Medical diagnosis support method, medical diagnosis support system, and program

Abstract

To improve estimation accuracy of functional abnormality.SOLUTION: A medical diagnosis support method includes acquiring a medical image of a subject, acquiring variant information including an estimation result of variants of a plurality of genes of the subject by analyzing the medical image, classifying the plurality of genes into one or more co-occurrence groups based on co-occurrence information indicating co-occurrence of variants between genes, determining variants of one or more genes belonging to each of the co-occurrence groups based on the variant information, and estimating a functional abnormality of the subject based on a determination result of the variants of the genes.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The embodiments disclosed in the present specification and drawings relate to a medical diagnosis support method, a medical diagnosis support system, and a program. [Background technology]

[0002] Conventionally, knowledge related to the functions of biological tissues and organs, known as gene ontology, has been known. Gene ontology includes knowledge of functional abnormalities that appear in biological tissues and organs. Functional abnormalities are thought to occur when gene mutations cause changes in gene expression levels. Gene mutations can be used to detect functional abnormalities.

[0003] On the other hand, for non-invasive use in current clinical workflows, it is desirable to be able to estimate functional abnormalities from medical images. However, it is difficult to detect, for example, irregularities in cardiac contraction from 3D medical images that lack time-series information. Furthermore, even when observing 3D medical images for a certain period of time, it is difficult to detect, for example, abnormalities in angiogenesis. Therefore, technology that improves the accuracy of estimating functional abnormalities is needed. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-174108 Summary of the Invention [Problem to be solved by the invention]

[0005] One of the problems to be solved by the embodiments disclosed in this specification and the drawings is to improve the accuracy of estimating functional abnormalities. However, the problems to be solved by the embodiments disclosed in this specification and the drawings are not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described below can also be positioned as other problems. [Means for solving the problem]

[0006] A medical diagnosis support method according to an embodiment includes acquiring a medical image of a subject, analyzing the medical image to acquire mutation information having estimated results of mutations in a plurality of genes of the subject, classifying the plurality of genes into one or more co-occurrence groups based on co-occurrence information indicating co-occurrence of mutations between genes, determining mutations in one or more genes belonging to each co-occurrence group for each co-occurrence group based on the mutation information, and estimating a functional abnormality of the subject based on the genetic mutation determination results. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 is a block diagram showing an example of the configuration of a medical diagnosis support system according to this embodiment. [Figure 2] FIG. 2 shows a data structure illustrating an example of a gene ontology table according to this embodiment. [Figure 3] FIG. 3 shows a data configuration of an example of a co-occurrence table according to this embodiment. [Figure 4] FIG. 4 is a diagram showing an example of a method for acquiring mutation information using the gene mutation estimation function according to this embodiment. [Figure 5] FIG. 5 is a diagram showing an example of a related gene graph according to this embodiment. [Figure 6] FIG. 6 is a diagram showing an example of a co-occurrence graph according to this embodiment. [Figure 7] FIG. 7 is a diagram showing the determination result of the first co-occurrence group according to this embodiment. [Figure 8] FIG. 8 is a diagram showing the determination result of the second co-occurrence group according to this embodiment. [Figure 9] FIG. 9 is a diagram showing an example of a functional abnormality estimation image according to this embodiment. [Figure 10] FIG. 10 is a flowchart showing an example of functional abnormality estimation processing executed by the medical image processing apparatus according to this embodiment. [Figure 11]FIG. 11 is a diagram showing the determination result of the first co-occurrence group according to the first modification. [Figure 12] FIG. 12 is a diagram showing the determination result of the second co-occurrence group according to the first modification. [Figure 13] FIG. 13 is a diagram showing the determination result of the first co-occurrence group according to the second modification. [Figure 14] FIG. 14 is a diagram showing the determination result of the second co-occurrence group according to the second modification. [Figure 15] FIG. 15 is a diagram showing an example of a confusion matrix of an estimation model for estimating a mutation of a gene F according to Modification 3. [Figure 16] FIG. 16 is a diagram showing an example of a confusion matrix of an estimation model for estimating a mutation of a gene G according to Modification 3. [Figure 17] FIG. 17 is a diagram showing an example of a confusion matrix of an estimation model for estimating a mutation of gene H according to Modification 3. DETAILED DESCRIPTION OF THE INVENTION

[0008] Hereinafter, a medical diagnosis support method, a medical diagnosis support system, and a program according to the present embodiment will be described with reference to the drawings. In the following embodiments, parts with the same reference numerals perform similar operations, and redundant explanations will be omitted as appropriate.

[0009] 1 is a block diagram showing an example of the configuration of a medical diagnosis support system 1 according to this embodiment. The medical diagnosis support system 1 is a system that detects functional abnormalities of biological tissues and internal organs from medical images L1 (see FIG. 4) based on gene mutations. The medical diagnosis support system 1 includes a medical image management system (Picture Archiving and Communication Systems: PACS) 10, a gene ontology server 20, and a medical information processing device 30.

[0010] Furthermore, each system and each device are connected to each other so that they can communicate with each other via a network 40. Note that the configuration shown in Fig. 1 is an example, and the number of each system and each device may be changed as desired. Furthermore, devices not shown in Fig. 1 may be connected to the network 40.

