Method, system, and program for analyzing functional connection of brain

Abstract

To provide a novel method to analyze functional connection of the brain capable of improving accuracy of estimating a state of an object.SOLUTION: There is provided a method to analyze functional connection of the brain. The method includes: receiving an image of the brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; estimating, for each region of the plurality of regions, a plurality of signal values of each of a plurality of other regions based on the plurality of signal values of the region; calculating, for each region of the plurality of regions, a correlation value between the plurality of signal values estimated for the region and the plurality of signal values of the region in the image; and deriving a pattern of functional connection of the brain based on the correlation value.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a method, system, and program for analyzing functional connectivity of the brain, and further to a method for estimating the state of a subject using the same. [Background technology]

[0002] Research is being conducted to analyze the functional connectivity of the brain. Functional connectivity refers to the correlation between neural activity in one area of ​​the brain and neural activity in another area of ​​the brain. In particular, recent research has shown that human characteristics and behavior can be inferred from resting-state functional connectivity (correlation of neural activity in the brain when not performing a task) (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-37397 Summary of the Invention [Problem to be solved by the invention]

[0004] The applicant faced the problem that existing techniques for analyzing the functional connectivity of the brain to estimate the state of a subject (e.g., human characteristics or behavior, or mental illness, etc.) do not provide sufficient estimation accuracy and have limitations in their analysis methods.

[0005] The present invention has been made in consideration of the above-mentioned problems, and aims to provide a novel method for analyzing functional connectivity of the brain that can improve the accuracy of estimating the state of a subject. [Means for solving the problem]

[0006] The present invention provides, for example, a method for analyzing brain functional connectivity by utilizing localized, fine-grained information from brain images.

[0007] The present invention provides, for example, the following items. (Item 1) 1. A method for analyzing brain functional connectivity, comprising: receiving an image of the brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; Deriving a pattern of functional connectivity of the brain based on the correlation value. A method comprising: (Item 2) said receiving including receiving a plurality of images of the brain; estimating a plurality of signal values ​​for each of the other plurality of regions includes estimating one signal value for each of the other plurality of regions, for each of the plurality of signal values; estimating the one signal value calculating a regression coefficient of a linear regression using a portion of the plurality of images, with each of the plurality of signal values ​​of the region as an explanatory variable and the one signal value as a response variable; calculating the one signal value by multiplying each of a plurality of signal values ​​of the region by the regression coefficient using the remainder of the plurality of images; The method according to the above item, comprising: (Item 3) 2. The method according to claim 1, wherein the plurality of images are time-series images. (Item 4) Calculating the correlation value calculating a correlation value between a pattern formed by the estimated plurality of signal values ​​and a pattern formed by the plurality of signal values ​​of the region in the image; The method according to any one of the preceding items, comprising: (Item 5) Deriving the pattern comprises: creating a correlation matrix representing correlation values ​​for each of the plurality of regions; The method according to any one of the preceding items, comprising: (Item 6) The method according to any one of the preceding items, wherein the brain image is an image of the brain at rest. (Item 7) A method for creating a classifier for estimating a state of an object, comprising: receiving brain images acquired from each of a plurality of subjects; Deriving a pattern of functional brain connectivity from each of the images of the plurality of subjects according to the method described in any one of the preceding items; learning the state of each of the plurality of subjects and the pattern of functional connectivity of the brain; A method comprising: (Item 8) A method for creating a classifier for estimating a state of an object, comprising: Preparing a first classifier created by the method according to any one of the above items; preparing a second classifier, the second classifier comprising: deriving a second pattern of functional brain connectivity from each of the images of the plurality of subjects according to a second method; learning a state of each of the plurality of subjects and a second pattern of functional connectivity of the brain; and creating a composite classifier by integrating the first classifier and the second classifier; wherein the second method comprises: calculating an average value for each of the plurality of regions by averaging a plurality of signal values ​​for the region; calculating a second correlation value between the average value of each of the plurality of regions and the average values ​​of each of the other plurality of regions; deriving a second pattern of brain functional connectivity based on the second correlation value; and A method comprising: (Item 9) 1. A method for estimating a state of a subject, comprising: Preparing a classifier created according to the method described in item 7 or 8; Inputting a brain functional connectivity pattern derived from a subject's brain image according to the method described in any one of items 1 to 6 into the classifier; obtaining an output from the classifier; A method comprising: (Item 10) A system for analyzing brain functional connectivity, comprising: a receiving means for receiving an image of the brain, the image being divided into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; an estimation means for estimating, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of the region; a correlation means for calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; a derivation means for deriving a pattern of functional connectivity of the brain based on the correlation value; A system comprising: (Item 10A) Item 11. A system according to item 10, comprising the features according to any of the preceding items. (Item 11) A system for estimating a state of an object, comprising: receiving means for receiving an image of the subject's brain, the image being divided into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; an estimation means for estimating, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of the region; a correlation means for calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; a derivation means for deriving a pattern of functional connectivity of the brain based on the correlation value; a classifier that estimates a state based on the derived pattern; A system comprising: (Item 11A) Item 12. A system according to item 11, comprising the features according to any of the preceding items. (Item 12) A system for estimating a state of an object, comprising: receiving means for receiving an image of the subject's brain, the image being divided into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; an estimation means for estimating, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of the region; a first correlation means for calculating, for each of the plurality of regions, a first correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; a first derivation means for deriving a first pattern of functional connectivity of the brain based on the first correlation value; a calculation means for calculating an average value for each of the plurality of regions by averaging a plurality of signal values ​​for the region; second correlation means for calculating a second correlation value between the average value of each of the plurality of regions and the average values ​​of the other plurality of regions; a second derivation means for deriving a second pattern of functional connectivity of the brain based on the second correlation value; a discriminator that estimates a state based on the first pattern and the second pattern; A system comprising: (Item 12A) Item 13. A system according to item 12, comprising the features according to any of the preceding items. (Item 13) A program for analyzing brain functional connectivity, the program being executed in a computer system having a processor unit, the program comprising: receiving an image of the brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; Deriving a pattern of functional connectivity of the brain based on the correlation value. A program that causes the processor to execute a process including the steps of: (Item 13A) Item 14. A program according to item 13, comprising the features according to any one of the preceding items. (Item 14) A program for estimating a state of an object, the program being executed in a computer system including a processor unit, the program comprising: receiving an image of the subject's brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; Deriving a pattern of brain functional connectivity based on the correlation value; estimating a state based on the derived pattern; and A program that causes the processor to execute a process including the steps of: (Item 14A) Item 15. A program according to item 14, comprising the features according to any one of the preceding items. (Item 15) A program for estimating a state of an object, the program being executed in a computer system including a processor unit, the program comprising: receiving an image of the subject's brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a first correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; deriving a first pattern of brain functional connectivity based on the first correlation value; calculating an average value for each of the plurality of regions by averaging a plurality of signal values ​​for the region; calculating a second correlation value between the average value of each of the plurality of regions and the average values ​​of each of the other plurality of regions; deriving a second pattern of brain functional connectivity based on the second correlation value; estimating a state based on the first pattern and the second pattern; A program that causes the processor to execute a process including the steps of: (Item 15A) Item 16. A program according to item 15, comprising the features according to any one of the preceding items. [Effects of the Invention]

[0008] According to the present invention, a novel method for analyzing functional connectivity of the brain can be provided, and the analysis results of this method can be used to estimate the state of a subject with higher accuracy than existing analytical methods. Furthermore, it can also clarify aspects of functional connectivity of the brain that could not be determined by existing analytical methods. [Brief explanation of the drawings]

