Imaging analysis device and imaging analysis method
Patent Information
- Application Number
- JP2025552493
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Filing Date
- 2026-04-07
- Publication Date
- 2026-07-08
Abstract
Description
Imaging analysis device and imaging analysis method
[0001] The present invention relates to an imaging analysis device and an imaging analysis method.
[0002] In recent years, imaging analysis techniques that acquire and visualize two-dimensional (or in some cases, more multidimensional) information on various samples, such as biological samples, have been widely used. For example, imaging mass spectrometers extract signal intensity values at the mass-to-charge ratio (m / z) of ions derived from a specific compound based on mass spectrum data obtained for each of many microregions (measurement points) within a two-dimensional region on a sample, such as a biological tissue section, and create an image in which the signal intensity values are arranged according to the position of each microregion on the sample, thereby obtaining an MS imaging image showing the distribution of the specific compound.
[0003] One known data analysis technique for such imaging mass spectrometers involves clustering each microregion using statistical analysis based on the mass spectrum pattern obtained for each microregion, dividing the two-dimensional region being analyzed into multiple regions based on the clustering results, and depicting each divided region in a different color (see Patent Document 1, etc.). In other words, this is region segmentation using cluster analysis. This allows regions on a sample that are likely to contain the same compound to be divided and presented in an easy-to-understand manner to the user.
[0004] International Publication No. 2020 / 166008 Japanese Patent Application Laid-Open No. 2021-196260
[0005] Among several cluster analysis methods, the K-means method is widely used when dealing with large amounts of data due to its short computational time. In the K-means method, the number of clusters is set by the user or automatically, and a large amount of data is clustered according to that number. The number of clusters is an important parameter for performing appropriate clustering. In segmentation using the K-means method, users may need to adjust the number of clusters as needed and view the segmentation images obtained under each cluster number on a display screen, performing a trial-and-error process to find the optimal number of clusters. When samples are organ slices from laboratory animals, for example, the boundary shapes of each part on the sample are complex, and the sample contains a large number of chemical compounds, resulting in highly complex segmentation images. Therefore, it is difficult for users to accurately determine whether the number of clusters is appropriate by reviewing the segmentation images, making it difficult to improve the efficiency of such tasks.
[0006] The present invention has been made to solve these problems, and its main purpose is to provide an imaging analysis device and method that allows a user to accurately and efficiently determine cluster analysis conditions, such as the number of clusters, when performing segmentation using cluster analysis.
[0007] One aspect of the imaging analysis device of the present invention comprises: a measurement unit that performs a predetermined analysis on each of a plurality of micro-regions set within a measurement region on a target sample; an analysis unit that performs a cluster analysis based on data for each micro-region obtained by the measurement unit under set cluster analysis conditions to classify the plurality of micro-regions within the measurement region into a plurality of clusters; an image creation unit that creates a segmentation image corresponding to the measurement region based on the results of the cluster analysis; and a display processing unit that arranges and displays, on the same screen of a display unit, a plurality of segmentation images obtained for one target sample by the analysis unit and the image creation unit under mutually different cluster analysis conditions.
[0008] Furthermore, one aspect of the imaging analysis method according to the present invention includes: a measurement step of performing a predetermined analysis on each of a plurality of micro-regions set within a measurement region on a target sample; an analysis step of performing a cluster analysis based on data for each micro-region obtained in the measurement step, under set cluster analysis conditions, to classify the plurality of micro-regions within the measurement region into a plurality of clusters; an image creation step of creating a segmentation image corresponding to the measurement region based on the results of the cluster analysis; and a display step of arranging and displaying, on the same screen of a display unit, a plurality of segmentation images obtained in the analysis step and the image creation step under mutually different cluster analysis conditions for one target sample.
[0009] According to the above aspect of the present invention, a user can easily select more appropriate cluster analysis conditions by visually comparing and evaluating multiple segmentation images obtained under different cluster analysis conditions (typically, the number of clusters) displayed on the screen of the display unit, thereby efficiently searching for appropriate cluster analysis conditions even for complex segmentation images of biological samples, etc.
