JPWO2019150573A1 - Data analyzer - Google Patents

Abstract

The first multivariate analysis processing unit (24) uses the mass spectrum data obtained for each of the plurality of measurement points in the measurement region as an explanatory variable for each sample, and a fluorescent image taken by fluorescently labeling a specific compound. In, the first-stage PLS regression analysis is performed with the pixel value of the pixel corresponding to each measurement point as the explained variable, and the regression coefficient matrix is obtained. The second multivariate analysis processing unit (25) aggregates a plurality of one-dimensional regression coefficient matrices obtained for a plurality of samples into explanatory variables, and uses the severity of cancer of each sample as information indicating the state of each sample. Perform a second-stage PLS regression analysis with the information shown as the dependent variable, and find the regression coefficient matrix again. This makes it possible to obtain new information that is difficult to obtain by conventional methods, such as a compound showing a distribution similar to a specific compound on a fluorescence image and a compound showing a change in the amount according to the severity of cancer. ..

Description

The present invention relates to a data analyzer that analyzes data obtained by measuring or observing a sample or a subject, and in particular, based on measurement data of a large number of samples obtained under various conditions. To extract useful information about these samples, or to extract useful information about the same sample based on multiple multidimensional measurement data obtained by different measurement methods and observation methods. With respect to a suitable data analyzer.

In recent years, multivariate analysis, which is one of the statistical analysis methods, has been actively used for analysis of data obtained by various analysis devices such as mass spectrometers, chromatograph devices, and optical analyzers. There is. Multivariate analysis is very useful, for example, in finding differences between samples based on the vast amount of data obtained by analyzing different samples, and in finding factors related to such differences. It is a method.
An example of using multivariate analysis to analyze the data obtained by the analyzer will be briefly described.

In recent years, a mass spectrometer capable of performing mass spectrometry for each of a large number of measurement points in a two-dimensional measurement region on a sample such as a biological tissue section has been known. This type of mass spectrometer is called an imaging mass spectrometer, and a specific mass-to-charge ratio (m / z) on the same sample surface can be measured while observing the fine morphology of the sample surface with an attached optical microscope. It is possible to acquire a mass spectrometric imaging image showing a two-dimensional intensity distribution of the ions.

In the imaging mass spectrometer described in Patent Document 1, when an analyst confirms a microscopic image obtained by an optical microscope and specifies a color that characterizes a region of interest (region of interest), the portion showing that color in the microscopic image A mass-to-charge ratio showing an ionic strength distribution similar to the distribution is extracted by correlation analysis or regression analysis. Thereby, if the compound can be specified based on the mass-to-charge ratio, the compound showing a distribution similar to the site of interest of the analyst on the sample can be specified.

Of course, not only mass spectrometric imaging images showing distributions similar to those observed in microscopic images, but also multiple mass-to-charge ratios with similar ionic strength distributions can be found relatively easily. This makes it possible to find a plurality of compounds showing the same dynamics, or to find a plurality of partial structures derived from the same compound and estimate the structure of the compound.

International Publication No. 2017/002226 Pamphlet International Publication No. 2016/092608 Pamphlet

The conventional analysis as described above is an analysis of only one sample, but if it is possible to appropriately link the analysis results of a large number of samples, the cause of a specific disease such as cancer can be investigated and more accurately. There is a possibility that very useful information can be obtained for research on various diagnostics. For example, from the analysis results of a large number of samples with various degrees of cancer progression, a compound that is specifically expressed not after the cancer has progressed but at the initial stage thereof can be found, or a compound that reflects the degree of therapeutic effect. You may be able to find it. However, there has been no conventional method that can easily realize data analysis for this purpose.

In addition, in general, it is difficult to grasp the analysis results obtained by a conventional data analysis device that performs multivariate analysis unless the person has specialized knowledge, and there is a problem that sufficiently useful information cannot be provided to the user. was there.

