A method of locating electroencephalography electrodes

Abstract

This application relates to the field of brain power imaging technology and discloses a brain power imaging electrode localization method. The method includes: acquiring a set of scalp vertices in an individual head model; for an electrode geodesic to be constructed, determining the anatomical constraint points and two endpoints of the electrode geodesic based on the plurality of anatomical landmarks; generating a reference plane based on the anatomical constraint points and the two endpoints; extracting candidate vertices that are compatible with the reference plane from the set of scalp vertices, and constructing the electrode geodesic that matches the individual head model based on the candidate vertices; and determining the actual position of each electrode based on the total length of the electrode geodesic, the preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices, thereby achieving the technical effect of accurately fitting the individual head shape and strictly conforming to the electrode topology layout of the international standard lead system.

Description

Technical Field

[0001] This application relates to the field of brain power imaging technology, and in particular to a brain power imaging electrode localization method. Background Technology

[0002] EEG (Electroencephalography) reconstructs cortical neural activity from electrical signals recorded from the scalp. Its accuracy is highly dependent on the precise positioning of electrodes on an individual head model. Ideal electrode positioning requires meeting two core requirements simultaneously: shape fit (the overall shape of the electrode array must precisely conform to the individual's scalp surface) and topological fidelity (the relative distances between electrodes must strictly adhere to the International Electrode Placement System (IEPS) specifications). Relevant electrode positioning methods are primarily based on registration techniques, achieving positioning by transferring a standard electrode template onto the individual head model. However, these registration methods cannot simultaneously achieve shape fit and topological fidelity; the resulting geometric distortion reduces the positioning accuracy of EEG. Summary of the Invention

[0003] This application provides a method for positioning electrodes in brain power imaging, which solves the technical problem that related registration methods cannot simultaneously take into account the shape adaptation of the electrode array and the topological fidelity, and achieves the technical effect of both accurately fitting the individual's head shape and strictly conforming to the electrode topological layout of the international standard lead system.

[0004] To achieve the above objectives, the main technical solutions adopted in this application include: In a first aspect, embodiments of this application provide a method for locating electrodes in brain power imaging. The method includes: acquiring a set of scalp vertices in an individual head model, the set of scalp vertices including multiple anatomical landmarks; for an electrode geodesic to be constructed, determining anatomical constraint points and two endpoints of the electrode geodesic based on the multiple anatomical landmarks; generating a reference plane based on the anatomical constraint points and the two endpoints, the reference plane being used to provide anatomical constraints for the orientation of the electrode geodesic; extracting candidate vertices adapted to the reference plane from the set of scalp vertices, and constructing the electrode geodesic matching the individual head model based on the candidate vertices; and determining the actual position of each electrode based on the total length of the electrode geodesic, a preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices.

[0005] The brain power imaging electrode localization method provided in this application provides a data foundation for shape adaptation of the individual head model by acquiring a set of scalp vertices containing anatomical landmarks. Based on the anatomical landmarks, the anatomical constraint points and two endpoints of the electrode geodesics are determined, and a reference plane is generated to provide anatomical constraints on the direction of the electrode geodesics, ensuring compliance with the specifications of international standard lead systems. Vertices that adapt to the reference plane are extracted as candidate vertices, and electrode geodesics are constructed based on these candidate vertices. Since the candidate vertices originate from the set of scalp vertices and are subject to the anatomical constraints of the reference plane, the constructed geodesics conform to the geometry of the individual head model and extend along the anatomical path specified by international standards. The actual positions of each electrode are determined based on the total length of the electrode geodesics, a preset relative distance ratio, and the set of scalp vertices, ensuring that the electrodes are precisely distributed along the geodesics according to international standard proportions and conform to the scalp surface. This achieves both shape adaptation and topological fidelity on the individual head model.

[0006] Optionally, the candidate vertices are determined as follows: extracting vertices from the scalp vertex set whose distance from the reference plane is less than a preset threshold, and determining the extracted vertices as candidate vertices.

[0007] By extracting vertices whose distance from the reference plane is less than a preset threshold as candidate vertices, it is ensured that the candidate vertices are located near the anatomical path, and the global search of scalp vertices is reduced to narrowband filtering, thus improving computational efficiency.

[0008] Optionally, the plurality of anatomical landmarks include the nasal root point, the external occipital protuberance, the left preauricular point, and the right preauricular point. Based on the plurality of anatomical landmarks, the anatomical constraint point and two endpoints of the electrode geodesic are determined, including: for the sagittal electrode geodesic to be constructed, determining the nasal root point and the external occipital protuberance as the two endpoints of the sagittal electrode geodesic; calculating the midpoint between the left preauricular point and the right preauricular point, and determining the midpoint as the anatomical constraint point of the sagittal electrode geodesic.

[0009] By defining the nasal root point and the external occipital protuberance as the two endpoints of the sagittal electrode geodesic, it conforms to the start and end definitions of the sagittal reference curve in IEPS. Using the midpoint between the left and right preauricular points as anatomical constraint points, the electrode geodesic is forcibly constrained to extend near the midsagittal plane of the head, achieving precise anatomical constraint on the geodesic's direction.

[0010] Optionally, the candidate vertices include the two endpoints. Based on the candidate vertices, constructing the electrode geodesic that matches the individual head model includes: taking the two endpoints as the starting path vertex and the ending path vertex of the electrode geodesic, respectively; sequentially determining each path vertex constituting the electrode geodesic from the candidate vertices according to the direction from the starting path vertex to the ending path vertex; and constructing the electrode geodesic based on each path vertex.

[0011] Following the direction from the starting path vertex to the ending path vertex, each path vertex is determined sequentially to ensure that the electrode geodesics form a continuous and smooth path on the scalp surface. Simultaneously, the path is generated only within candidate vertices near the reference plane, ensuring it does not deviate from anatomical constraints and adheres strictly to the scalp surface of the individual head model, thus providing a stable and accurate reference path for subsequent electrode positioning.

