A brain mapping method based on point constraint optimal transmission and related equipment

Abstract

The application discloses a brain mapping method based on point-constrained optimal transmission and related equipment, the method obtains a functional signal image by processing functional magnetic resonance imaging data, extracts average time sequences by using different brain atlases, calculates Montreal Neurological Institute coordinates, and then combines optimal transmission algorithm to calculate a constructed cost matrix and a mask matrix, obtains an optimal transmission transfer matrix between atlases, inputs the average time sequences, obtains new average time sequences, obtains a functional connection body of the brain, and finally realizes mapping between two different brain atlases; the optimal transmission model of point-line relationship adopted by the method introduces a point-line distance, so that the model increases an index, describes local structure constraints of a data domain, effectively realizes extraction of structure information of the data domain, and makes the converted atlas also well reflect individual information of the original atlas; and the accuracy of conversion between atlases is improved, and time cost is saved.

Description

Technical Field

[0001] This invention belongs to the field of image analysis, specifically to the field of brain atlas mapping, and particularly to a brain atlas mapping method and related equipment based on point-constrained optimal transmission. Background Technology

[0002] Connectomes are a popular approach to modeling the brain as a graph-like structure, playing a significant role in studying individual differences in brain function, brain-behavioral associations, and understanding brain alterations in neuropsychiatric disorders. Due to data privacy concerns, some datasets are only released as fully processed connectomes; however, different maps divide the brain into regions of varying sizes and topologies, making it impossible to directly compare connectomes created from different maps. To address this, researchers in this field have proposed remapping across maps using point-constrained optimal transport models, which would help increase the reusability of existing connectomes.

[0003] In most optimal transmission models, the primary constraint is minimizing the total transmission distance of all samples. Without any additional guidance, this can lead to sample matching errors. However, in many practical applications, cross-domain annotation of paired keypoints is relatively easy and reasonable. To address this, a relation-preserving Keypoint Guided Matching (KPG-RL) model has been proposed, demonstrating significantly better performance than traditional optimal transmission models.

[0004] However, the KPG-RL model preserves local properties by only considering distance constraints when pairing samples, and lacks discussion on the relative positions between samples, which affects the accuracy of conversion between different spectra. Furthermore, the time cost of using the KPG-RL model for spectra conversion is relatively large. Summary of the Invention

[0005] To overcome the shortcomings of the above-mentioned technologies, this invention provides a brain map mapping method and related equipment based on point-constrained optimal transmission, which can solve the technical problem that the lack of consideration for the relative positions between samples in the existing mapping methods leads to low accuracy in the conversion between maps.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A brain map mapping method based on point-constrained optimal transmission includes:

[0008] S1: Process the functional magnetic resonance imaging data to obtain functional signal images;

[0009] S2: Two different brain maps were used to extract functional signal images, and the corresponding average time series were obtained respectively;

[0010] S3: Calculate the corresponding coordinates of the Montreal Institute of Neuroscience using the two brain maps mentioned above to obtain data points;

[0011] S4: Construct the cost matrix and mask matrix based on the data points, and combine them with the optimal transmission algorithm to obtain the optimal transmission transfer matrix;

[0012] S5: Based on the average time series and the optimal transfer matrix, a new average time series is calculated;

[0013] S6: Based on the new average time series, the functional connectome of the brain is obtained, realizing the mapping between two different brain maps.

[0014] Furthermore, the specific steps of S1 include:

[0015] S1.1: Format conversion of functional magnetic resonance imaging data;

[0016] S1.2: The functional magnetic resonance imaging data after format conversion is sequentially processed by data filtering, time-level correction, eddy current correction, spatial standardization, smoothing, delinear drift removal, regression covariate, and low-frequency filtering to obtain the functional signal image.

[0017] Furthermore, in S3, the centroid coordinates of the frontal lobe, temporal lobe, occipital lobe, and parietal lobe of the brain are selected as key points from the calculated data points.

[0018] Furthermore, in S4, the cost matrix and mask matrix are constructed based on the data points, and the optimal transmission transfer matrix is ​​calculated by combining the optimal transmission model of point-line relationship.

[0019] Furthermore, the specific steps of S4 are characterized by:

[0020] The optimal transmission model based on point-line relationships calculates the cost matrix and mask matrix; then, the optimal transmission transfer matrix is ​​obtained by using the logarithmic field Sinkhorn iterative algorithm.