[0010] 1 is a block diagram of the main parts of an imaging mass spectrometer according to one embodiment of the present invention; 2 is a flowchart illustrating an example of the procedure for analysis processing in the imaging mass spectrometer of this embodiment; 3 is a schematic diagram for explaining analysis processing in the imaging mass spectrometer of this embodiment; 4 is a schematic diagram illustrating an example of an analysis result display screen in the imaging mass spectrometer of this embodiment; 5 is a diagram illustrating an example of displaying clustering results for multiple samples; 6 is a diagram illustrating another example of displaying clustering results for multiple samples.
[0011] An embodiment of an imaging analyzer according to the present invention will be described with reference to the accompanying drawings. The following embodiment is an example in which the present invention is applied to an imaging mass spectrometer, but as will be described later, the scope of application of the present invention is not limited thereto and can be applied to various imaging analyzers.
[0012] Figure 1 is a block diagram of the main components of an imaging mass spectrometer according to this embodiment. As shown in Figure 1, this imaging mass spectrometer broadly comprises an imaging mass analyzer 1, a data processing unit 2, an input unit 3, and a display unit 4. The input unit 3 and display unit 4 are user interfaces.
[0013] The imaging mass spectrometry unit 1 may be, for example, an imaging mass microscope "iMSope" manufactured by Shimadzu Corporation. TM However, as disclosed in Patent Document 2, a laser microdissection device may be used which combines a laser microdissection device with a mass spectrometer which performs mass analysis on a sample prepared from minute sample pieces collected from the sample by the device.
[0014] The data processing unit 2 includes, as functional blocks, a data storage unit 21, a data matrix creation unit 22, a nonlinear dimension reduction processing unit 23, a cluster analysis condition setting unit 24, a cluster analysis unit 25, an analysis result processing unit 26, a segmentation image creation unit 27, and a display processing unit 28.
[0015] In the imaging mass spectrometer of this embodiment, the data processing unit 2 is typically configured around a personal computer or a more powerful workstation, and the above-mentioned functional blocks can be realized by running dedicated data processing software installed on the computer. In this case, the input unit 3 is a keyboard and pointing device (such as a mouse) attached to the computer, and the display unit 4 is a display monitor.
[0016] The analytical processing procedure in the imaging mass spectrometer of this embodiment will be described with reference to Figures 2 and 3. Figure 2 is a flowchart showing an example of the analytical processing procedure, and Figure 3 is a schematic diagram for explaining the analytical processing. The object of analysis by the imaging mass spectrometer 1 is, for example, a thinly sliced sample of biological tissue such as an organ from a laboratory animal. Such a sample is placed on a sample plate, the surface of which is coated with a matrix for MALDI (matrix-assisted laser desorption ionization), and the plate is then set in a predetermined position in the imaging mass spectrometer 1.
[0017] As shown in Fig. 3(A), the imaging mass spectrometer 1 performs mass analysis on each of the microregions 52 obtained by dividing a predetermined measurement region 51 on the sample 50 into small, grid-like sections. As a result, mass spectrum data covering a predetermined m / z range is obtained for each microregion 52, as shown in Fig. 3(B) (step S10). In Fig. 3(B), pxL-1, pxL-2, ... are numbers assigned to the microregions 52. The data collected in this manner is sent from the imaging mass spectrometer 1 to the data processing unit 2 and stored in the data storage unit 21. Note that, instead of ordinary mass analysis, MS / MS analysis may be performed using ions having a specific m / z or falling within the m / z range as precursor ions, or MS analysis where n is 3 or more. n Analysis may be performed to obtain product ion spectral data.
[0018] After data collection, when a user issues an instruction to execute analysis via the input unit 3, the data matrix creation unit 22 sequentially reads data from the data storage unit 21, detects peaks for each mass spectrum data set according to a predetermined criterion, and calculates the m / z value and signal intensity value of each peak. The data matrix creation unit 22 collects the m / z values and signal intensity values of the peaks detected in the mass spectrum data for all microregions 52 and creates a data matrix as shown in Figure 3(C) (step S11). In Figure 3(C), the data matrix is a matrix in which the numbers of all microregions are arranged vertically, the m / z values of all peaks (M1, M2, M3, ...) are arranged horizontally, and the signal intensity value for a certain m / z value in a certain microregion is used as an element.
