Imaging data analysis device

Abstract

Disclosed is an imaging data analysis device (1) which comprises: a measurement data acquisition unit (31) that measures responses to N2 parameter values in N1 micro-regions which are set in a measurement region of a sample, and acquires N2 measurement data for each micro-region; a measurement data group classification unit (32) that classifies the measurement data of the measurement region into N3 measurement data groups by classifying the N1 micro-regions or the N2 parameter values into N3 groups; a parameter dimension aggregation unit (33) that three-dimensionally aggregates the dimensions of the N2 parameter values by applying a nonlinear dimensionality reduction method to the N2 measurement data for each micro-region for the N3 measurement data groups; a corresponding color determination unit (34) that allocates the density of three primary colors to each three-dimensional axis and determines a color corresponding to each measurement data; and an image generation unit (35) that generates a mapping image of the measurement region in which the measurement data for each micro-region is represented by the color for the N3 measurement data groups.

Description

Imaging data analysis device

[0001] The present invention relates to an apparatus for analyzing imaging data obtained by measuring, with an analyzer, responses (objective variables) for respective ones of a plurality of parameter values (explanatory variables) in each of a plurality of minute regions set on the surface of a sample.

[0002] Imaging mass spectrometry is performed to examine the distribution of a target substance present on the surface of a sample such as a section of biological tissue. In imaging mass spectrometry, a plurality of minute regions are set within a measurement region on the sample surface, and mass spectrometry is performed in each minute region to acquire mass spectrum data for each minute region. By extracting the intensity of ions specific to the target substance from the mass spectrum data for each minute region and creating a mapping image representing the intensity distribution, information on the distribution of the target substance in the measurement region on the sample surface can be obtained.

[0003] There are cases where imaging mass spectrometry is performed without prior determination of a target substance, such as when searching for biomarkers related to a specific disease. In that case, a mapping image of the measurement region is created from the measurement intensities for each of a large number (for example, 1000 to 100,000 points) of mass-to-charge ratio values that make up the mass spectrum data for each minute region, and those numerous mapping images have to be observed individually, which is time-consuming and laborious.

[0004] Patent Document 1 describes a technique in which, by using UMAP (Uniform Manifold Approximation and Projection for dimension reduction; for example, Non-Patent Document 1), the number (number of dimensions) of mass-to-charge ratio values that make up the mass spectrum data is aggregated into three dimensions, and the mass spectrum data acquired in each minute region is represented as a single point in that three-dimensional space. By assigning the three primary colors to each axis of the three-dimensional space, a two-dimensional mapping image representing the mass spectrum acquired in each minute region with a single color is created. By using such a technique, the distribution of characteristic substances can be easily grasped without creating and observing numerous mapping images.

[0005] Japanese Patent Publication No. 2021-196260

[0006] Leland McInnes, et al., "UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction", [online], [December 7, 2018], arXiv:1802.03426v2 [stat.ML], Internet <URL: https: / / arxiv.org / pdf / 1802.03426.pdf>

[0007] In Patent Document 1, the dimensionality of the mass-to-charge ratio value is reduced to three dimensions, which clearly shows substances that are abundant on the sample surface, but it may cause information about the distribution of small amounts of substances to be disregarded, making it impossible to understand the distribution of such substances.

[0008] Here, we have explained the challenges of conventional techniques using imaging mass spectrometry data obtained by performing mass spectrometry on each micro-region as an example. However, imaging data obtained by performing other analytical methods such as spectroscopic measurements also face similar challenges.

[0009] The problem that this invention aims to solve is to provide a technology that can appropriately determine the distribution of substances present on the surface of a sample.