[0012] Optionally, the path vertices constituting the electrode geodesic are sequentially determined from the candidate vertices, including: for any current path vertex, determining candidate vertices whose Euclidean distance to the current path vertex is greater than or equal to a first threshold and less than or equal to a second threshold; extracting the candidate vertex with the smallest Euclidean distance to the current path vertex from the candidate vertices, using the extracted candidate vertex as the next path vertex, and updating the next path vertex as the current path vertex.

[0013] By setting a first threshold and a second threshold, candidate vertices are double-screened at each selection step. This avoids local path oscillations caused by selecting vertices that are too close together, and prevents long-distance jumps caused by selecting vertices that are too far apart. This results in electrode geodesics with uniform spacing between adjacent vertices, a continuous and smooth path, and close adherence to the scalp surface of the individual head model. By selecting the vertex closest to the current vertex from the candidate vertices as the next path vertex, optimal path selection within a local range is achieved, improving path fit accuracy while ensuring overall stability. The electrode geodesics constructed in this way not only strictly follow the anatomical path specified by international standards but also precisely adhere to the surface of the individual head model, providing a high-quality reference path for subsequent electrode positioning.

[0014] Optionally, the actual position of each electrode is determined based on the total length of the electrode geodesic, the preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices. This includes: for any electrode, calculating the product of the preset relative distance ratio and the total length of the electrode geodesic to obtain the target cumulative distance of the electrode along the electrode geodesic; determining the target path vertex among the path vertices based on the target cumulative distance; determining the theoretical position of the electrode based on the target path vertex; and determining the actual position of the electrode based on the theoretical position of the electrode and the set of scalp vertices.

[0015] The target cumulative distance is obtained by multiplying the relative distance ratio specified by IEPS by the total length of the electrode geodesics. This standardized electrode arrangement ratio is then mapped onto the head models of different individuals. Based on the target cumulative distance, the target path vertices and theoretical positions are determined, and the actual positions of the electrodes are obtained by combining this with the scalp vertex set. This ensures that the final determined electrode positions not only conform to the IEPS topological layout requirements of being distributed along the geodesics, but also accurately fit the geometry of the individual's scalp.

[0016] Optionally, determining the target path vertex among the path vertices based on the target cumulative distance includes: for any adjacent first path vertex and second path vertex among the path vertices, if the first cumulative distance of the first path vertex is less than the target cumulative distance, and the second cumulative distance of the second path vertex is greater than or equal to the target cumulative distance, then the first path vertex is determined as the first target path vertex, and the second path vertex is determined as the second target path vertex.

[0017] By comparing the cumulative distance between adjacent path vertices on the electrode geodesic line with the target cumulative distance, the adjacent vertex pairs into which the target cumulative distance falls are identified as the first target path vertex and the second target path vertex, respectively. This determination, based on the strict monotonicity of the cumulative distance, can accurately and uniquely pinpoint the target path segment containing the target cumulative distance on the electrode geodesic line, ensuring that the theoretical position always lies within the anatomical orientation of the electrode geodesic line. This provides a definite geometric boundary for subsequent high-precision electrode positioning.

[0018] Optionally, for any path vertex other than the starting path vertex, the cumulative distance of the path vertex is determined as follows: for any pair of adjacent path vertices from the starting path vertex to the path vertex, the path segment length between the adjacent path vertices is calculated; the path segment lengths between each pair of adjacent path vertices are summed to obtain the cumulative distance of the path vertex.

[0019] Using the starting path vertex of the electrode geodesic line as a reference, the cumulative distance of the corresponding path vertex is obtained by calculating and accumulating the path segment lengths of each adjacent path vertex from the starting path vertex to the target path vertex. This ensures that the calculated cumulative distance fits the actual shape of the scalp surface of the individual head model, providing a reliable and accurate length basis for the subsequent precise positioning of the electrode position based on the target cumulative distance.

[0020] Optionally, the method further includes: determining the cumulative distance of the terminating path vertex as the total length of the electrode geodesic.

[0021] Since the terminal path vertex is the end vertex of the electrode geodesic, its cumulative distance has completely accumulated the lengths of all path segments from the starting path vertex to the terminal path vertex, which can fully reflect the actual total length of the electrode geodesic on the scalp surface of the individual head model, providing a length basis for subsequent calculation of electrode positions according to standard proportions.

[0022] Optionally, determining the theoretical position of the electrode based on the target path vertex includes: calculating the difference between the target cumulative distance and the cumulative distance of the first target path vertex; calculating the target path segment length between the second target path vertex and the first target path vertex; calculating the ratio between the difference and the target path segment length, and determining the ratio as an interpolation ratio; and performing linear interpolation between the first target path vertex and the second target path vertex based on the interpolation ratio to obtain the theoretical position of the electrode.

[0023] By calculating the difference between the cumulative distance to the target and the cumulative distance to the first target path vertex, the specific distance that needs to be traveled along the geodesic from the first target path vertex can be determined. The precise length of the target path segment is obtained by calculating the Euclidean distance between the first and second target path vertices. Dividing the difference by the target path segment length yields the interpolation ratio, which accurately reflects the relative position of the theoretical location on that path segment. Linear interpolation is performed between the two target vertices based on this interpolation ratio, ensuring that the calculated theoretical position is strictly located on the electrode geodesic path segment without deviating from the anatomical orientation. This also overcomes the resolution limitations of the head model mesh vertex density, achieving higher positioning accuracy than directly selecting vertices. This lays a solid foundation for the final, globally geometrically accurate electrode array deployment on individual head models.

[0024] Optionally, determining the actual position of the electrode based on the theoretical position of the electrode and the set of scalp vertices includes: for any scalp vertex in the set of scalp vertices, calculating the spatial distance between the theoretical position and the scalp vertex; identifying the minimum value among the various spatial distances, and determining the scalp vertex corresponding to the minimum value as the actual position of the electrode.