[0021] Furthermore, the specific steps for calculating the cost matrix are as follows:

[0022] Draw auxiliary lines between every two key points to form the auxiliary line structure of the source domain and the auxiliary line structure of the target domain; select two key points on any auxiliary line and the data points to be calculated to form a triangle, and use Heron's formula to calculate the distance from the data points to each auxiliary line.

[0023] Furthermore, in S4, a cost matrix and a mask matrix are constructed based on the data points. Combined with the optimal transmission model of incomplete point-line relationships, the optimal transmission transfer matrix is ​​calculated.

[0024] A brain map mapping system based on point-constrained optimal transmission, comprising the following steps for implementing the aforementioned brain map mapping method based on point-constrained optimal transmission:

[0025] The data processing module is used to process functional magnetic resonance imaging data to obtain functional signal images;

[0026] The feature extraction module is used to extract functional signal images using two different brain maps, and obtain the corresponding average time series respectively;

[0027] The coordinate calculation module is used to calculate the corresponding coordinates of the Montreal Institute of Neuroscience using two different brain maps, and obtain data points.

[0028] The optimal transmission calculation module is used to construct the cost matrix and mask matrix based on data points, and combine them with the optimal transmission algorithm to obtain the optimal transmission transition matrix;

[0029] The transfer module is used to calculate a new average time series based on the average time series and the optimal transfer matrix;

[0030] The atlas mapping module is used to obtain the functional connectome of the brain based on the new average time series, and to realize the mapping between two different brain atlases.

[0031] An apparatus comprising:

[0032] Memory, used to store computer programs;

[0033] A processor is used to implement the steps of the above-described brain map mapping method based on point-constrained optimal transmission when executing the computer program.

[0034] A computer-readable storage medium storing a computer program, which, when executed by a processor, is used to implement the steps of the above-described brain map mapping method based on point-constrained optimal transmission.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] This invention also provides a brain atlas mapping method based on point-constrained optimal transfer. This method processes functional magnetic resonance imaging (fMRI) data to obtain functional signal images, then extracts average time series from different brain atlases and calculates the coordinates of the Montreal Neuroscience Institute. Finally, it combines an optimal transfer algorithm to calculate the constructed cost matrix and mask matrix, obtaining the optimal transfer matrix between atlases. This matrix is ​​then input into the average time series to obtain a new average time series, thus obtaining the brain's functional connectome. Ultimately, this achieves mapping between two different brain atlases. The optimal transfer model using point-line relationships incorporates point-line distance, adding an indicator to the model and characterizing the local structural constraints of the data domain. This effectively extracts structural information from the data domain, ensuring that the converted atlas accurately reflects the individual information of the original atlas. This improves the accuracy of atlas conversions and saves time. Attached Figure Description

[0037] Figure 1 A flowchart for calculating the optimal transmission transfer matrix based on the optimal transmission model of point-line relationship provided in an embodiment of the present invention;

[0038] Figure 2 A schematic diagram illustrating the construction of auxiliary lines and the calculation of the distance between points and lines for three key points provided in the embodiments of the present invention;

[0039] Figure 3 A flowchart of a brain map mapping method based on point-constrained optimal transmission provided by the present invention;

[0040] Figure 4 This is a schematic diagram of the structure of a brain mapping system based on point-constrained optimal transmission provided by the present invention. Detailed Implementation

[0041] This invention provides a brain map mapping method based on point-constrained optimal transmission, such as... Figure 3 As shown, it includes the following steps:

[0042] S1: Process the functional magnetic resonance imaging data to obtain functional signal images, specifically:

[0043] S1.1: Format conversion of functional magnetic resonance imaging data;

[0044] S1.2: The functional magnetic resonance imaging data after format conversion is sequentially processed by data filtering, time-level correction, eddy current correction, spatial standardization, smoothing, delinear drift removal, regression covariate, and low-frequency filtering to obtain the functional signal image.

[0045] S2: Two different brain maps were used to extract functional signal images, and the corresponding average time series were obtained respectively.

[0046] S3: Calculate the corresponding coordinates of the Montreal Institute of Neuroscience using the two brain maps mentioned above to obtain data points.

[0047] Among the calculated data points (coordinates of the Montreal Neurological Institute), the centroid coordinates of the frontal lobe, temporal lobe, occipital lobe, and parietal lobe were selected as key points.