[0011] The medical image management system 10 is a system that stores medical images L1. The medical images L1 are images of a subject. For example, the medical images L1 are radiation images such as CT images taken by an X-ray CT (Computed Tomography) device or X-ray images taken by an X-ray diagnostic device. Furthermore, the medical images L1 are not limited to radiation images, and may be MRI images taken by an MRI (Magnetic Resonance Imaging) device or other images.

[0012] The gene ontology server 20 is a server device that stores various types of information related to gene ontology. For example, the gene ontology server 20 is realized by a computer device such as a server or a workstation.

[0013] The medical information processing device 30 estimates functional abnormalities of an object to be examined, such as the biological tissues or internal organs of a subject, based on the medical image L1. For example, the medical information processing device 30 is realized by computer equipment such as a server or a workstation.

[0014] Next, a description will be given of the medical information processing device 30. The medical information processing device 30 has an NW (network) interface 31, an input interface 32, a display 33, a memory , and a processing circuit .

[0015] The NW interface 31 is connected to the processing circuit 35 and controls the transmission and communication of various data between each device connected via the network 40. For example, the NW interface 31 is realized by a network card, a network adapter, a NIC (Network Interface Controller), or the like.

[0016] The input interface 32 is connected to the processing circuitry 35, converts an input operation received from an operator (such as a medical professional) into an electrical signal, and outputs the electrical signal to the processing circuitry 35. Specifically, the input interface 32 converts the input operation received from the operator into an electrical signal and outputs the electrical signal to the processing circuitry 35.

[0017] For example, the input interface 32 may be realized by a trackball, a switch button, a mouse, a keyboard, a touchpad that performs input operations by touching the operation surface, a touchscreen that integrates a display screen and a touchpad, a non-contact input circuit using an optical sensor, and a voice input circuit.

[0018] In this specification, the input interface 32 is not limited to an interface having physical operation parts such as a mouse, a keyboard, etc. For example, an example of the input interface 32 also includes an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the device and outputs this electrical signal to a control circuit.

[0019] The display 33 is connected to the processing circuit 35 and displays various information and image data output from the processing circuit 35. For example, the display 33 is realized by a liquid crystal display, a CRT (Cathode Ray Tube) display, an organic EL display, a plasma display, a touch panel, or the like.

[0020] The memory 34 is connected to the processing circuitry 35 and stores various data. The memory 34 also stores various programs that are read and executed by the processing circuitry 35 to realize various functions. For example, the memory 34 may be realized by a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, a hard disk, an optical disk, or the like.

[0021] For example, the memory 34 stores a gene ontology table 341 , one or more co-occurrence tables 342 , and an estimated accuracy table 343 .

[0022] The gene ontology table 341, one or more co-occurrence tables 342, and the estimation accuracy table 343 may be stored in another device such as the gene ontology server 20. In this case, the medical information processing device 30 may receive necessary information from the other device each time, or may request the other device to perform processing using the necessary information.

[0023] The gene ontology table 341 has information about each biological process in gene ontology. FIG. 2 shows an example of the data configuration of the gene ontology table 341 according to this embodiment. The gene ontology table 341 has information about a site, a biological process, a description, related genes, and a functional abnormality. The site is information indicating the site targeted by the biological process. The biological process is identification information for identifying the biological process. The description is an explanation indicating the content of the biological process. The related genes are information indicating one or more genes related to the biological process. The functional abnormality is information indicating an abnormality that occurs when there is a mutation in a gene indicated in the related genes.

[0024] The co-occurrence table 342 is an information table showing the co-occurrence of mutations between genes. The co-occurrence table 342 is an example of co-occurrence information. Here, the co-occurrence of mutations between genes refers to the possibility that when a certain gene mutates, one or more other genes will mutate. For example, the memory 34 stores the co-occurrence table 342 for each related gene in the gene ontology table 341.

[0025] FIG. 3 is a data configuration showing an example of a co-occurrence table 342 according to this embodiment. FIG. 3 shows a case where related genes include gene A, gene B, gene C, gene D, and gene E. For example, the co-occurrence table 342 shown in FIG. 3 indicates that when a mutation occurs in gene A, the probability that a mutation occurs in gene B is 0.8. It also indicates that when a mutation occurs in gene A, the probability that a mutation occurs in gene C is 0.1. It also indicates that when a mutation occurs in gene A, the probability that a mutation occurs in gene D is 0.1. It also indicates that when a mutation occurs in gene A, the probability that a mutation occurs in gene E is 0.1.

[0026] The estimation accuracy table 343 is information indicating the accuracy of estimation of mutations in each gene. For example, the estimation accuracy table 343 is information indicating the accuracy of each estimation model M1 (see FIG. 4) that estimates mutations in a gene.

