RU2652898C1 - Method for investigation of brain activity according to fmrt data - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: invention relates to medicine, particularly to methods for investigation of brain activity. Method for investigation of brain comprises functional MRI recording of brain activity in the process of solving by the study subject at least three problems designed to investigate certain factors by using different stimuli, and obtaining a series of images of the values of the MP signal in time for each problem, cleaning the obtained images from artifacts and their standardization, forming, based on the processed images, a two-dimensional space-time matrix of the MR signal intensity (Z) values with normalization of its values, each column of the matrix characterizes a certain brain voxel, and each line characterizes the data obtained during the next brain scanning in the course of solving a particular problem by the study subject, factoring the resulting matrix by means of factor analysis, during which the matrix of coefficients of correlation R is calculated for all lines of the normalized intensity matrix, then its eigenvalues and eigenvectors are determined, on the basis of them factor load matrix (A), orthogonal rotation of factor load matrix (A) are formed, and obtaining matrix A(rot) characterizing the dynamics of each factor in time for each problem, and a meaningful interpretation of factors as separate independent elementary mental processes involved in solving presented problems, obtaining factor value matrix P on the basis of intensity value matrix (Z) and the matrix of factor loads after rotation A(rot), which characterizes the localization of each factor in the space of the brain and according to which a brain composition of the given functional system providing elementary mental processes, involved in solving the original problems, are judged, obtaining a linear mathematical model that is a neuroimaging model reflecting the formed functional brain systems that ensure the person's performance of a corresponding set of cognitive tasks.

EFFECT: use of the invention provides simultaneous characterization and interpretation of functional brain systems.

3 cl, 2 dwg

Description

FIELD OF TECHNOLOGY

The invention relates to medicine and neuroscience, in particular to a method for studying brain activity according to functional magnetic resonance imaging (fMRI). The invention can be used as a research tool in studies of the neurocognitive functioning of a person in normal conditions and in various diseases through visualization of brain activity systems in the course of solving various cognitive tasks by a person.

BACKGROUND

One of the most important problems of brain mapping and neuroimaging of a person’s cognitive functioning is the lack of a formal method (and appropriate computer tools) for direct integration of experimental data into a single model. Existing methods make it possible to visualize brain activity only during a person’s individual tasks, but at the same time a complex of interconnected cognitive processes is always updated immediately. There is no such cognitive task, for the solution of which, for example, only memory would be required without actualizing the processes of perception, decision making and control, awareness and emotional assessment. As a result, it turns out that a mixture of several different systems of brain activity is always visualized immediately, and isolating individual processes is already a complex “meta-task”, solved by pairwise enumeration and comparison of visualization results for the corresponding tasks. Meanwhile, the need for fMRI signal separation is expressed in the proposal of various methods, which, however, do not solve the main problem of visualization.

So from the prior art - US 2001031917 A1 (IPC A61B 5/05; publ. 18.10.2001) there is a method of analyzing the fMRI signal for the formation of a nonlinear function, which allows you to divide the signal into two parts, one of which can be considered a useful signal, and the other - noise . It is also assumed that the brain activity recorded by the fMRI signal may be composite, contain a reaction to the next stimulus, which may overlap with the reaction of the previous stimulus. In other words, the method is aimed at visualizing the reaction to several stimuli at once, the presentation of which overlaps in time. In this case, however, it is proposed to isolate signals for individual voxels or pre-allocated brain loci of the region of interest. In addition, the selection of basis vectors describing brain activity when performing various tasks in this method is not used to visualize the brain activity system, but only to identify the binary function of cutting off “active” and “inactive” voxels with the local goal of modeling a complex function that describes brain activity with the simultaneous action of several stimuli in each individual voxel.

The prior art - US 9072496 B2 (IPC A61B 5/00; publ. 07/07/2015) known system and method for analyzing fMRI images, which includes collecting 4-D data for the subject and calculating the correlation matrix in the time series between all voxels of the brain with the aim a subsequent analysis of the significance of correlations and the use of a binary matrix for clustering and constructing a classifier of combined data (for test and functional loads). However, this method solves the classification problem, rather than isolating systems and visualizing their brain activity.

The technical problem is:

creating a method for studying brain activity, which allows using fMRI data to obtain data that reflects: a) a person’s performance of a number of cognitive tasks that form various functional brain systems and b) brain distribution (three-dimensional visualization map) of activity for each of the selected functional systems.