[0009] [Figure 1] A diagram showing an example of a flow for predicting the state of an object [Figure 2A] FIG. 1 is a diagram for explaining an example of the method of the present invention. [Figure 2B] FIG. 1 is a diagram for explaining an example of the method of the present invention. [Figure 3A] A diagram showing an example of the functional connection pattern created [Figure 3B] Functionally connected brain regions are color-coded based on the functional connectivity patterns created. [Figure 4A] FIG. 1 shows the results of estimation by a classifier constructed using functional connection patterns created according to a conventional method, the results of estimation by a classifier constructed using functional connection patterns created according to the method of the present invention, and the results of estimation by a composite classifier. [Figure 4B] 1 is a diagram showing the results of estimation by a classifier constructed using a functional connection pattern created according to a conventional method and the results of estimation by a composite classifier. [Figure 5] FIG. 1 is a block diagram showing an example of the configuration of a system 100 according to the present invention. [Figure 6A] A block diagram showing an example of the configuration of a processor unit 120. [Figure 6B] 1 is a block diagram showing another example of the configuration of the processor unit 120. [Figure 6C] 10 is a block diagram showing another example of the configuration of the processor unit 120. [Figure 7A] A block diagram showing an example of the configuration of a processor unit 120. [Figure 7B] 1 is a block diagram showing another example of the configuration of the processor unit 120. [Figure 8] A flowchart showing an example of processing (process 800) by the system 100 of the present invention. [Figure 9] A flowchart showing an example of processing (process 900) by the system 100 of the present invention. [Figure 10] A flowchart showing an example of processing (process 1000) by the system 100 of the present invention. [Figure 11] A flowchart showing an example of processing (process 1100) by the system 100 of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0010] (1.Definition) As used herein, the term "target" refers to any entity whose condition is to be estimated. The target is preferably a human being, but may also be an animal.

[0011] In this specification, the term "subject" refers to any entity from which training data is obtained to be used in a learning process for constructing a classifier that is used to estimate the state of an object. The subject is preferably a human being, but may also be an animal.

[0012] As used herein, the "condition" of a subject or test subject includes the psychological condition of the subject or test subject (e.g., the state of a mental disease, the state of performance in a psychological task, etc.), the physical condition of the subject or test subject (e.g., the state of a physical disease, the state of performance in a physical task, etc.), and the nature of the subject or test subject (e.g., age, sex, etc.).

[0013] In this specification, "estimating" a state may be a concept that includes predicting a future state in addition to estimating a current state.

[0014] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0015] (2. Prediction of the target state) Figure 1 shows an example of a flow for predicting the state of a subject. In this example, we will explain how to predict the state of a subject T from brain images acquired from the subject T (specifically, brain images acquired by fMRI (functional MRI)).

[0016] First, an image of the brain of subject T is acquired. The image of the brain of the subject is acquired, for example, by an MRI device. In step S1, the acquired image is imported into a terminal device of doctor D. The image is provided from the MRI device to the terminal device in any manner.

[0017] Once the image is captured, in step S2, doctor D provides the image via the terminal device to the system 100. The image may be provided from the terminal device to the system 100 in any manner.

[0018] The system 100 performs a process of predicting the state of a subject based on the provided image. The system 100 can analyze the functional connectivity of the brain using brain images and predict the state of the subject from the analysis results using a novel method described below. Furthermore, the novel method of the present invention can also reveal new aspects of the functional connectivity of the brain that could not be revealed by conventional methods.

[0019] In step S3, the predicted state of the subject is provided to the terminal device of Doctor D. Doctor D can then make a diagnosis by referring to the prediction of the subject's state. This can facilitate the diagnosis of mental illnesses and the like that are difficult for inexperienced doctors to diagnose, for example.

[0020] In the above example, the MRI apparatus, the terminal device of Doctor D, and the system 100 are separate entities, but the present invention is not limited to the above example. For example, the MRI apparatus and the terminal device may be integrated, or the terminal device and the system 100 may be integrated, or the MRI apparatus, the terminal device, and the system 100 may be integrated.

[0021] Unlike conventional methods that average local fine-scale information, the novel method of the present invention utilizes local fine-scale information from brain images to analyze brain functional connectivity.

[0022] In conventional methods, as described in the aforementioned Patent Document 1 (JP 2021-37397 A), the average waveform of each region of interest is calculated, and the correlation of these average waveforms between regions of interest is calculated. This is done for all region pairs to create a correlation matrix or a pattern of brain functional connectivity. Figure 3A(a) shows a functional connectivity pattern created by the conventional method. In the conventional method, the information from each region is averaged, so local information within each region is suppressed.

[0023] The method of the present invention utilizes multiple signal values ​​in multiple regions of a brain image without averaging them. Specifically, a regression equation for estimating multiple signal values ​​in one region from multiple signal values ​​in another region is derived from the brain image, and whether the regression equation can accurately estimate the multiple signal values ​​in the other region is evaluated. Based on the evaluation, a correlation matrix or a pattern of brain functional connectivity is created.

[0024] 2A and 2B are diagrams for explaining an example of the method of the present invention.

[0025] First, brain images 10 are acquired in time series as shown in Figure 2A. The brain images 10 are preferably represented as three-dimensional data.

[0026] The brain image 10 has multiple regions (e.g., region A, region B, region C, region D, etc.). The multiple regions of the brain image 10 correspond to multiple regions of the brain. The regions may be three-dimensional regions (i.e., space). In one example, the brain image 10 may be divided into 360 regions based on anatomical and functional knowledge. Each of the multiple regions is made up of multiple vertices, and each vertex has a signal value (note that the vertices may also be expressed as voxels). The signal value may represent the signal strength (e.g., the strength of neural activity) of the part corresponding to the vertex.

[0027] In the example shown in FIG. 2A, attention is focused on areas A and B, but similar explanations apply to other areas.

[0028] Region A and region B each have multiple vertices. For example, region A has X vertices, and region B has Y vertices, each of which has a signal value. In the example shown in FIG. 2A, for ease of illustration, the X vertices and the Y vertices are represented one-dimensionally, i.e., represented by X pieces of data and Y pieces of data. For example, a1 represents the signal value of vertex 1 in region A, a2 represents the signal value of vertex 2 in region A, and so on. X represents the signal value of vertex X in region A. For example, b1 represents the signal value of vertex 1 in region B, b2 represents the signal value of vertex 2 in region B, and so on. Y represents the signal value of vertex Y in region B.

[0029] The set of time-series brain images shown in Figure 2A may be images acquired in a single measurement, but the method of the present invention may also acquire brain images in multiple measurements and utilize multiple sets of brain images.

[0030] For example, a set of brain images is divided into two groups by separating them at appropriate time points: one group is used to derive a regression equation, and the other group is used to evaluate the regression equation.

[0031] FIG. 2B is a diagram for explaining the derivation and evaluation of a regression equation. For simplicity of explanation, in the example shown in FIG. 2B, each of the multiple regions has three vertices, and the time series has three time points. Each of the three vertices has a signal value, and in FIG. 2A, the signal values ​​are represented by different shadings.

[0032] First, a regression equation is derived. As mentioned above, the first group of brain images is used to derive the regression equation.

[0033] The left diagram of Figure 2A illustrates the derivation of a regression equation for estimating multiple signal values ​​in a second region from multiple signal values ​​in a first region of a brain image. That is, at a certain time point, the signal value of the first vertex (Target vertex 1) in the second region is related to each of the signal values ​​of multiple vertices (Seed vertices) in the first region by their respective regression coefficients β n,1 The regression coefficient β n,1 The regression equation is derived by solving for

[0034] Similarly, at a certain time point, the signal value of the second vertex (Target vertex 2) in the second region is related to each of the signal values ​​of the multiple vertices (Seed vertices) in the first region by the respective regression coefficients β n,2 The regression coefficient β n,2 The regression equation is derived by calculating the regression coefficient β n,3 The regression coefficient β n,3 A regression equation can be derived by solving for: A regression equation can be derived for each time point.

[0035] If we generalize the regression equation to be derived according to the example of Figure 2A, the signal value b of vertex J (1 ≤ J ≤ Y) in region B is J is the signal value a in region A i Using

number

[0036] The regression equation is then evaluated. As mentioned above, the second set of brain images is used to evaluate the regression equation.