[0048] S4: Construct the cost matrix and mask matrix based on the data points, and combine them with the optimal transmission algorithm to obtain the optimal transmission transfer matrix.

[0049] Specifically, a cost matrix and a mask matrix are constructed based on data points, and the optimal transmission transfer matrix is ​​calculated using either the optimal transmission model with point-line relationships or the optimal transmission model with incomplete point-line relationships.

[0050] The optimal transmission model based on point-line relationships calculates the cost matrix and mask matrix. The specific steps for calculating the cost matrix are as follows: draw auxiliary lines between every two key points to form the auxiliary line structure of the source domain and the auxiliary line structure of the target domain; select two key points on any auxiliary line and the data points to be calculated to form a triangle, and use Heron's formula to calculate the distance from the data points to each auxiliary line.

[0051] The optimal transfer matrix is ​​then calculated using the logarithmic field Sinkhorn iterative algorithm.

[0052] S5: Based on the average time series and the optimal transfer matrix, a new average time series is calculated.

[0053] S6: Based on the new average time series, the functional connectome of the brain is obtained, realizing the mapping between two different brain maps.

[0054] like Figure 4 As shown, this invention also provides a brain atlas mapping system based on point-constrained optimal transfer, comprising: a data processing module for processing functional magnetic resonance imaging data to obtain functional signal images; a feature extraction module for extracting functional signal images using two different brain atlases to obtain corresponding average time series; a coordinate calculation module for calculating the corresponding Montreal Neuroscience Institute coordinates using the two brain atlases to obtain data points; an optimal transfer calculation module for constructing a cost matrix and a mask matrix based on the data points, and combining them with an optimal transfer algorithm to obtain an optimal transfer matrix; a transfer transfer module for calculating a new average time series based on the average time series and the optimal transfer matrix; and an atlas mapping module for obtaining the functional connectome of the brain based on the new average time series, thereby realizing the mapping between the two different brain atlases.

[0055] The present invention also provides an apparatus comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the brain map mapping method based on point-constrained optimal transmission.

[0056] When the processor executes the computer program, it implements the above-mentioned steps for brain atlas mapping based on point-constrained optimal transfer, for example: processing functional magnetic resonance imaging data to obtain functional signal images; extracting functional signal images using two different brain atlases to obtain corresponding average time series; calculating the corresponding Montreal Neuroscience Institute coordinates using the two brain atlases to obtain data points; constructing a cost matrix and a mask matrix based on the data points, and combining them with the optimal transfer algorithm to obtain the optimal transfer matrix; calculating a new average time series based on the average time series and the optimal transfer matrix; and obtaining the brain's functional connectome based on the new average time series to achieve mapping between the two different brain atlases.

[0057] Alternatively, when the processor executes the computer program, it implements the functions of each module in the above system, such as: a data processing module for processing functional magnetic resonance imaging data to obtain functional signal images; a feature extraction module for extracting functional signal images using two different brain maps to obtain corresponding average time series; a coordinate calculation module for calculating the corresponding coordinates of the Montreal Institute of Neuroscience using the two brain maps to obtain data points; an optimal transfer calculation module for constructing a cost matrix and a mask matrix based on the data points, and combining them with the optimal transfer algorithm to obtain an optimal transfer matrix; a transfer transfer module for calculating a new average time series based on the average time series and the optimal transfer matrix; and a map mapping module for obtaining the functional connectome of the brain based on the new average time series to achieve mapping between the two different brain maps.

[0058] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a preset function, the instruction segments describing the execution process of the computer program in the point-constrained optimal transmission brain mapping device. For example, the computer program can be divided into a data processing module, a feature extraction module, a coordinate calculation module, an optimal transfer calculation module, a transfer transfer module, and a map mapping module. The specific functions of each module are as follows: The data processing module processes functional magnetic resonance imaging (fMRI) data to obtain functional signal images; the feature extraction module extracts functional signal images using two different brain maps to obtain corresponding average time series; the coordinate calculation module calculates the corresponding coordinates of the Montreal Neuroscience Institute using the two brain maps to obtain data points; the optimal transfer calculation module constructs a cost matrix and a mask matrix based on the data points, and combines them with the optimal transfer algorithm to obtain the optimal transfer matrix; the transfer transfer module calculates a new average time series based on the average time series and the optimal transfer matrix; and the map mapping module obtains the functional connectome of the brain based on the new average time series, realizing the mapping between the two different brain maps.