SUMMARY OF THE INVENTION

The technical result achieved by using the invention is to enable the simultaneous characterization of functional brain systems corresponding to elementary cognitive processes, and the interpretation of these functional brain systems from the point of view of participation in cognitive functioning. The inventive method allows to identify the relationship of changes in the activity of each particular voxel (locus) of the brain, not only with the change in load in the corresponding cognitive tests, but also with the activity of other voxels (loci) of the brain, which in turn allows you to select the basic components using factor analysis that can considered as elementary functional brain systems.

The technical result is achieved due to the method of studying brain activity, including

- fMRI registration of brain activity in the process of solving by the object of study at least three tasks intended for the study of certain factors using various stimuli, and obtaining a series of images of the MP signal values in time for each task;

- cleaning the received images from artifacts and bringing them to a standard form;

- the formation on the basis of processed images of a two-dimensional space-time matrix of values of intensities (Z) of the MR signal with the normalization of its values, each column of the matrix characterizes a specific voxel of the brain, and each row characterizes the data obtained during the next brain scan in the course of solving by the object of study a specific task;

- factorization of the obtained matrix by means of factor analysis, during which the matrix of correlation coefficients R is calculated for all rows of the normalized intensity matrix, after which its eigenvalues and eigenvectors are determined based on which the matrix of factor loads (A) is formed;

- orthogonal rotation of the factor load matrix (A) in various ways, for example, to find a simple data structure or to bring it to a predetermined form and obtain a matrix A (rot) characterizing the dynamics of each factor in time for each task, and a meaningful interpretation of factors as separate independent elementary mental processes involved in solving the problems presented;

- obtaining a matrix of factor values of P based on a matrix of intensity values (Z) and a matrix of factor loads after rotation A (rot), which characterizes the localization of each factor in the space of the brain and by which one judges the brain composition of this functional system (involved brain loci) that provide elementary mental (neurocognitive) processes included in the solution of initial problems;

- obtaining a linear mathematical model, which is a neuroimaging model that reflects the formed functional brain systems that ensure that a person fulfills the corresponding series of cognitive tasks.

The tasks solved by the object of study provide the possibility of involving several different mental processes of the object of study.

Normalization can be performed for all tasks simultaneously or for each task separately, while in the normalization process, the arithmetic mean value is determined respectively for each column of the intensity matrix or each column of each task, after which the obtained value is subtracted from the corresponding intensities of the same column.

A two-dimensional space-time matrix can be optimized by removing uninformative voxels characterized by a critical dispersion and their anatomical location.

BRIEF DESCRIPTION OF THE DRAWINGS

In FIG. 1 and 2 are diagrams illustrating the inventive method.

The notation in FIG. 1 and 2 mean the following:

n is the number of columns in the matrix equal to the number of analyzed voxels of the brain;

k is the number of factors allocated;

m is the number of rows in the matrix, equal to the product of the number of fMRI scans in one task, by the number of tasks;

Z is the original matrix;

Z 'is the transposed original matrix;

A - matrix of factor loads;

A 'is the transposed matrix of factor loads;

N is the number of experimental tasks;

A (rot) - matrix obtained after orthogonal rotation of the factor load matrix (A);

P - matrix of values of factors;

R is the matrix of correlation coefficients.

DETAILED DESCRIPTION OF THE INVENTION

The term “task” in this application means: 1) the problem situation reflected in the consciousness or objectified in a symbolic model, containing data and conditions that are necessary and sufficient for its resolution in cash with knowledge and experience; 2) the form of structuring and presentation of experimental material in studies of cognitive processes and practical activities.

The term “factor” in this application refers to the concept of mathematical statistics, which means the common cause of many random changes in the totality of variables, events, etc. The factor is interpreted as a separate independent elementary mental (neurocognitive) process involved in solving the problems presented, for example, the perception of a certain type of image extracting a visual image from the memory, etc.

The term "factor analysis" in this application refers to a multidimensional method used to study the relationships between the values of variables. Known variables are assumed to depend on fewer unknown variables and random error. Factor analysis allows us to solve two important research problems: to describe the measurement object comprehensively and at the same time compactly. Using factor analysis, it is possible to identify hidden variables - factors responsible for the presence of linear statistical correlations between observed variables (URL: https://ru.wikipedia.org/wiki/%D0%A4%D0%B0%D0%BA%D1% 82% D0% BE% D1% 80% D0% BD% D1% 8B% D0% B9_% D0% B0% D0% BD% D0% B0% D0% BB% D0% B8% D0% B7). Factor analysis is a mathematical-statistical method that allows you to find the desired system of mental (neurocognitive) processes that are hidden from direct observation, which are involved in solving a person's corresponding tasks.