[0037] The right diagram of Figure 2A explains that the signal values ​​of multiple vertices in the second region are estimated by substituting the signal values ​​of multiple vertices in the first region into the derived regression equation. That is, according to the regression equation for the signal value of the first vertex (Target vertex 1) at a certain point in time, each of the signal values ​​of multiple vertices in the first region and their respective regression coefficients β n,1 The sum of products with estimates the signal value of the first vertex in the second region.

[0038] Similarly, according to the regression equation for the signal value of the second vertex, each of the signal values ​​of the multiple vertices in the first region and the respective regression coefficients β n,2 The signal value of the second vertex in the second region is estimated by the sum of products of the respective regression coefficients β n,3 The signal value of the third vertex in the second region can be estimated by the sum of products with

[0039] Following the example of Figure 2A, the signal value a at the vertex of region A i of

number

[0040] The signal values ​​of the vertices estimated for the second region may form an estimated activity pattern, as shown in the lower right diagram of FIG. 2A . The estimated activity pattern is correlated with the activity pattern of the actual signal values ​​(i.e., the signal values ​​of the vertices in the second region of the second group of images) to calculate a correlation coefficient. For example, a high correlation coefficient (e.g., close to 1) indicates that the signal values ​​estimated by the regression equation (and thus the estimated activity pattern) are correlated with the actual signal values ​​(and thus the actual activity pattern), and the regression equation has a sufficiently high accuracy. This may indicate that the first region and the second region are correlated, i.e., that there is a high degree of functional coupling between the first region and the second region. For example, a low correlation coefficient (e.g., close to 0) indicates that the signal values ​​estimated by the regression equation (and thus the estimated activity pattern) are not correlated with the actual signal values ​​(and thus the actual activity pattern), and the regression equation has a low accuracy. This may indicate that the first and second regions are not correlated, i.e., the degree of functional coupling between the first and second regions is low. Thus, these correlation coefficients may represent the degree of functional coupling between the first and second regions.

[0041] By deriving and evaluating the regression equations described above for each of the multiple regions and each of the other multiple regions, the functional connectivity between each of the multiple regions and each of the other multiple regions can be clarified. By arranging the correlation coefficients between each of the multiple regions and each of the other multiple regions in a matrix, a correlation matrix or a pattern of functional connectivity in the brain can be created.

[0042] For example, following the example of Figure 2A, the correlation coefficient when estimating the signal value of region B from the signal value of region A, the correlation coefficient when estimating the signal value of region C from the signal value of region A, the correlation coefficient when estimating the signal value of region D from the signal value of region A,...the correlation coefficient when estimating the signal value of region A from the signal value of region B, the correlation coefficient when estimating the signal value of region C from the signal value of region B, the correlation coefficient when estimating the signal value of region D from the signal value of region B,...the correlation coefficient when estimating the signal value of region A from the signal value of region C, the correlation coefficient when estimating the signal value of region B from the signal value of region C, the correlation coefficient when estimating the signal value of region D from the signal value of region C,...are estimated, and by arranging these correlation coefficients in a matrix, a correlation matrix or a pattern of functional connectivity in the brain can be created.

[0043] For example, if a correlation matrix or functional connectivity pattern is created for each time point, the correlation matrices or functional connectivity patterns for each time point can be averaged and combined into a single correlation matrix or functional connectivity pattern.

[0044] Figure 3A is a diagram showing an example of a created functional connection pattern. As described above, Figure 3A(a) is a pattern created according to the conventional method. Figure 3A(b) is a pattern created according to the method of the present invention. Figure 3A is an example of a brain image having 360 regions.

[0045] In Figure 3A, the vertical and horizontal axes represent each region, and the shading represents the correlation coefficient between the two regions. In Figure 3A(a), values ​​closer to 1 indicate a positive correlation, values ​​closer to -1 indicate a negative correlation, and values ​​closer to 0 indicate no correlation. In Figure 3A(b), values ​​closer to 1 indicate a strong correlation, and values ​​closer to 0 indicate no correlation. The example shown in Figure 3A utilized data obtained from 995 subjects.

[0046] As shown in Figure 3A, the brain functional connectivity pattern created according to the method of the present invention is different from the brain functional connectivity pattern created according to the conventional method. In other words, by analyzing brain functional connectivity according to the method of the present invention, it is possible to discover functional connectivity patterns that could not be found by the conventional method. This is because the present invention derives the brain functional connectivity pattern by utilizing local, fine-grained information that is averaged in the conventional method.

[0047] Figure 3B shows functionally connected brain regions clustered and color-coded using a hierarchical clustering method based on the functional connectivity patterns created. Figure 3B(a) shows the color-coded pattern based on the conventional method, and Figure 3B(b) shows the color-coded pattern based on the method of the present invention.

[0048] As can be seen from Figure 3B, the brain functional connectivity pattern created according to the method of the present invention shows a different pattern from the pattern created according to the conventional method, although it has a tendency similar to that of the pattern created according to the conventional method. This means that the brain functional connectivity pattern created according to the method of the present invention can adequately capture the actual brain functional connectivity pattern, while also being able to find functional connectivity patterns that could not be found by the conventional method.

[0049] In this way, the method of the present invention can find functional connection patterns that could not be found by conventional methods. By using these functional connection patterns to estimate the state of an object, it is possible to estimate the state of an object that could not be estimated by conventional methods.

[0050] For example, a classifier that estimates the state of a target can be constructed by deriving a pattern of brain functional connectivity from brain images acquired from each of a plurality of subjects according to the technique of the present invention and performing a learning process using the states of each of the plurality of subjects as labels. For example, the classifier can be implemented by learning using a support vector machine (SVM). For example, the classifier can be implemented by learning using a neural network.

[0051] In addition to inferring the state of an object using a functional connection pattern created according to the technique of the present invention, it is also possible to infer the state of an object using, for example, both a functional connection pattern created according to the technique of the present invention and a functional connection pattern created according to a conventional technique.

[0052] For example, a first classifier that estimates the state of a target is constructed by deriving a pattern of functional brain connectivity from brain images acquired from each of a plurality of subjects according to the method of the present invention and performing a learning process using the states of each of the plurality of subjects as labels. Next, a second classifier that estimates the state of a target is constructed by deriving a pattern of functional brain connectivity from brain images acquired from each of the plurality of subjects according to a conventional method and performing a learning process using the states of each of the plurality of subjects as labels. Next, the first classifier and the second classifier are integrated by performing ensemble learning (weighted averaging) to construct a composite classifier.

[0053] 4A is a diagram showing the results of estimation by a classifier constructed using functional connection patterns created according to a conventional method, the results of estimation by a classifier constructed using functional connection patterns created according to the method of the present invention, and the results of estimation by a composite classifier. FIG. 4A shows the results of estimation of the performance state when a subject performed a task. The horizontal axis indicates the performance evaluation item, and the vertical axis indicates the estimation accuracy. Three bar graphs are shown for each item. From left to right, the three bar graphs show the results of estimation by a classifier constructed using functional connection patterns created according to a conventional method (lightest gray), the results of estimation by a classifier constructed using functional connection patterns created according to the method of the present invention (darkest gray), and the results of estimation by a composite classifier (medium gray).

[0054] 4A, the accuracy of estimation by the composite classifier is generally higher than that of estimation by a classifier constructed using functional connection patterns created according to conventional techniques. Furthermore, the accuracy of estimation by a classifier constructed using functional connection patterns created according to the technique of the present invention is higher in specific items such as processing speed, delay discounting, and verbal episodic memory than that of estimation by a classifier constructed using functional connection patterns created according to conventional techniques.

[0055] Figure 4B shows the results of estimation by a classifier constructed using functional connection patterns created according to a conventional method, and the results of estimation by a composite classifier. Figure 4B shows the results of estimating the subject's gender (A) and the results of estimating the subject's age (B). The vertical axis indicates the estimation accuracy. "Traditional" indicates the results of estimation by a classifier constructed using functional connection patterns created according to a conventional method, and "ensemble" indicates the results of estimation by the composite classifier.