[0025] By calculating the spatial distance between the theoretical electrode position and each vertex in the scalp vertex set, and selecting the scalp vertex with the smallest distance as the actual electrode position, the theoretical position obtained by high-precision interpolation can be mapped to the actual scalp vertices in the individual head model. This process ensures that the electrode position is as close as possible to the theoretical calculation target and does not deviate from the anatomical constraint path, while making the final electrode landing point a discrete mesh vertex on the individual scalp surface, thus ensuring that the electrode position perfectly fits the scalp surface of the individual head model.

[0026] Secondly, embodiments of this application provide a brain power imaging electrode localization system, the system comprising: an acquisition module, configured to acquire a set of scalp vertices in an individual head model, the set of scalp vertices including multiple anatomical landmarks; a generation module, configured to, for an electrode geodesic to be constructed, determine anatomical constraint points and two endpoints of the electrode geodesic based on the multiple anatomical landmarks; generate a reference plane based on the anatomical constraint points and the two endpoints, the reference plane being used to provide anatomical constraints for the orientation of the electrode geodesic; a construction module, configured to extract candidate vertices adapted to the reference plane from the set of scalp vertices, and construct the electrode geodesic matching the individual head model based on the candidate vertices; and a determination module, configured to determine the actual position of each electrode based on the total length of the electrode geodesic, a preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices.

[0027] Thirdly, embodiments of this application provide a brain-computer interface device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the above-described brain power imaging electrode localization method by executing the computer instructions.

[0028] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the above-described brain power imaging electrode localization method.

[0029] Fifthly, embodiments of this application provide a computer program product, including computer instructions, which are used to cause a computer to execute the above-described brain power imaging electrode localization method. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0031] Figure 1 A flowchart illustrating a brain power imaging electrode localization method provided in this application embodiment; Figure 2 A schematic diagram of the electrode geodesic line provided in the embodiments of this application; Figure 3 A schematic diagram illustrating how AELE changes with parameters, provided in an embodiment of this application; Figure 4 A schematic diagram illustrating how AELE changes with parameters, provided in an embodiment of this application; Figure 5 A schematic diagram of a brain power imaging electrode localization system provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a brain-computer interface device provided in an embodiment of this application. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0033] Electroencephalography (EEG), a non-invasive electroencephalography technique, records electrical potential signals on the scalp surface in real time, indirectly reflecting the electrical activity of neurons in the cerebral cortex. Source imaging of the brain (ESI), a key extension of EEG technology, aims to reconstruct the source of neural activity in the cerebral cortex from the recorded electrical potential signals on the scalp, providing precise spatial localization for brain function assessment and disease diagnosis. Accurate electrode localization on an individual head mold is the core spatial computational basis for reliable ESI, directly determining the source localization accuracy and clinical application value of ESI. Electrode localization must simultaneously meet two core requirements: first, the geometry of the electrode array must be highly adapted to the individual head mold, conforming to the physiological structure of different subjects' heads; second, the relative distance between electrodes must strictly adhere to the International Electrode Placement System (IEPS) to ensure the universality and comparability of the test data. However, current mainstream electrode localization methods are based on standard template registration paradigms, making it difficult to simultaneously meet both requirements and resulting in unavoidable geometric distortion problems. The core logic of relevant registration methods is to adapt a pre-set standard electrode template to an individual head model. This can be divided into two categories: one is a rigid transformation method based on anatomical landmarks, which aligns the standard template with the scalp grid using anatomical landmarks such as the nasal root point and bilateral preauricular points. However, rigid transformation can only achieve a globally uniform adjustment and cannot adapt to the local geometric differences of the individual head model, resulting in insufficient adhesion between the electrode array and the scalp. The other is a nonlinear registration method based on deformable models, which generates an average electrode template and performs local deformation to adapt to the individual head model. However, nonlinear local adjustments can disrupt the topological relationships between electrodes specified by IEPS, causing the relative positions of the electrodes to deviate from the standard and affecting data consistency. The fundamental contradiction between these two methods lies in the fact that maintaining topological fidelity requires a globally invariant uniform transformation, while achieving shape adaptation requires flexible local adjustments. These two methods are mathematically incompatible, ultimately leading to inherent deviations in electrode positioning and reducing the reconstruction accuracy of ESI. To improve the initial accuracy of template registration, related research has introduced sensor technologies such as 3D scanning and photogrammetry to obtain more refined standard templates. However, these methods still do not deviate from the core framework of template adaptation and cannot fundamentally solve the problem of geometric distortion. In addition, although related geodesic path calculation methods have been applied to fields such as mathematics and computer science, they mostly rely on global surface calculations. For high-resolution individual head models containing millions of vertices, these methods suffer from high computational costs and low efficiency, making them difficult to directly apply to electrode positioning scenarios.

[0034] Therefore, developing an electrode positioning method that can simultaneously achieve electrode array shape adaptation to individual head models and topological fidelity (i.e., global geometric fidelity) and is computationally efficient has become a key requirement for breaking through the current ESI technology bottleneck.

[0035] This application provides a method for locating electrodes in brain power imaging. By calculating precise individual electrode positions on an individual head model, the method improves electrode positioning accuracy while achieving global geometric fidelity of the electrode array.

[0036] This application provides a method for locating electrodes for brain power imaging, specifically a global geometry preservation electrode localization (GPEL) method based on the geodesy of an individual head model. It should be noted that the steps shown in the flowcharts can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0037] Please refer to Figure 1 , Figure 1 A flowchart of a brain power imaging electrode localization method provided in this application embodiment is shown below. Figure 1 As shown, the process includes the following steps: Step S1: Obtain the set of scalp vertices in the individual head model, wherein the set of scalp vertices includes multiple anatomical landmarks.