The present invention has been made to solve these problems, and one of its main purposes is to perform an analysis process for comparing and associating a plurality of data obtained by different types of measurements and observations. This is to provide a data analysis device capable of acquiring useful information that cannot be obtained by a conventional analysis method.

The data analysis apparatus of the first aspect according to the present invention, which has been made to solve the above problems, aggregates the data obtained for each predetermined parameter value by performing predetermined measurements on each of a plurality of samples. The plurality of data groups are analyzed by analysis processing based on the first data group and the second data group in which the data obtained by executing the measurement by a method different from the predetermined measurement for each of the plurality of samples. It is a data analysis device that seeks information on all samples in Japan.
a) Perform a predetermined statistical analysis process with the first data group as the explanatory variable and the second data group as the explained variable (objective variable), and show the similarity or difference of the data in both data groups. A first statistical analysis processing unit that calculates multiple index values,
b) Predetermined statistical analysis processing using a plurality of index values calculated by the first statistical analysis processing unit as explanatory variables and other information related to the plurality of samples as explained variables (objective variables). And a second statistical analysis processing unit that calculates a plurality of index values indicating the similarity or difference between the index value and the other information.
c) The relationship between the other information and the first data group, or the other information and the second data group, using a plurality of index values calculated by the second statistical analysis processing unit. A display processing unit that creates and displays a graph showing the relationship with
It is characterized by having.

As a typical embodiment in the data analysis device of the first aspect according to the present invention,
The first data group is a mass having a mass-to-charge ratio as a predetermined parameter at each of a plurality of measurement points in a two-dimensional measurement region obtained by performing imaging mass spectrometry on a plurality of samples. Spectral data
The second data group is image data showing a predetermined signal intensity distribution in a two-dimensional measurement region obtained by a measurement method different from imaging mass spectrometry for a plurality of samples.
The first statistical analysis processing unit resembles one or more images of the second data group and a distribution image of the signal intensity at each mass-to-charge ratio of the first data group for each sample. It can be configured to calculate an index value indicating gender or difference.

The above-mentioned "measurement method different from imaging mass analysis" includes Raman spectroscopic measurement, fluorescence measurement, and measurement of electromagnetic emission intensity of various wavelengths (terahertz region, near-infrared region, visible region, ultraviolet region, X-ray region, etc.). And absorption measurement, PET (Positron Emission Tomography) measurement, MRI (Magnetic Resonance Imaging) measurement, ESR (Electron Spin Resonance) measurement, CT (Computed Tomography) measurement, surface analysis by EPMA (Electron Probe MicroAnalyser), etc. It can be either analyzed or observed with an electronic microscope or an optical microscope. For example, when the second data group includes the image data obtained by the optical measurement method as described above, the image data may be image data constituting different images at each of a plurality of wavelengths, or may be one specific wavelength. It may be the image data constituting one image in.

Now, for example, when the second data group is image data constituting an optical microscopic image obtained by monochromeizing one color image, that is, a data group showing pixel values of pixels corresponding to measurement points, the first statistical analysis For each sample, the processing unit calculates an index value indicating the similarity or difference between the pixel value data and the image in the sample and the distribution image showing the distribution of the signal intensity of each mass charge ratio. The predetermined statistical analysis processing by the first statistical analysis processing unit may be, for example, partial least squares regression (PLS regression) analysis.

By applying the PLS regression analysis to the above data, the regression coefficient can be obtained for each mass-to-charge ratio as the index value. If the regression coefficient for each mass-to-charge ratio obtained for one sample is used as a coefficient matrix, a plurality of coefficient matrices can be obtained for a plurality of samples.

In the data analysis apparatus of the first aspect according to the present invention, the second statistical analysis processing unit further uses the plurality of coefficient matrices as explanatory variables and other information related to the plurality of samples as explained variables. Perform statistical analysis processing of. Here, "other information related to each of a plurality of samples" is not particularly limited as long as it can be quantified, and appropriate information can be used according to the purpose of analysis and the like. Further, the predetermined statistical analysis processing by the second statistical analysis processing unit may be the same method as the statistical analysis processing by the first statistical analysis processing unit, or may be a different method. Specifically, for example, partial least squares regression (PLS regression) analysis or principal component regression analysis may be used.

