、イオン選択性電極、及びイオンセンサ |
JP2019056664A (ja) * | 2017-09-22 | 2019-04-11 | 東海電子株式会社 | におい判定システム、および、におい判定プログラム |
JP2019086475A (ja) * | 2017-11-09 | 2019-06-06 | 富士通株式会社 | 学習プログラム、検出プログラム、学習方法、検出方法、学習装置および検出装置 |
WO2020070786A1 (ja) * | 2018-10-02 | 2020-04-09 | 株式会社島津製作所 | 識別器の生成方法 |
JP2020165847A (ja) | 2019-03-29 | 2020-10-08 | 株式会社島津製作所 | 食品の品質判定方法、及び、食品品質判定装置 |
Non-Patent Citations (1)
Title |
---|
ANONYMOUS: "Analysis sequence", 20 March 2015 (2015-03-20), XP055967952, Retrieved from the Internet <URL:www.lcdez.com/_userdata/Support/ChromatoStage/Help/%E6%B3%A2%E5%BD%A2%E5%87%A6%E7%90%86%E3%81%AE%E5%8E%9F%E7%90%86%EF%BC%8D%E8%A7%A3%E6%9E%90%E3%82%B7%E3%83%BC%E3%82%B1%E3%83%B3%E3%82%B9.htm> [retrieved on 20221004] * |
Also Published As
Publication number | Publication date |
---|---|
EP4310494A1 (en) | 2024-01-24 |
JPWO2022195935A1 (ja) | 2022-09-22 |
CN117083523A (zh) | 2023-11-17 |
US20240167988A1 (en) | 2024-05-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP2020165847A (ja) | 食品の品質判定方法、及び、食品品質判定装置 | |
JP4594154B2 (ja) | 少なくとも1つの成分および生成する生成物の観点でサンプルを特性付けし、特性付けデータを提供するための2つ以上の技術に基づいた少なくとも1つのサンプルの分析;方法、システムおよび指示プログラム | |
JP5365579B2 (ja) | クロマトグラフ質量分析用データ処理装置 | |
Burton et al. | Instrumental and experimental effects in LC–MS-based metabolomics | |
CN110214271B (zh) | 分析数据解析方法以及分析数据解析装置 | |
US11442047B2 (en) | Chromatograph mass spectrometry data processing device and chromatograph mass spectrometry data processing program | |
CN103134850A (zh) | 一种基于特征香气的茶叶品质快速检测装置及检测方法 | |
US20070094196A1 (en) | Manufacture data analysis method and manufacture data analyzer apparatus | |
US9582909B2 (en) | Chromatograph mass spectrometry data processing device | |
JP6308107B2 (ja) | クロマトグラフ質量分析データ処理装置 | |
JPWO2020044435A1 (ja) | データ解析方法、データ解析装置、及びデータ解析用の学習モデル作成方法 | |
JP4470505B2 (ja) | クロマトグラフ質量分析用データ処理装置 | |
US10444206B2 (en) | Chromatography/mass spectrometry data processing device | |
WO2022195935A1 (ja) | 学習用データ作成方法及び学習用データ作成装置 | |
US11474086B2 (en) | Measurement data processing method, measurement data processing device, and program for processing measurement data | |
JP2022052158A (ja) | 学習用データの生成装置、モデル学習装置、試料の特性推定装置、及びクロマトグラフ質量分析装置 | |
JP3663140B2 (ja) | 質量分析方法および質量分析装置 | |
JP2016133339A (ja) | 質量分析データ処理装置および質量分析データ処理方法 | |
US20220373522A1 (en) | Waveform Analytical Method and Waveform Analytical Device | |
CN110069039B (zh) | 测量结果分析装置、测量结果分析方法及计算机可读介质 | |
CN114199989A (zh) | 一种基于质谱数据融合的广陈皮识别方法及*** | |
US20230288381A1 (en) | Sample analyzing device | |
US11946916B2 (en) | Waveform processing assistance device and waveform processing assistance method | |
US20230280316A1 (en) | Learning data producing method, waveform analysis device, waveform analysis method, and recording medium | |
CN116868273A (zh) | 质谱数据的分析方法 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21931678 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2023506724 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 202180094903.9 Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 18282755 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2021931678 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2021931678 Country of ref document: EP Effective date: 20231019 |