Information processing device, analysis method, and analysis program

Abstract

An information processing device (1) is provided with: a determination unit (103) that uses a ploidy determination model (111) generated by machine learning using an image of a biological tissue including a polyploidy tumor to determine the ploidy of the tumor from a target image obtained by imaging the target tumor; and an output control unit (105) that causes an output device to output the determination result. The information processing device (1) makes it possible to easily know the ploidy of the tumor for each tumor region.

Description

Information processing device, analysis method, and analysis program

[0001] The present disclosure relates to an information processing device or the like that performs tumor analysis.

[0002] Techniques for analyzing tumor properties through image analysis have been known for some time. For example, Patent Document 1 listed below discloses a discrimination system that discriminates whether a cancer is a hypermutated cancer based on an image of a pathological section of the cancer.

[0003] International Publication No. 2019-159821

[0004] It is known that polyploidy, or genome doubling, occurs frequently in tumor cells. This polyploidy is observed not only in specific organs. Recent studies have revealed that polyploidy in tumors is the basis for chromosomal instability, which causes tumor metastasis and drug resistance, and is associated with poor tumor prognosis.

[0005] However, the technology of Patent Document 1 cannot determine the ploidy of a tumor. Currently, determining the ploidy of a tumor requires the application of complicated techniques such as flow cytometry, chromosomal FISH (Fluorescence In Situ Hybridization), or genome sequence analysis. It is difficult to apply such techniques in ordinary medical settings.

[0006] One aspect of the present invention has been made in consideration of the above-mentioned problems, and one example of its purpose is to realize an information processing device, etc. that makes it possible to easily determine the ploidy of a tumor for each region of the tumor.

[0007] An information processing device according to one aspect of the present disclosure includes a determination unit that determines the ploidy of a tumor from a target image of the tumor using a ploidy determination model generated by machine learning using an image of biological tissue containing a polyploid tumor, and an output control unit that outputs the determination result of the determination unit to an output device.

[0008] Furthermore, an analysis method according to one aspect of the present disclosure is an analysis method executed by at least one information processing device, and includes a determination step of determining the ploidy of a tumor from a target image of the tumor using a ploidy determination model generated by machine learning using an image of biological tissue containing a polyploidized tumor, and an output control step of outputting the result of the determination to an output device.

[0009] According to one aspect of the present invention, it becomes possible to easily determine the ploidy of a tumor for each region of the tumor.

[0010] Fig. 1 is a block diagram illustrating an example of a main configuration of an information processing device according to an embodiment of the present disclosure; Fig. 2 is a diagram illustrating a method for generating a ploidy determination model; Fig. 3 is a diagram illustrating an image of a tumor tissue and a determination result image showing a ploidy determination result for each region of the tumor tissue; Fig. 4 is a flowchart illustrating an example of an analysis method according to an embodiment of the present disclosure;

[0011] [Configuration of Information Processing Device] The configuration of an information processing device 1 according to this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the main parts of the information processing device 1. As shown in the figure, the information processing device 1 includes a control unit 10 that controls each part of the information processing device 1, and a storage unit 11 that stores various data used by the information processing device 1. The information processing device 1 also includes a communication unit 12 that allows the information processing device 1 to communicate with other devices, an input unit 13 that accepts input to the information processing device 1, and an output unit 14 that allows the information processing device 1 to output various data.

[0012] The control unit 10 includes a data acquisition unit 101, a division unit 102, a determination unit 103, a result image generation unit 104, and an output control unit 105. The storage unit 11 stores a ploidy determination model 111.

[0013] The data acquisition unit 101 acquires an image of a tumor to be determined for ploidy. For example, the data acquisition unit 101 may acquire an image generated by taking a specimen of a tumor excised from a living body, staining it with hematoxylin and eosin (HE), and photographing the specimen. Note that any staining method may be used, and an image of an unstained tumor may also be acquired.