[0056] As can be seen from Figure 4B, the accuracy of estimating the subject's gender was high for both the classifier constructed using the functional connectivity pattern created according to conventional methods and the composite classifier, with no significant difference. However, the accuracy of estimating the subject's age was shown to be significantly higher for the composite classifier.

[0057] In this way, by using the analysis results of the method of the present invention, it is generally possible to predict the state of a subject with higher accuracy than when the analysis results of conventional methods are used to predict the state of a subject, and it is also possible to estimate the state of a subject that could not be estimated by conventional methods. This is thought to be due to the fact that the analysis by the method of the present invention can discover functional connectivity patterns that could not be found by conventional methods.

[0058] The above-described system 100 may have, for example, the configuration described below.

[0059] (3. System Configuration) FIG. 5 shows an example of the configuration of a system 100 of the present invention.

[0060] The system 100 of the present invention may be implemented, for example, as a server device that communicates with a user's terminal device (for example, the terminal device used by Doctor D in FIG. 1), or may be implemented as the user's terminal device itself.

[0061] When the system 100 is implemented as a server device, the system 100 can communicate with a user's terminal device via any network. The network may be, for example, the Internet or a LAN. The network may be a wired network or a wireless network. The system 100 can receive and analyze brain images (specifically, fMRI images) from the user's terminal device.

[0062] When the system 100 is implemented as a user terminal device itself, the system 100 can communicate with an MRI device via any network, and can receive and analyze brain images from the MRI device.

[0063] The system 100 includes at least an interface unit 110, a processor unit 120, and a memory unit 150. The system 100 is connected to a database unit 200.

[0064] The interface unit 110 exchanges information with the outside of the system 100. The processor unit 120 of the system 100 can receive information from the outside of the system 100 and can send information to the outside of the system 100 via the interface unit 110. The interface unit 110 can exchange information in any format.

[0065] The interface unit 110 includes, for example, an input unit that allows information to be input to the system 100. It does not matter how the input unit allows information to be input to the system 100. For example, if the input unit is a receiver, the receiver may input information by receiving information from outside the system 100 via a network. Alternatively, if the input unit is a data reading device, the input unit may input information by reading information from a storage medium connected to the system 100.

[0066] The interface unit 110 includes, for example, an output unit that enables information to be output from the system 100. It does not matter in what manner the output unit enables information to be output from the system 100. For example, if the output unit is a transmitter, the transmitter may output information by transmitting it to an external device outside the system 100 via a network. Alternatively, if the output unit is a data writing device, the output unit may output information by writing it to a storage medium connected to the system 100.

[0067] The system 100 may also transmit information to and / or receive information from the database portion 200 via, for example, the interface portion 110 .

[0068] The system 100 can receive brain images, for example, via the interface unit 110. The brain images may be received from a user's terminal device, an MRI device, or the database unit 200, for example.

[0069] The system 100 can transmit an output (e.g., a derived brain functional connection pattern, a constructed classifier, or an estimated state of a subject) from the system 100 via, for example, the interface unit 110. The output from the system 100 may be transmitted to, for example, a user's terminal device or to the database unit 200.

[0070] The processor unit 120 executes the processing of the system 100 and controls the overall operation of the system 100. The processor unit 120 reads out a program stored in the memory unit 150 and executes the program. This allows the system 100 to function as a system that executes desired steps. The processor unit 120 may be implemented by a single processor or by multiple processors.

[0071] The memory unit 150 stores programs required to execute the processing of the system 100, data required to execute the programs, and the like. The memory unit 150 may store a program for causing the processor unit 120 to perform processing for analyzing the brain's functional connectivity (e.g., a program for realizing the processing shown in FIG. 8 or FIG. 9 described later) or a program for causing the processor unit 120 to perform processing for estimating the state of a subject (e.g., a program for realizing the processing shown in FIG. 10 or FIG. 11 described later). Here, how the program is stored in the memory unit 150 is not important. For example, the program may be pre-installed in the memory unit 150. Alternatively, the program may be installed in the memory unit 150 by being downloaded via a network. In this case, the type of network does not matter. The memory unit 150 may be implemented by any storage means. Alternatively, the program may be stored in a non-transitory computer-readable storage medium and installed in the memory unit 150 by reading the medium.

[0072] For example, the state of a subject and a pattern of functional connectivity of the brain for each of a plurality of subjects can be stored in association with each other in the database unit 200. The data stored in the database unit 200 can be used, for example, to create a classifier for estimating the state of the subject.

[0073] The database unit 200 may store the functional connection pattern of the subject's brain output from the system 100, the classifier, the subject's state, and the like.

[0074] In the example shown in FIG. 5 , the database unit 200 is provided outside the system 100, but the present invention is not limited to this. At least a portion of the database unit 200 can also be provided inside the system 100. In this case, at least a portion of the database unit 200 may be implemented by the same storage means as the storage means that implements the memory unit 150, or by a storage means different from the storage means that implements the memory unit 150. In either case, at least a portion of the database unit 200 is configured as a storage unit for the system 100. The configuration of the database unit 200 is not limited to a specific hardware configuration. For example, the database unit 200 may be configured as a single hardware component or multiple hardware components. For example, the database unit 200 may be configured as an external hard disk drive for the system 100, as cloud storage connected via a network, or as a distributed network using blockchain technology or the like.

[0075] The processor unit 120 may have the following configurations. Note that the following configurations are not mutually exclusive, and a combination of the following configurations is also possible, and the processor unit 120 may also have configurations other than those described below.

[0076] (3.1 Configuration of a system for analyzing brain functional connectivity) 6A shows an example of the configuration of the processor unit 120. In this example, the processor unit 120 has a configuration for analyzing functional connectivity of the brain, specifically, a configuration for analyzing functional connectivity of the brain by the method described above with reference to FIGS. 2A to 3B.

[0077] The processor unit 120 comprises receiving means 121, estimating means 122, correlating means 123 and deriving means .

[0078] The receiving means 121 is configured to receive an image of the brain of the subject or test subject. The receiving means 121 can receive the brain image received from outside the system 100 via the interface unit 110, for example. The brain image is divided into a plurality of regions corresponding to a plurality of regions of the brain, as described above with reference to FIG. 2A, and each of the plurality of regions has a plurality of signal values. The plurality of signal values ​​may represent the signal strength (e.g., the strength of neural activity) of portions corresponding to a plurality of vertices.

[0079] The receiving means 121 may receive, for example, a time series of brain images or multiple brain images obtained by multiple measurements. The brain images may preferably be images of the brain of the subject or test subject at rest.

[0080] The brain image received by the receiving means 121 is passed to the estimating means 122 .

[0081] The estimation means 122 is configured to estimate, based on the signal values ​​of one region among the plurality of regions of the brain image, the signal values ​​of the other regions. The estimation means 122 can estimate the signal values ​​of all the plurality of regions of the brain image by repeating the estimation for each of the plurality of regions of the brain image.

[0082] The estimation means 122 can estimate multiple signal values ​​in other regions by using multiple signal values ​​in one region of the brain image to estimate one signal value in another region for each of the multiple signal values. As described above with reference to FIG. 2B , estimating one signal value includes deriving a regression equation using some of the multiple brain images and calculating one signal value using the remaining multiple brain images and the regression equation. The regression equation is expressed, for example, by linear regression in which each of the multiple signal values ​​in one region is an explanatory variable and the single signal value in the other region is a response variable, and regression coefficients can be calculated by performing regression analysis.

[0083] The signal value of each of the multiple regions estimated by the estimation means 122 is passed to the correlation means 123 .

[0084] The correlation means 123 is configured to calculate correlation values ​​between the estimated signal values ​​and the actual signal values. The correlation means 123 can calculate correlation values ​​for all of the regions of the brain image by repeatedly calculating the correlation value for each of the regions of the brain image.