For example, when a plurality of samples are biological tissue sections collected from a subject suffering from a specific disease and the degree of progression or severity of the disease is given for each sample as other information, the second By performing PLS regression analysis in the statistical analysis processing unit, it is possible to obtain a regression coefficient indicating the similarity between the regression coefficient for each sample obtained by the first statistical analysis processing unit and the degree of disease progression. In this case, this regression coefficient indicates the mass-to-charge ratio showing a distribution similar to the specific distribution appearing in the optical microscopic image, that is, the relationship between the compound and the degree of disease progression. For example, the display processing unit creates a graph of an appropriate format showing the relevance and displays it on the screen of the display unit. Thereby, for example, the user can easily and visually confirm whether or not a specific compound having a distribution similar to the specific distribution appearing in the optical microscopic image is related to the degree of disease progression. ..

In the data analysis apparatus of the first aspect according to the present invention, the same plurality as the first data group in which the data obtained for each predetermined parameter value by performing predetermined measurements on each of a plurality of samples are aggregated. The second data group, which aggregates the data obtained by performing measurements by different methods for each of the samples, and the first-stage statistical analysis process are performed on the same samples, respectively. A second stage of statistical analysis was performed on the third data, which is other relevant information. On the other hand, the second-stage statistical analysis process may be performed in order to express the result obtained by the first-stage statistical analysis process in an easy-to-understand or broader manner.

That is, the data analysis apparatus of the second aspect according to the present invention includes a first data group that aggregates data obtained for each predetermined parameter value by performing predetermined measurements on a plurality of samples, respectively. A second data group in which data obtained by performing measurements by a method different from the predetermined measurement for each of the plurality of samples or data which is other information related to the plurality of samples is aggregated. It is a data analysis device that obtains information on the plurality of samples in general by analysis processing based on the above.
a) Perform a predetermined statistical analysis process with the first data group as the explanatory variable and the second data group as the explained variable (objective variable), and show the similarity or difference of the data in both data groups. A first statistical analysis processing unit that calculates multiple index values,
b) A second statistical analysis processing unit that performs a predetermined statistical analysis processing for aggregating a plurality of index values calculated by the first statistical analysis processing unit into lower-dimensional information.
c) A display processing unit that creates and displays a graph showing the relationship between the second data group and each parameter value in the first data group based on the result of the second statistical analysis processing unit.
It is characterized by having.

In the data analysis device of the second aspect according to the present invention, similarly to the data analysis device of the first aspect, the predetermined statistical analysis process by the first statistical analysis processing unit is, for example, partial minimum square regression analysis. You should do it. On the other hand, as the predetermined statistical analysis processing by the second statistical analysis processing unit, principal component analysis, factor analysis, cluster analysis and the like can be used.

In the data analysis apparatus of the second aspect according to the present invention, for example, the sample is a sample derived from a living body, the first data group is mass spectrum data obtained for each of a plurality of samples, and the second data group is a sample. It shall be the information indicating each state. When PLS regression analysis is performed on such data by the first statistical analysis processing unit, each regression coefficient is calculated as an index value indicating the degree of relevance between the mass-to-charge ratio value and the state of each sample. can do. Then, when the principal component analysis is performed on the plurality of regression coefficients obtained in this way and the display processing unit creates a score plot or a loading plot based on the results, the relationship between the state of each sample and the mass-to-charge ratio value is overlooked. You will be able to observe it.

According to the data analysis apparatus according to the present invention, for example, by comparing or associating a plurality of data obtained by different types of measurement or observation, useful information that cannot be obtained by a conventional analysis method can be obtained. can do. Specifically, for example, finding a compound that is specifically expressed at an early stage of a specific disease such as cancer, not after it has progressed, or finding a compound that reflects the degree of therapeutic effect on such a disease. May be possible.