[0014] The division unit 102 divides the image acquired by the data acquisition unit 101 into predetermined division units to generate multiple partial images. The division units may be set as appropriate, but it is preferable that each partial image captures an area that occupies a portion of the biological tissue and includes multiple tumorigenic cells. This allows the ploidy to be determined for each area based on pathological information that takes into account the structure of the tissue formed by the collection of multiple tumorigenic cells and the presence of various cells present alongside the tumorigenic cells. In this way, the ploidy determination result for each area is based on more information and is therefore more useful as a basis for predicting a patient's prognosis, etc., than the ploidy determination result for each cell. The data acquisition unit 101 may also acquire the divided images, in which case the division unit 102 is omitted.

[0015] The determination unit 103 uses the ploidy determination model 111 to determine the ploidy of a tumor from a target image of the tumor whose ploidy is to be determined. Here, "tumor ploidy" refers to whether or not the genome of a tumorous cell is polyploidized, or the likelihood that it is polyploidized. Therefore, the determination result of the determination unit 103 indicates whether or not the genome of a tumor cell captured in the target image is polyploidized, or the likelihood that it is polyploidized. The determination unit 103 may also output an index value (e.g., standard deviation, variance, etc.) indicating the variance in the value of the likelihood that it is polyploidized. This index value is useful for determining the reliability of the determination result and may be used to represent the properties of the tumor.

[0016] Note that polyploid cells refer to cells whose genomes have doubled to triploid or more in an organism in which the genomes of normal cells are diploid, for example. Furthermore, how to evaluate the "tumor ploidy" is arbitrary. For example, the proportion of tumor cells whose genomes have become polyploid among tumorigenic cells may be used as the "tumor ploidy." In this case, the determination unit 103 predicts the proportion of polyploid tumor cells among tumor cells captured in the target image. The proportion of polyploid tumor cells may be expressed, for example, by the ratio of the area of ​​the area in which the polyploid tumor cells are captured to the area in which the tumor cells are captured.

[0017] In this embodiment, an example will be described in which a segmented image generated by the segmentation unit 102 from an image acquired by the data acquisition unit 101 is used as a target image of a tumor whose ploidy is to be determined. Note that the target image may be any image of a tumor whose ploidy is to be determined, and does not necessarily have to be a segmented image.

[0018] The polyploidy determination model 111 is a trained model generated by machine learning using images of biological tissue containing polyploid tumors. For example, the polyploidy determination model 111 may be a one-class classification model that classifies an input target image as polyploid or not. In this case, the output value of the polyploidy determination model 111 is, for example, a numerical value between 0 and 1 indicating the likelihood of polyploidy. The determination unit 103 may use this numerical value as the polyploidy determination result. Furthermore, the determination unit 103 may output a determination result indicating polyploidy if this numerical value is equal to or greater than a predetermined threshold, and may output a determination result indicating no polyploidy if this numerical value is less than the threshold.

[0019] Furthermore, for example, the ploidy determination model 111 may be a classification model with two or more classes. Furthermore, for example, the ploidy determination model 111 may be a prediction model that predicts an evaluation value of ploidy (e.g., the proportion of polyploid tumor cells among tumorigenic cells). Note that the data input to the ploidy determination model 111 may be a target image or feature amounts extracted from the target image.

[0020] It is preferable that the above-mentioned polyploidy determination model 111 is a model that has been machine-learned to determine whether or not the tumor cells contained in a region that occupies a portion of biological tissue and contains multiple tumorous cells are polyploid as a whole.

[0021] Because the ploidy of tumor cells varies from cell to cell, when making a medical decision, it is more important to determine whether the tumor cells in a region as a whole are polyploid than whether individual tumor cells are polyploid. Furthermore, pathological information that takes into account the tissue structure formed by the aggregation of multiple tumor-forming cells and the presence of various cells present alongside the tumor-forming cells is also an important factor in determining the characteristics of a tumor. Therefore, the above configuration makes it possible to output a determination result that is useful for making a medical decision. Preferably, the ploidy determination model 111 is generated by machine learning using images of biological tissues containing both polyploid and non-polyploid tumors. This makes it possible to easily determine the ploidy of tumors in a region where polyploid and non-polyploid tumors coexist.