[0059] The point-constrained optimal transmission-based brain mapping device can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. The point-constrained optimal transmission-based brain mapping device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above are examples of point-constrained optimal transmission-based brain mapping devices and do not constitute a limitation on such devices. The device may include more components than described above, or combine certain components, or use different components. For example, the point-constrained optimal transmission-based brain mapping device may also include input / output devices, network access devices, buses, etc.

[0060] The processor referred to can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or any conventional processor. The processor is the control center of the point-constrained optimal transmission-based brain mapping system, connecting various parts of the entire point-constrained optimal transmission-based brain mapping device through various interfaces and lines.

[0061] The memory can be used to store the computer program and / or modules. The processor implements various functions of the point-constrained optimal transmission brain mapping device by running or executing the computer program and / or modules stored in the memory and calling the data stored in the memory.

[0062] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function (such as sound playback or image playback). The data storage area may store data created based on the use of the mobile phone (such as audio data and phonebook entries). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0063] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the brain map mapping method based on point-constrained optimal transmission.

[0064] If the modules / units integrated by the brain mapping system based on point-constrained optimal transmission are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0065] Based on this understanding, the present invention can implement all or part of the processes in the above-described brain mapping method based on point-constrained optimal transmission, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-described brain mapping method based on point-constrained optimal transmission. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or a preset intermediate form, etc.

[0066] The computer-readable storage medium may include any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0067] It should be noted that the content contained in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0068] The present invention will be further described below with reference to embodiments and accompanying drawings:

[0069] Example

[0070] As described in the background section, in most optimal transmission models, the main constraint is to minimize the total transmission distance of all samples. Without any additional guidance, this may lead to sample matching errors. However, in many practical applications, cross-domain annotation of paired keypoints is relatively easy and reasonable. To address this, a relation-preserving keypoint guided matching model (KPG-RL) has been proposed, which significantly outperforms traditional optimal transmission models.

[0071] However, the KPG-RL model preserves local properties by only considering distance constraints when pairing samples, and lacks discussion on the relative positions between samples, which affects the accuracy of conversion between different spectra. Furthermore, the time cost of using the KPG-RL model for spectra conversion is relatively large.

[0072] To address the aforementioned issues, this implementation provides a brain map mapping method based on point-constrained optimal transmission. The idea behind this method is to establish an optimal transmission model with local constraints around key points within the domain, thereby overcoming the shortcomings of existing optimal transmission methods for brain map mapping.

[0073] This implementation provides a brain map mapping method based on point-constrained optimal transmission, with the following specific steps:

[0074] Step 1: Obtain functional magnetic resonance imaging (fMRI) data of the adult brain and high-resolution structural magnetic resonance imaging (T1w) data from the ICBM website.

[0075] Step 2: Convert the format of functional magnetic resonance imaging (fMRI) data and structural magnetic resonance imaging (SMRI) data.

[0076] Step 3: Remove the first 5 time points from the functional magnetic resonance imaging data after format conversion in Step 2 to eliminate the instability of the magnetic field in the early stage of scanning, and then perform time-layer correction and head motion correction.

[0077] Step 4: For each subject's functional magnetic resonance imaging (fMRI) image corrected in Step 3, perform image registration with its T1 structural image. Then, select the T1 template of the Montreal Neurological Institute (MNI) standard space as the reference standard, spatially transform the subject's T1 image to the MNI standard space, and apply the same image transformation parameters to its corresponding fMRI image to complete the spatial standardization of the fMRI image.

[0078] Step 5: Perform smoothing, delinear shifting, regression covariate, and low-frequency filtering on the spatially standardized functional magnetic resonance images from Step 4.

[0079] Step 6: Extract the average time series of different brain regions from the functional magnetic resonance images processed in Step 5 based on the Brainnetome map and Power map templates.

[0080] The following is the solution process for the optimal transport transition matrix, such as... Figure 1 As shown:

[0081] Step 7: Calculate the centroid MNI coordinates of different brain regions under the two brain maps used in Step 6 as data points, and select the centroid MNI coordinates of the frontal lobe, temporal lobe, occipital lobe and parietal lobe as key points.

[0082] Step 8: Based on the MNI coordinates obtained in Step 7, construct the cost matrix and mask matrix, and calculate the transition matrix using the optimal transmission model (PL-LC-OT model) based on point-line relationships.

