Cell information acquisition method, cell manufacturing method, cell inspection method, cell abnormality detection method, cell information acquisition device, and program

Abstract

To provide a cell information acquisition method which allows a cell type or a cell state to be evaluated with regard to the whole cell area.SOLUTION: A cell information acquisition method includes: an image acquisition step for acquiring a cell image containing an image of a cell mass; a cell area extraction step for extracting a cell area indicating the area of the cell mass, from the cell image; a first feature amount distribution acquisition step for calculating a texture feature amount as a first feature amount for a pixel in the cell area, and for acquiring the distribution of the first feature amount in the cell area; an area division step for dividing the cell area into two or more division areas; a second feature amount calculation step for calculating a statistical quantity of the first feature amount for the division areas respectively, and for calculating a second feature amount indicating a relation of a statistical quantity of the first feature amount between the division areas; and a cell information acquisition step for acquiring information related to a cell type or information related to a cell state on the basis of the second feature amount.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The disclosure of this specification relates to a cell information acquiring method, a cell manufacturing method, a cell inspection method, a cell abnormality detection method, a cell information acquiring device, and a program. [Background technology]

[0002] In recent years, in the field of cell culture, it is expected that by analyzing images of the cells being cultured, information about the cells, such as the cell line type and colony state, can be obtained and used for quantitative analysis of the cell state and for testing of cell culture. A specific method of cell image analysis that has been proposed involves extracting image features such as colony size, shape, and texture characteristics from images of the colony, and using these for statistical analysis and machine learning. For example, Patent Document 1 discloses a technique for calculating local texture features in a cellular region. In addition, it is known that the characteristics of cells in a colony vary within the colony region, so it is important to calculate not only the characteristics of the entire cell region, but also the characteristics that reflect the differences in characteristics depending on the region. Patent Document 2 discloses a technique for dividing an area into partial areas and calculating a feature amount for each partial area. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 5941674 [Patent Document 2] JP 2016-007270 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, while the technology described in Patent Document 2 can evaluate the differences in characteristics of each partial region using feature values, the differences in feature values ​​between partial regions mean that it cannot be said to be a sufficient image analysis method for evaluating the cell type and cell state for the entire cellular region.

[0005] In view of the above problems, an object of the present invention is to provide a method for acquiring cell information that makes it possible to evaluate the cell type and state of cells over an entire cellular region. [Means for solving the problem]

[0006] A cell information acquisition method according to one aspect of the disclosure of this specification is characterized by comprising an image acquisition step of acquiring a cell image including an image of a cell cluster, a cell region extraction step of extracting a cell region indicating the region of the cell cluster from the cell image, a first feature distribution acquisition step of calculating a texture feature as a first feature for pixels in the cell region and acquiring a distribution of the first feature in the cell region, a region division step of dividing the cell region into two or more partitioned regions, a second feature calculation step of calculating statistics of the first feature for each of the partitioned regions and calculating a second feature indicating a relationship between the statistics of the first feature between the partitioned regions, and a cell information acquisition step of acquiring information regarding a cell type or information regarding a cell state based on the second feature.

[0007] Furthermore, a cell manufacturing method according to another aspect of the disclosure of this specification is characterized by comprising a cell status information acquisition step of acquiring information regarding the state of the cell using the above-mentioned cell information acquisition method, and a cell processing step of performing at least one selected from the group consisting of cell sorting, cell removal, and transition to the next step of cell manufacturing based on the information regarding the state of the cell acquired in the cell status information acquisition step.

[0008] Furthermore, a cell inspection method according to another aspect of the disclosure of this specification is characterized by having a cell status information acquisition step of acquiring information regarding the state of the cell using the above-mentioned cell information acquisition method, and a cell inspection step of inspecting the cell based on the information regarding the state of the cell acquired in the cell status information acquisition step.

[0009] a first feature amount distribution acquisition step of calculating a texture feature amount as a first feature amount for pixels in the cell region and acquiring a distribution of the first feature amount in the cell region; a region division step of dividing the cell region into two or more partitioned regions; a second feature amount calculation step of calculating statistics of the first feature amount for each of the partitioned regions and calculating a second feature amount indicating a relationship between the statistics of the first feature amount between the partitioned regions; a trained model acquisition step of acquiring a trained model that has been machine-learned to determine the probability that the second feature amount may occur by using a data group of the second feature amount obtained from a plurality of the cell images prepared for machine learning; an abnormality score calculation step of inputting the second feature amount obtained from the cell image to be evaluated into the trained model and calculating an abnormality score based on the output probability; and a detection step of detecting an abnormality in the cell mass based on the abnormality score.

[0010] Furthermore, a cell information acquisition device according to another aspect of the disclosure of this specification is characterized in having an image acquisition unit that acquires a cell image including an image of a cell cluster, a cell region extraction unit that extracts a cell region indicating the region of the cell cluster from the cell image, a first feature distribution acquisition unit that calculates a texture feature as a first feature for pixels in the cell region and acquires a distribution of the first feature in the cell region, a region partitioning unit that partitions the cell region into two or more partitioned regions, a second feature calculation unit that calculates statistics of the first feature for each of the partitioned regions and calculates a second feature indicating a relationship between the statistics of the first feature between the partitioned regions, and a cell information acquisition unit that acquires information regarding a cell type or information regarding a cell state based on the second feature.

[0011] A program according to another aspect of the disclosure of this specification is a program for causing a computer to execute the above-mentioned cell information acquisition method or the above-mentioned anomaly detection method. Effect of the Invention

[0012] According to the disclosure of the present specification, a method for acquiring cell information is provided that makes it possible to evaluate the cell type and cell state for an entire cellular region. [Brief description of the drawings]

[0013] [Figure 1] 1 is a functional block diagram of a cell information acquisition device according to a first embodiment. FIG. [Diagram 2] FIG. 2 is a flow diagram of a cell information acquisition method according to the first embodiment. [Diagram 3] FIG. 1 is a flow diagram showing a method for extracting a cell region. [Figure 4] FIG. 13 is a diagram showing an example of an extracted cell region. [Diagram 5] FIG. 13 is a functional block diagram of a cell information acquisition device according to a third embodiment. [Figure 6] FIG. 13 is a flowchart of a cell information acquisition method according to a third embodiment. [Figure 7]FIG. 13 is a functional block diagram of a cellular abnormality detection device according to a fourth embodiment. [Figure 8] FIG. 13 is a flowchart of a method for detecting a cellular abnormality according to a fourth embodiment. [Figure 9] FIG. 2 is a block diagram showing a hardware configuration of the information processing device. [Figure 10] FIG. 1 shows the results obtained in Example 1, where (a) is a box plot showing the distribution of the second feature by strain, and (b) is a diagram showing the output results of the F value and P value in the analysis of variance. [Figure 11] FIG. 13 shows the results obtained in Comparative Example 1, where (a) is a box plot showing the distribution of the average value of texture features across the entire cell region for each strain, and (b) is a diagram showing the output results of F and P values ​​in analysis of variance. [Figure 12] FIG. 11 is a scatter plot showing the analysis results by the principal component analysis obtained in Example 2. [Figure 13] FIG. 11 is a scatter diagram showing the analysis results by principal component analysis obtained in Comparative Example 2. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, the embodiments will be described in detail with reference to the attached drawings. Note that the following embodiments do not limit the invention according to the claims. Although the embodiments describe a number of features, not all of these features are essential to the invention, and the features may be combined in any manner. Furthermore, in the attached drawings, the same reference numbers are used for the same or similar configurations, and duplicated descriptions are omitted.