[0038] An individual head model is a personalized three-dimensional geometric model of the head reconstructed from medical imaging data of a specific individual's head. This medical imaging data can be obtained through magnetic resonance imaging (MRI). A set of scalp vertices characterizes the scalp surface geometry of the individual head model and is obtained by extracting a discrete set of three-dimensional points from the scalp surface. Anatomical landmarks, also part of the scalp vertices set, constrain the extension of electrode geodesics. Electrode geodesics are scalp surface paths that provide a basis for electrode placement. Anatomical landmarks include the nasal root point (Nz), the external occipital protuberance point (Iz), the left preauricular point (Lpa), and the right preauricular point (Rpa).

[0039] Step S3: For the electrode geodesic to be constructed, based on the multiple anatomical landmarks, determine the anatomical constraint points and two endpoints of the electrode geodesic; based on the anatomical constraint points and the two endpoints, generate a reference plane, which is used to provide anatomical constraints for the orientation of the electrode geodesic.

[0040] Please refer to Figure 2 , Figure 2 This is a schematic diagram of an electrode geodesic provided in an embodiment of this application. The two endpoints define the starting and ending positions of the electrode geodesic. The anatomical constraint point can be calculated from anatomical landmarks. The two endpoints and the anatomical constraint point uniquely define a reference plane. The reference plane constrains the extension of the electrode geodesic on the scalp surface, ensuring it conforms to the electrode placement specifications stipulated by IEPS. Taking a sagittal plane geodesic as an example, the nasal root point and the external occipital protuberance are determined as the two endpoints of the sagittal plane electrode geodesic. The midpoint between the left and right preauricular points is calculated and determined as the anatomical constraint point of the sagittal plane electrode geodesic.

[0041] Step S5: Extract candidate vertices that are compatible with the reference plane from the scalp vertex set, and construct the electrode geodesics that match the individual head model based on the candidate vertices.

[0042] Since the candidate vertices originate from the scalp vertex set, the constructed electrode geodesics can accurately conform to the actual geometry of the individual head model, thus eliminating shape adaptation errors caused by template deformation in related registration methods. Furthermore, because the candidate vertices are constrained by the anatomical reference plane, the constructed electrode geodesics extend along the anatomical path specified by IEPS, laying a foundation for topologically accurate electrode placement in the subsequent process.

[0043] Step S7: Based on the total length of the electrode geodesic, the preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices, determine the actual position of each electrode.

[0044] The relative distance ratio refers to the percentage of each electrode relative to the total length of the entire scalp reference curve, as specified in IEPS. Combining the relative distance ratio specified in IEPS with the total length of the electrode geodesics facilitates precise mapping of electrode positions onto the individualized head mold. Because the electrodes are positioned strictly according to the IEPS ratio along the electrode geodesics, the topology of the electrode array conforms to international standards. Furthermore, the precise fit between the electrode geodesics and the scalp surface of the individual head mold ensures that the final electrode positions conform to the individual's head shape.

[0045] The brain power imaging electrode localization method provided in this application provides a data foundation for shape adaptation of the individual head model by acquiring a set of scalp vertices containing anatomical landmarks. Based on the anatomical landmarks, the anatomical constraint points and two endpoints of the electrode geodesics are determined, and a reference plane is generated to provide anatomical constraints on the direction of the electrode geodesics, ensuring compliance with the specifications of international standard lead systems. Vertices that adapt to the reference plane are extracted as candidate vertices, and electrode geodesics are constructed based on these candidate vertices. Since the candidate vertices originate from the set of scalp vertices and are subject to the anatomical constraints of the reference plane, the constructed geodesics conform to the geometry of the individual head model and extend along the anatomical path specified by international standards. The actual positions of each electrode are determined based on the total length of the electrode geodesics, a preset relative distance ratio, and the set of scalp vertices, ensuring that the electrodes are precisely distributed along the geodesics according to international standard proportions and conform to the scalp surface. This achieves both shape adaptation and topological fidelity on the individual head model.

[0046] In some specific embodiments, the candidate vertices are determined as follows: vertices in the scalp vertex set whose distance from the reference plane is less than a preset threshold are extracted, and the extracted vertices are determined as candidate vertices.

[0047] By filtering vertices close to the reference plane from a large number of scalp vertices using a preset threshold, the candidate vertices are restricted to a narrow band near the reference plane, which significantly reduces the subsequent path search range and improves computational efficiency. By excluding vertices with excessively large distances from the reference plane, the subsequently constructed electrode geodesics are ensured to be constrained to the vicinity of the reference plane and not deviate from the anatomical orientation, thereby improving the accuracy of electrode localization.

[0048] In some specific embodiments, the plurality of anatomical landmarks includes the nasal root point, the external occipital protuberance, the left preauricular point, and the right preauricular point. Based on the plurality of anatomical landmarks, the anatomical constraint point and two endpoints of the electrode geodesic are determined, including: for the sagittal electrode geodesic to be constructed, determining the nasal root point and the external occipital protuberance as the two endpoints of the sagittal electrode geodesic; calculating the midpoint between the left preauricular point and the right preauricular point, and determining the midpoint as the anatomical constraint point of the sagittal electrode geodesic.

[0049] By defining the nasal root point and the external occipital protuberance as the two endpoints of the sagittal electrode geodesic, it conforms to the start and end definitions of the sagittal reference curve in IEPS. Using the midpoint between the left and right preauricular points as anatomical constraint points, the electrode geodesic is forcibly constrained to extend near the midsagittal plane of the head, achieving precise anatomical constraint on the geodesic's direction.

[0050] In some specific embodiments, the candidate vertex includes the two endpoints. Based on the candidate vertex, constructing the electrode geodesic that matches the individual head model includes: taking the two endpoints as the starting path vertex and the ending path vertex of the electrode geodesic, respectively; sequentially determining each path vertex constituting the electrode geodesic from the candidate vertices in the direction from the starting path vertex to the ending path vertex; and constructing the electrode geodesic based on each path vertex.