[0010] The imaging data analysis apparatus according to the present invention, which is made to solve the above problems, measures, by an analyzer, the response to each of N2 parameter values (N2 is an integer of 2 or more) in each of N1 micro regions (N1 is an integer of 2 or more) set in a measurement region of a sample, to obtain N2 measurement data for each micro region; a measurement data acquisition unit; classifies the N1 micro regions or the N2 parameter values into N3 (N3 < N1, N2) groups to classify the measurement data of the measurement region into N3 measurement data groups; a measurement data group classification unit; for each of the N3 measurement data groups, applies a non-linear dimensionality reduction method using manifold learning to the N2 measurement data for each micro region to aggregate the dimensions of the N2 parameter values into three dimensions; a parameter dimension aggregation unit; a corresponding color determination unit that determines a color corresponding to each measurement data by assigning the densities of the three primary colors to each axis of the three dimensions aggregated by the parameter dimension aggregation unit; and for each of the N3 measurement data groups, an image generation unit that generates a mapping image of the measurement region in which the measurement data for each micro region is represented by the color determined by the corresponding color determination unit.

[0011] The imaging data analysis apparatus according to the present invention analyzes N2 measurement data for each micro region obtained by measuring, by an analyzer, the response (objective variable) to each of N2 parameter values (explanatory variables) in each of N1 micro regions set in a measurement region of a sample. In the imaging data analysis apparatus according to the present invention, first, the N1 micro regions or the N2 parameter values are classified into N3 (N3 < N1, N2) groups to classify the N1×N2 measurement data of the measurement region into N3 measurement data groups. Then, for each measurement data group, a process of aggregating the dimensions of the N2 parameter values into three dimensions is performed individually, and the densities of the three primary colors are assigned to each axis of the aggregated three dimensions. As a result, the measurement data for each micro region is represented by one color. Thereafter, for each measurement data group, a mapping image of the measurement region in which the measurement data for each micro region is represented by the above one color is generated.

[0012] The imaging data analysis device according to the present invention classifies the measurement data acquired in the measurement area into N3 measurement data groups corresponding to the number of characteristic substance distributions present on the sample surface, and creates a mapping image for each of these N3 measurement data groups, thereby enabling an appropriate understanding of the distribution of substances present on the sample surface.

[0013] A diagram showing the main components of an imaging data analysis system including an imaging data analysis device according to the present invention. A flowchart illustrating the procedure for analysis processing using the imaging data analysis system of this embodiment. A conceptual diagram of the analysis processing in this embodiment. A graph plotting the mass-to-charge ratio values ​​in a three-dimensional space defined by three feature axes relating to a minute region in an example of analysis using the imaging analysis system of this embodiment. A graph plotting minute regions in a three-dimensional space defined by three feature axes relating to the mass-to-charge ratio values ​​in one of the groups classified in this embodiment. A graph plotting minute regions in a three-dimensional space defined by three feature axes relating to the mass-to-charge ratio values ​​in another group classified in this embodiment. Mapping images for each of the two groups obtained in this embodiment. A graph plotting minute regions in a three-dimensional space defined by three feature axes relating to the mass-to-charge ratio values ​​in a comparative example. A mapping image obtained in the comparative example.

[0014] One embodiment of the imaging data analysis device according to the present invention will be described below with reference to the drawings.

[0015] Figure 1 is a diagram showing the main components of the imaging data analysis system 1 of this embodiment. The imaging data analysis system 1 of this embodiment comprises an imaging mass spectrometer 10 and a control / processing device 20. The control / processing device 20 corresponds to the imaging data analysis device according to the present invention.

[0016] The imaging mass spectrometer 10 comprises a sample observation unit 11 and a mass spectrometry unit 12. The sample observation unit 11 is used to observe the sample to be analyzed and set the measurement area. For example, an optical microscope can be used for the sample observation unit 11. The mass spectrometry unit 12 is used to perform mass spectrometry in each of a plurality of micro-regions set in the measurement area. For example, the mass spectrometry unit 12 can have a MALDI ion source that coats the sample surface with a matrix material and ionizes the material present in each micro-region coated with the matrix material by irradiating the micro-region with laser light, and a ToF type mass spectrometry unit. The above specific configurations for the sample observation unit 11 and the mass spectrometry unit 12 are examples, and configurations that are appropriate can be used depending on the sample to be analyzed and the substance to be analyzed.