[0015] <First embodiment> In this embodiment, a method for dividing a cell region and calculating new feature values ​​based on the division results will be described using stem cells as a suitable example of cells that form a cell colony (hereinafter simply referred to as a colony) that is the basis for extracting a cell region. Examples of stem cells that can be used include iPS cells (induced pluripotent stem cells), which are pluripotent stem cells, ES cells (embryonic stem cells), and mesenchymal stem cells, which are tissue stem cells. Note that a colony in this embodiment refers to a group of cells that exist in contact with each other (hereinafter simply referred to as a cell group), and refers to a cell group that exists independently in a culture vessel. In other words, a cell group that is not attached to other surrounding cell groups is regarded as one colony.

[0016] FIG. 1 shows an example of the configuration of an apparatus for carrying out the cell information acquiring method according to the first embodiment. The cell information acquisition device 1 has an image acquisition unit 11, a cell area extraction unit 12, a first feature distribution acquisition unit 13, an area division unit 14, a second feature calculation unit 15, a cell information acquisition unit 16, an input unit 17, an output unit 18 and a memory unit 19.

[0017] FIG. 2 shows a flow diagram of the cell information acquiring method according to the first embodiment. (Step S100: Acquire cell image) Step S100 is an image acquiring step in which an image of a cell cluster is acquired by the image acquiring unit 11. In this embodiment, the cell cluster is a colony formed by culturing stem cells. In step S100, the image acquisition unit 11 acquires cell image data captured by the cell culture observation device. Here, the cell image is image data showing the morphology of the cell. The image data showing the morphology of the cell is, for example, phase contrast image data captured by a phase contrast microscope.

[0018] The cell image acquired in this step is not limited to phase contrast image data, and may be an image having a feature that the contrast of the cell outline is enhanced relative to the non-cellular region. For example, an image captured in a differential interference optical system, an image captured in an oblique illumination system, an image captured in an optical system in which the object side and the image side are telecentric (bilateral telecentric optical system), a defocused image, etc. can be used as the cell image. The cell information acquisition method according to this embodiment can be suitably applied to the images exemplified above. Note that a defocused image refers to a bright-field image captured at a certain distance from the focal position in the optical axis direction.

[0019] In addition, the cell image is preferably taken at a time when a colony of sufficient size has been formed. For example, in the case of cultured cells in a maintenance culture process of iPS cells, the image is taken on the seventh day after seeding. The cell image data may be acquired in real time from the cell culture observation device, or may be acquired from an external storage area such as the storage unit 19 or cloud storage.

[0020] (Step S110: Cell area extraction) Step S110 is a cell region extraction step in which the cell region extraction unit 12 extracts a cell region indicating a cell cluster region from the cell image acquired in step S100. The cell region in the phase contrast image can be extracted using a known image analysis technique based on image processing or machine learning. For example, the cell region is extracted using a differential filter and binarization processing.

[0021] 3 is a flow diagram showing a flow of extracting a cellular region by using a differential filter and binarization processing as an example of the processing in step S110. Hereinafter, a method of extracting a cellular region will be described with reference to FIG.

[0022] In step S111, a differential image is generated by applying a differential filter to the cell image. The differential image is an image in which the amount of change in luminance value between each pixel and its surrounding pixels is calculated and expressed as an image. The differential image is an image with a high amount of change in luminance value at the contours of the cell region and at the contours of the cells within the cell region.

[0023] In step S112, a binarization process is performed on the differential image generated in step S111 to extract an area having a high amount of change in luminance value in the differential image. In the binarization process, an arbitrary threshold is set, and the amount of change in luminance value of each pixel in the differential image is replaced with a value of 1 if the amount of change is equal to or greater than the threshold, and with a value of 0 if the amount of change is less than the threshold.

[0024] Here, the method of binarization is not limited to the method of setting an arbitrary threshold. For example, a method of automatically determining a threshold such as Otsu's binarization or Li's binarization may be used. When an arbitrary threshold is set, the threshold is set according to the shooting conditions of the device, such as the exposure time and focus setting. A method of determining a threshold for each pixel of an image, such as adaptive binarization, may also be used. The binarization process creates a binary image (hereinafter referred to as an edge image) in which pixel values ​​in areas with large changes in luminance value are expressed as 1 and other pixel values ​​are expressed as 0.

[0025] In step S113, a mask image of the cellular region is generated based on the edge image generated in step S112. Here, the mask image is a binary image in which the cellular region is represented by a pixel value of 1 and the other regions by a pixel value of 0. The mask image is generated by extracting regions in the edge image where pixel values ​​of 1 are connected, and replacing the pixel values ​​inside each connected region by 1.

[0026] In step S114, a labeling process is performed on the mask image generated in step S113 to generate a labeled image in which each cell region is identified. In the labeling process, an area in which pixel values ​​of 1 are connected is extracted from the mask image, and the pixels in the area are replaced with different pixel values ​​for each connected area.

[0027] It is also possible to set a threshold value for the size of the cell region and exclude those below the threshold. For example, in the case of stem cells, an area of ​​10,000 μm 2 is set as the threshold. When labeling each connected component of the mask image, the area of ​​the connected region is calculated, and if the area is equal to or smaller than the threshold, the pixel value of the connected region is set to 0. As described above, by carrying out the processes from step S111 to step S114, a cell region is extracted from a cell image.

[0028] Fig. 4 shows an example of cell regions extracted from a cell image 401. The cell image 401 has three cell regions 410, 420, and 430. The mask image 402 is an image generated in step S113, with white indicating the background region and black indicating the cell regions. By performing a labeling process in step S114 on the mask image, isolated individual cell regions such as cell regions T01, T02, and T03 shown in Fig. 4 can be identified and the cell regions can be extracted.

[0029] Here, if the cell region is in contact with the outer periphery of the culture vessel or the imaging field of view, it is preferable to exclude the cell region. Whether or not the cell region is in contact with the outer periphery of the culture vessel is determined by setting a contour representing the culture vessel in advance or extracting the contour of the culture vessel using a method such as Hough transform, and determining whether or not the cell region overlaps the contour. Similarly, whether or not the cell region is in contact with the outer periphery of the imaging field of view is determined by determining whether or not the cell region overlaps the outer periphery of the captured image.

[0030] In this step, a mask image capable of identifying individual independent cell regions by labeling processing and contour coordinate information corresponding to each individual cell region are obtained as information indicating a plurality of cell regions.

[0031] (Step S120: Obtaining the first feature distribution) Step S120 is a first feature distribution acquisition step in which the first feature distribution acquisition unit 13 calculates texture features as first features for pixels in a cellular region and acquires the distribution of the first features in the cellular region.

[0032] The distribution of the first feature can be calculated by calculating the texture feature of a small region centered on an arbitrary pixel position in a cell region in a cell image, and repeating the process of mapping the texture feature to the pixel position for all pixels in the cell region. Note that the texture feature as the first feature does not necessarily have to be mapped to all pixels, but it is preferable to map the texture feature to a higher proportion (density) of pixels to which the texture feature is mapped, since this results in a more accurate result.