The schematic block diagram of an Example of the imaging mass spectrometric system using the data analysis apparatus which concerns on this invention. The conceptual diagram of the data analysis processing in the imaging mass spectrometry system of this Example. The flowchart which shows the procedure of data analysis in the imaging mass spectrometry system of this Example. The explanatory view of the modification of the data analysis in the imaging mass spectrometry system of this Example. The schematic block diagram of an Example of the mass spectrometry system using the data analysis apparatus which concerns on this invention. The conceptual diagram of the data analysis processing in the mass spectrometry system of this Example.

[First Example]
An embodiment of an imaging mass spectrometric system using the data analysis apparatus according to the present invention will be described with reference to the accompanying drawings.
FIG. 1 is a schematic configuration diagram of the imaging mass spectrometry system of this embodiment.
This system includes a measuring unit 1 including an optical microscopic observation unit 11 and an imaging mass spectrometric unit 12, a data analysis unit 2, an input unit 3, and a display unit 4.

The optical microscopic observation unit 11 acquires an optical microscopic observation image of the measurement region on the sample. On the other hand, the imaging mass spectrometric unit 12 performs mass spectrometry over a predetermined mass-to-charge ratio range for each minute region (measurement point) in which the measurement region on the same sample is finely divided, and collects mass spectrum data for each. Is.

The data analysis unit 2 includes a data input unit 21 that receives optical microscopic image data acquired by the optical microscopic observation unit 11 and mass spectrum data acquired by the imaging mass analysis unit 12, and a data storage unit that stores the received data. 22, a calculation data preprocessing unit 23, a first multivariate analysis processing unit 24, a second multivariate analysis processing unit 25, and an analysis result display processing unit 26 are provided as functional blocks.
In general, the substance of the data analysis unit 2 is a personal computer or a workstation which is a higher-performance computer, and the functions of each unit are realized by executing the dedicated analysis software installed on the computer on the computer. Can be.

FIG. 2 is a conceptual diagram of a characteristic data analysis process in the imaging mass spectrometry system of this embodiment, and FIG. 3 is a flowchart showing a procedure of the data analysis process.
Here, a section cut out from the same biological tissue (for example, liver) of a large number of subjects (for example, mouse) is used as a sample. This large number of samples are sections of approximately the same site of the same biological tissue of different subjects. Although the subject has developed cancer, the degree of progression (severity) of the cancer varies, and the severity is quantified in advance and is known for each sample.

The measurement unit 1 acquires mass spectrum data (this is referred to as MS imaging data) of a large number of measurement points in the measurement region by the imaging mass spectrometric analysis unit 12 for each sample. In addition, a sample that can be regarded as the same as each of the above samples (a sample that is the same subject and a continuous section in the same biological tissue) is fluorescently labeled with a specific protein and then photographed by the optical microscopic observation unit 11. To acquire fluorescent image data. The data input unit 21 summarizes the MS imaging data and the fluorescent image data of the sample that can be regarded as the same obtained in this way, and further adds information indicating the severity of the cancer of the sample, specifically one. It is stored in the data storage unit 22 as a data file. Therefore, the data storage unit 22 stores the same number of data files as the number of samples (two samples that can be regarded as the same are one).

As described above, the analysis process is performed as follows with the MS imaging data and the fluorescence image data of the plurality of samples stored in the data storage unit 22.
When the user instructs the data to be processed from the input unit 3, the arithmetic data preprocessing unit 23 in the data analysis unit 2 inputs the MS imaging data and the fluorescent image data of one sample to be processed from the data storage unit 22. Read (step S1). When those data are stored in one data file as described above, the data file may be read.