[0019] In the data matrix, the number of m / z values is the number of dimensions. Therefore, the number of dimensions of this data matrix is usually very large, ranging from several hundred to several thousand. Next, the nonlinear dimension reduction processor 23 performs a process to reduce the number of dimensions in the m / z direction of this data matrix (step S12).
[0020] Here, we use a nonlinear dimensionality reduction method called manifold learning, specifically UMAP (Uniform Manifold Approximation and Projection), to reduce the dimensionality of the data. A detailed theoretical explanation of UMAP is provided in various papers, so it will be omitted here. UMAP can be implemented using publicly available programs. By applying dimensionality reduction using UMAP to multidimensional data (the same number of dimensions as the number of m / z values) such as that shown in Figure 3(C), data with hundreds to thousands of dimensions is compressed into three-dimensional data. In other words, the data matrix shown in Figure 3(C) is converted into the three-column data matrix shown in Figure 3(D).
[0021] The nonlinear dimension reduction processor 23 further assigns three primary colors, R (red), G (green), and B (blue), to each of the three-dimensional axes (X, Y, and Z) after dimension reduction, and converts the data values (intensity values) into brightness values of the colors (converting them appropriately as necessary) to assign colors to points corresponding to each microregion. This results in a virtual three-dimensional image in which colored points corresponding to the microregions 52 within the measurement region 51 are arranged in a three-dimensional space with X, Y, and Z as the three mutually orthogonal axes (step S13). Note that this three-dimensional image is merely an image obtained during the calculation process and is not actually displayed, although it may be displayed on the display unit 4. The data constituting this virtual three-dimensional image (hereinafter referred to as "dimensionally reduced data") is stored in the data storage unit 21 as data to be subjected to cluster analysis.
[0022] Next, the cluster analysis condition setting unit 24 sets the number of clusters, which is a parameter in cluster analysis using the K-means method, based on input from the user via the input unit 3 (step S14). Specifically, the cluster analysis condition setting unit 24 displays an input box on the screen of the display unit 4 for inputting the number of clusters as a number. The user inputs a desired number in this input box and clicks the "Execute Analysis" button. In response to this operation, the cluster analysis condition setting unit 24 reads the input number and can set it as a parameter for performing cluster analysis.
[0023] When an instruction to execute cluster analysis is issued, the cluster analysis unit 25 reads the dimension-reduced data stored in the data storage unit 21 and performs clustering using the K-means method based on the three-dimensional brightness values of each point corresponding to the infinitesimal region 52, with the set number of clusters as the condition (step S15). All points corresponding to different infinitesimal regions 52 within the measurement region 51 are classified into one of the clusters. During the clustering process, the position of the center of gravity (coordinate position in the virtual three-dimensional image) for each cluster is calculated. For example, if the number of clusters is set to "5," all points are classified into one of five clusters.
[0024] When the clustering is completed, the analysis result processing unit 26 counts the number of points classified into each cluster (that is, the number of minute regions) (step S16).
[0025] Next, the segmentation image creation unit 27 creates a segmentation image corresponding to the measurement region 51 on the sample 50, reflecting the results of the clustering in step S15 (step S17). Specifically, since the position of the center of gravity in the virtual three-dimensional image for each cluster has been determined, a color is identified from the R, G, and B luminance values for that center of gravity. The segmentation image is then created by coloring the microregions classified into that cluster with that color. Therefore, the number of regions separated on the segmentation image is the same as the number of clusters set in step S14, and the display color of each cluster depends on the position of the center of gravity of each cluster obtained by clustering. In other words, if the position of the center of gravity changes, the display color on the segmentation image also changes.
[0026] The display processing unit 28 creates an analysis result display screen in which the segmentation image created by the segmentation image creation unit 27 and information on the number of microregions per cluster calculated by the analysis result processing unit 26 are arranged in a predetermined format, and displays this on the display unit 4 (step S18). Figure 4 is a schematic diagram showing an example of the analysis result display screen 100. In this example, the analysis result display screen 100 includes a cluster information display area 105, a segmentation image display area 101 corresponding to the current number of clusters, a data point list table area 102 corresponding to the current number of clusters, a segmentation image display area 103 corresponding to the number of clusters before change, and a data point list table area 104 corresponding to the number of clusters before change. The cluster information display area 105 includes a cluster change instruction input box 106 and an analysis execution button 107.