[0033] The texture feature is a scalar value calculated based on the luminance distribution of an image, and for example, contrast is calculated by a texture analysis method using GLCM (Gray Level Co-Occurrence Matrix). Contrast is an example of a texture feature calculated by a texture analysis method using GLCM, and is not limited to contrast, and may be a feature such as entropy, uniformity, or non-uniformity.

[0034] The size of the small region for calculating the texture feature is preferably an area that contains a certain number of cells. For example, if the diameter of one cell in the acquired cell image is equivalent to 8 pixels, the small region is a square area with one side of 128 pixels. In this step, the distribution of the first feature amount corresponding to each cellular region acquired in step S110 is acquired.

[0035] (Step S130: Area division) Step S130 is a region dividing step in which the region dividing unit 14 divides the cellular region into two or more divided regions. In colony-forming cells such as iPS cells, multiple regions formed by cell groups with different morphological or functional characteristics may occur within the colony region (hereinafter, this phenomenon is referred to as domain formation). For example, in a colony of grown pluripotent stem cells, the cell density in the inner region of the colony may be higher than that in the outer region of the colony. In addition, for example, in staining observation with an undifferentiated marker, strong marker expression may be observed in the outer region of the colony. In addition, domain formation may be observed, for example, when cells that have differentiated from stem cells partially occur within the colony due to poor culture, or when a colony is formed derived from multiple cells. Therefore, in this step, it is desirable to divide the colony regions so that the differences in characteristics within the colony regions are reflected.

[0036] In this embodiment, in step S130, each cellular region extracted in step S110 is divided into two regions, an inner region and an outer region, based on the area of ​​the cellular region. Specifically, for each cellular region obtained in step S110, a contraction process is performed on the cellular region to divide the cellular region into an inner region and an outer region. At this time, the contraction process is performed so that the area ratio between the inner region and the outer region becomes a predetermined ratio. For example, the contraction process is performed so that the area ratio between the inner region and the outer region becomes 1:1.

[0037] The cell region may be divided into two regions, an inner region and an outer region, based on the distance image. The distance image is an image in which each pixel value in the cell region is replaced with a calculated value of the distance from the center of gravity of the cell region to the contour. Specifically, a distance image is first generated for each cell region acquired in step S110. At this time, the calculated distance value is normalized so that the maximum value is 1.0 and the minimum value is 0.0 in each cell region. A threshold is set for the pixel values ​​of each generated distance image, and values ​​less than the threshold are defined as the inner region and values ​​equal to or greater than the threshold are defined as the outer region, thereby dividing the cell region into two regions, an inner region and an outer region. The threshold is set to, for example, 0.5. In this step, a segmented region is created for each cellular region acquired in step S110.

[0038] (Step S140: Calculation of second feature amount) Step S140 is a second feature calculation step in which the second feature calculation unit 15 calculates statistics of the first feature for each of the partitioned areas, and calculates second feature indicating a relationship between the statistics of the first feature between the partitioned areas. Specifically, for each cell region, the statistics of the first feature amount in the inner region and the statistics of the first feature amount in the outer region divided in step S130 are calculated, and the ratio between them is calculated as the second feature amount. The statistics is calculated, for example, as an average value.

[0039] The statistic is not limited to the average value, and may be any of other types of statistics such as the median, standard deviation, and coefficient of variation (CV value), or a feature vector corresponding to one cell region may be obtained by calculating a plurality of types of second feature amounts using a plurality of types of statistics. Hereinafter, even a feature vector in which a plurality of types of second feature amounts are calculated will be simply referred to as a second feature amount. In this step, a second feature amount expressing the degree of change in luminance characteristics between the inner region and the outer region is further calculated for each cellular region.

[0040] (Step S150: Acquire cell information) Step S150 is a cell-information acquiring step in which the cell-information acquiring unit 16 acquires information regarding the cell type or the state of the cell based on the second feature amount. The information regarding the cell type acquired in the information acquisition process of step S150 is not limited to biological classification of cell types, but includes, for example, information regarding strain types (cells with different origins, production lots, clones, etc.) of the same cell type.

[0041] Information on the state of cells includes information on cell quality and disease such as viability, proliferation, degree of undifferentiation, remaining state of exogenous genes and reprogramming factors for iPS cell generation, presence or absence of karyotype or genomic abnormalities, production efficiency of useful substances, presence or absence of cancer, etc. Furthermore, in cell manufacturing, information on the state of cells also includes information on the suitability of moving to the next process such as medium replacement, cell recovery, and differentiation induction. It is preferable to obtain information regarding these cell types or cell states in combination with a comparison of different observations of cultured cells according to the second embodiment described below, or with analysis using a trained model according to the third embodiment.

[0042] In this embodiment, an example is described in which the number of cellular regions extracted in the cellular region extraction process is three or more, and the cellular information acquisition unit 16 acquires a clustering result in which the cellular regions are clustered into two or more clusters. In step S150, the cellular information acquisition unit 16 acquires the analysis result by unsupervised clustering using the second feature amount corresponding to each of the multiple cellular regions acquired in step S140, as information on the state of the cells.

[0043] The unsupervised clustering method uses, for example, KMeans with the number of clusters set to 2. Note that the number of clusters may be set by the user at any value, or the number of likely clusters may be automatically determined based on an evaluation index for the clustering results, such as AIC or BIC. Furthermore, the unsupervised clustering method is not limited to KMeans, and other clustering methods such as hierarchical clustering or a mixture of Gaussian distributions may be used. Clustering may be performed using a plurality of methods, and the results of each may be obtained.

[0044] In this step, each cellular region is classified based on the second feature amount, and information indicating the classification result of each cellular region is obtained as cellular information. The analysis result may be output as a graph in which the second feature amount of each cell region is color-coded according to the classification result, or may be displayed in a format in which the contour line indicating the cell region is color-coded according to the classification result and superimposed on the cell image acquired in step S100.

[0045] According to the cell information acquisition method of this embodiment, the colony area is divided and new feature amounts are calculated based on the division results, making it possible to acquire more appropriate analysis results. This embodiment can also be suitably applied to a case where a plurality of cell images are acquired, such as a case where a plurality of cell images are taken in a plurality of fields of view in observing cultured cells. In this case, steps S110 to S140 are applied to each cell image, and the second feature amount in each cell region of each cell image is calculated, and then analysis is performed by step S150.

[0046] (Modification 1 of the first embodiment: Texture feature quantity by frequency analysis) In the first embodiment, an example was shown in which contrast was calculated as the texture feature in step S120 by a texture analysis method using GLCM, but the texture feature to be calculated is not limited to the method using GLCM. For example, the texture feature in step S120 may be a feature calculated by a method of analyzing spatial frequency characteristics. For example, after converting a small region of interest into spatial frequency by Fourier analysis, the average value of the power spectrum in a specific frequency band is set as the texture feature. The specific frequency band is preferably a band that includes a frequency corresponding to the cell diameter size, and for example, a frequency band corresponding to 12 to 24 μm is set. The feature quantity obtained by the method of analyzing spatial frequency characteristics can be suitably used for cell types that form colonies having a luminance distribution with a periodic pattern, such as iPS cell colonies, for example. The process from step S130 onwards is similar to that in the first embodiment.

[0047] It is also possible to calculate both the texture feature distributions by GLCM and frequency analysis. In that case, in step S140, the second feature is calculated using each of the texture feature distributions, and the feature vector obtained by integrating these is defined as the second feature. According to the cell information acquisition method of this modified example, it is possible to obtain more appropriate analysis results when the colony is a cultured cell in which domains are formed by cells having different morphological characteristics within the colony.