[0027] The processing circuitry 35 controls the overall operation of the medical information processing device 30. The processing circuitry 35 has, for example, an image acquisition function 351, a genetic mutation estimation function 352, a co-occurrence classification function 353, a genetic mutation determination function 354, a functional abnormality estimation function 355, and a display control function 356. In the embodiment, each processing function performed by the components of the image acquisition function 351, the genetic mutation estimation function 352, the co-occurrence classification function 353, the genetic mutation determination function 354, the functional abnormality estimation function 355, and the display control function 356 is stored in the memory 34 in the form of a computer-executable program. The processing circuitry 35 is a processor that reads and executes the programs from the memory 34 to realize the function corresponding to each program. In other words, the processing circuitry 35 in a state in which each program has been read has each function shown in the processing circuitry 35 of FIG. 1.

[0028] 1 is described as realizing the image acquisition function 351, gene mutation estimation function 352, co-occurrence classification function 353, gene mutation determination function 354, functional abnormality estimation function 355, and display control function 356 with a single processor, but it is also possible to combine multiple independent processors to configure the processing circuit 35 and have each processor execute a program to realize the function. Also, in FIG. 1, it is described as a single memory such as memory 34 storing programs corresponding to each processing function, but it is also possible to configure multiple memories to be distributed and have the processing circuit 35 read out corresponding programs from individual memories.

[0029] The term "processor" used in the above description refers to a circuit such as a CPU (Central Processing Unit), a GPU (Graphical Processing Unit), an Application Specific Integrated Circuit (ASIC), a programmable logic device (e.g., a Simple Programmable Logic Device (SPLD), a Complex Programmable Logic Device (CPLD), and a Field Programmable Gate Array (FPGA)). The processor realizes its functions by reading and executing a program stored in memory 34. Note that instead of storing a program in memory 34, the processor may be configured so that the program is directly embedded in its circuitry. In this case, the processor realizes its functions by reading and executing the program embedded in its circuitry.

[0030] The image acquisition function 351 acquires a medical image L1 of a subject. The image acquisition function 351 is an example of a first acquisition unit. More specifically, the image acquisition function 351 acquires the medical image L1 from the medical image management system 10. Note that the image acquisition function 351 may acquire the medical image L1 not only from the medical image management system 10 but also from other devices or storage media.

[0031] The genetic mutation estimation function 352 analyzes the medical image L1 acquired by the image acquisition function 351 to acquire mutation information containing the estimation results of mutations in multiple genes of the subject. The genetic mutation estimation function 352 is an example of a second acquisition unit. The mutation information is information indicating whether or not a mutation has occurred in each gene. Note that the mutation information is not limited to whether or not a mutation has occurred in a gene, but may also be information indicating the probability that a mutation has occurred.

[0032] FIG. 4 is a diagram showing an example of a method for acquiring mutation information by the genetic mutation estimation function 352 according to this embodiment. As shown in FIG. 4, for example, the genetic mutation estimation function 352 acquires mutation information by inputting a medical image L1 into an estimation model M1. The estimation model M1 is a trained model that estimates genetic mutations. For example, the estimation model M1 is generated by machine learning such as deep learning. The estimation model M1 may be a trained model that estimates the mutation of one gene, or may be a trained model that estimates the mutations of multiple genes.

[0033] The estimation model M1 outputs a gene mutation estimation result, which is an estimation result of a gene mutation, when a medical image L1 acquired by the image acquisition function 351 is input. Then, the gene mutation estimation function 352 acquires mutation information based on the gene mutation estimation result output from the estimation model M1.

[0034] The genetic mutation estimation function 352 may also receive a gene specification. The genetic mutation estimation function 352 estimates the mutation of the specified gene. That is, the genetic mutation estimation function 352 inputs a medical image L1 into an estimation model M1 corresponding to the specified gene. The genetic mutation estimation function 352 then acquires mutation information based on the genetic mutation estimation result output from the estimation model M1.

[0035] The genetic mutation estimation function 352 may accept a site designation as a gene designation. When the site designation is accepted, the genetic mutation estimation function 352 estimates a genetic mutation included in the associated genes corresponding to the designated site in the gene ontology table 341 shown in Fig. 2. That is, the genetic mutation estimation function 352 analyzes the medical image L1 acquired by the image acquisition function 351 to acquire mutation information including the estimation results of mutations in multiple genes corresponding to the designated site.

[0036] The genetic mutation estimation function 352 may accept a biological process designation as a gene designation. When the designation of a biological process is accepted, the genetic mutation estimation function 352 estimates mutations in genes included in the associated genes corresponding to the designated biological process in the gene ontology table 341 shown in Fig. 2. That is, the genetic mutation estimation function 352 analyzes the medical image L1 acquired by the image acquisition function 351 to acquire mutation information including the estimation results of mutations in multiple genes corresponding to the designated biological process.

[0037] The genetic variation estimation function 352 may accept a description indicating the details of a biological process as a gene specification. When the description specification is accepted, the genetic variation estimation function 352 estimates a genetic variation included in the related genes corresponding to the specified description in the gene ontology table 341 shown in Fig. 2. That is, the genetic variation estimation function 352 analyzes the medical image L1 acquired by the image acquisition function 351 to acquire variation information including an estimation result of estimating mutations in multiple genes corresponding to the specified description.