The calculation data preprocessing unit 23 converts the data of each pixel of the fluorescence image into simple pixel value data according to preset conditions. For example, in a fluorescent image using hematoquilicin / eosin (HE) or the like, the intensity value of the fluorescent color is converted into data as a pixel value. Further, the arithmetic data preprocessing unit 23 adjusts the resolution to match the resolution with the alignment process for correcting the size, orientation, distortion, etc. of the image of the same object on the fluorescence microscopic image and the MS imaging image, if necessary. The process is executed (step S2).

In the alignment process, specifically, for example, the MS imaging image is enlarged / reduced, rotated, and further deformed according to a predetermined algorithm based on the microscopic image to obtain the positional relationship between the two images on the sample. Approximately match. For this alignment, for example, the method described in Patent Document 1 can be used.

Further, the resolution (positional resolution) of imaging in the optical microscopic observation unit 11 is usually determined by the resolution of the imaging camera, whereas the resolution of the MS imaging image is determined by the spot diameter of the laser beam irradiated to the sample for ionization. It is decided. Therefore, the resolution of the MS imaging image is often lower than the resolution of the optical image. Therefore, in order to perform multivariate analysis of images, which will be described later, a resolution adjustment process for aligning the resolutions of both images is performed. A simple method of aligning the resolutions is to reduce the resolution of the higher resolution image to match the lower resolution image. For example, a binning process is useful as such a method. Further, the resolution of the image having the lower resolution may be increased to match the image having a higher resolution. For that purpose, after upsampling a low-resolution image to apparently match the number of pixels, upsampling is performed by interpolation processing using a plurality of pixel values adjacent to or close to a certain pixel. The pixel value of the newly inserted pixel may be calculated and filled.

By the alignment processing and the resolution adjustment processing described above, it is possible to associate pixels at the same position two-dimensionally between the image of the brightness value distribution obtained from the fluorescence image and the MS imaging image. After that, the first multivariate analysis processing unit 24 created a two-dimensional matrix created based on the MS imaging data for one sample with the explanatory variable (X) and the pixel value data of the fluorescent image. PLS regression analysis is performed with the dimension matrix as the dependent variable (Y). The two-dimensional matrix is a matrix in which the mass-to-charge ratio value is taken in one direction and the position of the measurement point (pixel) is taken in the other direction orthogonal to the one direction, and the signal strength value is an element. The PLS regression analysis creates a regression model represented by Y = Bpis · X + B ₀ . Here, Bpis is a regression coefficient. The regression coefficient is obtained for each mass-to-charge ratio, and the absolute value of the regression coefficient Bpls is the absolute value of the regression coefficient Bpls in the pattern of the fluorescent image, that is, the mass-to-charge ratio showing a mass-to-charge ratio similar to the spatial distribution of proteins or a distribution inverted from the distribution. growing. This regression coefficient becomes a one-dimensional matrix (step S3).

The data analysis unit 2 determines whether or not data processing has been performed on all the samples to be processed (step S4), and if there are unprocessed samples, the process returns from steps S4 to S1. Therefore, by repeating the processing of steps S1 to S4, a regression coefficient matrix based on the data for all the samples designated as the processing target can be obtained. The number of regression coefficient matrices is the same as the total number of samples. The value of one regression coefficient in this regression coefficient matrix is the distribution of ionic strength at the corresponding mass-to-charge ratio (that is, the distribution of compounds corresponding to that mass-to-charge ratio) and the distribution of fluorescently labeled compounds in the fluorescent image. Shows the similarity of. Therefore, when it is desired to investigate a compound having a distribution similar to that of a fluorescently labeled compound in one sample, the treatment up to this point is sufficient.