[0027] If the processing of steps S14 to S18 is performed for the first time after the dimension-reduced data is stored in the data storage unit 21 in step S13, the "number of clusters before the change" does not exist. In this case, for example, the segmentation image display area 103 corresponding to the number of clusters before the change and the data point list table area 104 corresponding to the number of clusters before the change on the analysis result display screen 100 are blank, the segmentation image created in step S17 is displayed in the segmentation image display area 101 corresponding to the current number of clusters, and the table aggregating the number of microregions for each cluster calculated in step S16 is displayed in the data point list table area 102 corresponding to the current number of clusters. Of course, the analysis may be ended in this state.
[0028] The number of clusters is the most important parameter in cluster analysis, and if the number of clusters is inappropriate, the clustering may be biased or otherwise inappropriate. However, because the state of clustering that is appropriate depends on the type of sample, the purpose of the analysis, and other factors, it is not necessarily easy to automatically determine the appropriate number of clusters, and in many cases the user must make the decision. Therefore, if the user wishes to change the number of clusters and continue the analysis, the user enters the desired value in the cluster change instruction input box 106 and clicks the analysis execution button 107.
[0029] 2, the process returns to step S14, where the cluster analysis condition setting unit 24 sets the specified numerical value as the new number of clusters. Then, under the newly set number of clusters, the cluster analysis unit 25, the analysis result processing unit 26, and the segmentation image creation unit 27 execute the processes of steps S15 to S17 described above based on the dimension-reduced data stored in the data storage unit 21, and create a segmentation image corresponding to the new number of clusters. While the dimension reduction process requires more calculation time than cluster analysis, the cluster analysis and subsequent processes can be performed again using the saved dimension-reduced data, thereby saving the calculation time.
[0030] Thereafter, the display processing unit 28 displays the segmentation image and list that were previously displayed in the segmentation image display area 101 corresponding to the current number of clusters and the data point list table area 102 corresponding to the current number of clusters, respectively, in a segmentation image display area 103 corresponding to the pre-change number of clusters and a data point list table area 104 corresponding to the pre-change number of clusters. Furthermore, the segmentation image display area 101 corresponding to the current number of clusters and the data point list table area 102 corresponding to the current number of clusters display a segmentation image newly created under the changed number of clusters and a table aggregating the number of microregions for each cluster.
[0031] FIG. 4 shows an example in which the number of clusters before the change is "6" and the number of clusters after the change is "5." That is, as shown in FIG. 4 , two segmentation images corresponding to the same measurement region 51 obtained with different numbers of clusters are displayed side by side on the analysis result display screen 100. This allows the user to compare the left and right segmentation images and select the number of clusters that is deemed more appropriate depending on the type of sample, the purpose of analysis, etc. If necessary, the number of clusters can be changed again and analysis performed with the changed number of clusters, thereby ultimately determining the number of clusters that is deemed most appropriate. Note that changing the number of clusters usually changes the position of the center of gravity of each cluster, and therefore the color of each region in the segmentation image also changes accordingly.
[0032] In the example of Fig. 4, two segmentation images corresponding to a certain number of clusters and the number of clusters before a single change are displayed side by side, but three or more segmentation images each having a different number of clusters may be displayed side by side. Furthermore, when the number of segmentation images to be displayed is large, the segmentation images may be displayed as a list of thumbnail images, and a selected segmentation image may be enlarged and displayed. Furthermore, instead of displaying the segmentation images side by side, multiple segmentation images may be displayed by switching tabs, or the user may be able to switch between tab switching and list display through settings.
[0033] In the tables displayed in the area 102 listing the number of data points corresponding to the current number of clusters and the area 104 listing the number of data points corresponding to the number of clusters before change on the analysis result display screen 100, the number of data points for each cluster is shown numerically, allowing the user to know the quantitative ratio of the number of data points for each cluster. Furthermore, in order to more intuitively show the degree of distribution of each cluster within the measurement area 51, the proportion of the number of data points that each cluster occupies in the entire measurement area 51 (e.g., displayed as a percentage) may be shown.