[0048] (Modification 2 of the First Embodiment: Region Segmentation Using Feature Distribution) In the first embodiment, a method for dividing a cell region into two regions, an inner region and an outer region, based on the area of ​​the cell region or a distance image in step S130 has been described. The method for dividing the region in step S130 is not limited to this, and the region may be divided in step S130 based on the texture feature distribution acquired in step S120. Specifically, a threshold is set for the texture feature acquired in step S120, and the texture feature distribution is divided into a region below the threshold and a region equal to or greater than the threshold. The threshold is set to 0.5, for example. Here, the texture feature in the texture feature distribution is min-max normalized to be in the range of 0.0 to 1.0.

[0049] Alternatively, regions may be divided by using texture features corresponding to each pixel position in the texture feature distribution as input data for an unsupervised clustering method. For example, KMeans with the number of clusters set to 2 may be used. The process from step S140 onwards is similar to that in the first embodiment. According to the cell information acquisition method of this modified example, it is possible to acquire appropriate analysis results by calculating new features based on the texture feature distribution and the results of dividing the cell region.

[0050] (Modification 3 of the First Embodiment: Area Division into Three or More Areas) In the first embodiment or the second variant of the first embodiment, an example has been described in which the cell region is divided into two regions and the second feature amount is calculated, but the cell region may be divided into three or more regions and the second feature amount may be calculated. Specifically, two thresholds a and b are set for the distance image in the first embodiment, and the image is divided into three regions. For example, the thresholds a and b are set to 0.25 and 0.5, respectively, and the image is divided into three regions: a region less than a, a region between a and b, and a region greater than or equal to b.

[0051] Furthermore, as described in the second modification of the first embodiment, when dividing regions based on the distribution of texture features, the number of clusters in the division of regions by unsupervised clustering may be set to three or more, thereby dividing the regions. In the first embodiment, in step S140, the ratio of the statistics of the texture features in two divided regions is calculated as the second feature, but when the region is divided into three or more regions, the following procedure may be followed. That is, an index indicating the variation of the statistics of the texture features in each region is calculated, and the second feature is calculated. The index indicating the variation is, for example, the variance, standard deviation, or coefficient of variation (CV value). In addition, the second feature may be a feature vector, which is a set of scalar values ​​obtained by calculating the ratio of statistics of texture features for every possible combination of two different regions out of the three or more regions divided into the region. The processing from step S150 onwards is similar to that in the first embodiment, and therefore a description thereof will be omitted.

[0052] This modified example can be suitably applied to the analysis of cultured cells such as iPS cells, in which multiple cell states, such as mature, differentiated, and undifferentiated states, appear in cell images and which form complex domains.

[0053] (Modification 4 of the First Embodiment: Use of Morphological Features for Analysis) In step S150, information on the cell type or cell state is obtained by analysis using the second feature amount. However, the morphological feature amount of each cell region may be further calculated and used in the analysis. Specifically, the cellular information acquisition unit 16 calculates morphological features from each cellular region extracted in step S110, and uses the feature vector obtained by integrating the morphological features with the second feature calculated in step S140 for analysis. When integrating the morphological features with the second feature, the cellular information acquisition unit 16 may acquire the results of principal component analysis. Here, the morphological feature amount is, for example, any one of a scalar value of area, diameter, circularity, perimeter, area envelopment ratio, and perimeter envelopment ratio, or a vector obtained by calculating a plurality of values. According to the cell information acquisition method of this modified example, it is possible to acquire appropriate analysis results even in the analysis of cultured cells in which significant differences in the morphological characteristics of cell regions are observed.

[0054] (Modification 5 of the First Embodiment: Principal Component Analysis) In step S150, an example of outputting the analysis result by the unsupervised clustering method has been described. In step S150, when the feature used in the analysis is a two-dimensional or higher dimensional feature vector, the cell information acquisition unit 16 may acquire the result of analyzing the two-dimensional or higher dimensional feature vector including the second feature by principal component analysis. In this case, the first feature amount calculated in the first feature amount distribution acquisition step may be of two or more types, and the second feature amount calculation step may calculate a second feature amount for each of the first feature amounts. Also, as described in Modification 4, a morphological feature amount of each cell region may be calculated and integrated with the second feature amount for use.

[0055] Principal component analysis is a statistical analysis method that aggregates data with many variables to create principal components, making it possible to represent data represented by many variables (features) using fewer variables. Principal component analysis outputs principal components, which are elements that represent the characteristics of the data, principal component scores, which indicate the numerical values ​​in the principal component space of the feature vector of the input data, and principal component loadings, which indicate the degree to which the elements of the original feature vector contribute to each principal component. When performing analysis by principal component analysis, for example, a two-dimensional scatter plot is created and output, with the horizontal axis representing the principal component score in the first principal component and the vertical axis representing the principal component score in the second principal component. Furthermore, the principal component loadings may also be displayed.

[0056] With this modification, when the features used in the analysis are two- or higher-dimensional feature vectors, the user can grasp which elements of the feature vector are important from the principal component loadings, while at the same time obtaining an overview of the characteristics of the cellular region from the scatter plot. According to the cell information acquisition method of this modified example, even when the feature amounts used in the analysis are three- or higher-dimensional feature amount vectors, it is possible to acquire analysis results that the user can easily grasp. It is also possible to further perform unsupervised clustering in the first embodiment using the principal component scores of each cell region obtained by this modification as an input.

[0057] Second Embodiment In the first embodiment, a method for acquiring cell images in arbitrary cultured cell observation and acquiring information on the states of a plurality of cell regions has been described. In the second embodiment, a method for acquiring a plurality of mutually different cell images and acquiring information on the cell type or cell state will be described. Here, the multiple cell images that are different from one another are, for example, cell images acquired during the culture of multiple cell types that are different from one another, and the multiple cell types that are different from one another are, for example, iPS cell strains.

[0058] In the second embodiment, an example is described in which cell images obtained during the culture of iPS cells of different strains are used as the multiple different cell images, but the present embodiment is not limited thereto. For example, the present embodiment can be suitably applied to cell images obtained during the culture using different culture vessels or cell images obtained at different imaging times.

[0059] In step S100, the image acquisition unit 11 acquires two or more different cell images in response to a user request input from the input unit 17. For example, cell images of two types of iPS cell lines C01 and C02 are acquired. The cell images are images showing the morphology of the cells, as in the first embodiment. Thereafter, the processes in steps S110 to S140 are applied to the cell images of each of the strains C01 and C02, thereby acquiring a data group D01 of the second feature amount corresponding to a plurality of cell regions in the strain C01 and a data group D02 of the second feature amount corresponding to a plurality of cell regions in the strain C02. The process contents from step S110 to step S140 are the same as those in the first embodiment, and therefore the description thereof will be omitted.

[0060] Here, the data of the cell region and the second feature amount data corresponding to the cell region are assigned an identification ID so that it is possible to know from which cell image the data was acquired. In this embodiment, the cell information acquisition process in step S150 includes statistically analyzing the difference between the second feature amounts calculated from each of two or more cell images. Specifically, in step S150, the data groups D01 and D02 are used to perform an analysis using a statistical test method, and the analysis result is output. The statistical test method is, for example, a t-test, and a p-value from the t-test is output as the analysis result. The statistical test method is not limited to the t-test, and methods such as an F-test or, in the case of three or more data groups, analysis of variance can be used.