In the imaging mass analysis system of this embodiment, as shown in FIG. 2, the second multivariate analysis processing unit 25 further sets a regression coefficient matrix, which is the same number as the number of samples, as an explanatory variable (X), and accompanies each sample. PLS regression analysis is performed with the information indicating the severity of the cancer given as the specific information as the dependent variable (Y). The severity of cancer is an index value indicating the degree of severity of cancer at a certain point in time. Since the severity information is given for each sample, this severity matrix is a one-dimensional matrix having the same number of elements as the number of samples. In this PLS regression analysis, as in the process of step S3, a regression coefficient matrix having the same number of elements as the number of samples is obtained (step S5). The regression coefficients in this regression coefficient matrix indicate the degree of severity and the degree of association between compounds similar to the distribution of fluorescence images. That is, it is presumed that the compound corresponding to the mass-to-charge ratio in which the absolute value of the regression coefficient at this time becomes large is strongly related to the severity of cancer.

The analysis result display processing unit 26 creates a predetermined appropriate graph based on the regression coefficient matrix obtained in step S5, and displays this on the screen of the display unit 4 (steps S6 and S7). For example, as the simplest graph, it is preferable to create a mass spectrum-like graph in which the horizontal axis is the mass-to-charge ratio and the vertical axis is the coefficient value (however, there are positive and negative polarities) based on the regression coefficient matrix obtained in step S5. .. As a result, among the compounds showing a two-dimensional distribution similar to the fluorescently labeled compound, the compound strongly related to the severity of cancer (mass-to-charge ratio corresponding to the compound) can be confirmed at a glance.

In the above processing, the fluorescence image is converted into simple pixel value data, but the same processing can be performed even if the fluorescence image is a color image or an image containing a plurality of wavelength components. In this case, there is not one fluorescent image for one sample, and there are as many fluorescent images as there are wavelengths. When there was only one fluorescent image, there was only one explained variable (Y) in the PLS regression analysis in step S3 (one one-dimensional matrix), but when there are multiple fluorescent images, step S3 There are a plurality of explained variables (Y) in the PLS regression analysis in. Therefore, the PLS regression analysis at this time may use the PLS2 algorithm.

The regression coefficient obtained at this time is a coefficient of each mass-to-charge ratio at each wavelength as shown in FIG. 4A, that is, a two-dimensional tabular form. Therefore, as shown in FIG. 4B, this is rearranged into a one-dimensional array in which the mass-to-charge ratio and the wavelength are combined, and in step S5, the coefficient matrix of the one-dimensional array prepared for the number of samples is used as an explanatory variable. (X), PLS regression analysis may be performed with the information indicating the severity of cancer as the dependent variable (Y).
As a matter of course, the regression coefficient matrix thus obtained is a matrix of coefficients for the set of mass-to-charge ratio and wavelength.

Further, in the above example, instead of the fluorescence image, an image obtained by another measurement method such that the distribution of a specific compound in the sample can be visualized may be used.
For example, data showing the degree of absorption of electromagnetic waves such as light having a specific wavelength and X-rays and distribution of radiation intensity, data showing intensity distribution such as Raman scattered light and fluorescence, stained image data, PET (positron emission tomography). ), CT (computer tomography), MRI (magnetic resonance imaging), ESR (electron spin resonance), etc., image data using radioisotope labeling, EPMA, SPM (Scanning Probe Microscope), etc. The surface unevenness image data acquired in the above can be used as the two-dimensional image data.

[Second Example]
In the first embodiment, multivariate analysis was performed on MS imaging data (mass spectrum data obtained at each of a plurality of measurement points in a measurement region) obtained from a plurality of samples, but one from one sample. The present invention can also be applied when one mass spectrum data can be obtained. A schematic configuration diagram of the mass spectrometry system according to this embodiment is shown in FIG. This system includes an imaging mass analysis unit 100, a data analysis unit 200, an input unit 300, and a display unit 400, and the data analysis unit 200 is a data input unit that receives mass spectrum data obtained by the imaging mass analysis unit 100. 201, data storage unit 202 for storing received data, average mass spectrum calculation unit 203, first multivariate analysis processing unit 204, second multivariate analysis processing unit 205, and analysis result display processing unit 206. And are provided as functional blocks.

As an example of the analysis in the mass spectrometry system according to this example, an example of investigating the relationship between non-alcoholic steatohepatitis and diet will be described. FIG. 6 is a conceptual diagram of characteristic data analysis processing in the mass spectrometry system of this embodiment.