The term "meaningful interpretation" in this application refers to a scientific interpretation in which the model is theoretical and the explanation is a characteristic of reality.

The term "stimulus" in this application is understood to be: 1) the effect that determines the dynamics of the mental states of the individual and is in a causal relationship with it; 2) this is an impulse, the effect of which is mediated by the human psyche, his views, feelings, mood, interests, aspirations, etc. In the experiment, it is set in the form of visual, auditory, tactile, and other influences. Incentives can be individual objects, actions of other people, promises, carriers of obligations and opportunities, opportunities provided, and much more that can be offered to a person in compensation for his actions or that he would like to receive as a result of certain actions.

Modality (with respect to stimuli) - perceived by certain senses (vision, hearing, touch, taste, smell).

The term "standard view" in this application refers to the orientation and scale of the images (slices) of the brain obtained in the experiment that best matches the corresponding image (slice) of a particular atlas of the brain (usually use the MNI atlas or the Taleyraha atlas (https: // en .wikipedia.org / wiki / Talairach_coordinates)).

The main idea of the proposed approach is to combine in a special way the indicators of brain activity (according to fMRI data) when a person performs a system of cognitive tasks and build a vector multidimensional model. As a result, correlations of changes in the activity of each particular voxel (locus) of the brain are revealed, not only with changes in the load in the corresponding cognitive tests (for example, with a block experimental paradigm), but also with the activity of other voxels (loci) of the brain. It is proposed to use the Q-technique of factor analysis for the combined data matrix when factors that describe the correlation of the BOLD signal (Blood-oxygen-level dependent contrast imaging, https://en.wikipedia.org/wiki/Blood-oxygen-level_dependent) are not distinguished in time (scans), and in spatial distribution (in contrast to the usual R-technique, the input matrix is transposed). Then the dimension of the obtained correlation matrix is quite small, equal to the number of time slices (for all tests together), but the spatial characteristic of the factors identified can be restored by calculating the values of the factors. This will allow (when some special tests are included in the experimental data) in one algorithm to solve several interrelated tasks at once - to identify factors that describe physiological noise and factors that describe how a person performs certain tasks. But for this it is necessary to prepare the initial data in a special way. It is impossible to directly calculate the correlation coefficients of signal changes in the brain space (a three-dimensional array of MRI data can always be represented as a linear sequence) between different scans (time slices), because they will all be very large, since the differences in signal intensity for different areas of the brain are much larger than the differences in the change in the BOLD signal for the same area of the brain. However, this can be corrected by subtracting from the initial signal for each individual voxel the arithmetic mean over the entire studied range of time slices, as would be done automatically when calculating the correlations between voxels using standard factor analysis (R-technique). In addition, usually, component extraction algorithms (in particular, ICA) are used within the same experimental session, the comparison of the results for different cognitive tasks occurs after the selection of the necessary components by other methods - using atlases and statistics in the selected areas of interest. In the claimed method, it is proposed to combine in a single array all the primary fMRI data obtained in the course of a person performing a specially prepared set of various cognitive tasks. This becomes possible precisely thanks to the proposed technology, since all tasks are solved by the same person, then all the anatomical characteristics of his brain remain relatively constant for different experimental series (the spatial dimension of the data is the same). Therefore, time slices (scans) when a person performs different tasks in different series (including periods of "rest") using a simple block design can be directly combined into one data array. Then the correlation matrix will contain not only information about the interconnected changes of the BOLD signal within the same task, but also between the entire set of tasks. In this case, of course, the dimension of the data can significantly increase, but still, the proposed algorithm can analyze it quickly enough.

In this regard, the resulting solution can be directly compared with the components distinguished by ICA (Independent Component Analysis), which is widely used including for the analysis of individual fMRI data (to describe the relationships between different voxels of the brain) in order to remove artifacts and identifying “resting state” systems (https://en.wikipedia.org/wiki/Resting_state_fMRI). In the claimed method, in contrast to the method of independent components (ICA), an orthogonal model is obtained, the dimension of which can be estimated statistically and applied immediately to all tasks. Another important advantage of this method compared to ICA is the ability to rotate the coordinate system to achieve a more preferable solution. In this regard, effective algorithms used in some methods for cleaning the fMRI signal from physiological noise can be used: rotation based on a previously known spatial pattern (Perlbarg et al., 2007) and / or analysis of spatio-temporal qualities (Tohka et al., 2008 ; De Martino et al., 2007).