[0061] In addition to the analysis results of the statistical testing method, information showing the statistics of the data group or graphs such as box plots and histograms showing the data distribution may be output. When the second feature is a feature vector, an analysis result may be calculated for each element of the feature vector and the result may be output. As a result, the user can quantitatively grasp the degree of difference between cultured cells due to strain types such as strain C01 and strain C02.

[0062] As described above, according to the cell information acquisition method of the second embodiment, when at least two or more mutually different cell images are acquired, the user can quantitatively grasp the degree of difference between the cells captured in the cell images. In the second embodiment, an example of acquiring information about differences in cell strains has been described, but the information acquired from the differences between the second feature amounts analyzed in the cell information acquiring process of step S150 in this embodiment is not limited to this. For example, it is also possible to analyze two or more cell images acquired in the process of culturing cells of the same strain under different culture conditions, and evaluate the influence of differences in culture conditions on the cell culture.

[0063] It is also preferable to obtain control cell images in advance using appropriately controlled cultured cell samples. The degree of difference compared to the control cell image can provide information that can be used to determine whether cross-contamination or sample mix-ups have occurred in the cell manufacturing facility. It can also provide information that can be used to determine whether the cells have been cultured while maintaining the expected state, and whether it is appropriate to move on to the next process of cell manufacturing (medium replacement, cell recovery, differentiation induction, etc.).

[0064] (Modification 1 of the second embodiment: Principal component analysis) In the second embodiment, an example has been described in which an analysis result based on a statistical testing method is output in step S150. However, as described in the fifth modified example of the first embodiment, when the feature values ​​used in the analysis are two- or higher-dimensional feature vectors, the analysis may be based on principal component analysis. When performing analysis using principal component analysis, for example, a two-dimensional scatter plot is created and output, with the horizontal axis representing the principal component score in the first principal component and the vertical axis representing the principal component score in the second principal component, and color-coded according to the data group. Furthermore, the principal component loadings may also be displayed. Also, one data group may be regarded as one cluster, and the results of evaluating the cohesion within the cluster and the discreteness between the clusters may be output. This enables the user to grasp which elements of the feature vector are important from the principal component loadings, while at the same time obtaining an overview of the characteristics of each cell region and data group from the scatter plot. According to the cell information acquisition method of this modified example, even when the features used in the analysis are two- or higher-dimensional feature vectors, it is possible to acquire analysis results that the user can easily grasp.

[0065] <Third embodiment> In the third embodiment, a method for performing analysis using a trained model that uses the second feature as input and obtaining information regarding the cell type or cell state is described. FIG. 5 shows an example of the configuration of an apparatus for carrying out the cell information acquiring method according to the third embodiment. The cell information acquiring device 3 includes the same components as the cell information acquiring device 1, and further includes a trained model acquiring unit 31.

[0066] FIG. 6 shows a flow diagram of a cell information acquiring method according to the third embodiment. The process contents from step S100 to step S140 are the same as those in the first embodiment, and therefore the description thereof will be omitted. As described in the first embodiment or the first modification of the first embodiment, the second feature may be a feature vector obtained by integrating features calculated from a texture feature distribution based on an analysis method using the GLCM or a plurality of texture feature distributions based on frequency analysis. Furthermore, as described in the fourth modification of the first embodiment, the morphological feature amount of the cellular region may be integrated with the second feature amount.

[0067] In the trained model acquisition process of step S300, the trained model acquisition unit 31 acquires a trained model through machine learning using the second feature amount and information on the cell type or information on the cell state. Here, the trained model is a machine learning model optimized to input second feature data corresponding to a cell region and output a classification result indicating the cell type or cell state for each cell region. For example, a support vector machine (SVM) can be used as the machine learning model. Note that the machine learning model is not limited to SVM, and models such as random forest and neural network can be used.

[0068] The classification result indicating the cell type or state of the cells is, for example, the result of classifying whether a colony in an iPS cell culture is in an undifferentiated state or in a differentiated state. Note that the classification is not limited to the result of classifying whether a colony is in an undifferentiated state or in a differentiated state, and may be the result of classifying whether a colony has cells with abnormal karyotypes.

[0069] The cell information acquisition step in step S150 includes using the trained model acquired in step S300. That is, the cell information acquisition unit 16 acquires information on the cell type or the cell state by inputting the second feature calculated in step S140 to the trained model. When there are multiple trained models, results may be acquired from each trained model.

[0070] Note that the trained model acquisition process in step S300 may be performed before or after any of steps S100 to S140, so long as it is performed before step S150. In step S150, information on whether the colonies in the iPS cell culture are in an undifferentiated state or a differentiated state is output as cell information. The user can check information relating to the cell type or cell state of each cellular region by outputting and checking the information acquired in step S150 from the output unit 18. As described above, according to the cell information acquisition method of the third embodiment, it is possible to acquire information regarding the cell type or cell state using a trained model with the second feature as input.

[0071] (Fourth embodiment) In the third embodiment, an example has been described in which a trained model that outputs a classification result indicating a cell type or a cell state is acquired in step S300, but a trained model for anomaly detection may also be acquired. In this embodiment, an example of a method for detecting an abnormality in a cell using a trained model for detecting an abnormality will be described.

[0072] FIG. 7 shows an example of the configuration of an apparatus for carrying out the method for detecting a cellular abnormality according to the fourth embodiment. The cellular abnormality detection device 4 has an abnormality score calculation unit 41 and a detection unit 42 instead of the cellular information acquisition unit 16 in the cellular information acquisition device 3.

[0073] FIG. 8 shows a flow diagram of a cellular abnormality detection method according to the fourth embodiment. The process contents from step S100 to step S140 are the same as those in the first embodiment, and therefore the description thereof will be omitted.

[0074] In the trained model acquisition step of step S300, the trained model acquisition unit 31 acquires a trained model. Here, the trained model is a trained model that has been machine-learned to determine the probability that the second feature will occur, using a data group of the second feature obtained from a plurality of cell images prepared for machine learning. The probability that the second feature learned by the trained model in this embodiment may occur is estimated by a method such as kernel density estimation or a mixed Gaussian distribution.

[0075] In the abnormality score calculation process of step S400, the abnormality score calculation unit 41 inputs the second feature obtained from the cell image to be evaluated to the trained model, and calculates the abnormality score based on the output probability. Here, the abnormality score can be appropriately determined according to the output probability. For example, the abnormality score may be 0 if the probability is equal to or greater than a predetermined threshold, and 1 if it is less than the predetermined threshold, or the abnormality score may be the reciprocal of the output probability.

[0076] As the trained model, for example, in the case of iPS cell culture, a trained model trained using a data group of the second feature amount of a colony in an undifferentiated state as a training data group can be used. In this case, a scalar value indicating how far each input iPS colony deviates from an undifferentiated state is output as an abnormality score in the abnormality score calculation step. Note that the trained model is not limited to a trained model trained using a data group of the second feature amount of a colony in an undifferentiated state as a training data group. For example, a trained model trained using a data group of the second feature amount of a colony that does not have abnormal karyotype cells as a training data group may be used. In addition, when there are multiple trained models, the results of each of them may be obtained.