[0038] The genetic mutation estimation function 352 may accept a designation of a functional abnormality as a gene designation. When the designation of a functional abnormality is accepted, the genetic mutation estimation function 352 estimates a genetic mutation included in the associated genes corresponding to the designated functional abnormality in the gene ontology table 341 shown in Fig. 2. That is, the genetic mutation estimation function 352 analyzes the medical image L1 acquired by the image acquisition function 351 to acquire mutation information including an estimation result of estimating mutations in multiple genes corresponding to the designated functional abnormality.

[0039] The co-occurrence classification function 353 classifies multiple genes into at least one co-occurrence group based on the co-occurrence table 342 that indicates the co-occurrence of mutations between genes. The co-occurrence classification function 353 is an example of a classification unit. More specifically, the co-occurrence classification function 353 classifies multiple genes that are the targets of estimation by the genetic mutation estimation function 352 into at least one co-occurrence group based on the co-occurrence table 342.

[0040] Fig. 5 is a diagram showing an example of a related gene graph G1 according to this embodiment. The related gene graph G1 shown in Fig. 5 is a graph showing the co-occurrence of mutations between genes belonging to the related genes shown in Fig. 3. The related gene graph G1 shown in Fig. 5 also shows that the co-occurrence between gene A and gene B is 0.8, the co-occurrence between gene C and gene D is 0.6, the co-occurrence between gene C and gene E is 0.7, the co-occurrence between gene A and gene C is 0.1, and the co-occurrence between gene B and gene C is 0.3.

[0041] Fig. 6 is a diagram showing an example of a co-occurrence graph according to this embodiment, in which the related gene graph G1 shown in Fig. 5 is divided into a first co-occurrence graph G21, which is a graph of the first co-occurrence group, and a second co-occurrence graph G22, which is a graph of the second co-occurrence group.

[0042] The co-occurrence classification function 353 divides genes belonging to the related genes into one or more co-occurrence groups based on the co-occurrence table 342 that shows the co-occurrence of mutations between genes. Here, a co-occurrence group is a group to which genes whose co-occurrence of mutations between genes is equal to or greater than a threshold belong.

[0043] The co-occurrence classification function 353 associates genes whose co-occurrence of mutations between genes is equal to or greater than a threshold value by grouping them. For example, the co-occurrence classification function 353 groups genes based on a co-occurrence of 0.6 or greater. As a result, the co-occurrence classification function 353 generates co-occurrence groups to which the grouped genes belong. In other words, the co-occurrence classification function 353 does not use co-occurrence between genes whose co-occurrence is less than 0.6 for grouping.

[0044] For example, as shown in FIG. 5, the co-occurrence between gene A and gene B is 0.8, so co-occurrence classifying function 353 groups gene A and gene B. In other words, co-occurrence classifying function 353 generates a first co-occurrence group to which gene A and gene B, which are associated by grouping, belong. Also, as shown in FIG. 5, the co-occurrence between gene C and gene D is 0.6, and the co-occurrence between gene C and gene E is 0.7, so co-occurrence classifying function 353 groups gene C and gene D, and groups gene C and gene E. Then, co-occurrence classifying function 353 generates a second co-occurrence group to which gene C, gene D, and gene E, which are associated by grouping, belong.

[0045] The genetic mutation determination function 354 determines mutations of genes belonging to one or more co-occurrence groups classified by the co-occurrence classification function 353, based on mutation information including the estimation results estimated by the genetic mutation estimation function 352. The genetic mutation determination function 354 is an example of a determination unit.

[0046] Here, a co-occurrence group is a group to which genes with a co-occurrence of mutations between genes equal to or greater than a threshold belong. Therefore, if a target gene is estimated to have mutated, it is highly likely that other genes in the co-occurrence group to which the target gene belongs have also mutated.

[0047] That is, when a gene to be determined is presumed to have mutated, if other genes in the co-occurrence group to which the gene to be determined belongs are also presumed to have mutated, there is a high possibility that the gene to be determined is actually mutated. On the other hand, when a gene to be determined is presumed to have mutated, if other genes in the co-occurrence group to which the gene to be determined belongs are also presumed not to have mutated, there is a high possibility that the gene to be determined is not actually mutated.

[0048] Therefore, the gene mutation determination function 354 determines gene mutations for each co-occurrence group.

[0049] For example, the genetic mutation determination function 354 determines mutations of one or more genes belonging to a co-occurrence group based on the estimation result with the highest accuracy among the estimation results of one or more genes belonging to a co-occurrence group included in the mutation information. In other words, the genetic mutation determination function 354 acquires the accuracy of the estimation model M1 for each of the one or more genes included in the co-occurrence group based on the estimation accuracy table 343 indicating the accuracy of the estimation model M1 that estimates genetic mutations. The genetic mutation determination function 354 determines the mutation of the gene to which the estimation model M1 with the highest accuracy belongs based on the genetic mutation estimation result corresponding to the estimation model M1 with the highest accuracy.

[0050] Here, the first co-occurrence group and the second co-occurrence group will be described as examples.

[0051] 7 is a diagram showing the determination result of the first co-occurrence group according to this embodiment. As shown in FIG. 7, the accuracy of the estimation model M1 that estimates the mutation of gene A is 0.9, and the accuracy of the estimation model M1 that estimates the mutation of gene B is 0.5. The genetic mutation estimation result of the estimation model M1 of gene A, which has the highest accuracy in the first co-occurrence group, is 0 (no mutation). Therefore, the genetic mutation determination function 354 determines that genes A and B belonging to the first co-occurrence group are not mutated.