[0017] The control and processing unit 20 includes a storage unit 21. The storage unit 21 stores observation images of the sample surface acquired by the imaging mass spectrometer 10, information on the measurement area of ​​the sample set on the sample surface and the minute areas set within the measurement area, and measurement data acquired in each minute area (measurement data of the measurement area).

[0018] The control and processing unit 20 comprises, as functional blocks, a measurement data acquisition unit 31, a measurement data group classification unit 32, a parameter dimension aggregation unit 33, a corresponding color determination unit 34, an image generation unit 35, and a display processing unit 36. The control and processing unit 20 is, for example, a general personal computer, and these functional blocks are realized by executing a pre-installed imaging data analysis software program on the processor. The control and processing unit 20 is also connected to an input unit 41, consisting of a keyboard and mouse, for the user to perform appropriate input operations, and a display unit 42, consisting of a liquid crystal display, for displaying appropriate information.

[0019] The measurement data acquisition unit 31 acquires measurement data by measuring a sample using the imaging mass spectrometer 10, or by reading previously acquired measurement data from the storage unit 21.

[0020] The measurement data group classification unit 32 classifies the measurement data of the sample's measurement area into multiple measurement data groups by classifying the minute regions set in the sample's measurement area, or the mass-to-charge ratio values ​​included in the measurement range in mass spectrometry, into multiple groups.

[0021] The parameter dimension aggregation unit 33 aggregates the dimensions of the mass-to-charge ratio values ​​into three dimensions by applying a nonlinear dimensionality reduction method using manifold learning to each of the multiple measurement data groups classified by the measurement data group classification unit 32.

[0022] The corresponding color determination unit 34 determines the color corresponding to each measurement data by assigning the density of the three primary colors to each axis of the three dimensions aggregated by the parameter dimension aggregation unit 33.

[0023] The image generation unit 35 generates a mapping image of the measurement area for each of the multiple measurement data groups classified by the measurement data group classification unit 32, representing the measurement data for each minute region with the color determined by the corresponding color determination unit 34.

[0024] The display processing unit 36 ​​displays the mapping image, etc., generated by the image generation unit 35 on the screen of the display unit 42.

[0025] Next, the procedure for analyzing the distribution of material on the surface of a sample using the imaging data analysis system 1 of this embodiment will be explained with reference to the flowchart in Figure 2 and the conceptual diagram in Figure 3.

[0026] First, the user places the sample to be analyzed in the sample observation unit 11 of the imaging mass spectrometer 10 and observes the sample surface to set the measurement area (region of interest) (conceptual diagram A). Once the user sets the measurement area and gives a predetermined input instruction to set the micro-regions, the measurement data acquisition unit 31 sets multiple (N1) micro-regions (pxL-1 to pxL-N1) in a two-dimensional manner (typically in a grid pattern) at predetermined intervals within the measurement area. These predetermined intervals and the number of micro-regions may be predetermined and stored in the storage unit 21, or they may be set by the user each time. Observation data (image data) of the sample surface and positional information of the measurement area and micro-regions are stored in the storage unit 21.

[0027] When the measurement area is set, the measurement data acquisition unit 31 transports the sample to the ion source of the mass spectrometer 12. When the ion source is a MALDI source, in the process, a pretreatment of applying a matrix substance to the sample surface is performed.