[0022] As described above, the target image for which ploidy is to be determined may be one of a plurality of partial images obtained by dividing an original image of a target tumor into predetermined division units. In this case, the determination unit 103 determines the ploidy for each of the plurality of partial images obtained from the original image.

[0023] According to the above configuration, it is possible to obtain a ploidy assessment result for each division unit. This also makes it possible to precisely detect the boundary between polyploid regions and non-polyploid regions. Furthermore, the area to be assessed is narrower than when the entire original image is assessed, and more appropriate correct labels can be assigned to each region, making it possible to obtain highly accurate assessment results. Furthermore, it is possible to obtain results regarding the proportion of polyploid regions in the tumor and the heterogeneity of ploidy.

[0024] The result image generating unit 104 generates a determination result image that indicates the determination result of the determining unit 103. The determination result image will be described later in the section "Method for generating a determination result image."

[0025] The output control unit 105 outputs the determination result of the determination unit 103 to a predetermined output device. For example, the output control unit 105 may cause the output device to display and output the determination result image generated by the result image generation unit 104. The output device may be a device external to the information processing device 1, or may be a device (e.g., the output unit 14) included in the information processing device 1. The output mode of the output device is arbitrary. That is, the output control unit 105 may display and output the determination result, or may output it in other output modes such as print output or audio output.

[0026] As described above, the information processing device 1 includes a determination unit 103 that determines the ploidy of a tumor from a target image of the target tumor using a ploidy determination model 111 generated by machine learning using an image of biological tissue containing a polyploid tumor, and an output control unit 105 that outputs the determination result of the determination unit 103 to an output device.

[0027] According to the above configuration, if a target image of a target tumor is available, the determination result of the ploidy of the tumor can be output. Therefore, according to the above configuration, it is possible to easily determine the ploidy of the tumor for each region of the tumor without using complicated techniques such as flow cytometry, chromosome FISH, or genome sequence analysis.

[0028] Furthermore, recent research has suggested that tumor ploidy correlates with the tumor's plasticity and evolutionary potential. In other words, the above configuration is expected to facilitate the assessment of the future risk of tumor malignant evolution. Furthermore, as described above, the determination unit 103 preferably determines the tumor ploidy using the ploidy determination model 111 generated by machine learning using images of biological tissues containing both polyploid tumors and non-polyploid tumors. This makes it possible to easily determine the ploidy of tumors in regions where polyploid tumors and non-polyploid tumors coexist.

[0029] [Method of Generating the Polyploidy Determination Model] A method of generating the polyploidy determination model 111 will be described with reference to Fig. 2. Fig. 2 is a diagram illustrating a method of generating the polyploidy determination model 111. Because the polyploidy determination model 111 is generated by supervised learning, in order to generate the polyploidy determination model 111, it is first necessary to generate training data.

[0030] To achieve this, biological tissues containing both non-polyploid tumors and polyploid tumors are first obtained and sliced ​​to prepare specimens. Figure 2 shows an example of specimens prepared from liver tumors. Furthermore, to improve the visibility of polyploid tumor cells, each specimen is stained with HE staining. Similar specimens can also be prepared from tumors in organs other than the liver. Below, we will explain the process of generating training data from specimen a1, one of these specimens, but training data can also be generated from other specimens in a similar manner.

[0031] Next, the prepared specimen a1 is photographed. The image may be a color image or a grayscale image. FIG. 2 shows image a2 obtained by photographing specimen a1. Then, the region in image a2 that contains tumor cells is set as a region of interest, and the region of interest is cut out from image a2 to obtain image a3. In the example of FIG. 2, two regions of interest are set in image a2, and each region of interest is shown as a rectangle. Image a3 is obtained by cutting out the lower of the two regions of interest. Although not shown, the other region of interest is also cut out in the same manner.