The basics of the proposed method:

1) to obtain fMRI data for several (at least three) different experimental series when the subject performs the corresponding cognitive tasks in the standard block research paradigm (in which periods of testing and rest are alternated);

2) using a special set of cognitive tasks to obtain fMRI data, including some specific tests, for example, the task of hyperventilation of the lungs, for the purpose of automatically determining factors associated with physiological processes that are noise in the context of this method;

3) by combining primary fMRI data of different experimental series into a common two-dimensional space-time matrix of primary data;

4) on a special normalization of the initial data, in particular, by subtracting from the initial signal for each individual voxel the arithmetic mean or over the entire studied range of time slices, or for each individual series;

5) on the application of the Q-technique of factor analysis with the assessment of the necessary dimension of the obtained normalized data array;

6) on the application of both standard techniques aimed at finding a simple data structure, for example, the varimax method, and special methods of rotation of the resulting factor solution: rotation based on a previously known spatial pattern and / or analysis of spatio-temporal qualities;

7) on the possibility of a meaningful interpretation of the resulting vector model, which formally characterizes simultaneously two aspects:

a) functional brain systems that are formed when a person solves this set of cognitive tasks for spatial localization, reflected in the calculated values of the selected factors, and

b) components of cognitive processes based on the values of factor loads for the corresponding experimental series, since it will be possible to formally evaluate the contribution of each of the selected components (factors) to the performance of a particular task.

The functional brain organs of the subject of activity are formed in vivo, they are carriers of the ability to carry out certain activities and are responsible for its occurrence and course. This concept under varying names (“physiological organs of the nervous system”, “functional organs of the brain”, “brain organs”, “functional brain systems”, etc.) was used and developed in the domestic physiology of activity and the psychological theory of activity - in A.A. . Ukhtomsky, N.A. Bernstein, A.V. Zaporozhets, A.R. Luria, A.N. Leontiev, P.Ya. Halperin; L.S. Vygotsky used a similar concept - “psychological functional system”.

An example implementation of the method.

A method for studying brain activity involves registering a BOLD signal of fMRI (for example, in the standard block paradigm) of brain activity in the process of solving at least three tasks by the object of study using stimuli (external stimuli perceived by the senses of the object of study) and obtaining a series of images over time each task. It is preferable that, when obtaining primary data, as much as possible various test loads are used, which involve the complex of the studied cognitive processes as fully as possible. Tasks and incentives are formed, taking into account the ability to involve several different mental processes.

In the fMRI experiment (in the tomograph from 1.5 T and above), a set of tasks was used related to probabilistic forecasting - the implementation of the process of choosing the optimum among many alternatives. For this, the test subject (object of study) was presented with a quasi-random sequence of three digits (1-3) in three tasks.

In task 1 (actually forecasting), the test subject needs to press a key with the number that, in his opinion, should follow the presented stimulus. As control tasks on memory and perception are used.

The memory task 2 represents the classic 1-back test, when the test subject must memorize the presented numbers and press the key if the present number is the same as the one that came before it.

In task 3, perception is simply required to press a key with the number that was presented at the moment.

As a result, for each of the tasks, in accordance with the block paradigm of fMRI research, an array of 50 sets (scans, k = 50) was obtained, each of which contains 33 T2 * images (sections) of the brain, in which the intensity of the BOLD signal is reflected. Each slice is a 64 × 64 plane image of the brain (the dimension is determined by the fMRI parameters and represents the resolution of the device in this mode). In this case, the scans alternate: during five scans, the corresponding task is performed, and during the next five scans, rest, then again this sequence is repeated. Next, they use image cleaning from various artifacts and bring to a standard view in accordance with the selected atlas of the brain by conventional means (most often the FSL or SPM software packages are used). Typical artifacts associated with both technical errors and unwanted movements of the object of study are removed. Preliminary standardization of the orientation of the obtained images of the brain by searching for the rotation of the image to the reference (by the atlas of the brain) allows us to further compare the results of the analysis with other data. As a result, the data set remains the same in size, but in it each image is already brought to the standard form (see Fig. 1).