[0051] Following the direction from the starting path vertex to the ending path vertex, each path vertex is determined sequentially to ensure that the electrode geodesics form a continuous and smooth path on the scalp surface. Simultaneously, the path is generated only within candidate vertices near the reference plane, ensuring it does not deviate from anatomical constraints and adheres strictly to the scalp surface of the individual head model, thus providing a stable and accurate reference path for subsequent electrode positioning.

[0052] In some specific embodiments, the path vertices constituting the electrode geodesic are sequentially determined from the candidate vertices, including: for any current path vertex, determining candidate vertices whose Euclidean distance to the current path vertex is greater than or equal to a first threshold and less than or equal to a second threshold; extracting the candidate vertex with the smallest distance to the current path vertex from the candidate vertices, using the extracted candidate vertex as the next path vertex, and updating the next path vertex as the current path vertex.

[0053] By setting a first threshold and a second threshold, candidate vertices are double-screened at each selection step. This avoids local path oscillations caused by selecting vertices that are too close together, and prevents long-distance jumps caused by selecting vertices that are too far apart. This results in electrode geodesics with uniform spacing between adjacent vertices, a continuous and smooth path, and close adherence to the scalp surface of the individual head model. By selecting the vertex closest to the current vertex from the candidate vertices as the next path vertex, optimal path selection within a local range is achieved, improving path fit accuracy while ensuring overall stability. The electrode geodesics constructed in this way not only strictly follow the anatomical path specified by international standards but also precisely adhere to the surface of the individual head model, providing a high-quality reference path for subsequent electrode positioning.

[0054] In some specific implementations, constructing the electrode geodesic includes the following steps: 1) For the electrode geodesic to be constructed, generate the corresponding reference plane S. The normal vector n of S can be expressed as: In the above formula, and These are the coordinates of the starting and ending vertices of the electrode geodesic line, respectively. yes Anatomical constraint points other than the starting path vertex and the ending path vertex.

[0055] 2) Extract candidate vertices. For the scalp vertex set... any scalp vertex Calculate the perpendicular distance from the vertex to the reference plane S. If this distance is less than a preset threshold t, then include the vertex in the candidate vertex set. : In the above formula, t is a preset threshold; This represents the total number of scalp vertices contained in the set of scalp vertices.

[0056] The value of the threshold t is related to the resolution of the head model mesh. If the value of t is too small, there will be too few candidate vertices, which may destroy the continuity of the path; if the value of t is too large, too many vertices far away from the anatomical plane will be introduced, which will weaken the effect of anatomical constraints.

[0057] Through this screening process, the candidate vertex set... The candidate vertices are all located in a narrow band region near the reference plane, which satisfies the requirements of the anatomical constraints and significantly reduces the number of vertices for subsequent path search.

[0058] 3) Determine the vertices of each path that constitute the electrode geodesic. (Set the starting point...) As the initial current path vertex (k=1). In each iteration, from the candidate vertex set... Filter out all vertices on the current path Vertices whose distances satisfy the double threshold constraint constitute the candidate vertex set. : In the above formula, This is the minimum connection threshold (i.e., the first threshold). This is the maximum jump threshold (i.e., the second threshold). This represents the number of candidate vertices that meet the above conditions.

[0059] The minimum connectivity threshold is used to prevent local oscillations in the path caused by selecting vertices that are too close together. When the head model mesh is dense, allowing the selection of extremely close vertices may cause the path to oscillate back and forth within a small range, affecting the overall smoothness. The maximum jump threshold is used to prevent long jumps in the path caused by selecting vertices that are too far apart. When the head model mesh is sparse, allowing the selection of vertices that are too far apart may cause the path to cross a large curved area, deviating from the true curvature of the head model.

[0060] After obtaining the set of candidate vertices Then, a greedy search strategy is used to select the optimal vertex for the next path. The core idea of ​​greedy search is to make a locally optimal choice in the current step, that is, to select the vertex that is closest to the current path from the set of candidate vertices. The vertex with the smallest distance is selected as the next path vertex. : This selection strategy ensures that the path extends forward with the minimum step size locally, thus maximizing its approximation to the natural curvature of the headform surface. The next path vertex is then selected. Update the current path vertex and repeat the filtering, selection, and updating steps above until the destination is reached. .

[0061] As a robust measure, if a candidate vertex satisfying the double threshold constraint cannot be found during the iteration process (i.e., C is an empty set), then the endpoint is forcibly removed. Set it as the next path vertex. This mechanism ensures that the algorithm can still terminate successfully under special circumstances, constructing a complete geodesic path.

[0062] Through the above iterative process, the discrete path vertex sequence constituting the electrode geodesic is finally obtained: In the above formula, It is the total number of vertices on the path. , This vertex sequence is the geometric representation of electrode geodesics that match the shape of the individual head model and extend strictly along the anatomical path.

[0063] In some specific embodiments, the actual position of each electrode is determined based on the total length of the electrode geodesic, a preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices. This includes: for any electrode, calculating the product of the preset relative distance ratio and the total length of the electrode geodesic to obtain the target cumulative distance of the electrode along the electrode geodesic; determining the target path vertex among the path vertices based on the target cumulative distance; determining the theoretical position of the electrode based on the target path vertex; and determining the actual position of the electrode based on the theoretical position of the electrode and the set of scalp vertices.

[0064] The target cumulative distance refers to the path length that the electrode should traverse from the starting path vertex to its theoretical position along the electrode geodesic. Target path vertices are the two adjacent path vertices on the electrode geodesic path where the cumulative distance first exceeds the target cumulative distance. These two target path vertices allow for the location of the specific path segment where the theoretical electrode position is situated. The theoretical position lies between the two target path vertices, but not necessarily on the scalp vertex. By mapping the theoretical position to the scalp vertex, the actual scalp vertex position where the electrode is placed is obtained.