[0085] As described above with reference to Fig. 2B, the correlation means 123 can calculate a correlation value between a pattern formed by a plurality of signal values ​​estimated for one region and a pattern formed by a plurality of actual signal values ​​for one region. The correlation means 123 can calculate the similarity between the patterns as a correlation value by, for example, pattern matching.

[0086] The correlation value calculated by the correlation means 123 is passed to the derivation means 124 .

[0087] The deriving means 124 is configured to derive a pattern of functional connectivity of the brain based on the correlation values. The deriving means 124 can create a correlation matrix or pattern representing the correlation values ​​of each of the multiple regions, for example, as described above with reference to FIG. 3A. The correlation matrix can be, for example, a color map that represents the correlation values ​​between regions by color.

[0088] The derivation means 124 may derive, for example, a diagram in which the functionally connected brain regions described above with reference to FIG. 3B are color-coded as the pattern of functional connectivity of the brain.

[0089] The output from the derivation means 124 is provided to the outside of the system 100 via, for example, the interface unit 110 .

[0090] The brain functional connectivity patterns derived in this way may include functional connectivity patterns that could not be found by conventional methods. The derived brain functional connectivity patterns can be used for any purpose. For example, as described below, they can be used to build a classifier for estimating the state of a subject.

[0091] 6B shows another example of the configuration of the processor unit 120. In this example, the processor unit 120 has a configuration for analyzing the functional connectivity of the brain and constructing a classifier using the analysis results. The same reference numerals are used to designate components similar to those described above with reference to FIG. 6A, and detailed description thereof will be omitted here.

[0092] The processor unit 120 comprises a receiving means 121 , an estimating means 122 , a correlating means 123 , a deriving means 124 and a learning means 125 .

[0093] The receiving means 121 is configured to receive an image of the brain of a subject. The receiving means 121 can receive an image of the brain of each of the multiple subjects. The receiving means 121 can also receive data representing the state of each of the multiple subjects.

[0094] The brain image received by the receiving means 121 is passed to the estimating means 122. The data received by the receiving means 121 and representing the state of each of the multiple subjects is passed to the learning means 125.

[0095] The estimation means 122 is configured to estimate, for each of the multiple regions of the brain image, multiple signal values ​​of other regions based on the multiple signal values ​​of the region. The signal values ​​of each of the multiple regions estimated by the estimation means 122 are passed to the correlation means 123.

[0096] The correlation means 123 is configured to calculate a correlation value between the estimated signal values ​​and the actual signal values ​​for each of the regions of the brain image. The correlation value calculated by the correlation means 123 is passed to the derivation means 124.

[0097] The deriving means 124 is configured to derive a pattern of functional connection of the brain based on the correlation value. The pattern of functional connection of the brain derived by the deriving means 124 is passed to the learning means 125.

[0098] The learning means 125 is configured to learn the states of each of the multiple subjects and the brain functional connection pattern of each of the multiple subjects. The learning means 125 can learn the brain functional connection pattern, for example, using the states of each of the multiple subjects as labels. That is, the learning means 125 learns training data including (brain functional connection pattern of a first subject, state of the first subject), (brain functional connection pattern of a second subject, state of the second subject), ... (brain functional connection pattern of an nth subject, state of the nth subject).

[0099] This allows the construction of a classifier capable of estimating the state of the subject. When the functional connectivity pattern of the subject's brain is input to the classifier constructed in this manner, the classifier will estimate and output the state of the subject. By using the state of the subject at the time the brain image used to derive the functional connectivity pattern of the subject's brain was acquired in the above-mentioned training data, a classifier capable of estimating the current state of the subject can be constructed. Alternatively, by using the state of the subject after a predetermined period has elapsed since the brain image used to derive the functional connectivity pattern of the subject's brain was acquired in the above-mentioned training data, a classifier capable of estimating the future state of the subject after a predetermined period has elapsed can be constructed.

[0100] The learning means 125 can use any machine learning model, and typically can use a support vector machine (SVM) to learn the states of each of the multiple subjects and the functional connection patterns of the brains of each of the multiple subjects. Alternatively, for example, the learning means 125 can use a neural network to learn the states of each of the multiple subjects and the functional connection patterns of the brains of each of the multiple subjects.

[0101] A classifier constructed in this manner can estimate with greater accuracy a specific state of a target that could not be estimated by a classifier constructed by a conventional method, as described above with reference to FIG. 4A, for example.

[0102] 6C shows another example of the configuration of the processor unit 120. In this example, the processor unit 120 has a configuration for analyzing the brain's functional connectivity by a first method to construct a first classifier, analyzing the brain's functional connectivity by a second method to construct a second classifier, and integrating the first classifier and the second classifier to construct a composite classifier. The same reference numerals are used to designate the same components as those described above with reference to FIGS. 6A and 6B, and detailed description thereof will be omitted here.

[0103] The processor unit 120 includes a receiving means 121, an estimating means 122, a first correlating means 123, a first deriving means 124, a learning means 125, a calculating means 127, a second correlating means 128, and a second deriving means 129.

[0104] The receiving means 121 is configured to receive an image of the brain of a subject. The receiving means 121 can receive an image of the brain of each of the multiple subjects. The receiving means 121 can also receive data representing the state of each of the multiple subjects.

[0105] The brain image received by the receiving means 121 is passed to the estimating means 122 and the calculating means 127. The data received by the receiving means 121 and representing the state of each of the multiple subjects is passed to the learning means 125.

[0106] The estimation means 122 is configured to estimate, for each of the multiple regions of the brain image, multiple signal values ​​of other regions based on multiple signal values ​​of the region. The signal values ​​of each of the multiple regions estimated by the estimation means 122 are passed to the first correlation means 123.

[0107] The first correlation means 123 is configured to calculate a correlation value (first correlation value) between the estimated signal values ​​and the actual signal values ​​for each of the multiple regions of the brain image. The first correlation value calculated by the first correlation means 123 is passed to the first derivation means 124.

[0108] The first derivation means 124 is configured to derive a pattern (first pattern) of functional connectivity of the brain based on the first correlation value. The first pattern of functional connectivity of the brain derived by the first derivation means 124 is passed to the learning means 125.

[0109] The calculation means 127 is configured to calculate an average value for one of the multiple regions of the brain image by averaging multiple signal values ​​for the region. The calculation means 127 can calculate average values ​​for all of the multiple regions of the brain image by repeating the calculation of the average value for each of the multiple regions of the brain image. If the brain image is a time-series image, the average value may be a time-series value and may be represented, for example, by a waveform that changes over time.

[0110] The average value of each of the plurality of regions calculated by the calculation means 127 is passed to the second correlation means 128 .

[0111] The second correlation means 128 is configured to calculate a correlation value (second correlation value) between an average value calculated for one region and an average value calculated for another region. The second correlation means 128 can calculate, for example, a correlation value between waveforms that change over time. The second correlation means 128 can calculate the second correlation value for all region pairs in the brain image by repeatedly calculating the correlation value for each pair of a plurality of regions in the brain image and another region.

[0112] The second correlation value calculated by the second correlation means 128 is passed to the second derivation means 129 .

[0113] The second derivation means 129 is configured to derive a pattern (second pattern) of brain functional connectivity based on the second correlation value. The second derivation means 129 can create a correlation matrix or pattern representing the correlation value of each of the multiple regions, for example, as described above with reference to FIG. 3A. The correlation matrix can be, for example, a color map that represents the correlation value between regions by color.

[0114] The second derivation means 129 may derive, for example, a diagram in which the functionally connected brain regions described above with reference to FIG. 3B are color-coded as the pattern of functional connectivity of the brain.

[0115] The second pattern of functional connectivity of the brain derived by the second derivation means 129 is passed to the learning means 125 .

[0116] The learning means 125 is configured to learn the states of each of the multiple subjects and the functional connection patterns of the brains of each of the multiple subjects. For example, the functional connection patterns of the brains can be learned using the states of each of the multiple subjects as labels.