[0052] 8 is a diagram showing the determination results of the second co-occurrence group according to this embodiment. As shown in FIG. 8, the accuracy of the estimation model M1 that estimates the mutation of gene C is 0.9, the accuracy of the estimation model M1 that estimates the mutation of gene D is 0.6, and the accuracy of the estimation model M1 that estimates the mutation of gene E is 0.5. The genetic mutation estimation result of the estimation model M1 of gene C, which has the highest accuracy in the second co-occurrence group, is 0 (no mutation). Therefore, although the genetic mutation estimation result of the estimation model M1 of gene D is 1 (mutation), the genetic mutation determination function 354 prioritizes the genetic mutation estimation result of the estimation model M1 of gene C and determines that genes C, D, and E belonging to the second co-occurrence group are not mutated.

[0053] The functional abnormality estimation function 355 estimates a functional abnormality of the subject based on the genetic mutation determination result by the genetic mutation determination function 354. The functional abnormality estimation function 355 is an example of an estimation unit. That is, the functional abnormality estimation function 355 estimates a functional abnormality of the subject contained in the medical image L1 acquired by the image acquisition function 351.

[0054] More specifically, the functional abnormality inferring function 355 infers, based on the gene ontology table 341, that there is a functional abnormality corresponding to the related gene to which the gene determined to have a mutation by the gene mutation determining function 354 belongs.

[0055] The display control function 356 displays on the display 33 a functional abnormality estimation image L2 that shows the functional abnormality of the subject estimated by the functional abnormality estimation function 355.

[0056] 9 is a diagram showing an example of a functional abnormality estimation image L2 according to this embodiment. The functional abnormality estimation image L2 is an image that notifies the content of the functional abnormality of the subject estimated by the functional abnormality estimation function 355. The functional abnormality estimation image L2 has a medical image L1 and an estimation result image L3. The medical image L1 is an image acquired by the image acquisition function 351. In other words, the medical image L1 is an image input to the estimation model M1. The estimation result image L3 is an image that distinguishes between content with functional abnormality estimated by the functional abnormality estimation function 355 and content without functional abnormality.

[0057] The estimation result image L3 includes the content of the functional abnormality and the associated genes related to the functional abnormality as the content of the functional abnormality, and the estimation result image L3 includes the content of the estimated functional abnormality and the associated genes related to the estimated functional abnormality as the content of the functional abnormality not present.

[0058] The estimation result image L3 does not have to include information indicating that there is no functional abnormality. Furthermore, if there is no item estimated to have a functional abnormality by the functional abnormality estimation function 355, the estimation result image L3 may be an image indicating that there is no functional abnormality. In this case, the estimation result image L3 may not include information about the estimated functional abnormality or any associated genes related to the estimated functional abnormality.

[0059] Next, the functional abnormality estimation process executed by the medical information processing apparatus 30 according to this embodiment will be described.

[0060] FIG. 10 is a flowchart showing an example of a functional abnormality estimation process executed by the medical information processing apparatus 30 according to this embodiment.

[0061] The image acquisition function 351 acquires a medical image L1 of a subject (step S1).

[0062] The gene mutation estimation function 352 estimates a gene mutation by inputting the acquired medical image L1 into an estimation model M1 that estimates a gene mutation (step S2).

[0063] The co-occurrence classification function 353 classifies genes belonging to related genes into one or more co-occurrence groups based on the co-occurrence table 342 that indicates the co-occurrence of gene mutations (step S3).

[0064] The gene mutation determination function 354 determines, for each co-occurrence group, mutations in the genes belonging to the co-occurrence group based on the gene mutation information estimated by the gene mutation estimation function 352 (step S4).

[0065] The functional abnormality inferring function 355 infers the presence or absence of a functional abnormality based on the presence or absence of a gene mutation determined by the gene mutation determining function 354 (step S5).

[0066] The display control function 356 displays a functional abnormality estimation image L2 on the display 33 based on the presence or absence of a functional abnormality estimated by the functional abnormality estimation function 355 (step S6).

[0067] With the above, the medical information processing device 30 ends the functional abnormality estimation process.

[0068] As described above, the medical information processing device 30 according to this embodiment analyzes a medical image L1 of a subject to obtain mutation information containing estimated results of mutations in multiple genes of the subject. The medical information processing device 30 also classifies multiple genes into one or more co-occurrence groups based on a co-occurrence table 342 that indicates the co-occurrence of mutations between genes. The medical information processing device 30 also determines, for each co-occurrence group, mutations in one or more genes belonging to the co-occurrence group based on the mutation information. The medical information processing device 30 then estimates functional abnormalities in the subject based on the genetic mutation determination results.

[0069] Here, the genetic mutation determination function 354 determines genetic mutations for each co-occurrence group. Therefore, even if the accuracy of the estimation model M1 for estimating a mutation in a certain gene is insufficient, the genetic mutation determination function 354 can compensate for this by using the estimation model M1 for estimating mutations in other genes. Therefore, the medical information processing device 30 can improve the accuracy of estimating functional abnormalities.