[0065] The target cumulative distance is obtained by multiplying the relative distance ratio specified by IEPS by the total length of the electrode geodesics. This standardized electrode arrangement ratio is then mapped onto the head models of different individuals. Based on the target cumulative distance, the target path vertices and theoretical positions are determined, and the actual positions of the electrodes are obtained by combining this with the scalp vertex set. This ensures that the final determined electrode positions not only conform to the IEPS topological layout requirements of being distributed along the geodesics, but also accurately fit the geometry of the individual's scalp.

[0066] In some specific embodiments, determining the target path vertex among the path vertices based on the target cumulative distance includes: for any adjacent first path vertex and second path vertex among the path vertices, if the first cumulative distance of the first path vertex is less than the target cumulative distance, and the second cumulative distance of the second path vertex is greater than or equal to the target cumulative distance, then the first path vertex is determined as the first target path vertex, and the second path vertex is determined as the second target path vertex.

[0067] By comparing the cumulative distance between adjacent path vertices on the electrode geodesic line with the target cumulative distance, the adjacent vertex pairs into which the target cumulative distance falls are identified as the first target path vertex and the second target path vertex, respectively. This determination, based on the strict monotonicity of the cumulative distance, can accurately and uniquely pinpoint the target path segment containing the target cumulative distance on the electrode geodesic line, ensuring that the theoretical position always lies within the anatomical orientation of the electrode geodesic line. This provides a definite geometric boundary for subsequent high-precision electrode positioning.

[0068] In some specific embodiments, for any path vertex other than the starting path vertex, the cumulative distance of the path vertex is determined as follows: for any pair of adjacent path vertices from the starting path vertex to the path vertex, the path segment length between the adjacent path vertices is calculated; the path segment lengths between each pair of adjacent path vertices are summed to obtain the cumulative distance of the path vertex.

[0069] The path segment length between adjacent path vertices refers to the Euclidean distance between two adjacent path vertices. Taking the starting path vertex of the electrode geodesic as a reference, the cumulative distance of the corresponding path vertices is obtained by calculating and accumulating the path segment lengths of each adjacent path vertex from the starting path vertex to the target path vertex. This ensures that the calculated cumulative distance closely matches the actual shape of the scalp surface of the individual head model, providing a reliable and accurate length basis for subsequent precise positioning of the electrode based on the target cumulative distance.

[0070] In some specific embodiments, the method further includes: determining the cumulative distance of the terminating path vertex as the total length of the electrode geodesic.

[0071] Since the terminal path vertex is the end vertex of the electrode geodesic, its cumulative distance has completely accumulated the lengths of all path segments from the starting path vertex to the terminal path vertex, which can fully reflect the actual total length of the electrode geodesic on the scalp surface of the individual head model, providing a length basis for subsequent calculation of electrode positions according to standard proportions.

[0072] In some specific embodiments, determining the theoretical position of the electrode based on the target path vertex includes: calculating the difference between the target cumulative distance and the cumulative distance of the first target path vertex; calculating the target path segment length between the second target path vertex and the first target path vertex; calculating the ratio between the difference and the target path segment length, and determining the ratio as an interpolation ratio; and performing linear interpolation between the first target path vertex and the second target path vertex based on the interpolation ratio to obtain the theoretical position of the electrode.

[0073] By calculating the difference between the cumulative distance to the target and the cumulative distance to the first target path vertex, the specific distance that needs to be traveled along the geodesic from the first target path vertex can be determined. The precise length of the target path segment is obtained by calculating the Euclidean distance between the first and second target path vertices. Dividing the difference by the target path segment length yields the interpolation ratio, which accurately reflects the relative position of the theoretical location on that path segment. Linear interpolation is performed between the two target vertices based on this interpolation ratio, ensuring that the calculated theoretical position is strictly located on the electrode geodesic path segment without deviating from the anatomical orientation. This also overcomes the resolution limitations of the head model mesh vertex density, achieving higher positioning accuracy than directly selecting vertices. This lays a solid foundation for the final, globally geometrically accurate electrode array deployment on individual head models.

[0074] In some specific embodiments, determining the actual position of the electrode based on the theoretical position of the electrode and the set of scalp vertices includes: for any scalp vertex in the set of scalp vertices, calculating the spatial distance between the theoretical position and the scalp vertex; identifying the minimum value among the various spatial distances, and determining the scalp vertex corresponding to the minimum value as the actual position of the electrode.

[0075] By calculating the spatial distance between the theoretical electrode position and each vertex in the scalp vertex set, and selecting the scalp vertex with the smallest distance as the actual electrode position, the theoretical position obtained by high-precision interpolation can be mapped to the actual scalp vertices in the individual head model. This process ensures that the electrode position is as close as possible to the theoretical calculation target and does not deviate from the anatomical constraint path, while making the final electrode landing point a discrete mesh vertex on the individual scalp surface, thus ensuring that the electrode position perfectly fits the scalp surface of the individual head model.

[0076] In some specific embodiments, electrode positioning specifically includes the following steps: 1) Calculate the total length of the electrode geodesic line.

[0077] In obtaining the discrete path vertex sequence constituting the electrode geodesic line Then, the total length of the geodesic line needs to be calculated.

[0078] For any two adjacent path vertices and ( The Euclidean distance between any two adjacent path vertices is the length of that path segment. The total length of the electrode geodesic is obtained by summing the lengths of the path segments between all adjacent path vertices. : Total length It reflects the actual geometric length of the reference curve defined by IEPS on the individual head model.

[0079] 2) Determine the theoretical location of the electrodes. For each electrode, calculate its precise theoretical location on the electrode geodesic line according to the relative distance ratio specified by IEPS. This theoretical location may lie between two path vertices, thus achieving sub-vertex level positioning accuracy. For the first... Each electrode, and the target cumulative distance along the electrode geodesic line. By using the relative distance ratio specified by IEPS Total length of geodesic lines Multiplying them together gives: in, This is the percentage of the electrode relative to the total length of the electrode geodesic, as specified in IEPS. This product translates the abstract, standardized proportion into a concrete, measurable distance on the individual head mold.