[0032] In addition, to determine the correct data in the training data, the ploidy of each region of interest is determined in advance. Any method for determining the ploidy may be used. For example, the ploidy of each region of interest may be determined by chromosome FISH.

[0033] Next, image a3 is divided into predetermined division units to generate multiple partial images a4. The size of the division units may be set so that the ploidy of the tumor cells depicted in each partial image a4 can be determined. For example, image a3 may be divided into division units such that each partial image a4 contains approximately 3 to 10 tumor cells. Then, training data for generating the ploidy determination model 111 is generated by associating the generated partial images a4 with correct data indicating the ploidy of the tumor cells depicted therein. When ploidy is determined in units of regions of interest, the ploidy determination result for each region of interest may be associated with each partial image obtained from the image of that region of interest as correct data. For example, the ploidy determination result for image a3 may be associated with partial image a4 as correct data. Note that the ploidy of tumor cells depicted in each partial image may also be determined. In this case, the process of associating the determination result of a certain partial image with that partial image and using it as training data may be performed for each of the multiple partial images. For example, if partial image a4 contains approximately 3 to 10 tumor cells, and the majority of those tumor cells are polyploid, partial image a4 can be associated with a judgment result indicating that the tumor cells are polyploid.

[0034] Machine learning is performed using the training data generated as described above to generate the ploidy determination model 111. Note that the information processing device 1 may perform the processes from generating the partial images to generating the ploidy determination model 111. In other words, the information processing device 1 may be provided with a training data generation unit that generates training data and a learning unit that performs machine learning.

[0035] In this case, the data acquisition unit 101 acquires image a3, and the segmentation unit 102 generates partial image a4 from image a3. Then, the training data generation unit generates training data by associating correct data with the generated partial image a4, and the learning unit generates the ploidy determination model 111 through machine learning using the generated training data. Note that either the generation of the training data or the generation of the ploidy determination model 111 may be executed by another information processing device. For example, the information processing device 1 may acquire training data generated by another information processing device and generate the ploidy determination model 111 using the training data.

[0036] Any algorithm that can generate a model capable of outputting a ploidy determination result from an image can be applied as the algorithm of the ploidy determination model 111. For example, the ploidy determination model 111 may be a vision transformer model. Furthermore, the ploidy determination model 111 may be a convolutional neural network model or the like. Specific examples include HIPT (Hierarchical Image Pyramid Transformer), EfficientNet, DINO (self-DIstillation with NO labels), ResNet (Residual Neural Networks), DenseNET (Densely Connected Convolutional Networks), or the like.

[0037] Experiments conducted by the inventors of the present application have confirmed that the ploidy determination model 111 of the various algorithms described above can accurately determine the ploidy of tumors in many cases other than the training data. For example, in experiments conducted on TCGA (The Cancer Genome Atlas) cases, particularly the HIPT ploidy determination model 111 provided favorable determination results. Furthermore, experiments conducted by the inventors of the present application have confirmed that the ploidy of tumors can be robustly determined even in pathological images containing various modifications such as hemorrhage and necrosis.

[0038] As described above, the training data generation method according to this embodiment includes the steps of: acquiring images of biological tissue containing polyploid tumors; and generating training data by associating the acquired images with ground truth data indicating the ploidy of the tumors depicted in the images. Using the training data generated by this training data generation method, a ploidy determination model can be generated that enables easy determination of the ploidy of tumors in each tumor region. In particular, the image (image of biological tissue containing polyploid tumors) is preferably an image of biological tissue that contains both polyploid tumors and non-polyploid tumors. This makes it possible to easily determine the ploidy of tumors in regions where polyploid tumors and non-polyploid tumors coexist.