Then, based on the obtained images, a two-dimensional space-time matrix of MP signal intensity values is formed, in which each column characterizes a certain brain voxel and contains the BOLD signal intensity values in all time slices (scans) and tasks, and each row characterizes the data obtained during the next scanning the brain in the course of solving a particular problem by the subjects. At the same time, they carry out optimization and normalization (see Fig. 1) of values for each column of this matrix, which can be done in several ways: 1) by subtracting from each value the arithmetic average of the entire column (i.e., the total average of all tasks), or 2) by subtracting from each value the arithmetic mean for each time block corresponding to a separate task (which allows you to align the condition of activity of a given voxel in various tasks). In the course of optimization, variances (standard deviations) of the values for each voxel are also calculated (for each task separately or as a whole) and voxels with insignificant dispersion are excluded from further analysis in accordance with a given threshold of significance (for example, 5% of the total variance of all data) . Also, for optimization purposes, voxels located outside the brain structures of interest for study (outside the subject’s head or in body tissues not related to the brain) may be excluded. For normalization purposes, normalization (dividing all values) by the corresponding standard deviation of each voxel separately is also possible. In general, normalization is carried out either for each task separately, or for all at the same time.

In this example, a matrix of processed images containing n = 64 * 64 * 33 = 135168 columns and m = 50 * 3 = 150 rows was prepared for the study. The resulting matrix was normalized using the first type of normalization. For this, the arithmetic mean of the values of each column for each of the three tasks was determined in the obtained matrix separately. After that, the arithmetic mean obtained is subtracted from the values in the corresponding columns of the matrix. The data obtained as a result of normalization and reduction — folding voxels with a small dispersion (less than 5% of the total dispersion) —the data were formed into a two-dimensional array — a matrix of relative intensities.

Further, this matrix is factorized by calculating the eigenvalues and eigenvectors (for this purpose, an expansion algorithm for singular numbers can be used) (http://en.wikipedia.org/wiki/%D0%90%D0%BB%D0%B3%D0 % BE% D1% 80% D0% B8% D1% 82% D0% BC_% D0% B2% D1% 8B% D1% 87% D0% B8% D1% 81% D0% BB% D0% B5% D0% BD % D0% B8% D1% 8F_% D1% 81% D0% BE% D0% B1% D1% 81% D1% 82% D0% B2% D0% B5% D0% BD% D0% BD% D1% 8B% D1 % 85_% D0% B7% D0% BD% D0% B0% D1% 87% D0% B5% D0% BD% D0% B8% D0% B9 and https://ru.wikipedia.org/wiki/%D0% A1% D0% B8% D0% BD% D0% B3% D1% 83% D0% BB% D1% 8F% D1% 80% D0% BD% D0% BE% D0% B5_% D1% 80% D0% B0% D0% B7% D0% BB% D0% BE% D0% B6% D0% B5% D0% BD% D0% B8% D0% B5).

In the process of factorization, the matrix of correlation coefficients R (the dimension of which is determined by the total number of all time slices (m = 150 in Fig. 2)) is calculated over all rows of the normalized intensity matrix. Based on the eigenvalues, a factorial dimension (k, in Fig. 2) is quantified, which characterizes the number of elementary mental (neurocognitive) processes involved in the study. In this example, 3 factors were identified.

Based on the matrix of eigenvectors and eigenvalues, a factor load matrix (A, in Fig. 2) is obtained, which is subjected to the orthogonal rotation procedure to obtain A (rot), a factor load matrix after rotation, which contains the desired components of cognitive processes (Fig. 2) . For this, it is possible to use both standard (to search for a simple data structure, for example, the varimax / varimax method), and special rotation methods based on a previously known spatial pattern. In this example, the varimax algorithm was used.

Moreover, each factor is interpreted as a separate independent elementary mental (neurocognitive) process involved in solving the problems presented. Since the matrix A (rot) characterizes the dynamics of each factor in time for each task, on this basis it is possible to carry out a meaningful interpretation of these factors. For example, in the task of studying probabilistic forecasting, memory and perception, the load on one of the identified factors in time slices turned out to be greater for the forecasting problem, and their change in scans inside the task reproduced the experimental block paradigm, i.e. corresponded to the moments of the task and the state of rest. At the same time, for other tasks, the loads on this factor were insignificant. Then this factor can be interpreted as a system of advancing and testing hypotheses. At the same time, another factor had high loads only when the subjects solved the task of keeping in memory and recognizing the presented figures with similar dynamics. The third factor was associated with the perception of the presented figures. The dynamics of factor loadings of this factor was greater in the third task, but was also present in the first two, since visual stimuli were also presented there.