[0083]

[0084] In the formula: Let M be the feasible solution set of the PL-LC-OT model, where M is the mask matrix; and the cost matrix C = (C). ij =c(x i ,y j), where c(·,·) is the Jensen-Shannon divergence (JS divergence), x i ,y j The corresponding local structures include:

[0085] S1.1 Calculate the local structural constraints and cost matrix based on the point-line distance using the centroid MNI coordinates between the Brainnetome map and the Power map, where the Brainnetome map is the source domain and the Power map is the target domain.

[0086] S1.2, Calculate the mask matrix.

[0087] S1.3 calculates the transmission scheme using the logarithmic Sinkhorn iterative algorithm.

[0088] For cases where the transmission quality in the source and destination domains is incomplete, a corresponding Partial-PL-LC-OT model (the optimal transmission model for incomplete point-line relationships) is proposed based on the PL-LC-OT model to calculate the transition matrix:

[0089] In the formula:

[0090]

[0091] Let M be the feasible solution set corresponding to the model, where M is the mask matrix and m is the partial quality of transmission; the cost matrix C = (C) ij =c(x i ,y j ), where c(·,·) is the JS divergence, x i ,y j This corresponds to the local structure.

[0092] Step 9, S1.1 specifically includes the following steps:

[0093] S1.1.1: Draw an auxiliary line connecting two non-collinear key points. There are n non-collinear key points. Auxiliary lines, auxiliary line structure S of the source domain s Auxiliary line structure S of the target domain t As shown below, s and t are used to label the source domain and the target domain, respectively:

[0094]

[0095]

[0096] The auxiliary lines will divide the area into several smaller regions, such as Figure 2 As shown;

[0097] S1.1.2: For the auxiliary lines in S1.1.1, any two key points on an auxiliary line and the data point to be calculated can form a triangle. Use Heron's formula to calculate the distance from the data point to each auxiliary line:

[0098]

[0099] Where a, b, and c are the side lengths of the triangle, S is half the perimeter of the triangle, and S is the area of ​​the triangle;

[0100] S1.1.2.1: Calculate the lengths of the three sides respectively:

[0101]

[0102]

[0103]

[0104] in, Let i represent any key point within the domain. For any data point;

[0105] S1.1.2.2: Calculate any data point using Heron's formula To the auxiliary line S γ Distance:

[0106]

[0107] Among them, key points

[0108] S1.1.3: After the calculation in S1.1.2, the source domain data points can be obtained. Corresponding local structure and target domain data points Corresponding local structure

[0109]

[0110]

[0111] S1.1.4: To address the interference (non-rigid transformation) caused by changes in the spatial structure of the source and target domains, the source domain data points are defined as... With auxiliary linear structure Evaluation metrics for relation preservation (RL) properties:

[0112]

[0113] in, Dividing the point-to-line distance by its maximum value normalizes it to [0,1], increasing the robustness of the relationship score to the distance scale. The value of parameter ρ is often set to 0.1 in the softmax function.

[0114] Similarly, the target domain data points can be obtained. With auxiliary linear structure Evaluation metrics for the properties of maintaining relationships:

[0115]

[0116] in, Dividing the point-to-line distance by its maximum value normalizes it to [0,1], increasing the robustness of the relationship score to the distance scale. The value of parameter ρ is often set to 0.1 in the softmax function.

[0117] S1.1.5: Based on the relation preservation property evaluation metric defined in S1.1.4, calculate S based on the key points in the source domain and the corresponding auxiliary line structure. s The resulting local structure:

[0118]

[0119] Similarly, we can obtain the structure S based on the key points and corresponding auxiliary lines within the target domain. t The resulting local structure:

[0120]

[0121] S1.1.6: Calculate the cost matrix C = (C) ij =c(x i ,y j ), where c(·,·) is the JS divergence, x i ,y j For the corresponding local structure;

[0122] The specific method for calculating the mask matrix M in S1.2 is as follows:

[0123]

[0124] in, Let Q be the index of the keypoint pair, and Q be the number of keypoint pairs. Let represent the key point index sets in the source and target domains, respectively. s and t are used to label the source and target domains, respectively, and i and j represent the row and column indices of the matrix elements, respectively.