The samples are the liver of a mouse that has developed non-alcoholic steatohepatitis (NASH) by feeding a special diet (hereinafter referred to as "special diet"), and the normal diet (hereinafter "normal diet"). The livers (control samples) of the mice fed with (), and the feeding period is 4 weeks, 8 weeks, 16 weeks, respectively, and there are a total of 6 types. Since it is empirically known that the longer the feeding period of the special diet is, the more severe the inflammation of NASH is, the 4th, 8th, and 16th weeks of the feeding period are described as severity: 1, 2, and 3, respectively. I will do it. In addition, the state in which NASH has not developed is described as severity: 0.

Mass spectrometry is executed by the imaging mass spectrometer 100 for each of the above six types of samples, and mass spectrum data is acquired for each measurement point. In the data analysis unit 200, the average mass spectrum calculation unit 203 calculates the average mass spectrum by calculating the average of the mass spectrum data at all the measurement points for each sample. As a result, the same number of average mass spectra as the number of samples can be obtained. The data constituting the average mass spectrum thus obtained is used as an explanatory variable (X) for PLS regression analysis.

On the other hand, the pattern of change in the amount of compounding according to the progression of severity is set as the explained variable (Y), but since it is unclear how the amount of compound changes, the following five types of patterns are used here. Suppose.
Y1 = (1,2,3): Compound amount increases or decreases in proportion to severity Y2 = (1,2,1): Compound amount increases or decreases temporarily during the feeding period Y3 = (1) , 0,0): Compound is present only at the beginning of the feeding period, or compound is not present only at the beginning of the feeding period Y4 = (0,1,0): Compound is present only in the middle of the feeding period, or Compound is not present only in the middle of the feeding period Y5 = (0,0,1): Compound is present only at the end of the feeding period, or is not present only at the end of the feeding period.

Characteristic values (Y) of Y1 to Y5 are set for each of a plurality of average mass spectra (X), and a data set for PLS regression is created. Then, the first multivariate analysis processing unit 204 executes PLS regression analysis for each data set thus created, and calculates the regression coefficient Bpls. Therefore, this regression coefficient Bpls can be obtained for each characteristic value and for each mass-to-charge ratio value. That is, a one-dimensional matrix in which the regression coefficients for each mass-to-charge ratio are arranged for each characteristic value can be obtained. However, it is difficult to grasp the overall picture of the characteristic values (Y1 to Y5) and each mass-to-charge ratio only by looking at this result.

Therefore, next, the second multivariate analysis processing unit 205 performs principal component analysis on the regression coefficient matrix for all the characteristic values. As is well known, by performing principal component analysis, score plots and loading plots can be obtained as analysis results. In this case, in the score plot, the characteristic values Y1 to Y5 are positioned on the graph having the two principal components (usually the first principal component and the second principal component) as orthogonal axes. Further, in the loading plot, each mass-to-charge ratio value is positioned on a graph having two principal components as orthogonal axes. The analysis result display processing unit 206 creates such a graph and displays it on the screen of the display unit 400. As a result, the user can grasp the relationship between the characteristic value and the characteristic mass-to-charge ratio value from a bird's-eye view.

It should be noted that all of the above examples are merely examples of the present invention, and in points other than those described above, even if appropriate modifications, modifications, additions, etc. are made within the scope of the present invention, they are included in the claims of the present application. It is clear that.