[0028] In the mass spectrometer 12, in each of a plurality of minute regions, by ionizing the substances contained in the minute region and performing mass spectrometry, measurement data (mass spectrometry data) for each minute region is acquired (step 1, conceptual diagram B. The numbers attached to the rows and columns represent the number of dimensions. The same applies hereinafter). The mass spectrometry is performed by measuring the intensity of ions at each of a plurality (N2 pieces) of mass-to-charge ratio values within a predetermined mass-to-charge ratio range, and mass spectrum data for each minute region is obtained as the measurement data. The measurement data thus obtained is stored in the storage unit 21. Here, the measurement data is obtained by actually measuring the sample to be analyzed with the imaging mass spectrometer 10, but the measurement data acquisition unit 31 may also acquire the measurement data by having the user specify a file of the acquired measurement data and reading it from the storage unit 21 (step 1). The measurement data thus obtained is data in a matrix format with the mass-to-charge ratio (m / z) as the rows and the minute regions (pxL) as the columns, as shown in conceptual diagram C.

[0029] After the measurement data is acquired, when the user instructs the execution of the analysis process by a predetermined input operation, the measurement data group classification unit 32 reads out the measurement data of the measurement area of the sample (data of the measurement intensities at N2 mass-to-charge ratio values acquired in each of the N1 minute regions). Then, the mass spectrometry data of the N1 minute regions is classified into N3 (N3 < N1) measurement data groups (step 2). This classification can be executed, for example, by using UMAP (Uniform Manifold Approximation and Projection for dimension reduction), a non-linear dimensionality reduction method using manifold learning described in Non-Patent Document 1.

[0030] Specifically, based on N1 micro-regions, N4 (N4 < N2) characteristic axes (feature axes) are determined from the measurement data of the measurement region, and matrix transformation is performed (reducing the dimensionality from N1 to N4) (conceptual diagram D. In this figure, N4 = 3). N2 mass-to-charge ratio values are plotted in the N4-dimensional space defined by the N4 axes. As a result, mass-to-charge ratio values with similar ion distributions in the measurement region are plotted at positions close to each other in the N4-dimensional space. Then, a set of points (cluster) in the N4-dimensional space is regarded as one group, and the measurement data of the measurement region (mass spectrometry data for each of the N1 micro-regions) is classified into N3 groups (conceptual diagram E. Gr.1 to Gr.3 in this figure). As shown in conceptual diagram F, the measurement data of each group after classification is also data in a matrix form with the mass-to-charge ratio as rows and the micro-regions as columns, similar to conceptual diagram C. However, due to the classification of the N2 mass-to-charge ratio values into N4 groups, the number of rows (the number n of mass-to-charge ratio values) is smaller than the number of rows of the data matrix in conceptual diagram C.

[0031] The classification of the measurement data (points) by the measurement data group classification unit 32 may be performed based on a predetermined criterion, for example, classifying those with a separation distance between points being less than or equal to a predetermined value into the same group, or the user designates the number of groups (N3) to be classified, and the points distributed in the N4-dimensional space are clustered in ascending order of the separation distance and classified into N3 groups, so as to classify the measurement data group of the measurement region into the designated number of groups.

[0032] The number of axes (N4) is arbitrary as long as it is less than Nl. However, by setting the number of axes to 2 or 3, the two-dimensional plane or three-dimensional space in which the points representing each mass-to-charge ratio value are plotted is visualized, and the measurement data (points) of each micro-region classified by the measurement data group classification unit 32 are displayed on the screen of the display unit 42 in different modes (for example, in different colors) for each group, so that the user can visually grasp the classification result.

[0033] Once the measurement data group classification unit 32 has completed the classification of the measurement data groups, the parameter dimension aggregation unit 33, for each of the N3 classified groups, determines three characteristic axes (characteristic axes X, Y, Z) for the measurement data of the minute regions belonging to that group from N2 mass-charge ratio values ​​(as described above, the actual number of mass-charge ratio values ​​included in the matrix is ​​less than N2), performs a matrix transformation (reducing the number of dimensions from N2 to 3) (conceptual diagram G), and plots points representing the measurement data of each minute region in the three-dimensional space defined by these three axes (step 3). Since the mass-charge ratio values ​​actually included in the measurement data groups constituting the N3 groups are all different, in step 3, three different axes are determined for each group.