[0039] As described above, the learning method according to this embodiment includes the steps of: acquiring training data in which images of biological tissue containing polyploid tumors are associated with ground truth data indicating the ploidy of the tumors depicted in the images; and performing machine learning using the acquired training data to generate a ploidy determination model for determining the ploidy of a tumor from a target image of the tumor. This allows the generation of a ploidy determination model that enables the ploidy of each tumor region to be easily determined. In particular, the images in the training data (images of biological tissue containing polyploid tumors) are preferably images of biological tissue that contain both polyploid tumors and non-polyploid tumors. This allows the generation of a ploidy determination model 111 that enables the ploidy of tumors to be easily determined for regions where polyploid tumors and non-polyploid tumors coexist.

[0040] The procedure for determining the ploidy of a tumor of unknown ploidy is similar to the procedure described above. That is, if a tumor of unknown ploidy is found, it can be sampled and photographed, and a region of interest can be extracted and input to the information processing device 1. The data acquisition unit 101 acquires the input image, and the division unit 102 divides the image to generate partial images. The determination unit 103 then inputs the generated partial images to the trained ploidy determination model 111 and determines the ploidy of the tumor depicted in the partial images based on the output value of the ploidy determination model 111. The result image generation unit 104 then generates a determination result image indicating the determination result by the determination unit 103, and the output control unit 105 outputs the determination result image to the output unit 14.

[0041] [Method for generating a determination result image] A method for generating a determination result image will be described with reference to Fig. 3. Fig. 3 shows image b1 obtained by capturing an image of tumor tissue and determination result image b2 showing the ploidy determination result for each region of the tumor tissue. Image b1 shown in Fig. 3 is divided into multiple sections by a dashed grid. Each divided section corresponds to the partial image described above. For example, image b1 is an image acquired by the data acquisition unit 101, and b11 at the upper right corner of image b1 is one of the partial images generated by the division unit 102 from image b1.

[0042] The determination unit 103 inputs each of the multiple partial images generated from image b1, including partial image b11, into the polyploidy determination model 111 and obtains a determination result of the polyploidy for each of the multiple partial images. For example, when the polyploidy determination model 111 outputs a numerical value in the range of 0 to 1 indicating the likelihood of polyploidization, a numerical value in the range of 0 to 1 is output for each of the multiple partial images.

[0043] The result image generation unit 104 generates a determination result image by arranging the determination results of the determination unit 103 for each partial image in each area corresponding to the arrangement of each partial image in image b1. When using the polyploidy determination model 111 that outputs a numerical value ranging from 0 to 1 indicating the likelihood of polyploidization, the determination result image is an image in which numerical values ​​ranging from 0 to 1 are arranged in a grid pattern, as in determination result image b2 shown in Figure 3.

[0044] The numerical values ​​in the determination result image b2 indicate the determination result of the partial image corresponding to the region that the numerical value is located in. For example, the numerical value "0.03" displayed in the region b21 at the top right of the determination result image b2 is the numerical value output by inputting the partial image b11 to the ploidy determination model 111, and indicates the likelihood that the portion of the tumor tissue depicted in the partial image b11 is polyploid as a whole.

[0045] Furthermore, in the determination result image b2, the color of each area where a numerical value is placed is a color corresponding to that numerical value. More specifically, the color of each area included in the determination result image b2 is closer to white the closer the placed numerical value is to 0, and closer to black the closer the placed numerical value is to 1. For example, the result image generation unit 104 may determine the display color of each area by converting a numerical value ranging from 0 to 1 indicating the likelihood of polyploidization into a gradation value ranging from 0 to 256. Furthermore, the result image generation unit 104 may determine the display mode of each area by converting a numerical value ranging from 0 to 1 indicating the likelihood of polyploidization into a brightness value or lightness value.