[0070] (Variation 1) The genetic mutation determination function 354 determines whether or not a gene belonging to a co-occurrence group is mutated based on the proportion of estimation results for each of the multiple genes belonging to the co-occurrence group, based on the genetic mutation information estimated by the genetic mutation estimation function 352. That is, the genetic mutation determination function 354 determines mutations in one or more genes belonging to the co-occurrence group based on the proportion of estimation results that indicate genetic mutations among the estimation results for each of the one or more genes belonging to the co-occurrence group, which are included in the mutation information.

[0071] More specifically, the gene mutation determination function 354 determines that the genes belonging to a co-occurrence group are mutated if the genes estimated to be mutated based on the gene mutation information estimated by the gene mutation estimation function 352 are the majority of the genes in the co-occurrence group.

[0072] The gene mutation determination function 354 uses the gene mutation information estimated by the gene mutation estimation function 352 to determine whether the number of genes estimated to be mutated is the majority of the co-occurrence group using the following formula (1).

[0073]

number

[0074] Here, the first co-occurrence group and the second co-occurrence group will be described as examples.

[0075] 11 is a diagram showing the determination result of the first co-occurrence group according to Modification 1. As shown in FIG. 11, according to the mutation information, gene A is estimated to have no mutation, and gene B is estimated to have no mutation. Therefore, in the first co-occurrence group, the estimated number of mutations, which indicates the total number of genes estimated to have mutations, is 0. In addition, the threshold for the majority in the first co-occurrence group is 1. Therefore, since the estimated number of mutations is 0, which is less than the threshold of 1, the genetic mutation determination function 354 determines that the genes belonging to the first co-occurrence group have no mutation.

[0076] 12 is a diagram showing the determination result of the second co-occurrence group according to Modification 1. As shown in FIG. 12, according to the mutation information, gene C is estimated to have no mutation, gene D is estimated to have a mutation, and gene E is estimated to have no mutation. Therefore, in the second co-occurrence group, the estimated number of mutations, which indicates the total number of genes estimated to have a mutation, is 1. The threshold for the majority in the second co-occurrence group is 1.5. Therefore, since the estimated number of mutations is 1, which is less than the threshold of 1.5, the genetic mutation determination function 354 determines that the genes belonging to the second co-occurrence group have no mutation.

[0077] (Variation 2) The genetic mutation determination function 354 weights the estimation results of one or more genes belonging to a co-occurrence group, which are included in the genetic mutation information estimated by the genetic mutation estimation function 352, according to the accuracy of each estimation result. The genetic mutation determination function 354 then determines mutations in one or more genes belonging to a co-occurrence group based on the proportion of estimation results that indicate genetic mutations among the weighted estimation results of one or more genes belonging to a co-occurrence group. In other words, the genetic mutation determination function 354 determines whether or not a gene belonging to a co-occurrence group is mutated based on the genetic mutation information estimated by the genetic mutation estimation function 352, using a combination of the weighted estimation results of the multiple genes belonging to the co-occurrence group.

[0078] More specifically, the genetic mutation determination function 354 weights the inference result of each genetic mutation included in the mutation information by the accuracy of the inference model M1 that inferred the genetic mutation, based on the estimation accuracy table 343 that indicates the accuracy of the inference model M1. Then, if the weighted inference results are in the majority of the co-occurrence groups, the genetic mutation determination function 354 determines that the genes belonging to the co-occurrence groups are mutated.

[0079] The gene mutation determination function 354 uses the gene mutation information estimated by the gene mutation estimation function 352 and the estimation accuracy table 343 to determine whether the weighted estimation result is the majority of the co-occurrence groups using the following formula (2).

[0080]

number

[0081] Here, the first co-occurrence group and the second co-occurrence group will be described as examples.

[0082] Fig. 13 is a diagram showing the determination result of the first co-occurrence group according to Modification Example 2. As shown in Fig. 13, according to the mutation information, gene A is estimated to have no mutation, and gene B is estimated to have no mutation. Furthermore, according to the estimation accuracy table 343, the accuracy of the estimation model M1 that estimated the mutation of gene A is 0.9, and the accuracy of the estimation model M1 that estimated the mutation of gene B is 0.5.

[0083] Since gene A has no mutations and gene B has no mutations, the estimated number of mutations after weighting is 0. Furthermore, in the first co-occurrence group, the accuracy of the estimation model M1 that estimated the mutation of gene A is 0.9, and the accuracy of the estimation model M1 that estimated the mutation of gene B is 0.5, so the threshold for the majority is 0.7. Since the estimated number of mutations is 0, which is less than the threshold of 0.7, the genetic mutation determination function 354 determines that the genes belonging to the first co-occurrence group have no mutations.

[0084] Fig. 14 is a diagram showing the determination results of the second co-occurrence group according to Modification Example 2. As shown in Fig. 14, according to the mutation information, gene C is estimated to have no mutation, gene D is estimated to have a mutation, and gene E is estimated to have no mutation. Furthermore, according to the estimation accuracy table 343, the accuracy of the estimation model M1 that estimated the mutation of gene C is 0.9, the accuracy of the estimation model M1 that estimated the mutation of gene D is 0.6, and the accuracy of the estimation model M1 that estimated the mutation of gene E is 0.5.