[0117] The learning means 125 can construct a first classifier by learning the state of each of the multiple subjects and a first pattern of functional connectivity of the brains of each of the multiple subjects, construct a second classifier by learning the state of each of the multiple subjects and a second pattern of functional connectivity of the brains of each of the multiple subjects, and construct a composite classifier by integrating the first classifier and the second classifier.

[0118] For example, training data for constructing a first classifier includes (first pattern of functional connectivity in the brain of a first subject, state of the first subject), (first pattern of functional connectivity in the brain of a second subject, state of the second subject), ... (first pattern of functional connectivity in the brain of an nth subject, state of the nth subject). The first classifier can estimate the state of the subject by itself, and when a pattern of functional connectivity in the brain of the subject is input, the first classifier will estimate and output the state of the subject.

[0119] For example, the training data for constructing the second classifier includes (second pattern of functional connectivity in the brain of the first subject, state of the first subject), (second pattern of functional connectivity in the brain of the second subject, state of the second subject), ... (second pattern of functional connectivity in the brain of the nth subject, state of the nth subject). The second classifier can estimate the state of the subject by itself, and when the pattern of functional connectivity in the brain of the subject is input, the second classifier will estimate and output the state of the subject.

[0120] The learning means 125 can integrate the first classifier and the second classifier using, for example, ensemble learning, and can employ weighted averaging as the ensemble learning.

[0121] For example, the learning means 125 can construct a composite classifier to output the sum of an output from a first classifier multiplied by a first weight and an output from a second classifier multiplied by a second weight. The first weight and the second weight can be adjusted to obtain an optimal output (e.g., an output with the highest accuracy).

[0122] The composite classifier constructed in this manner can estimate the state of the target with higher accuracy than a classifier constructed by a conventional method (i.e., the second classifier), as described above with reference to, for example, FIGS. 4A and 4B.

[0123] The classifier constructed by the configuration described above in FIGS. 6B and 6C can be used by a system for estimating the state of an object, which will be described later.

[0124] (3.2 Configuration of a system for estimating the state of an object) Fig. 7A shows an example of the configuration of the processor unit 120. In this example, the processor unit 120 has a configuration for estimating the state of the target using a classifier constructed by the system of Fig. 6B.

[0125] The processor unit 120 includes a receiving means 121 , an estimating means 122 , a correlating means 123 , a deriving means 124 , and a discriminator 126 .

[0126] The receiving means 121 is configured to receive an image of the brain of the subject. The image of the brain received by the receiving means 121 is passed to the estimating means 122.

[0127] The estimation means 122 is configured to estimate, for each of the multiple regions of the brain image, multiple signal values ​​of other regions based on the multiple signal values ​​of the region. The signal values ​​of each of the multiple regions estimated by the estimation means 122 are passed to the correlation means 123.

[0128] The correlation means 123 is configured to calculate a correlation value between the estimated signal values ​​and the actual signal values ​​for each of the regions of the brain image. The correlation value calculated by the correlation means 123 is passed to the derivation means 124.

[0129] The derivation means 124 is configured to derive a pattern of functional connection of the brain based on the correlation value. The pattern of functional connection of the brain derived by the derivation means 124 is passed to the classifier 126.

[0130] The classifier 126 is configured to estimate the state of a subject based on the pattern of functional connectivity of the brain. As described above, the classifier 126 has learned the states of each of a plurality of subjects and the pattern of functional connectivity of the brain of each of the plurality of subjects, and therefore, when the pattern of functional connectivity of the brain of a subject is input, the classifier 126 can estimate and output the state of the subject.

[0131] The output from the discriminator 126 is provided to the outside of the system 100, for example, via the interface unit 110. For example, it can be provided to the terminal device of the doctor D, as described above with reference to FIG.

[0132] For example, as described above with reference to FIG. 4A, it is possible to estimate with greater accuracy a particular state of a subject that could not be estimated by a classifier constructed by a conventional method.

[0133] Fig. 7B shows another example of the configuration of the processor unit 120. In this example, the processor unit 120 has a configuration for estimating the state of the target using a composite classifier constructed by the system of Fig. 6C.

[0134] The processor unit 120 includes a receiving means 121, an estimating means 122, a first correlating means 123, a first deriving means 124, a calculating means 127, a second correlating means 128, a second deriving means 129, and a composite discriminator 130.

[0135] The receiving means 121 is configured to receive an image of the brain of the subject. The image of the brain received by the receiving means 121 is passed to the estimating means 122 and the calculating means 127.

[0136] The estimation means 122 is configured to estimate, for each of the multiple regions of the brain image, multiple signal values ​​of other regions based on multiple signal values ​​of the region. The signal values ​​of each of the multiple regions estimated by the estimation means 122 are passed to the first correlation means 123.

[0137] The first correlation means 123 is configured to calculate a correlation value (first correlation value) between the estimated signal values ​​and the actual signal values ​​for each of the multiple regions of the brain image. The first correlation value calculated by the first correlation means 123 is passed to the first derivation means 124.

[0138] The first derivation means 124 is configured to derive a pattern (first pattern) of brain functional connectivity based on the first correlation value. The first pattern of brain functional connectivity derived by the first derivation means 124 is passed to the composite classifier 130.

[0139] The calculation means 127 is configured to calculate an average value for each of the plurality of regions of the brain image by averaging the plurality of signal values ​​of the region. The average value for each of the plurality of regions calculated by the calculation means 127 is passed to the second correlation means 128.

[0140] The second correlation means 128 is configured to calculate, for each of the multiple regions of the brain image, a correlation value (second correlation value) between the average value calculated for one region and the average value calculated for another region. The second correlation value calculated by the second correlation means 128 is passed to the second derivation means 129.

[0141] The second derivation means 129 is configured to derive a pattern (second pattern) of functional connectivity of the brain based on the second correlation value. The second pattern of functional connectivity of the brain derived by the second derivation means 129 is passed to the composite classifier 130.

[0142] The composite classifier 130 is configured to estimate a state of a subject based on a first pattern and a second pattern of functional connectivity of the brain. As described above, the composite classifier 130 is an integration of a first classifier that has learned the states of each of a plurality of subjects and the first pattern of functional connectivity of the brain of each of the plurality of subjects, and a second classifier that has learned the states of each of the plurality of subjects and the second pattern of functional connectivity of the brain of each of the plurality of subjects. When the first pattern and the second pattern of functional connectivity of the brain of the subject are input to the composite classifier 130, the composite classifier 130 can estimate and output the state of the subject. The composite classifier 130 can provide an output calculated, for example, by taking a weighted average of the output from the first classifier and the output from the second classifier.

[0143] The output from the composite classifier 130 is provided to the outside of the system 100, for example, via the interface unit 110. For example, as described above with reference to FIG. 1, it can be provided to the terminal device of doctor D.

[0144] 4A and 4B, the composite classifier 130 can estimate the state of the object with higher accuracy than a classifier constructed by a conventional method (i.e., the second classifier). This is because the estimation by the second classifier is based on global information that averages local, detailed information, while the estimation by the first classifier is based on local information that utilizes local, detailed information, and therefore both global information and local information can be effectively reflected in the estimation.

[0145] In the above example, the use of a classifier constructed by the configuration described above in Figures 6B to 6C has been described as an example, but the classifier is not limited to the classifier constructed by the configuration described above in Figures 6B to 6C, and a classifier constructed differently by another system may be used as long as it can achieve similar functions.

[0146] Each component of the system 100 described above may be composed of a single hardware component or multiple hardware components. When composed of multiple hardware components, the manner in which the hardware components are connected does not matter. The hardware components may be connected wirelessly or by wire. The system 100 of the present invention is not limited to a specific hardware configuration. It is also within the scope of the present invention for the processor unit 120 to be configured using analog circuits rather than digital circuits. The configuration of the computer system 100 of the present invention is not limited to the one described above as long as it can realize its functions.