[0046] In this way, the result image generating unit 104 may generate a determination result image in which the display mode of each region containing a numerical value is set to a display mode corresponding to the numerical value, thereby making it possible to easily recognize regions in which tumor cells are likely to be polyploid.

[0047] The output control unit 105 may display and output the determination result image generated as described above, i.e., the determination results of the determination unit 103 for each partial image, on the output device as an image displayed in each region corresponding to the arrangement of each partial image in the original image (image b1 in FIG. 3 ). With this configuration, even if the region in the original image containing a polyploid tumor is divided into multiple partial images or is limited to a portion of the entire image, the region can be recognized from the determination results of each partial image. For example, from the determination result image b2 shown in FIG. 3 , it can be recognized that the region at the top of the image is unlikely to be polyploid, but that the remaining regions are widely polyploid. The determination results for each region may be displayed superimposed on image b1.

[0048] [Analysis Method] The analysis method according to this embodiment will be described with reference to Fig. 4. Fig. 4 is a flowchart showing an example of the analysis method.

[0049] In S1, the data acquisition unit 101 acquires an image of a tumor whose ploidy is to be determined. For example, the data acquisition unit 101 may acquire an image of a slice of biological tissue containing a tumor whose ploidy is to be determined, which has been stained with HE, or may acquire an image of a region of interest cut out from the sliced ​​image. Furthermore, the image before the region of interest is cut out may be stored in the storage unit 11 or the like. In this case, in S1, the data acquisition unit 101 may acquire an image by cutting out a region of interest specified via the input unit 13 or the like from the image stored in the storage unit 11 or the like.

[0050] In S2, the division unit 102 divides the image acquired in S1 into predetermined division units to generate a plurality of partial images. In the analysis method shown in FIG. 4, the partial images generated in S2 are used as target images for determining ploidy. Note that partial images generated by another device may be acquired in S1, in which case the process of S2 is omitted. The process of S2 is also omitted when using a ploidy determination model 111 generated by machine learning using an undivided image (e.g., an image like image a3 in FIG. 2).

[0051] In S3 (determination step), the determination unit 103 determines the ploidy of the tumor depicted in the partial image acquired in S2 using the ploidy determination model 111. Specifically, the determination unit 103 inputs the partial image acquired in S2 into the ploidy determination model 111 to obtain a numerical value indicating the ploidy determination result. As described above, the determination unit 103 may use the output value of the ploidy determination model 111 as the determination result directly, or may determine the ploidy based on the output value of the ploidy determination model 111. The process of determining the ploidy of the partial image is performed for each of the partial images acquired in S2.

[0052] In S4, the result image generation unit 104 generates a determination result image showing the result of the determination in S3. More specifically, the result image generation unit 104 generates the determination result image by arranging the determination results of the determination unit 103 for each of the partial images generated in S2 in each area corresponding to the arrangement of each partial image in the image acquired in S1.

[0053] In S5, the output control unit 105 causes the output device to output the result of the determination in S3. For example, the output control unit 105 may cause the output unit 14 to output the determination result image generated in S4. When the process of S5 ends, the analysis method shown in FIG. 4 ends.

[0054] 4 may be shared and executed by multiple information processing devices. For example, the processes of S1 and S2 may be executed by a first information processing device, and the process of S3 using the partial image generated by these processes may be executed by a second information processing device. Furthermore, the determination result of S3 may be notified to a third information processing device, and the processes of S4 and S5 may be executed by the third information processing device.

[0055] As described above, the analysis method according to this embodiment is an analysis method executed by at least one information processing device, and includes a determination step (S3) of determining the ploidy of a tumor from a target image of the tumor using a ploidy determination model 111 generated by machine learning using an image of biological tissue containing a polyploid tumor, and an output control step (S5) of outputting the determination result to an output device. This analysis method makes it possible to easily determine the ploidy of a tumor for each region of the tumor.