[0085] Since gene C has no mutation, gene D has a mutation, and gene E has no mutation, the estimated number of mutations after weighting is 0.6. Furthermore, in the second co-occurrence group, the accuracy of the estimation model M1 that estimated the mutation of gene C is 0.9, the accuracy of the estimation model M1 that estimated the mutation of gene D is 0.6, and the accuracy of the estimation model M1 that estimated the mutation of gene E is 0.5, so the majority threshold is 1.0. Since the estimated number of mutations is 0.6, which is less than the threshold of 1.0, the genetic mutation determination function 354 determines that the genes belonging to the second co-occurrence group have no mutations.

[0086] (Variation 3) The genetic mutation determination function 354 calculates the probability that the estimation result includes a false positive and the probability that the estimation result includes a false negative, based on each confusion matrix corresponding to the estimation result of one or more genes belonging to a co-occurrence group, which is included in the genetic mutation information estimated by the genetic mutation estimation function 352. Then, the genetic mutation determination function 354 determines the mutation of one or more genes belonging to a co-occurrence group based on the probability that the estimation result includes a false positive and the probability that the estimation result includes a false negative.

[0087] For example, the genetic mutation determination function 354 determines whether or not genes belonging to a co-occurrence group are mutated based on conditional probabilities calculated by Naive Bayes. For example, the genetic mutation determination function 354 calculates the probability that a false positive and a false negative are included in the mutation information including the estimation result of whether or not each gene is mutated.

[0088] Then, the gene mutation determination function 354 determines whether or not a gene belonging to a co-occurrence group is mutated based on the probability that the mutation information contains a false positive and the probability that the mutation information contains a false negative.

[0089] For example, the gene mutation determination function 354 determines that a gene belonging to a co-occurrence group is not mutated when the probability of including false positives is higher than the probability of including false negatives. On the other hand, the gene mutation determination function 354 determines that a gene belonging to a co-occurrence group is mutated when the probability of including false positives is lower than the probability of including false negatives.

[0090] More specifically, the genetic mutation determination function 354 substitutes the accuracy of the estimation model M1 corresponding to the genetic mutation estimation result included in the mutation information into the following mathematical formula (3) showing Naive Bayes.

[0091]

number

[0092] Here, an example will be described in which genes F, G, and H belong to a co-occurrence group. In addition, in the mutation information, the genetic mutation estimation result indicating the estimated result of the mutation of gene F indicates that there is no mutation, the genetic mutation estimation result indicating the estimated result of the mutation of gene G indicates that there is no mutation, and the genetic mutation estimation result indicating the estimated result of the mutation of gene H indicates that there is a mutation.

[0093] Furthermore, the gene mutation determination function 354 obtains the value to be substituted into the formula (3) from the confusion matrix of the estimation model M1 that estimates gene mutation.

[0094] Fig. 15 is a diagram showing an example of a confusion matrix of an estimation model M1 that estimates a mutation of a gene F according to Modification Example 3. Fig. 15 shows that the TN (True Negative) of the estimation model M1 that estimates a mutation of a gene F is 0.9, the FP (False Positive) is 0.1, the FN (False Negative) is 0.1, and the TP (True Positive) is 0.9.

[0095] Fig. 16 is a diagram showing an example of a confusion matrix of an estimation model M1 that estimates a mutation of a gene G according to Modification Example 3. Fig. 16 shows that the TN of the estimation model M1 that estimates a mutation of a gene G is 0.6, the FP is 0.4, the FN is 0.5, and the TP is 0.5.

[0096] 17 is a diagram showing an example of a confusion matrix of an estimation model M1 that estimates a mutation of a gene H according to Modification Example 3. In Fig. 17, the TN of the estimation model M1 that estimates a mutation of a gene H is 0.5, the FP is 0.5, the FN is 0.6, and the TP is 0.4.

[0097] Since the genetic mutation estimation result of gene F indicates no mutation, the genetic mutation estimation result of gene G indicates no mutation, and the genetic mutation estimation result of gene H indicates a mutation, the genetic mutation determination function 354 substitutes TN in Figure 15, TN in Figure 16, and FP in Figure 17 into formula (3). That is, the genetic mutation determination function 354 calculates the following formula (4).

[0098]

number

[0099] As a result, the genetic mutation determination function 354 calculates the probability that the genetic mutation estimation result of the gene F, the genetic mutation estimation result of the gene G, and the genetic mutation estimation result of the gene H contains a false positive.

[0100] Furthermore, the genetic mutation determination function 354 substitutes FN in Fig. 15, FN in Fig. 16, and TP in Fig. 17 into the formula (3). That is, the genetic mutation determination function 354 calculates the following formula (5).

[0101]

number

[0102] As a result, the genetic mutation determination function 354 calculates the probability that the mutation information including the genetic mutation estimation result of the gene F, the genetic mutation estimation result of the gene G, and the genetic mutation estimation result of the gene H contains a false negative.