[0034] Next, the corresponding color determination unit 34 assigns one of the three primary colors (R, G, B) to each of the three axes (feature axes X, Y, Z) and determines the color corresponding to the point representing the measurement data of each minute region (step 4). As described above, since different three axes are determined for each group, even if the colors corresponding to multiple minute regions are the same color, if the groups to which those multiple minute regions belong are different from each other, the content of the measurement data that the color represents will differ for each group.

[0035] Once a color is determined to represent each of the measurement data sets for the microregions belonging to each group, the image generation unit 35 creates a mapping image for each group, representing each microregion with the color associated with the measurement data for that microregion (step 5), and the display processing unit 36 ​​displays the raw mapping image on the screen of the display unit 42 (step 6). This allows the user to visually grasp the distribution of characteristic substances on the sample surface for each of the N3 groups on the screen.

[0036] <Examples> Here, the results of the inventors actually performing the process described in the above embodiment will be explained with reference to Figures 4 to 7.

[0037] Figure 4 is a graph showing the results of classifying (clustering) the N2 mass-charge ratio values ​​into two groups (equivalent to N3) by setting three feature axes (Dimension 1 to 3) from the N1 microregions using UMAP (consolidating the N2 dimensions into 3) and plotting the N2 mass-charge ratio values ​​in the three-dimensional space defined by these three feature axes. This process corresponds to step 2 in the above embodiment.

[0038] Figures 5 and 6 show that, for each of the two groups classified by the above process, three feature axes (dim_1 to 3) were set from N2 mass-to-charge ratio values ​​included in the mass spectrometry data set of the microregions classified into that group using UMAP, and the microregions classified into that group were plotted in the three-dimensional space defined by these three feature axes. This process corresponds to step 3 in the above embodiment. As described above, the feature axes of the mass-to-charge ratio values ​​defining the three-dimensional space in Figures 5 and 6 are different from each other, so they cannot be simply compared, but it can be seen that the distribution of microregions in the same space is different from each other.

[0039] Figure 7 is a mapping image in which the three primary colors (RGB) are assigned to each of the three axes in Figures 5 and 6, thereby assigning a color to each minute region plotted in the three-dimensional space defined by these three axes, and representing the position of each minute region with that color. This process corresponds to steps 4 to 6 in the above embodiment. The two mapping images are originally in color, and although it is somewhat difficult to discern from the monochrome Figure 6, it can be seen that the appearance of the two mapping images is different from each other.

[0040] For comparison with the above example, a comparative example in which the prior art described in Patent Document 1 is applied to the same imaging mass spectrometry data as in the above example will be described. Figure 8 shows the mass spectrometry data of all minute regions using UMAP, with three feature axes set from N2 mass-to-charge ratio values, and each minute region plotted in the three-dimensional space defined by these three feature axes. Figure 9 is a mapping image in which three primary colors are assigned to the three axes of Figure 8, and each minute region plotted in the three-dimensional space defined by these three axes is assigned a color, representing the position of the minute region with that color. In the comparative example, only one mapping image is generated, so the distribution of substances that are abundant on the sample surface is reflected in the mapping image.

[0041] Comparing the mapping image in Figure 6 obtained from the example with the mapping image in Figure 9 obtained from the comparative example, for example, the mapping image in the upper part of Figure 6 shows the distribution of a characteristic substance marked with a circle (distribution in red in the color image), whereas this characteristic substance distribution does not appear in the comparative example. This is thought to be because the distribution of the substance marked with an oval (distribution in green in the color image), shown in the lower part of Figure 6, was abundant on the sample surface, and in the comparative example, the distribution of the characteristic substance was obscured by the distribution of this substance.

[0042] As described above, by using the embodiments and examples, the measurement data acquired in the measurement area can be classified into multiple measurement data groups (corresponding to N3 in this invention) corresponding to the number of characteristic substance distributions present on the sample surface, and a mapping image can be created for each of these multiple measurement data groups, thereby allowing for an appropriate understanding of the distribution of substances present on the sample surface.