[0147] (4. System Processing) 8 is a flowchart showing an example of processing (process 800) by the system 100 of the present invention. Process 800 is processing for analyzing functional connectivity of the brain. Process 800 is executed by the processor unit 120 of the system 100, and in particular, can be executed by the processor unit 120 having the configuration described above with reference to FIG. 6A or 6B.

[0148] In step S801, the receiving means 121 of the processor unit 120 receives a brain image. For example, when analyzing the functional connectivity of a subject's brain, the receiving means 121 receives the brain image of the subject, and when analyzing the functional connectivity of the subject's brain to create a classifier, the receiving means 121 can receive the brain image of the subject. As described above with reference to FIG. 2A, the brain image is divided into a plurality of regions corresponding to a plurality of regions of the brain, and each of the plurality of regions has a plurality of signal values. The plurality of signal values ​​may represent the signal strength (e.g., the strength of neural activity) of portions corresponding to a plurality of vertices.

[0149] In step S802, the estimation means 122 of the processor unit 120 estimates, for each of the multiple regions, multiple signal values ​​for each of the other multiple regions based on multiple signal values ​​for one region. The estimation means 122 can estimate multiple signal values ​​for the other regions by using multiple signal values ​​for one region in the brain image to estimate one signal value for the other regions for each of the multiple signal values. As described above with reference to FIG. 2B , estimating one signal value includes deriving a regression equation using some of the multiple brain images and calculating one signal value using the remaining multiple brain images and the regression equation. The regression equation is expressed, for example, by linear regression in which each of the multiple signal values ​​for one region is an explanatory variable and one signal value for the other regions is a response variable, and regression coefficients can be calculated by performing regression analysis.

[0150] In step S803, the correlation means 123 of the processor unit 120 calculates, for each of the multiple regions, a correlation value between the multiple signal values ​​estimated for a region and the multiple signal values ​​of that region in the image. As described above with reference to Figure 2B, the correlation means 123 can calculate a correlation value between a pattern formed by the multiple signal values ​​estimated for a region and a pattern formed by the multiple actual signal values ​​of the region.

[0151] In step S804, the derivation means 124 of the processor unit 120 derives a pattern of functional connectivity of the brain based on the correlation values. The derivation means 124 can create a correlation matrix or pattern representing the correlation values ​​of each of the multiple regions, for example, as described above with reference to FIG. 3A.

[0152] The process 800 may end at step S804, or an additional step S805 may be performed.

[0153] In step S805, the learning means 125 of the processor unit 120 learns the states of each of the multiple subjects and the functional connection patterns of the brains of each of the multiple subjects. The learning means 125 can learn the functional connection patterns of the brains, for example, using the states of each of the multiple subjects as labels. This makes it possible to create a classifier that can estimate the state of the target. When the functional connection pattern of the brain of the target is input to the classifier constructed in this way, the classifier will estimate and output the state of the target.

[0154] 9 is a flowchart showing an example of processing (process 900) by the system 100 of the present invention. Process 900 is processing for analyzing functional connectivity of the brain and constructing a composite classifier. Process 900 is executed by the processor unit 120 of the system 100, and in particular, can be executed by the processor unit 120 having the configuration described above with reference to FIG. 6C.

[0155] In step S901, the receiving means 121 of the processor unit 120 receives an image of the subject's brain. Step S901 is the same as step S801.

[0156] In step S902, the estimation means 122 of the processor unit 120 estimates, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of one region. Step S902 is a step similar to step S802.

[0157] In step S903, the first correlation means 123 of the processor unit 120 calculates, for each of the multiple regions, a correlation value (first correlation value) between the multiple signal values ​​estimated for one region and the multiple signal values ​​of that region in the image. Step S903 is the same as step S803.

[0158] In step S904, the derivation means 124 of the processor unit 120 derives a first pattern of the brain's functional connectivity based on the first correlation value. Step S904 is a step similar to step S804.

[0159] In step S905, the calculation means 127 of the processor unit 120 calculates an average value for each of the plurality of regions by averaging a plurality of signal values ​​for each of the plurality of regions.

[0160] In step S906, the second correlation means 128 of the processor unit 120 calculates a correlation value (second correlation value) between the average value of each of the multiple regions and the average value of each of the other multiple regions. The second correlation means 128 can calculate a correlation value between waveforms that change over time, for example.

[0161] In step S907, the second derivation means 129 of the processor unit 120 derives a second pattern of brain functional connectivity based on the second correlation value. The second derivation means 129 can create a correlation matrix or pattern representing the correlation value of each of the multiple regions, for example, as described above with reference to FIG. 3A.

[0162] In step S908, the learning means 125 of the processor unit 120 constructs a first classifier by learning the states of each of the multiple subjects and the first pattern of functional connectivity of the brain of each of the multiple subjects. The learning means 125 can, for example, learn the first pattern of functional connectivity of the brain using the states of each of the multiple subjects as labels. This makes it possible to create a first classifier that can estimate the state of the target. When the first pattern of functional connectivity of the brain of the target is input to the first classifier constructed in this way, the classifier will estimate and output the state of the target.

[0163] In step S909, the learning means 125 of the processor unit 120 constructs a second classifier by learning the states of each of the multiple subjects and the second pattern of functional connectivity of the brain of each of the multiple subjects. The learning means 125 can, for example, learn the second pattern of functional connectivity of the brain using the states of each of the multiple subjects as labels. This makes it possible to create a second classifier that can estimate the state of the target. When the second pattern of functional connectivity of the brain of the target is input to the second classifier constructed in this way, the classifier will estimate and output the state of the target.

[0164] In step S910, the learning means 125 of the processor unit 120 creates a composite classifier by integrating the first classifier and the second classifier. The learning means 125 can integrate the first classifier and the second classifier using, for example, ensemble learning. Weighted averaging can be adopted as the ensemble learning.

[0165] The composite classifier constructed in this manner can estimate the state of the target with higher accuracy than a classifier constructed by a conventional method, as described above with reference to, for example, FIGS. 4A and 4B.

[0166] 10 is a flowchart showing an example of a process (process 1000) performed by the system 100 of the present invention. The process 1000 is a process for estimating a state of an object. The process 1000 is executed by the processor unit 120 of the system 100, and in particular, can be executed by the processor unit 120 having the configuration described above with reference to FIG. 7A.

[0167] In step S1001, the receiving means 121 of the processor unit 120 receives an image of the subject's brain. Step S1001 is the same as step S801.

[0168] In step S1002, the estimation means 122 of the processor unit 120 estimates, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of one region. Step S1002 is a step similar to step S802.

[0169] In step S1003, the correlation means 123 of the processor unit 120 calculates, for each of the multiple regions, a correlation value between the multiple signal values ​​estimated for one region and the multiple signal values ​​of that region in the image. Step S1003 is the same as step S803.

[0170] In step S1004, the derivation means 124 of the processor unit 120 derives a pattern of functional connectivity of the brain based on the correlation value. Step S1004 is the same as step S804.

[0171] In step S1005, the classifier 126 of the processor unit 120 estimates the state of the subject based on the pattern of functional connectivity of the brain. The classifier 126 is created in the learning process of step S805 of process 800, and has learned the states of each of the multiple subjects and the functional connectivity patterns of the brain of each of the multiple subjects. Therefore, when the functional connectivity pattern of the subject's brain is input, the classifier 126 can estimate and output the state of the subject.

[0172] 11 is a flowchart showing an example of a process (process 1100) performed by the system 100 of the present invention. The process 1100 is a process for estimating a state of an object. The process 1100 is executed by the processor unit 120 of the system 100, and in particular, can be executed by the processor unit 120 having the configuration described above with reference to FIG. 7B.

[0173] In step S1101, the receiving means 121 of the processor unit 120 receives an image of the subject's brain. Step S1101 is the same as step S801.

[0174] In step S1102, the estimation means 122 of the processor unit 120 estimates, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of one region. Step S1102 is a step similar to step S902.