[0077] In the detection step of step S410, the detection unit 42 detects a cellular abnormality based on the abnormality score. The cellular abnormality detected by the detection unit 42 is output via the output unit 18. Note that the trained model acquisition process in step S300 may be performed before or after any of steps S100 to S140, so long as it is performed before step S150. In the present embodiment, an example is described in which a trained model is used that has been machine-learned to estimate the probability of occurrence of the second feature amount, but the combination is not limited to this. For example, a trained model that has been machine-learned to estimate the probability of occurrence of a combination of the second feature amount and another feature amount, such as the morphological feature amount described in the fourth modification of the first embodiment, can also be used.

[0078] <Program for executing the cell information acquisition method> The program according to the present invention is a program for causing a computer to execute the cell information acquisition method according to the present invention or the cell abnormality detection method according to the present invention described above.

[0079] FIG. 9 is a block diagram showing an example of the hardware configuration of an information processing device 9 that can be used as the cell information acquiring devices 1 and 3 and the cell abnormality detecting device 4. The information processing device 9 has a function of a computer. For example, the information processing device 9 may be integrated with a desktop personal computer (PC), a laptop PC, a tablet PC, a smartphone, or the like. The information processing device 9 includes a central processing unit (CPU) 91, a random access memory (RAM) 92, a read only memory (ROM) 93, and a hard disk drive (HDD) 94 to realize the functions of a computer that performs calculations and storage. The information processing device 9 also includes a communication I / F (interface) 95, an input device 96, and an output device 97. The CPU 91, the RAM 92, the ROM 93, the HDD 94, the communication I / F 95, the input device 96, and the output device 97 are connected to each other via a bus 98. The input device 96 and the output device 97 may be connected to the bus 98 via a drive device (not shown) for driving these devices.

[0080] 9, the components constituting the information processing device 9 are illustrated as an integrated device, but some of these functions may be configured by external devices. For example, the input device 96 and the output device 97 may be external devices separate from the components constituting the functions of the computer, including the CPU 91, etc.

[0081] The CPU 91 performs predetermined operations according to programs stored in the RAM 92, HDD 94, etc., and also has the function of controlling each part of the information processing device 9. The RAM 92 is composed of a volatile storage medium, and provides a temporary memory area necessary for the operation of the CPU 91. The ROM 93 is composed of a non-volatile storage medium, and stores necessary information such as programs used in the operation of the information processing device 9. The HDD 94 is composed of a non-volatile storage medium, and is a storage device that stores information regarding the number and positions of the individual independent separation compartments, fluorescent light intensity, etc.

[0082] The communication I / F 95 is a communication interface based on standards such as Wi-Fi (registered trademark) and 4G, and is a module for communicating with other devices. The output device 97 is a liquid crystal display, an OLED (Organic Light Emitting Diode) display, etc. used for displaying moving images, still images, characters, etc., or a speaker used for outputting sound, etc. The input device 96 is a button, a touch panel, a keyboard, a pointing device, etc., and is used by the user to operate the information processing device 9. The input device 96 and the output device 97 may be integrally formed as a touch panel.

[0083] The hardware configuration shown in FIG. 9 is an example, and other devices may be added, or some devices may not be provided. Some devices may be replaced with other devices having similar functions. Some functions may be provided by other devices via a network, and the functions constituting this embodiment may be distributed and realized among a plurality of devices. For example, the HDD 94 may be replaced with an SSD (Solid State Drive) using semiconductor elements such as flash memory, or may be replaced with cloud storage.

[0084] The CPU 91 loads a program stored in the ROM 93 or the like into the RAM 92 and executes it. In this way, the CPU 91 realizes the functions of the image acquisition unit 11, the cellular region extraction unit 12, the first feature distribution acquisition unit 13, the region partition unit 14, the second feature calculation unit 15, the cellular information acquisition unit 16, the trained model acquisition unit 31, the abnormality score calculation unit 41, and the detection unit 42. The CPU 91 also realizes the functions of the input unit 17 and the output unit 18 by controlling the input device 96 and the output device 97. Moreover, the CPU 91 realizes the function of the storage unit 19 by controlling the HDD 94 .

[0085] The information processing device 9 may further have a function of controlling the operations of the above-mentioned cell culture observation device, the incubator for performing cell culture, and the like in accordance with a predetermined program. The present invention can also be realized by executing the following process. That is, software (programs) for causing a computer (or a CPU, MPU, etc.) to execute the functions of the above-described embodiments is supplied to a system or device via a network or various storage media. The computer (or a CPU, MPU, etc.) of the system or device then reads and executes the program. The present invention may also be realized, for example, by a circuit (e.g., ASIC) that implements one or more functions.

[0086] <Cell manufacturing method> The cell production method according to the present invention is a method for producing cells based on the information on the state of cells acquired by the cell information acquisition method according to the present invention described above. That is, the cell production method according to the present invention includes a cell state information acquisition step of acquiring information on the state of the cells using the cell information acquisition method described above, and a cell processing step. The cell processing step is a step of performing at least one step selected from the group consisting of cell sorting, cell removal, and transition to the next step in cell production, based on the information on the state of the cells acquired in the cell state information acquisition step. For example, in a process for culturing iPS cells to maintain undifferentiation, it is possible to control the cell manufacturing method by classifying colonies as differentiated or undifferentiated and removing differentiated colonies from the culture vessel based on the third embodiment. It is also possible to control the cell manufacturing method by determining whether a colony has a karyotype abnormality and removing a colony that has a karyotype abnormality from the culture vessel.

[0087] As described in the second embodiment, a control cell culture observation is performed in advance using an appropriate cultured cell sample. After that, it is also preferable to perform a statistical test based on the second embodiment on the cell image of the cultured cell sample to be manufactured and the cell image of the above-mentioned cultured cell sample. This makes it possible to determine, for example, that cross-contamination or mix-up between samples has not occurred in the cell manufacturing facility. It is also possible to determine whether the cells have been cultured while maintaining the expected state, and whether it is appropriate to move to the next process of cell manufacturing (medium replacement, cell recovery, differentiation induction, etc.), and to control the cell manufacturing process.

[0088] <Cell examination method> The cell inspection method according to the present invention is a method for inspecting cells based on information about the state of the cells acquired by the cell information acquiring method according to the present invention described above. That is, the cell inspection method according to the present invention includes a cell state information acquisition step of acquiring information on the state of a cell using the cell state information acquisition method described above, and a cell inspection step. The cell inspection step is a step of inspecting a cell based on the information on the state of the cell acquired in the cell state information acquisition step.

[0089] For example, as described in the first embodiment, the following information on the state of a cell is obtained based on the cell information obtaining method according to the present invention. That is, information such as the viability and proliferation of a cell, the degree of undifferentiation of a cell, the remaining state of an exogenous gene or a reprogramming factor for iPS cell formation, the presence or absence of karyotype or genomic abnormality, the production efficiency of a useful substance, and the presence or absence of cancer is obtained. This information can then be used in tests such as evaluation of cell quality and judgment of the presence or absence of disease. EXAMPLES

[0090] <Example 1> FIG. 10 shows the results obtained in Example 1 in which the second embodiment was applied to the culture of an iPS cell line. In Example 1, cell images were obtained in the culture of three types of iPS cell lines A, B, and C, and the second embodiment described above was applied. In step S120, the texture feature distribution of contrast was calculated using an analysis method based on GLCM. In step S130, the region was divided into an inner region and an outer region by performing a contraction process so that the area ratio of the inner region to the outer region was 1:1. In step S140, the average value of the texture feature of the inner region and the average value of the texture feature of the outer region were calculated as the second feature. Thereafter, in step S150, a statistical test was performed using analysis of variance.