[0043] The above embodiments and examples are merely examples and can be modified as appropriate in accordance with the spirit of the present invention.

[0044] In the above-described embodiments and examples, for the measurement data of the measurement region (measurement data of N1 minute regions), three characteristic axes related to the minute regions are set, and N2 mass-to-charge ratio values are plotted in the three-dimensional space defined by the three characteristic axes, whereby the N2 mass-to-charge ratio values are classified into two. However, the number of characteristic axes and the number of classifications can be changed as appropriate. Further, in the above embodiment, the N2 mass-to-charge ratio values are classified into N3 using UMAP. However, other dimensionality reduction methods such as principal component analysis (PCA), which is a projection method, cluster analysis, independent component analysis (ICA), and manifold learning methods such as LLE (Locally Linear Embedding), Isomap, t-SNE (t-distributed Stochastic Neighbor Embedding), etc. may be used for classification. Alternatively, without using such dimensionality reduction methods, the N2 mass-to-charge ratio values may be classified into N3 by mechanically dividing the range of the mass-to-charge ratio values obtained from the mass spectrometry data from the low mass-to-charge ratio side to the high mass-to-charge ratio side. For example, when a low molecule that generates ions with a low mass-to-charge ratio value and a high molecule that generates ions with a high mass-to-charge ratio value are distributed on the sample surface, mapping images representing the distribution of the low molecule and mapping images representing the distribution of the high molecule can be obtained individually by such a simple division of the mass-to-charge ratio range.

[0045] In the above embodiments, the measurement data of the measurement area was classified into N3 measurement data groups by setting three feature axes related to the minute region. However, the measurement data of the measurement area may also be classified into N3 measurement data groups by setting multiple feature axes related to the mass-to-charge ratio value. In that case, each minute region is plotted in the space defined by the multiple feature axes (or a plane if there are only two feature axes), and the set of plots of minute regions is classified into N3. Then, for the N3 measurement data groups of minute regions after classification, three feature axes based on N2 mass-to-charge ratio values ​​are set again, each minute region is plotted in the three-dimensional space defined by these feature axes, and a mapping image is created by assigning a color to each minute region as described above. In this case, each measurement data group after classification contains N2 mass-to-charge ratio values ​​as they are, but since the classification groups of measurement data have different characteristics from each other, three different feature axes are set for each group, as in the above embodiments. In this case as well, N1 minute regions may be classified into N3 by simply dividing the measurement area without using a dimensionality reduction method such as UMAP. For example, if there are substances distributed locally within a measurement area, conventional methods may make it difficult to confirm the distribution of these local substances because they are obscured by the distribution of substances that are more widely distributed throughout the measurement area. However, by dividing the measurement area, it is possible to obtain a mapping image showing the distribution of locally present substances for each divided measurement area.

[0046] In the above embodiments and examples, measurement data was obtained by performing mass spectrometry in each of the N1 minute regions within the measurement area. However, the measurement method is not limited to mass spectrometry, and the above embodiments and examples can be applied to various types of measurement data obtained by measuring the response (dependent variable) to each of the N2 parameter values ​​(explanatory variables) using an analytical instrument. For example, the same configuration as in the above embodiments and examples can be applied to measurement data obtained by performing spectroscopic measurements in each minute region using a spectroscopic measuring instrument. In this case, wavelength becomes the explanatory variable, and absorbance or emission intensity becomes the dependent variable.

[0047] [Aspect] It is clear to those skilled in the art that the above exemplary embodiments are specific examples of the following aspects.