[0034] Furthermore, to help the user determine the optimal number of clusters, one or more commonly known clustering evaluation indices may be calculated for each number of clusters and displayed together with the segmentation image. Known clustering evaluation indices include the sum of squared errors (Inertia), Silhouette Score, Davies-Bouldin Score, and Calinski-Harabasz Score.
[0035] In the imaging mass spectrometer of the above embodiment, the number of clusters, which is one of the cluster analysis conditions, can be specified by the user, and when the user changes the number of clusters, multiple segmentation images based on the number of clusters before and after the change can be displayed so that they can be easily compared. However, the cluster analysis conditions that can be changed are not limited to the number of clusters.
[0036] For example, in the k-means method, clustering is performed using the distance between the center of gravity calculated from the data points and each data point. This distance is usually calculated using Euclidean distance, but several other algorithms are known, such as Manhattan distance, Minkowski distance, and Mahalanobis distance. Changing the algorithm will change the clustering state even if the number of clusters remains the same. Therefore, it is possible to select the distance calculation algorithm as a cluster analysis condition, and display multiple segmentation images obtained as a result of cluster analysis using different calculation algorithms for comparison.
[0037] Furthermore, in the imaging mass spectrometer of the above embodiment, different segmentation images for the same measurement region 51 are displayed for comparison. However, it is also possible to display different segmentation images for different measurement regions 51 on a single sample 50 or for measurement regions on different samples, i.e., for multiple measurement regions, for each measurement region for comparison. In this case, the dimension-reduced data obtained from the mass spectral data obtained for the multiple samples (or measurement regions) is subjected to cluster analysis under the same cluster analysis conditions. Then, based on the clustering results, segmentation images for the different measurement regions are created. In this case, it is preferable to make it easy to compare the number of data points (or the proportion of data points) for each cluster in different segmentation images.
[0038] Figure 5 shows an example of a table showing the number of data points per cluster when cluster analysis was performed on eight samples (No. 1 to No. 8) with a cluster count of "5." Figure 6 shows an example of the same results displayed in graph format. In graph format, each cluster is indicated by its display color on the segmentation image. Such a table or graph display makes it easy to find, for example, samples among multiple samples that have a similar distribution of data points per cluster.
[0039] Although the above embodiment is an application of the present invention to an imaging mass spectrometer, the present invention can also be applied to imaging analyzers that use various analytical techniques other than mass spectrometry. For example, it is readily conceivable that the present invention can also be applied to imaging analyzers that use infrared spectroscopy, fluorescence analysis, etc.
[0040] Furthermore, the above-described embodiment and modified examples are merely examples of the present invention, and it goes without saying that any appropriate modifications, alterations, additions, etc. made within the spirit of the present invention will also be encompassed within the scope of the claims of the present application.
[0041] Various Aspects It will be appreciated by those skilled in the art that the exemplary embodiments described above are examples of the following aspects.
[0042] (Item 1) One aspect of the imaging analysis device of the present invention comprises: a measurement unit that performs a predetermined analysis on each of a plurality of micro-regions set within a measurement region on a target sample; an analysis unit that performs a cluster analysis based on data for each micro-region obtained by the measurement unit under set cluster analysis conditions to classify the plurality of micro-regions within the measurement region into a plurality of clusters; an image creation unit that creates a segmentation image corresponding to the measurement region based on the results of the cluster analysis; and a display processing unit that arranges and displays, on the same screen of a display unit, a plurality of segmentation images obtained by the analysis unit and the image creation unit under mutually different cluster analysis conditions for one target sample.
[0043] (Item 7) Another aspect of the imaging analysis method according to the present invention includes a measurement step of performing a predetermined analysis on each of a plurality of micro-regions set within a measurement region on a target sample; an analysis step of performing a cluster analysis based on data for each micro-region obtained in the measurement step, under set cluster analysis conditions, to classify the plurality of micro-regions within the measurement region into a plurality of clusters; an image creation step of creating a segmentation image corresponding to the measurement region based on the results of the cluster analysis; and a display step of arranging and displaying, on the same screen of a display unit, a plurality of segmentation images obtained in the analysis step and the image creation step under mutually different cluster analysis conditions for one target sample.