[0091] Figure 10(a) is a box plot showing the distribution of the second feature amount of each cell region by strain. Figure 10(b) is a diagram showing the output results of the F value and P value in the analysis of variance. These results allow the user to understand information such as the type of feature distribution of colonies of each strain type and the degree of difference between strain types. In addition, in obtaining cell information according to Comparative Example 1, cell images were obtained in the culture of three types of iPS cell lines as in Example 1. In obtaining cell information according to Comparative Example 1, region division in step S130 and calculation of the second feature in step S140 were not performed, and analysis of variance was performed using the average value of the texture feature over the entire cell region. Fig. 11(a) is a box plot showing the distribution of the average value of the texture feature over the entire cell region for each line. Fig. 11(b) is a diagram showing the output results of the F value and P value in the analysis of variance. It can be seen that it is difficult for the user to grasp the difference in cell state depending on the strain type in the result shown in Figure 11 compared with the result shown in Figure 10. As described above, from the comparison between Example 1 and Comparative Example 1, it can be seen that the second feature amount calculated by the present invention is a useful feature amount for distinguishing strain types.

[0092] <Example 2> In Example 2, cell images were obtained in the culture of three types of iPS cell lines A, B, and C, and the modified example 1 of the second embodiment described above was applied. Note that, for the texture feature in step S120, a plurality of texture feature distributions based on the GLCM and the frequency characteristics described in the modified example 1 of the first embodiment were calculated. Thereafter, in step S130, the region was divided into an inner region and an outer region by performing a contraction process so that the area ratio of the inner region to the outer region was 1:1. In addition, in step S140, the ratio of the statistics of the texture feature in the inner region to the statistics of the texture feature in the outer region was calculated as the second feature. The statistics used were the average value and the coefficient of variation (CV value). Furthermore, the morphological feature described in the modified example 5 of the first embodiment was integrated with the second feature and used as an input to the principal component analysis in step S150.

[0093] 12 is a scatter plot of the results obtained in Example 2, with the horizontal axis representing principal component score 1 and the vertical axis representing principal component score 2, plotting each input data. From this result, the user can grasp information such as the type of characteristic distribution of colonies of each strain type and the degree of difference between strain types.

[0094] In obtaining cell information according to Comparative Example 2, cell images were obtained in the culture of three types of iPS cell lines as in Example 2. In obtaining cell information according to Comparative Example 2, the second feature amount was not calculated in steps S130 and S140, and principal component analysis was performed using the average value of the texture feature amount over the entire cell region and the morphological feature amount.

[0095] Fig. 13 is a scatter plot of the results obtained in Comparative Example 2, with principal component score 1 on the horizontal axis and principal component score 2 on the vertical axis, plotting each input data. Compared with the results shown in Fig. 12, it can be seen that it is difficult for the user to grasp the difference in cell state depending on the strain type in Fig. 13. As described above, a comparison between Example 2 and Comparative Example 2 shows that the second feature calculated by the present invention is a useful feature for distinguishing strain types.

[0096] It should be noted that the above-mentioned embodiments and examples are merely examples of the embodiment of the present invention, and the technical scope of the present invention should not be interpreted as being limited by them. In other words, the present invention can be implemented in various forms without departing from its technical idea or main features. For example, it should be understood that an embodiment in which a part of the configuration of any of the embodiments is added to another embodiment, or an embodiment in which a part of the configuration of any of the embodiments is replaced with a part of the configuration of another embodiment, is also an embodiment to which the present invention can be applied.

[0097] Disclosure according to an embodiment of the present invention includes the following configurations and methods. (Method 1) an image acquiring step of acquiring a cell image including an image of a cell aggregate; a cell region extraction step of extracting a cell region indicating a region of the cell cluster from the cell image; a first feature amount distribution acquisition step of calculating a texture feature amount as a first feature amount for pixels in the cell region and acquiring a distribution of the first feature amount in the cell region; A region dividing step of dividing the cell region into two or more divided regions; a second feature calculation step of calculating statistics of the first feature amounts for each of the partitioned regions and calculating a second feature amount indicating a relationship between the statistics of the first feature amounts between the partitioned regions; a cell information acquiring step of acquiring information about a cell type or information about a cell state based on the second feature amount; having A method for acquiring cell information, comprising: (Method 2) The cell information acquisition method described in Method 1, wherein the region division step divides the cell region based on the area of ​​the cell region, the distance from the center of gravity of the cell region, or the distribution of the first feature amount. (Method 3) The cell information acquisition method described in Method 1, wherein the second feature calculation step includes calculating a difference or ratio of a statistical value of the first feature value between the divided regions. (Method 4) The cell information acquisition method described in Method 1, wherein the second feature calculation process includes calculating the variance, standard deviation, or coefficient of variation (CV value) of the statistics of the first feature between the divided areas. (Method 5) The number of the cell regions extracted in the cell region extraction step is 3 or more, 5. The method for acquiring cell information according to any one of Methods 1 to 4, wherein the cell information acquiring step includes acquiring a clustering result in which the cell regions are clustered into two or more clusters. (Method 6) The cell information acquisition method according to any one of methods 1 to 5, wherein the cell information acquisition step includes acquiring results of analyzing a two- or more dimensional feature vector including the second feature by principal component analysis. (Method 7) The image acquisition step includes acquiring two or more of the cell images that are different from each other, A cell information acquisition method described in any one of methods 1 to 6, wherein the cell information acquisition process includes statistically analyzing the difference between the second feature amounts calculated from each of two or more of the cell images. (Method 8) The method further includes a trained model acquisition step of acquiring a trained model that is machine-learned using the second feature amount and information on the cell type or information on the state of the cell; The cell information acquiring method according to any one of Methods 1 to 4, wherein the cell information acquiring step includes using the trained model. (Method 9) The method for obtaining cell information described in Method 8, wherein the information regarding the state of the cells includes at least one selected from the group consisting of information regarding viability, proliferation, degree of undifferentiation, remaining state of exogenous genes, presence or absence of genomic abnormalities, efficiency of production of useful substances, presence or absence of cancer, and suitability for transitioning to the next step in cell production. (Method 10) The method for acquiring cell information according to any one of Methods 1 to 9, wherein the cells constituting the cell mass are pluripotent stem cells. (Method 11) 11. The method for acquiring cell information according to any one of Methods 1 to 10, wherein the information relating to the cell type includes information indicating the strain type of the cells that constitute the cell mass. (Method 12) A cell state information acquisition step of acquiring information about the state of the cell using the cell information acquisition method according to any one of methods 1 to 11; a cell processing step of carrying out at least one step selected from the group consisting of cell sorting, cell removal, and transition to a next step in cell production based on the information on the state of the cells acquired in the cell state information acquisition step; A method for producing a cell, comprising: (Method 13) A cell state information acquisition step of acquiring information about the state of the cell using the cell information acquisition method according to any one of methods 1 to 11; a cell inspection step of inspecting cells based on the information on the state of the cells acquired in the cell state information acquisition step; A method for inspecting a cell, comprising: (Method 14) an image acquiring step of acquiring a cell image including an image of a cell aggregate; a cell region extraction step of extracting a cell region indicating a region of the cell cluster from the cell image; a first feature amount distribution acquisition step of calculating a texture feature amount as a first feature amount for pixels in the cell region and acquiring a distribution of the first feature amount in the cell region; A region dividing step of dividing the cell region into two or more divided regions; a second feature calculation step of calculating statistics of the first feature amounts for each of the partitioned regions and calculating a second feature amount indicating a relationship between the statistics of the first feature amounts between the partitioned regions; a trained model acquisition step of acquiring a trained model that machine-learns the probability of occurrence of the second feature amount by using a data group of the second feature amount obtained from a plurality of the cell images prepared for machine learning; an abnormality score calculation step of inputting the second feature amount obtained from the cell image to be evaluated to the trained model and calculating an abnormality score based on an output probability; A detection step of detecting an abnormality in the cell mass based on the abnormality score; A method for detecting an abnormality in a cell mass, comprising: (Configuration 1) an image acquisition unit for acquiring cell images including images of cell aggregates; a cell region extraction unit that extracts a cell region indicating a region of the cell cluster from the cell image; a first feature amount distribution acquisition unit that calculates a texture feature amount as a first feature amount for pixels in the cell region and acquires a distribution of the first feature amount in the cell region; A region dividing unit that divides the cell region into two or more divided regions; a second feature amount calculation unit that calculates statistics of the first feature amount for each of the partitioned regions and calculates a second feature amount that indicates a relationship between the statistics of the first feature amounts between the partitioned regions; A cell information acquisition device having a cell information acquisition unit that acquires information regarding a cell type or a cell state based on the second feature amount. (Configuration 2) A program for causing a computer to execute the cell information acquisition method according to any one of Methods 1 to 11 or the abnormality detection method according to Method 14. [Explanation of symbols]