[0175] In step S1103, the first correlation means 123 of the processor unit 120 calculates, for each of the multiple regions, a correlation value (first correlation value) between the multiple signal values ​​estimated for one region and the multiple signal values ​​of that region in the image. Step S1103 is the same as step S903.

[0176] In step S1104, the derivation means 124 of the processor unit 120 derives a first pattern of the brain's functional connectivity based on the first correlation value. Step S1104 is a step similar to step S904.

[0177] In step S1105, the calculation means 127 of the processor unit 120 calculates an average value for each of the plurality of regions by averaging a plurality of signal values ​​for each of the plurality of regions. Step S1105 is a step similar to step S905.

[0178] In step S1106, the second correlation means 128 of the processor unit 120 calculates a correlation value (second correlation value) between the average value of each of the multiple regions and the average values ​​of the other multiple regions. Step S1106 is the same as step S906.

[0179] In step S1107, the second derivation means 129 of the processor unit 120 derives a second pattern of the brain's functional connectivity based on the second correlation value. Step S1107 is a step similar to step S907.

[0180] In step S1108, the composite classifier 130 of the processor unit 120 estimates the state of the subject based on the first and second patterns of functional connectivity of the brain. The composite classifier 130 is created by process 900, and has learned the states of each of the subjects and the first and second patterns of functional connectivity of the brain of each of the subjects. Therefore, when the first and second patterns of functional connectivity of the subject's brain are input, the composite classifier 130 can estimate and output the state of the subject.

[0181] In the above example, the use of a classifier constructed by process 800 or process 900 has been described as an example, but the classifier is not limited to a classifier constructed by process 800 or process 900, and a classifier constructed differently by another system may be used as long as it can achieve similar functions.

[0182] In the examples described above with reference to Figures 8 to 11, the steps are described as being executed in a specific order, but the order shown is merely an example, and the order in which the steps are executed is not limited to this. The steps can be executed in any logically possible order or simultaneously. For example, steps S905 to S907 may be executed before steps S902 to S904, or simultaneously with steps S902 to S904.

[0183] 8 to 11, the processing of each step shown in Fig. 8 to 11 may be realized by the processor unit 120 and a program stored in the memory unit 150, but the present invention is not limited to this. At least one of the processing of each step shown in Fig. 8 to 11 may be realized by a hardware configuration such as a control circuit.

[0184] The present invention 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 invention. [Industrial Applicability]

[0185] The present invention is useful for providing a novel method for analyzing functional connectivity in the brain, and is also useful for providing a method for estimating the state of a subject by utilizing the analysis results of this method. [Explanation of symbols]

[0186] T target Doctor D 10 Brain Images 100 systems 110 Interface section 120 Processor Unit 150 Memory section

Claims

1. 1. A method for analyzing brain functional connectivity, comprising: receiving an image of the brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; Deriving a pattern of functional connectivity of the brain based on the correlation value. A method comprising:

2. said receiving including receiving a plurality of images of the brain; estimating a plurality of signal values ​​for each of the other plurality of regions includes estimating one signal value for each of the other plurality of regions, for each of the plurality of signal values; estimating the one signal value calculating a regression coefficient of a linear regression using a portion of the plurality of images, with each of the plurality of signal values ​​of the region as an explanatory variable and the one signal value as a response variable; calculating the one signal value by multiplying each of a plurality of signal values ​​of the region by the regression coefficient using the remainder of the plurality of images; The method of claim 1 , comprising:

3. The method of claim 2 , wherein the plurality of images is a time series of images.

4. Calculating the correlation value calculating a correlation value between a pattern formed by the estimated plurality of signal values ​​and a pattern formed by the plurality of signal values ​​of the region in the image; The method of claim 1 , comprising:

5. Deriving the pattern comprises: creating a correlation matrix representing correlation values ​​for each of the plurality of regions; The method of claim 4, comprising:

6. The method of claim 1 , wherein the brain image is a resting brain image.

7. A method for creating a classifier for estimating a state of an object, comprising: receiving brain images acquired from each of a plurality of subjects; deriving a pattern of functional brain connectivity from each of the images of the plurality of subjects according to the method of claim 1; learning the state of each of the plurality of subjects and the pattern of functional connectivity of the brain; A method comprising:

8. A method for creating a classifier for estimating a state of an object, comprising: providing a first classifier produced by the method of claim 7; preparing a second classifier, the second classifier comprising: deriving a second pattern of functional brain connectivity from each of the images of the plurality of subjects according to a second method; learning a state of each of the plurality of subjects and a second pattern of functional connectivity of the brain; and creating a composite classifier by integrating the first classifier and the second classifier; wherein the second method comprises: calculating an average value for each of the plurality of regions by averaging a plurality of signal values ​​for the region; calculating a second correlation value between an average value of each of the plurality of regions and an average value of each of the other plurality of regions; deriving a second pattern of brain functional connectivity based on the second correlation value; A method comprising:

9. 1. A method for estimating a state of a subject, comprising: preparing a classifier created according to the method of claim 7 or 8; inputting a brain functional connectivity pattern derived from an image of the subject's brain according to the method of any one of claims 1 to 6 into the classifier; obtaining an output from the classifier; A method comprising:

10. A system for analyzing brain functional connectivity, comprising: a receiving means for receiving an image of the brain, the image being divided into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; an estimation means for estimating, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of the region; a correlation means for calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; a derivation means for deriving a pattern of functional connectivity of the brain based on the correlation value; A system comprising:

11. A system for estimating a state of an object, comprising: receiving means for receiving an image of the subject's brain, the image being divided into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; an estimation means for estimating, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of the region; a correlation means for calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; a derivation means for deriving a pattern of functional connectivity of the brain based on the correlation value; a classifier that estimates a state based on the derived pattern; A system comprising:

12. A system for estimating a state of an object, comprising: receiving means for receiving an image of the subject's brain, the image being divided into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; an estimation means for estimating, for each of the plurality of regions, a plurality of signal values ​​of each of the other plurality of regions based on a plurality of signal values ​​of the region; a first correlation means for calculating, for each of the plurality of regions, a first correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; a first derivation means for deriving a first pattern of functional connectivity of the brain based on the first correlation value; a calculation means for calculating an average value for each of the plurality of regions by averaging a plurality of signal values ​​for the region; second correlation means for calculating a second correlation value between the average value of each of the plurality of regions and the average values ​​of the other plurality of regions; a second derivation means for deriving a second pattern of brain functional connectivity based on the second correlation value; a discriminator that estimates a state based on the first pattern and the second pattern; A system comprising:

13. A program for analyzing brain functional connectivity, the program being executed in a computer system having a processor unit, the program comprising: receiving an image of the brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; Deriving a pattern of functional connectivity of the brain based on the correlation value. A program that causes the processor to execute a process including the steps of:

14. A program for estimating a state of an object, the program being executed in a computer system including a processor unit, the program comprising: receiving an image of the subject's brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; Deriving a pattern of brain functional connectivity based on the correlation value; estimating a state based on the derived pattern; and A program that causes the processor to execute a process including the steps of:

15. A program for estimating a state of an object, the program being executed in a computer system including a processor unit, the program comprising: receiving an image of the subject's brain, the image being partitioned into a plurality of regions corresponding to a plurality of regions of the brain, each of the plurality of regions having a plurality of signal values; For each region among the plurality of regions, estimating a plurality of signal values ​​of each of the other regions based on a plurality of signal values ​​of the region; calculating, for each of the plurality of regions, a first correlation value between a plurality of signal values ​​estimated for the region and a plurality of signal values ​​of the region in the image; deriving a first pattern of brain functional connectivity based on the first correlation value; calculating an average value for each of the plurality of regions by averaging a plurality of signal values ​​for the region; calculating a second correlation value between an average value of each of the plurality of regions and an average value of each of the other plurality of regions; deriving a second pattern of brain functional connectivity based on the second correlation value; estimating a state based on the first pattern and the second pattern; A program that causes the processor to execute a process including the steps of:

Images