[0044] (Items 2 and 8) According to the imaging analysis device described in item 1 or the imaging analysis method described in item 7, the cluster analysis condition can be the number of clusters.
[0045] According to the imaging analysis apparatuses described in paragraphs 1 and 2 and the imaging analysis methods described in paragraphs 7 and 8, a user can easily select more appropriate cluster analysis conditions by visually comparing and evaluating multiple segmentation images obtained under different cluster analysis conditions, such as the number of clusters, displayed on the screen of the display unit. This makes it possible to efficiently search for appropriate cluster analysis conditions even for complex segmentation images of biological samples, etc.
[0046] (Item 3) In the imaging analysis device described in item 1, the predetermined analysis may be mass spectrometry.
[0047] (Item 9) Similarly, in the imaging analysis method described in item 7, the predetermined analysis may be mass spectrometry.
[0048] (4) In the imaging analysis device described in paragraph 3, the analysis unit may include a dimension reduction processing unit that processes the mass analysis data for each micro-region obtained by the measurement unit using a nonlinear dimension reduction method that utilizes manifold learning to obtain dimension-reduced data for each micro-region in which the dimensions corresponding to the number of mass-to-charge ratio values have been reduced to three dimensions, and that performs cluster analysis on the dimension-reduced data for each micro-region.
[0049] (Item 10) In the imaging analysis method described in Item 9, the analysis step may include a dimension reduction step in which the mass analysis data for each micro-region obtained in the measurement step is processed using a nonlinear dimension reduction method utilizing manifold learning to obtain dimension-reduced data for each micro-region in which the dimension corresponding to the number of mass-to-charge ratio values has been reduced to three dimensions, and cluster analysis may be performed on the dimension-reduced data for each micro-region.
[0050] Generally, the number of mass-to-charge ratio values in mass spectrum data obtained by performing mass analysis using a scan measurement is very large, and attempting to perform cluster analysis using this data as is results in complex calculations that require a long calculation time. In contrast, the imaging analysis apparatus described in paragraph 4 and the imaging analysis method described in paragraph 10 can significantly reduce the number of dimensions of the data, making cluster analysis extremely simple and reducing the overall time required for calculations.
[0051] (5) In the imaging analysis device described in 4, the image creation unit can determine a display color corresponding to a cluster in the segmentation image based on the position of the center of gravity of each cluster determined by cluster analysis in a three-dimensional space in which three primary colors are assigned to each axis of the three dimensions.
[0052] (Item 11) In the imaging analysis method described in Item 10, the image creation step can involve determining a display color corresponding to a cluster in the segmentation image based on the position of the center of gravity of each cluster determined by cluster analysis in a three-dimensional space in which three primary colors are assigned to each axis, with each axis corresponding to one of the three dimensions.
[0053] According to the imaging analysis device described in paragraph 5 and the imaging analysis method described in paragraph 11, the average state of the data included in the cluster corresponding to a region on a segmentation image can be grasped from the color of the region. This makes it easy for a user to evaluate the similarity of clusters, for example, when comparing different segmentation images of the same sample or when comparing segmentation images of different samples.
[0054] (6) In the imaging analysis device described in paragraph 1, the analysis unit may include a microregion number counting unit that determines the number of microregions contained in each cluster after cluster analysis, and the display processing unit may display information on the number of microregions for each cluster together with the segmentation image.
[0055] (Item 12) In the imaging analysis method described in Item 7, the analysis step can include a microregion number counting step for determining the number of microregions contained in each cluster after cluster analysis, and the display step can display information on the number of microregions for each cluster together with the segmentation image.
[0056] According to the imaging analysis device described in paragraph 6 and the imaging analysis method described in paragraph 12, the user can visually grasp the degree of distribution of each cluster.