[0048] (Item 1) An imaging data analysis device according to one aspect of the present invention includes: a measurement data acquisition unit that acquires N2 measurement data for each of N1 (N1 is an integer of 2 or more) minute regions set in a measurement region of a sample by measuring responses to each of N2 (N2 is an integer of 2 or more) parameter values with an analyzer; a measurement data group classification unit that classifies the measurement data of the measurement region into N3 (N3 < N1, N2) measurement data groups by classifying the N1 minute regions or the N2 parameter values into N3 groups; a parameter dimension aggregation unit that aggregates the dimensions of the N2 parameter values into three dimensions by applying a non-linear dimensionality reduction method using manifold learning to the N2 measurement data for each minute region for each of the N3 measurement data groups; a corresponding color determination unit that determines a color corresponding to each measurement data by assigning the concentrations of the three primary colors to each axis of the three dimensions aggregated by the parameter dimension aggregation unit; and an image generation unit that generates a mapping image of the measurement region in which the measurement data for each minute region is represented by the color determined by the corresponding color determination unit for each of the N3 measurement data groups.

[0049] The imaging data analysis device according to claim 1 analyzes N2 measurement data for each micro-region obtained by measuring the response (objective variable) for each of N2 parameter values (explanatory variables) in each of N1 micro-regions set in the measurement region of the sample using an analyzer. In the imaging data analysis device according to claim 1, first, the N1×N2 measurement data in the measurement region are classified into N3 measurement data groups by classifying the N1 micro-regions or the N2 parameter values into N3 (N3 < N1, N2) groups. Then, for each measurement data group, a process of aggregating the dimensions of the N2 parameter values into three dimensions is performed individually, and the concentrations of the three primary colors are assigned to each of the aggregated three-dimensional axes. As a result, the measurement data for each micro-region are represented by one color. After that, for each measurement data group, a mapping image of the measurement region in which the measurement data for each micro-region are represented by the above one color is generated.

[0050] In the imaging data analysis device according to claim 1, the measurement data acquired in the measurement region are classified into N3 measurement data groups corresponding to the number of distributions of characteristic substances present on the sample surface, and by creating a mapping image for each of these N3 measurement data groups, the distribution of substances present on the sample surface can be appropriately grasped.

[0051] (Claim 2) The imaging data analysis device according to claim 2 is the imaging data analysis device according to claim 1, wherein the measurement data are mass spectrometry data obtained by performing mass spectrometry in each of the N1 micro-regions.

[0052] In mass spectrometry, by measuring ions specific to a substance, measurement data with high substance discrimination can be obtained. In the imaging data analysis device according to claim 2, by using mass spectrometry data as the measurement data, a mapping image that more accurately reflects the distribution of specific substances in the measurement region of the sample can be obtained.

[0053] (Clause 3) The imaging data analysis device according to paragraph 3 is an imaging data analysis device according to paragraph 1 or paragraph 2, wherein the measurement data group classification unit classifies the N2 parameter values ​​into the N3 groups.

[0054] In the imaging data analysis device described in paragraph 1, the N2 parameter values ​​are for identifying the substances distributed in the measurement area of ​​the sample. Therefore, in the imaging data analysis device described in paragraph 3, by classifying these parameter values ​​into N3 categories, it becomes easier to obtain mapping images that reflect the characteristic distribution of various substances.

[0055] (Article 4) The imaging data analysis device relating to Article 4 is an imaging data analysis device relating to any of Articles 1 to 3, wherein the measurement data group classification unit classifies the N1 minute regions or the N2 parameter values ​​into the N3 groups by applying a nonlinear dimensionality reduction method utilizing manifold learning.

[0056] In the imaging data analysis device described in paragraph 4, multiple feature axes are defined based on either N1 minute regions or N2 parameter values. The other is plotted in the space defined by these feature axes, and the set of plotted points is classified into a single group. This allows for classification according to the characteristics of the distribution of materials, and N3 mapping images representing the distribution of materials with different characteristics can be obtained. UMAP can be suitably used as a nonlinear dimensionality reduction method utilizing manifold learning.