[0125] Step 10 and S1.3 are as follows:

[0126] S1.3.1, in the logarithmic field, the iteration of u in the Sinkhorn algorithm is transformed into the following form:

[0127]

[0128] S1.3.2, let Then the expression in S1.3.1 has the following equivalent form:

[0129]

[0130] Where, when M ij When = 0, log(M) ij )H(f (l) ,g (l) ) ij =-∞;

[0131] S1.3.3, To ensure the consistency of the algorithm, an extended definition of H(f,g) is given:

[0132]

[0133] S1.3.4, gives the log-sum-exp function LogSumExp: The definition of is:

[0134]

[0135] S1.3.5, combined with S1.3.1 and S1.3.3, can be simplified from the formula in S1.3.2 as follows:

[0136]

[0137] S1.3.6, similarly, regarding g (l+1) The iterative update formula is:

[0138]

[0139] S1.3.7, through the iterative formulas of S1.3.5 and S1.3.6, can iteratively update u and v in the Sinkhorn algorithm, and then solve for the optimal transmission scheme of the model, i.e. the optimal transmission transition matrix.

[0140] Step 11: Using the average time series obtained in Step 6 and the optimal transfer matrix in Step 10, transfer is performed on each time point to obtain a new average time series, and then the functional connectivity of the brain is calculated.

[0141] Step 12: Construct functional connectomes using the average time series obtained in Step 6, thereby obtaining brain functional connectomes from Brainnetome and Power map templates.

[0142] Step 13: Observe the data before and after transmission. When the data is converted from the Brainnetome map to the Power map, the correlation can be obtained (r = 0.5036, p < 0.001). For the functional connectomes obtained in Step 11 and Step 12, the gender is predicted using support vector machine. The original Power map (Step 12) has an accuracy of 58.69% in predicting gender, while the converted Power map (Step 11) has an accuracy of 49.67% in predicting gender.

[0143] Step 14: Following the steps above, calculate the transition matrix using the Power graph as the source domain and the Brainnetome graph as the target domain. When converting from the Power graph to the Brainnetome graph, the correlation can be obtained (r = 0.5632, p < 0.001). For the functional connectivity obtained in Steps 11 and 12, gender is predicted using a support vector machine. The accuracy of the original Brainnetome graph in predicting gender is 52.39%, and the accuracy of the converted Brainnetome graph in predicting gender is 51.94%.

[0144] Step 15: From the results of steps 13 and 14, it can be analyzed that the converted map can also reflect the individual information of the original map very well.

[0145] Therefore, the brain atlas mapping method based on point-constrained optimal transmission provided in this embodiment can remap different brain atlases. This method first obtains high-resolution structural (T1w) and functional (fMRI) images of the adult brain from publicly available datasets, and preprocesses the fMRI data to obtain time series. Then, it extracts the average time series of regions of interest based on Brainnetome and Power atlases respectively: using Brainnetome atlases, time series of 207 cortical regions can be obtained; using Power atlases, time series of 243 cortical regions can be obtained. However, existing mapping methods require fMRI data for conversion between different atlases, resulting in unnecessary time waste. Compared to traditional methods, this method considers that the cerebral cortex can be divided into 7 brain networks, and therefore uses their centroid coordinates as key points for optimal transmission guidance, thereby improving the accuracy of atlas conversion and reducing time costs.

[0146] In summary, the brain map mapping method based on point-constrained optimal transmission provided by this invention has the following advantages compared with existing mapping methods:

[0147] This method processes functional magnetic resonance imaging (fMRI) data to obtain functional signal images, then extracts average time series from different brain atlases and calculates the coordinates of the Montreal Neuroscience Institute. Next, it combines an optimal transfer algorithm to calculate the optimal transfer matrix between the atlases using the constructed cost matrix and mask matrix, and inputs this matrix into the average time series to obtain a new average time series. This yields the functional connectome of the brain, ultimately achieving mapping between two different brain atlases. The optimal transfer model using point-line relationships incorporates point-line distance, adding an indicator to the model and characterizing the local structural constraints of the data domain. This effectively extracts structural information from the data domain, ensuring that the converted atlas accurately reflects the individual information of the original atlas. This improves the accuracy of atlas conversion and saves time.