[0057] 1...imaging mass spectrometry section 2...data processing section 21...data storage section 22...data matrix creation section 23...nonlinear dimensionality reduction processing section 24...cluster analysis condition setting section 25...cluster analysis section 26...analysis result processing section 27...segmentation image creation section 28...display processing section 3...input section 4...display section 100...analysis result display screen 101...segmentation image display area corresponding to current number of clusters 102...data point number list table area corresponding to current number of clusters 103...segmentation image display area corresponding to number of clusters before change 104...data point number list table area corresponding to number of clusters before change 105...cluster information display area 106...cluster change instruction input box 107...analysis execution button
Claims
1. A measurement unit that performs a predetermined analysis on each of multiple minute regions set within the measurement area on the target sample, An analysis unit performs cluster analysis to classify multiple micro-regions within the measurement area into multiple clusters based on the data for each micro-region obtained by the measurement unit, under a set number of clusters. An image creation unit that creates a segmentation image corresponding to the measurement area based on the results of the cluster analysis, For a single target sample, multiple segmentation images obtained by the analysis unit and the image creation unit under different cluster counts are arranged and displayed on the same screen of the display unit, comprising: an input field for the user to specify the number of clusters; an operator for instructing the analysis unit to perform cluster reanalysis under the number of clusters entered in the input field; and a display processing unit that arranges and displays the segmentation images before and after the cluster reanalysis on the same screen of the display unit. An imaging analyzer equipped with the following features.
2. The imaging analyzer according to claim 1, wherein the predetermined analysis is mass spectrometry.
3. The imaging analyzer according to claim 2, wherein the analysis unit includes a dimensionality reduction processing unit that performs processing on the mass spectrometry data for each minute region obtained by the measurement unit using a nonlinear dimensionality reduction method utilizing manifold learning to obtain dimensionality-compressed data for each minute region, reducing the dimension corresponding to the number of mass-to-charge ratio values to three dimensions, and performs cluster analysis on the dimensionality-compressed data for each minute region.
4. The imaging analyzer according to claim 3, wherein the image creation unit determines the display color corresponding to the cluster in the segmentation image based on the position of the centroid of each cluster obtained by cluster analysis in a three-dimensional space in which the three primary colors are assigned to each axis with respect to each dimension of the three dimensions.
5. The imaging analysis apparatus according to claim 1, wherein the analysis unit includes a micro-region counting unit that determines the number of micro-regions included in each cluster after cluster analysis, and the display processing unit displays information regarding the number of micro-regions for each cluster together with the segmentation image.
6. A measurement step in which a predetermined analysis is performed on each of multiple microregions set within the measurement area on the target sample, An analysis step in which, based on the data for each minute region obtained in the measurement step, cluster analysis is performed to classify multiple minute regions within the measurement region into multiple clusters under a set number of clusters, An image creation step to create a segmentation image corresponding to the measurement area based on the results of the cluster analysis, A step of displaying multiple segmentation images obtained in the analysis step and the image creation step for a single target sample, each with a different number of clusters, arranged on the same screen of the display unit, comprising: an input field for the user to specify the number of clusters; an operator for instructing the execution of cluster reanalysis in the analysis step with the number of clusters entered in the input field; and a display step of displaying the segmentation images before and after the execution of the cluster reanalysis, arranged on the same screen of the display unit. An imaging analysis method having [a certain characteristic].
7. The imaging analysis method according to claim 6, wherein the predetermined analysis is mass spectrometry.
8. The imaging analysis method according to claim 7, wherein the analysis step includes a dimensionality reduction step in which the mass spectrometry data for each microregion obtained in the measurement step is processed by a nonlinear dimensionality reduction method using manifold learning to obtain dimensionality-reduced data for each microregion in which the dimension corresponding to the number of mass-to-charge ratio values is reduced to three dimensions, and cluster analysis is performed on the dimensionality-reduced data for each microregion.
9. The imaging analysis method according to claim 8, wherein in the image creation step, in a three-dimensional space in which the three primary colors are assigned to each axis with respect to each dimension of the three dimensions, the display color corresponding to that cluster in the segmentation image is determined based on the position of the centroid of each cluster obtained by cluster analysis.
10. The imaging analysis method according to claim 6, wherein the analysis step includes a micro-region counting step that determines the number of micro-regions included in each cluster after cluster analysis, and the display step displays information regarding the number of micro-regions for each cluster together with the segmentation image.