[0080] Along the electrode geodesic line, from the starting path vertex to the ending path vertex, the path segment lengths are accumulated sequentially to obtain the length of each path vertex. The cumulative distance. Find the minimum. Make: That is, find the first cumulative distance that is greater than or equal to the target cumulative distance. vertex .at this time, The first target path vertex, The second target path vertex. The path segments between them [ , [This refers to the target path segment where the theoretical location of the electrode is located.]

[0081] First, calculate the path from the starting vertex to... cumulative distance : The calculation also needs to be performed along the path segment [ , How far do I need to travel to reach the target (cumulative distance)? That is, the target offset distance : Length of the target path segment This is the Euclidean distance between the two target path vertices: The interpolation ratio t is and The ratio, that is: The ratio t represents the proportion of the target path segment length that needs to be traversed from the vertex of the first target path, along the direction pointing to the vertex of the second target path, in order to reach the point where the cumulative distance equals... The location.

[0082] The theoretical position of the electrode is calculated using linear interpolation. : Since interpolation can generate points at any position between two vertices, its accuracy is not limited by the vertex density of the head model mesh, thus achieving precise positioning at the sub-vertex level.

[0083] 3) Determine the actual position of the electrodes.

[0084] While the theoretical location is precisely pinpointed, it may lie on the line connecting two path vertices, rather than necessarily on discrete vertices of the head model mesh. To meet the requirement of subsequent brain energy imaging modeling that electrode points be located on mesh vertices, the theoretical locations need to be mapped to actual vertices in the scalp vertex set. For the scalp vertex set... For each vertex in the array, calculate its relationship to the theoretical position. The spatial distance between them. The vertex with the smallest distance is selected as the actual position of the electrode. : Process all electrodes defined by IEPS sequentially. For each electrode i, according to its corresponding relative distance ratio... Calculate the cumulative distance to the target The theoretical position is obtained through subvertice interpolation. The actual location is obtained through nearest neighbor mapping. Finally, the actual location set of all G electrodes is obtained: The actual set of electrode positions is a global geometrically accurate electrode array that simultaneously satisfies shape adaptation (i.e., electrode points fall on the scalp apex) and topological fidelity (i.e., electrodes are distributed along geodesics according to the IEPS ratio) on an individual head model.

[0085] In some specific implementations, the complete workflow from individual head model construction to determination of all electrode positions includes the following steps: 1) Individual Head Model Construction. The individual head model is the geometric basis for electrode positioning, and its construction accuracy directly affects the effectiveness of all subsequent steps. This step utilizes the individual's medical imaging data (preferably MRI) to construct a personalized three-dimensional geometric model of the head, specifically including the following sub-steps: 1.1) Coordinate System Alignment. After acquiring the individual's T1-weighted MRI scan data, the first step is to perform coordinate system alignment. The coordinate system of the MRI image is registered with the unified spatial coordinate system required for subsequent electrode positioning and source imaging to ensure that the individual head model, electrode positions, and source model are always represented in the same coordinate system, avoiding systematic errors introduced by coordinate system inconsistencies.

[0086] 1.2) Tissue Segmentation. Image segmentation algorithms were used to segment the aligned MRI data, extracting regions of different brain tissues such as the scalp, skull, cerebrospinal fluid, gray matter, and white matter. The scalp segmentation results formed the basis for subsequent extraction of the scalp vertex set.

[0087] 1.3) Geometric Modeling and Vertex Extraction. Based on the segmented scalp region, a 3D surface reconstruction is performed to generate a high-resolution scalp mesh model composed of triangular patches and vertices. The 3D coordinates of all vertices are extracted from this mesh model to form a scalp vertex set. .

[0088] Through the above steps, a personalized head model that is completely consistent with the individual's head geometry was obtained, providing a precise spatial structural basis for subsequent electrode positioning.

[0089] 2) Locating the reference electrode on the geodesic line. After obtaining the individual head model, the first step is to locate the reference electrode, which will serve as the reference for subsequent full-electrode localization. This step, according to IEPS specifications, involves constructing the sagittal and coronal geodesics sequentially and locating the reference electrode on them.

[0090] 2.1) Sagittal reference electrode localization. The sagittal reference electrode geodesic is a reference curve extending from the root of the nose to the external occipital protuberance along the midsagittal plane of the head. Sagittal electrodes such as Fpz (Frontopolar midline), Fz (Frontal midline), FCz (Frontocentral midline), Cz (Central midline), Pz (Parietal midline), and Oz (Occipital midline) are distributed on it.

[0091] First, the nasal root point (Nz) and the external occipital protuberance point (Iz) are identified from the scalp vertex set and determined as the starting path vertices of the sagittal electrode geodesics. and terminating path vertex Simultaneously, the left preauricular point (Lpa) and the right preauricular point (Rpa) were identified, and the coordinates of the midpoint between these two points were calculated. This midpoint was then used as the anatomical constraint point for the sagittal electrode geodesic. The midpoint naturally lies on the midsagittal plane of the head, and using it as a constraint point ensures that the constructed geodesic strictly extends along the midsagittal plane. Using the aforementioned method for constructing electrode geodesics, sagittal electrode geodesics are constructed based on the starting path vertex, the ending path vertex, and the anatomical constraint point. Specifically, candidate vertices are screened using a reference plane, and a greedy search strategy with the aforementioned double threshold constraint is used to construct a discrete path vertex sequence that both conforms to the individual head model surface and strictly extends along the midsagittal plane. After obtaining the sagittal electrode geodesics, the aforementioned geodesic electrode positioning method is used, based on the relative distance ratio of each sagittal electrode along the path specified by IEPS, to sequentially determine the actual position of each sagittal reference electrode through sub-vertex interpolation and scalp surface mapping.