Then, using the initial matrix (Z in Fig. 2) and the factor load matrix obtained after rotation (A (rot) in Fig. 2), a matrix of factor values (P, in Fig. 2) is calculated. The matrix P characterizes the localization of each factor in the space of the brain, which makes it possible to determine the composition of brain structures (according to the corresponding atlas, the Taleyraha atlas was used in this example), which provide elementary mental (neurocognitive) processes involved in solving the initial problems.

The results obtained - the matrices A (rot) and P - are stored in a form convenient for further use: in the form of digital values that allow systemic visualization of the processes under study, i.e. build for each factor (an elementary neurocognitive process): 1) a graph of the change in values over time and tasks, which characterizes the systemic contribution of each of the selected elementary processes to the solution of a particular problem (an example with conclusions is needed); 2) a brain activity map, which, after comparison with the atlas, allows you to identify the system (composition) of brain structures that ensure the operation of this elementary neurocognitive process.

So, in our example, the calculation of the values of factors made it possible to determine the brain localization of the corresponding systems. For example, a probabilistic forecasting system (factor 1) consists mainly of the following synchronously operating structures listed in the order of their contribution to the system: Right Midbrain, Right Inferior Frontal Gyrus, Left Superior Temporal Gyrus, Left Parahippocampal Gyrus, Right Posterior Cingulate, Right Parahippocampal Gyrus , Left Brodmann area 47, Left Orbital Gyrus, Right Brodmann area 25, Right Temporal Lobe Sub-Gyral, Right Brodmann area 30, Left Putamen, Left Amygdala, Left Brainstem, Right Cingulate Gyrus, Left Cerebellum. Apparently, this system evaluates the effectiveness of the previous forecast and the correction of the following hypothesis, which is consistent with the available data on the existence of a basic decision model consisting of two interacting schemes: assessing possible alternatives and choosing the best option from them. In other tasks, the activity of this system was four times less.

Thus, the proposed method allows, according to fMRI data, to combine in a single system (neuroimaging model) data reflecting: a) a person’s performance of a number of cognitive tasks that form various functional brain systems, and b) brain distribution (three-dimensional visualization map) of activity for each of the selected functional systems, as well as 1) to identify the relationship of changes in the activity of each specific voxel (locus) of the brain, not only with the change in load in the corresponding cognitive tests, but also with the asset by other voxels (loci) of the brain, and 2) using factor analysis, identify the basic components that can be considered as elementary functional brain systems.

Claims (11)

1. A method for studying brain activity, including - functional MRI recording of brain activity in the process of solving by the object of study at least three tasks designed to study certain factors using various stimuli, and obtaining a series of images of the MP signal values in time for each task; - cleaning the received images from artifacts and bringing them to a standard form; - the formation on the basis of processed images of a two-dimensional space-time matrix of values of intensities (Z) of the MR signal with the normalization of its values, each column of the matrix characterizes a specific voxel of the brain, and each row characterizes the data obtained during the next brain scan in the course of solving by the object of study a specific task; - factorization of the obtained matrix by means of factor analysis, during which the matrix of correlation coefficients R is calculated for all rows of the normalized intensity matrix, after which its eigenvalues and eigenvectors are determined based on which the matrix of factor loads (A) is formed; - orthogonal rotation of the matrix of factor loads (A) and obtaining a matrix A (rot) characterizing the dynamics of each factor in time for each task, and a meaningful interpretation of factors as separate independent elementary mental processes involved in solving the problems presented; - obtaining a matrix of factor values of P based on a matrix of intensity values (Z) and a matrix of factor loads after rotation A (rot), which characterizes the localization of each factor in the space of the brain and by which one judges the brain composition of this functional system that provides elementary mental processes included in solving the initial problems; - obtaining a linear mathematical model, which is a neuroimaging model that reflects the formed functional brain systems that ensure that a person fulfills the corresponding series of cognitive tasks. 2. The method according to p. 1, characterized in that the tasks solved by the object of study, provide the ability to use several different mental processes of the object of study. 3. The method according to p. 1, characterized in that the normalization is carried out for all tasks simultaneously or for each task separately, while in the normalization process, the arithmetic mean value is determined respectively for each column of the intensity matrix or each column of each task, after which the obtained value is subtracted from the corresponding intensities of the same column. 4. The method according to p. 1, characterized in that the two-dimensional space-time matrix is optimized by removing uninformative voxels characterized by the critical dispersion and their anatomical location.

Images