[0098] 1, 3 Cell information acquisition device 4. Cellular abnormality detection device 9. Information processing equipment 11 Image acquisition section 12 Cell area extraction part 13 First feature distribution acquisition unit 14 Area division 15 Second feature calculation unit 16 Cell information acquisition department 31 Acquisition of trained model 41 Anomaly score calculation unit 42 Detection unit

Claims

1. an image acquiring step of acquiring a cell image including an image of a cell aggregate; a cell region extraction step of extracting a cell region indicating a region of the cell cluster from the cell image; a first feature amount distribution acquisition step of calculating a texture feature amount as a first feature amount for pixels in the cell region and acquiring a distribution of the first feature amount in the cell region; A region dividing step of dividing the cell region into two or more divided regions; a second feature calculation step of calculating statistics of the first feature amounts for each of the partitioned regions and calculating a second feature amount indicating a relationship between the statistics of the first feature amounts between the partitioned regions; a cell information acquiring step of acquiring information about a cell type or information about a cell state based on the second feature amount; having A method for acquiring cell information, comprising:

2. The cell information acquisition method according to claim 1 , wherein the region dividing step divides the cell region based on an area of ​​the cell region, a distance from a center of gravity of the cell region, or a distribution of the first feature amount.

3. The cell information acquisition method according to claim 1 , wherein the second feature calculation step includes calculating a difference or ratio of a statistical amount of the first feature amount between the divided regions.

4. The cell information acquisition method according to claim 1 , wherein the second feature calculation process includes calculating the variance, standard deviation, or coefficient of variation (CV value) of the statistics of the first feature between the divided regions.

5. The number of the cell regions extracted in the cell region extraction step is 3 or more, The cell information acquisition method according to any one of claims 1 to 4, wherein the cell information acquisition step includes acquiring a clustering result in which the cell regions are clustered into two or more clusters.

6. The cell information acquisition method according to any one of claims 1 to 4, wherein the cell information acquisition step includes acquiring a result of analyzing a two- or more dimensional feature vector including the second feature by principal component analysis.

7. The image acquisition step includes acquiring two or more of the cell images that are different from each other, The cell information acquisition method according to any one of claims 1 to 4, wherein the cell information acquisition process includes statistically analyzing the difference between the second feature amounts calculated from each of two or more of the cell images.

8. The method further includes a trained model acquisition step of acquiring a trained model that is machine-learned using the second feature amount and information about the cell type or information about the state of the cell; The cell information acquisition method according to any one of claims 1 to 4, wherein the cell information acquisition step includes using the trained model.

9. The cell information acquisition method of claim 8, wherein the information regarding the state of the cells includes at least one selected from the group consisting of information regarding viability, proliferation, degree of undifferentiation, remaining state of foreign genes, presence or absence of genomic abnormalities, efficiency of production of useful substances, presence or absence of cancer, and suitability for transitioning to the next process of cell production.

10. The method for acquiring cell information according to claim 1 , wherein the cells constituting the cell mass are pluripotent stem cells.

11. The cell information acquiring method according to claim 1 , wherein the information on the cell type includes information indicating a strain type of the cells constituting the cell mass.

12. A cell state information acquiring step of acquiring information about the state of the cell using the cell information acquiring method according to any one of claims 1 to 4; A cell processing step of carrying out at least one step selected from the group consisting of cell sorting, cell removal, and transition to a next step in cell production based on the information on the state of the cells acquired in the cell state information acquisition step; A method for producing a cell, comprising:

13. A cell state information acquiring step of acquiring information about the state of the cell using the cell information acquiring method according to any one of claims 1 to 4; a cell inspection step of inspecting cells based on the information on the state of the cells acquired in the cell state information acquisition step; A method for inspecting a cell, comprising:

14. an image acquiring step of acquiring a cell image including an image of a cell aggregate; a cell region extraction step of extracting a cell region indicating a region of the cell cluster from the cell image; a first feature amount distribution acquisition step of calculating a texture feature amount as a first feature amount for pixels in the cell region and acquiring a distribution of the first feature amount in the cell region; A region dividing step of dividing the cell region into two or more divided regions; a second feature calculation step of calculating statistics of the first feature amounts for each of the partitioned regions and calculating a second feature amount indicating a relationship between the statistics of the first feature amounts between the partitioned regions; a trained model acquisition step of acquiring a trained model that machine-learns the probability of occurrence of the second feature amount by using a data group of the second feature amount obtained from a plurality of the cell images prepared for machine learning; an abnormality score calculation step of inputting the second feature amount obtained from the cell image to be evaluated to the trained model and calculating an abnormality score based on an output probability; A detection step of detecting an abnormality in the cell mass based on the abnormality score; A method for detecting an abnormality in a cell, comprising:

15. an image acquisition unit for acquiring cell images including images of cell aggregates; a cell region extraction unit that extracts a cell region indicating a region of the cell cluster from the cell image; a first feature amount distribution acquisition unit that calculates a texture feature amount as a first feature amount for pixels in the cell region and acquires a distribution of the first feature amount in the cell region; A region dividing unit that divides the cell region into two or more divided regions; a second feature amount calculation unit that calculates statistics of the first feature amount for each of the partitioned regions and calculates a second feature amount that indicates a relationship between the statistics of the first feature amounts between the partitioned regions; a cell information acquiring unit that acquires information about a cell type or information about a cell state based on the second feature amount; A cell information acquisition device comprising:

16. A program for causing a computer to execute the cell information acquisition method according to any one of claims 1 to 4, 10 and 11 or the abnormality detection method according to claim 14.

Images