1 ... Measuring unit 11 ... Optical microscopic observation unit 12, 100 ... Imaging mass analysis unit 2, 200 ... Data analysis unit 21, 201 ... Data input unit 22, 202 ... Data storage unit 23 ... Calculation data preprocessing unit 203 ... Average mass Spectrum calculation units 24, 204 ... 1st multivariate analysis processing unit 25, 205 ... 2nd multivariate analysis processing unit 26, 206 ... Analysis result display processing unit 3, 300 ... Input unit 4, 400 ... Display unit

Now, for example, when the second data group is image data constituting an optical microscopic image obtained by monochromeizing one color image, that is, a data group showing pixel values of pixels corresponding to measurement points, the first statistical analysis For each sample, the processing unit calculates an index value indicating the similarity or difference between the image based on the pixel value data in the sample and the distribution image showing the distribution of the signal intensity of each mass charge ratio. The predetermined statistical analysis processing by the first statistical analysis processing unit may be, for example, partial least squares regression (PLS regression) analysis.

Claims (7)

A first data group in which data obtained for each predetermined parameter value is aggregated by performing predetermined measurements on a plurality of samples, and measurement by a method different from the predetermined measurement for each of the plurality of samples. This is a data analysis device that obtains information on the plurality of samples in general by an analysis process based on the second data group that aggregates the data obtained by executing the above.
a) Perform a predetermined statistical analysis process with the first data group as the explanatory variable and the second data group as the explained variable (objective variable), and show the similarity or difference of the data in both data groups. A first statistical analysis processing unit that calculates multiple index values,
b) Predetermined statistical analysis processing using a plurality of index values calculated by the first statistical analysis processing unit as explanatory variables and other information related to the plurality of samples as explained variables (objective variables). And a second statistical analysis processing unit that calculates a plurality of index values indicating the similarity or difference between the index value and the other information.
c) The relationship between the other information and the first data group, or the other information and the second data group, using a plurality of index values calculated by the second statistical analysis processing unit. A display processing unit that creates and displays a graph showing the relationship with
A data analysis device characterized by comprising. In the data analysis apparatus according to claim 1,
The first data group is a mass obtained by performing imaging mass spectrometry on a plurality of samples, respectively, with the mass-to-charge ratio at each of the plurality of measurement points in the two-dimensional measurement region as the predetermined parameter. Spectral data
The second data group is image data showing a predetermined signal intensity distribution in a two-dimensional measurement region obtained by a measurement method different from imaging mass spectrometry for a plurality of samples.
The first statistical analysis processing unit includes one or a plurality of images from the second data group and a distribution image of signal intensity at one mass-to-charge ratio according to the first data group for each sample. A data analysis device characterized by calculating an index value indicating similarity or difference. The data analysis apparatus according to claim 2.
A data analysis apparatus characterized in that a predetermined statistical analysis process by the first statistical analysis processing unit is partial least squares regression analysis. The data analysis apparatus according to claim 2.
A data analysis apparatus characterized in that the predetermined statistical analysis processing by the second statistical analysis processing unit is partial least squares regression analysis or principal component regression analysis. A first data group in which data obtained for each predetermined parameter value is aggregated by executing a predetermined measurement for each of a plurality of samples, and a measurement by a method different from the predetermined measurement for each of the plurality of samples. Data analysis for obtaining information on the plurality of samples in general by analysis processing based on a second data group that aggregates the data obtained by executing the above or other information related to the plurality of samples. It ’s a device,
a) Perform a predetermined statistical analysis process with the first data group as the explanatory variable and the second data group as the explained variable (objective variable), and show the similarity or difference of the data in both data groups. A first statistical analysis processing unit that calculates multiple index values,
b) A second statistical analysis processing unit that performs a predetermined statistical analysis processing for aggregating a plurality of index values calculated by the first statistical analysis processing unit into lower-dimensional information.
c) A display processing unit that creates and displays a graph showing the relationship between the second data group and each parameter value in the first data group based on the result of the second statistical analysis processing unit.
A data analysis device characterized by comprising. The data analysis apparatus according to claim 5.
A data analysis apparatus characterized in that a predetermined statistical analysis process by the first statistical analysis processing unit is partial least squares regression analysis. The data analysis apparatus according to claim 5.
A data analysis apparatus characterized in that the predetermined statistical analysis processing by the second statistical analysis processing unit is principal component analysis, factor analysis or cluster analysis.

Images