[0056] [Example of implementation by software] The functions of the information processing device 1 can be realized by a program (analysis program) that causes a computer to function as the information processing device 1, and that causes a computer to function as each control block of the information processing device 1 (particularly each part included in the control unit 10).

[0057] In this case, the information processing device 1 includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The functions described in the above embodiment are realized by executing the program using the control device and storage device.

[0058] The program may be stored non-transitory on one or more computer-readable recording media. The recording media may or may not be included in the information processing device 1. In the latter case, the program may be supplied to the information processing device 1 via any wired or wireless transmission medium.

[0059] In addition, some or all of the functions of each of the control blocks can be realized by logic circuits. For example, integrated circuits in which logic circuits that function as each of the control blocks are formed are also included in the scope of the present disclosure. In addition, the functions of each of the control blocks can also be realized by, for example, a quantum computer.

[0060] The present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present disclosure.

[0061] [Summary] As described above, the information processing device according to aspect 1 of the present disclosure includes a determination unit that determines the ploidy of a tumor from a target image of the tumor using a ploidy determination model generated by machine learning using an image of biological tissue containing a polyploid tumor, and an output control unit that outputs the determination result of the determination unit to an output device.

[0062] In the information processing device according to aspect 2 of the present disclosure, in aspect 1, the polyploidy determination model may be a model that has been machine-learned to determine whether or not the tumor cells contained in a region that occupies a portion of biological tissue and contains multiple tumorigenic cells are polyploid as a whole.

[0063] In the information processing device of aspect 3 of the present disclosure, in aspect 1 or 2, the target image is one of a plurality of partial images obtained by dividing an original image of the target tumor into predetermined division units, and the judgment unit may perform the judgment on each of the plurality of partial images obtained from the original image.

[0064] In the information processing device of aspect 4 of the present disclosure, in aspect 3, the output control unit may cause the output device to display and output the judgment results of the judgment unit for each of the partial images as judgment result images displayed in each area corresponding to the arrangement of each partial image in the original image.

[0065] The information processing method according to aspect 5 of the present disclosure is an analysis method executed by at least one information processing device, and includes a determination step of determining the ploidy of a tumor from a target image of the tumor using a ploidy determination model generated by machine learning using an image of biological tissue containing a polyploid tumor, and an output control step of outputting the result of the determination to an output device.

[0066] The analysis program according to aspect 6 of the present disclosure is an analysis program for causing a computer to function as the information processing device described in aspect 1, and is an analysis program for causing a computer to function as the judgment unit and the output control unit.

[0067] 1 Information processing device 103 Determination unit 105 Output control unit 111 Ploidy determination model

Claims

1. An information processing device comprising: a determination unit that determines the ploidy of a tumor from a target image of the tumor using a ploidy determination model generated by machine learning using images of biological tissue containing a polyploid tumor; and an output control unit that outputs the determination result of the determination unit to an output device.

2. The information processing device described in claim 1, wherein the polyploidy determination model is a model that has been machine-learned to determine whether or not the tumor cells contained in an area that occupies a portion of biological tissue and contains multiple tumorous cells are polyploid as a whole.

3. An information processing device as described in claim 1 or 2, wherein the target image is one of a plurality of partial images obtained by dividing an original image of the target tumor into predetermined division units, and the judgment unit makes the judgment for each of the plurality of partial images obtained from the original image.

4. An information processing device as described in claim 3, wherein the output control unit causes the output device to display and output the judgment results of the judgment unit for each of the partial images as judgment result images displayed in each area corresponding to the arrangement of each partial image in the original image.

5. An analytical method executed by at least one information processing device, comprising: a determination step of determining the ploidy of a tumor from a target image of the tumor using a ploidy determination model generated by machine learning using an image of biological tissue containing a polyploid tumor; and an output control step of outputting the result of the determination to an output device.

6. An analysis program for causing a computer to function as the information processing device according to claim 1, the analysis program causing the computer to function as the determination unit and the output control unit.

Images