[0057] (Clause 5) The imaging data analysis device according to Clause 5 is an imaging data analysis device according to Clause 4, wherein the measurement data group classification unit sets two or three feature axes based on the N1 minute regions, plots the N2 parameter values ​​in a plane or three-dimensional space defined by the feature axes, and classifies the N2 parameter values ​​into the N3 based on the set of plotted points, or sets two or three feature axes based on the N2 parameter values, plots the N1 minute regions in a plane or three-dimensional space defined by the feature axes, classifies the N1 minute regions into the N3 based on the set of plotted points, and displays the plane or three-dimensional space on the screen.

[0058] In the imaging data analysis device described in paragraph 5, the user can visually understand that N1 minute regions or N2 parameter values ​​have been appropriately classified into N3 groups.

[0059] (Clause 6) The imaging data analysis device according to Clause 6 is an imaging data analysis device according to any of Clauses 1 to 5, wherein the measurement data group classification unit classifies the N1 minute regions into the N3 groups by dividing the measurement region into the N3, or mechanically classifies the N2 parameter values ​​into the N3 groups.

[0060] In the imaging data analysis device described in paragraph 6, when the measurement area is divided into N3 parts, thereby classifying N1 minute regions into N3 groups, a mapping image of substances locally distributed in the measurement area of ​​the sample can be obtained. Furthermore, when N2 parameter values ​​are mechanically classified into N3 groups, a mapping image representing the distribution of substances that respond to small parameter values ​​and substances that respond to large parameter values ​​can be obtained, respectively.

[0061] 1…Imaging data analysis system 10…Imaging mass spectrometer 11…Sample observation unit 12…Mass spectrometry unit 20…Control and processing unit 21…Storage unit 31…Measurement data acquisition unit 32…Measurement data group classification unit 33…Parameter dimension aggregation unit 34…Corresponding color determination unit 35…Image generation unit 41…Input unit 42…Display unit

Claims

1. A measurement data acquisition unit that acquires N2 measurement data for each of the N1 (N1 is an integer of 2 or more) micro regions set in the measurement region of the sample by measuring the response to each of the N2 (N2 is an integer of 2 or more) parameter values with an analyzer; A measurement data group classification unit that classifies the measurement data of the measurement region into N3 (N3 < N1, N2) measurement data groups by classifying the N1 micro regions or the N2 parameter values into N3 groups; For each of the N3 measurement data groups, a parameter dimension aggregation unit that aggregates the dimensions of the N2 parameter values into three dimensions by applying a non-linear dimensionality reduction method using manifold learning to the N2 measurement data for each micro region; A corresponding color determination unit that determines the color corresponding to each measurement data by assigning the concentrations of the three primary colors to each axis of the three dimensions aggregated by the parameter dimension aggregation unit; and an image generation unit that generates a mapping image of the measurement region in which the measurement data for each micro region is represented by the color determined by the corresponding color determination unit for each of the N3 measurement data groups. An imaging data analysis device comprising:

2. The imaging data analysis device according to claim 1, wherein the measurement data is mass spectrometry data obtained by performing mass spectrometry on each of the N1 micro regions.

3. The imaging data analysis device according to claim 1, wherein the measurement data group classification unit classifies the N2 parameter values into the N3 groups.

4. The imaging data analysis device according to claim 1, wherein the measurement data group classification unit classifies the N1 micro regions or the N2 parameter values into the N3 groups by applying a non-linear dimensionality reduction method using manifold learning.

5. The imaging data analysis device according to claim 4, wherein the measurement data group classification unit sets two or three feature axes based on the N1 minute regions, plots the N2 parameter values ​​in a plane or three-dimensional space defined by the feature axes, and classifies the N2 parameter values ​​into the N3 based on the set of plotted points, or sets two or three feature axes based on the N2 parameter values, plots the N1 minute regions in a plane or three-dimensional space defined by the feature axes, classifies the N1 minute regions into the N3 based on the set of plotted points, and displays the plane or three-dimensional space on a screen.

6. The imaging data analysis apparatus according to claim 1, wherein the measurement data group classification unit classifies the N1 minute regions into N3 groups by dividing the measurement area into N3, or mechanically classifies the N2 parameter values ​​into N3 groups.

Images