[0148] The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

Claims

1. A brain map mapping method based on point-constrained optimal transmission, characterized in that, include: S1: Process the functional magnetic resonance imaging data to obtain functional signal images; S2: Two different brain maps were used to extract functional signal images, and the corresponding average time series were obtained respectively; S3: Calculate the corresponding coordinates of the Montreal Institute of Neuroscience using the two brain maps mentioned above to obtain data points; S4: Construct the cost matrix and mask matrix based on the data points, and combine them with the optimal transmission algorithm to obtain the optimal transmission transfer matrix; S5: Based on the average time series and the optimal transfer matrix, a new average time series is calculated; S6: Based on the new average time series, the functional connectome of the brain is obtained, and the mapping between two different brain maps is realized; In S4, the cost matrix and mask matrix are constructed based on the data points, and the optimal transmission transfer matrix is ​​calculated by combining the optimal transmission model of point-line relationship. The specific steps of S4: The cost matrix and mask matrix are calculated based on the optimal transmission model of point-line relationship; then the optimal transmission transfer matrix is ​​calculated using the logarithmic field Sinkhorn iterative algorithm. The specific steps for calculating the cost matrix are as follows: S1.1.1: Draw an auxiliary line connecting two non-collinear key points. There are n non-collinear key points. Auxiliary lines, auxiliary line structure of the source domain Auxiliary line structure of the target domain They are shown below, Used to mark the source domain and the target domain respectively: The auxiliary lines divide the area into several smaller regions; S1.1.2: For the auxiliary lines in S1.1.1, any two key points on an auxiliary line and the data point to be calculated can form a triangle. Use Heron's formula to calculate the distance from the data point to each auxiliary line: in, These are the side lengths of the triangle. It is half the perimeter of the triangle. Let be the area of ​​the triangle; S1.1.2.1: Calculate the lengths of the three sides respectively: in, As a key point within the domain, Represented as any key point, For any data point; S1.1.2.2: Calculate any data point using Heron's formula To the auxiliary line Distance: Among them, key points ; S1.1.3: After the calculation in S1.1.2, the source domain data points can be obtained. Corresponding local structure and target domain data points Corresponding local structure : S1.1.4: Define the source domain data points With auxiliary linear structure Evaluation metrics for the properties of maintaining relationships: in, The distance between the point and the line is normalized by dividing by its maximum value. ; Calculate the target domain data points With auxiliary linear structure Evaluation metrics for the properties of maintaining relationships: in, Divide the distance between the points and lines by their maximum value to normalize to... ; S1.1.5: Based on the relation preservation property evaluation metric defined in S1.1.4, calculate the structure based on key points in the source domain and their corresponding auxiliary lines. The resulting local structure: The calculation yields the structure based on key points and corresponding auxiliary lines within the target domain. The resulting local structure: S1.1.6: Calculate the cost matrix ,in Let JS divergence be the metric. This corresponds to the local structure.

2. The brain map mapping method based on point-constrained optimal transmission according to claim 1, characterized in that, The specific steps of S1 include: S1.1: Format conversion of functional magnetic resonance imaging data; S1.2: The functional magnetic resonance imaging data after format conversion is sequentially processed by data filtering, time-level correction, eddy current correction, spatial standardization, smoothing, delinear drift removal, regression covariate, and low-frequency filtering to obtain the functional signal image.

3. The brain map mapping method based on point-constrained optimal transmission according to claim 1, characterized in that, In S3, the centroid coordinates of the frontal lobe, temporal lobe, occipital lobe, and parietal lobe of the brain are selected as key points from the calculated data points.

4. A brain map mapping system based on point-constrained optimal transmission, used to implement the steps of the brain map mapping method based on point-constrained optimal transmission as described in any one of claims 1-3, characterized in that, include: The data processing module is used to process functional magnetic resonance imaging data to obtain functional signal images; The feature extraction module is used to extract functional signal images using two different brain maps, and obtain the corresponding average time series respectively; The coordinate calculation module is used to calculate the corresponding coordinates of the Montreal Institute of Neuroscience using two different brain maps, and obtain data points. The optimal transmission calculation module is used to construct the cost matrix and mask matrix based on data points, and combine them with the optimal transmission algorithm to obtain the optimal transmission transition matrix; The transfer module is used to calculate a new average time series based on the average time series and the optimal transfer matrix; The atlas mapping module is used to obtain the functional connectome of the brain based on the new average time series, and to realize the mapping between two different brain atlases.

5. A device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the brain map mapping method based on point-constrained optimal transmission as described in any one of claims 1-3 when executing the computer program.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it is used to implement the steps of the brain map mapping method based on point-constrained optimal transmission as described in any one of claims 1-3.

Images