[0103] Furthermore, the genetic mutation determination function 354 compares the formula (4) with the formula (5). That is, the genetic mutation determination function 354 compares whether the probability that the mutation information contains a false positive or the probability that the mutation information contains a false negative is higher.

[0104] When the probability that the mutation information contains false positives is higher than the probability that the mutation information contains false negatives, the genetic mutation estimation result indicating the presence of a mutation contained in the mutation information is likely to be incorrect. Therefore, the genetic mutation determination function 354 determines that a gene belonging to a co-occurrence group is not mutated when the probability that the mutation information contains false positives is higher than the probability that the mutation information contains false negatives.

[0105] On the other hand, when the probability that the mutation information contains a false positive is lower than the probability that the mutation information contains a false negative, the genetic mutation estimation result that the mutation information contains no mutation is likely to be incorrect. Therefore, the genetic mutation determination function 354 determines that a gene belonging to a co-occurrence group is mutated when the probability that the mutation information contains a false positive is lower than the probability that the mutation information contains a false negative.

[0106] According to at least one of the embodiments described above, it is possible to improve the accuracy of estimating a functional abnormality.

[0107] Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, modifications, and combinations of embodiments can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, as well as within the scope of the invention and its equivalents as defined in the claims. [Explanation of symbols]

[0108] 1. Medical diagnosis support system 10 Medical Image Management Systems (PACS: Picture Archiving and Communication Systems) 20 Gene Ontology Server 30 Medical information processing device 31 NW (Network) Interface 32 input interfaces 33 Display 34 memory 35 Processing circuit 40 Network 341 Gene Ontology Table 342 Co-occurrence Table 343 Estimated Accuracy Table 351 Image Acquisition Function 352 Gene mutation estimation function 353 Co-occurrence classification function 354 Gene mutation detection function 355 Malfunction Estimation Function 356 Display Control Function L1 Medical Imaging L2 dysfunction estimation image L3 Estimation result image M1 estimation model G1 related gene graph G21 First Co-occurrence Graph G22 Second Co-occurrence Graph

Claims

1. acquiring a medical image of the subject; By analyzing the medical image, mutation information including estimated results of mutations in a plurality of genes of the subject is obtained; classifying the plurality of genes into one or more co-occurrence groups based on co-occurrence information indicating co-occurrence of mutations between genes; determining, for each of the co-occurrence groups, mutations in one or more genes belonging to the co-occurrence group based on the mutation information; Inferring functional abnormality of the subject based on the result of the genetic mutation determination. A medical diagnosis support method comprising:

2. determining mutations in one or more genes belonging to the co-occurrence group based on the estimation result with the highest accuracy among the estimation results for the one or more genes belonging to the co-occurrence group included in the mutation information; The medical diagnosis support method according to claim 1 .

3. determining mutations in one or more genes belonging to the co-occurrence group based on a proportion of the estimation results indicating gene mutations among the estimation results of each of the one or more genes belonging to the co-occurrence group, which are included in the mutation information; The medical diagnosis support method according to claim 1 .

4. weighting each of the estimation results of one or more genes belonging to the co-occurrence group included in the mutation information according to the accuracy of each of the estimation results; determining mutations of one or more genes belonging to the co-occurrence group based on a proportion of the estimation results indicating gene mutations among the weighted estimation results of each of the one or more genes belonging to the co-occurrence group; The medical diagnosis support method according to claim 1 .

5. calculating a probability that the estimation result includes a false positive and a probability that the estimation result includes a false negative based on confusion matrices corresponding to the estimation results of one or more genes belonging to the co-occurrence group included in the mutation information; determining mutations in one or more genes belonging to the co-occurrence group based on the probability that the estimation result includes a false positive and the probability that the estimation result includes a false negative; The medical diagnosis support method according to claim 1 .

6. Displaying the estimated content of the functional abnormality of the subject. The medical diagnosis support method according to claim 1 .

7. Displaying the genes estimated to be associated with the functional abnormality of the subject. The medical diagnosis support method according to claim 1 .

8. Accepting designation of a part of the subject; acquiring the mutation information including the estimated results of mutations in a plurality of genes corresponding to the site by analyzing the medical image; The medical diagnosis support method according to any one of claims 1 to 7.

9. a first acquisition unit that acquires a medical image of a subject; a second acquisition unit that acquires mutation information including estimated results of mutations in a plurality of genes of the subject by analyzing the medical image; a classification unit that classifies the plurality of genes into one or more co-occurrence groups based on co-occurrence information indicating co-occurrence of mutations between genes; a determination unit that determines, for each of the co-occurrence groups, mutations in one or more genes belonging to the co-occurrence group based on the mutation information; an estimation unit that estimates a functional abnormality of the subject based on the determination result of the gene mutation; A medical diagnosis support system comprising:

10. To the computer acquiring a medical image of the subject; By analyzing the medical image, mutation information including estimated results of mutations in a plurality of genes of the subject is obtained; classifying the plurality of genes into one or more co-occurrence groups based on co-occurrence information indicating co-occurrence of mutations between genes; determining, for each of the co-occurrence groups, mutations in one or more genes belonging to the co-occurrence group based on the mutation information; Inferring functional abnormality of the subject based on the result of the genetic mutation determination. A program to make it happen.

Images