[0092] 2.2) Coronal reference electrode localization. The coronal electrode geodesic is a reference curve extending along the coronal plane of the head from the left preauricular point to the right preauricular point and passing through the Cz point at the midpoint of the top of the head. Coronal electrodes such as T3 / T7 (Temporal 3 / Temporal 7, left temporal electrode), C3 (Central 3, left central electrode), Cz (Central midline, central midline electrode), C4 (Central 4, right central electrode), and T4 / T8 (Temporal 4 / Temporal 8, right temporal electrode) are distributed on it.

[0093] First, according to the IEPS definition, the Cz electrode is located at the intersection of the sagittal line (the line connecting the nasal root and the external occipital protuberance) and the coronal line (the line connecting the left and right anterior auricular points). The actual position of Cz has been precisely located in step 2.1, and this position can be directly used as the anatomical constraint point for the coronal electrode geodesy. The left anterior auricular point (Lpa) and the right anterior auricular point (Rpa) are respectively determined as the starting path vertices of the coronal electrode geodesy. and terminating path vertex The actual position of Cz located in step 2.1 is used as the anatomical constraint point. Similarly, using the aforementioned method for constructing electrode geodesics and the electrode positioning method, the actual positions of the reference electrodes on each coronal plane can be determined sequentially.

[0094] 3) Full Electrode Positioning. After completing the positioning of the reference electrode, this step further positions all remaining electrodes, including the lower peripheral electrode and the peripheral electrode.

[0095] 3.1) Location of Lower Peripheral Electrodes. Lower peripheral electrodes refer to electrodes located below the head and close to the neck, such as those below T3 / T7 and T4 / T8. The location of these electrodes requires consideration of the already located coronal reference electrodes. Using Fpz and Oz as the starting and ending points, and T3 and T4 of the coronal reference electrodes as anatomical constraint points, geodesic paths for the lower peripheral electrodes are constructed. Specifically, for the left lower peripheral electrode, Fpz is used as the starting point, Oz as the ending point, and T3 as the anatomical constraint point, and the aforementioned method is used to construct the geodesic path. For the right lower peripheral electrode, T4 is used as the anatomical constraint point to construct the geodesic path. Subsequently, based on the relative distance ratio of the lower peripheral electrodes along this path specified by IEPS, the aforementioned geodesic electrode location method is used to determine the actual position of each lower peripheral electrode.

[0096] 3.2) Peripheral electrode positioning. Peripheral electrodes refer to electrodes distributed along the midline on both sides of the scalp, such as F7 / F8, T3 / T4, T5 / T6, etc. The positioning of these electrodes needs to be combined with the already positioned lower peripheral electrodes and sagittal reference electrodes.

[0097] According to the definition of IEPS, the distribution of peripheral electrodes along the midline of the left and right scalp follows a specific proportional relationship. Taking the left side as an example, starting from Fpz and ending at Oz (or the lower peripheral electrode), and using the already located F7, T3, and other electrodes as anatomical constraint points, the geodesic path of the left peripheral electrodes is constructed. The same method is used to construct the electrode geodesic path, and the actual position of each peripheral electrode is determined according to the IEPS proportions. The positioning method for the right peripheral electrodes is similar and will not be elaborated further here.

[0098] 3.3) Electrode Integration. All electrode positions obtained in the above steps are integrated, including the sagittal reference electrode, coronal reference electrode, lower peripheral electrode, and peripheral electrode. Through integration, the actual positions of all G electrodes defined by IEPS on the individual head model are obtained. Each electrode position in this actual position set is a true vertex on the scalp surface of the individual head model and strictly conforms to the topological layout specified by IEPS.

[0099] In some specific implementations, in order to quantify the preset threshold t and the minimum connection threshold (First threshold), maximum jump threshold The impact of the second threshold on electrode positioning accuracy is evaluated by introducing the average electrode location error (AELE) as an evaluation metric. In the above formula, N represents the total number of electrodes in the standard lead system. E is the Euclidean norm. si E represents the gold standard position of the i-th electrode. ci The value represents the position of the i-th electrode calculated using the GPEL method provided in this application. A smaller AELE value indicates higher electrode positioning accuracy.

[0100] To fully explore the relationship between the three parameters and the electrode positioning accuracy, the following parameter scanning experiment was designed: 1) The range of parameter t is from 0 mm to 1 mm, and scanning is performed with an appropriate step size.

[0101] 2) Minimum connection threshold Maximum jump threshold The value range is from 18mm to 28mm, with scanning performed at appropriate step sizes.

[0102] Using the standard electrode positions provided by the FieldTrip toolkit as the gold standard E si This standard position is based on the International Standard Leading System (IEPS) definition and has recognized accuracy. For each set of parameter combinations (t, , ), execute the complete electrode positioning process, and calculate the electrode position E. ci Then, the AELE value corresponding to this set of parameters is calculated.

[0103] The above parameter scanning experiment revealed the variation of AELE with parameters. Please refer to [the relevant documentation]. Figure 3 and Figure 4 , Figure 3 and Figure 4 This is a schematic diagram illustrating how AELE varies with parameters, as provided in an embodiment of this application. Based on these experimental results, the preferred values ​​for each parameter are determined as follows: 1) Optimal value of the preset threshold t. Experimental results show that when t is 0.5 mm, the electrode positioning achieves the highest accuracy (i.e., minimum AELE). When t is too small (e.g., less than 0.3 mm), the number of candidate vertices obtained through the reference plane screening is significantly reduced. Due to the sparse candidate vertices, there are insufficient vertices to choose from when constructing the geodesic, resulting in the path not extending smoothly in a local area, exhibiting jagged or discontinuous patterns, and the path curvature deviating from the true surface of the head model, thus increasing the positioning error. When t is too large (e.g., greater than 0.7 mm), a large number of vertices far from the reference plane are included in the candidate vertex set. Although these vertices are still located on the scalp surface, they have deviated from the anatomical plane that the electrode geodesic should follow. In the subsequent greedy search process, the path may drift to these vertices that deviate from the anatomical constraints, causing the constructed geodesic to no longer strictly extend along the anatomical path specified by international standards, also resulting in significant positioning errors. When t is 0.5mm, the candidate vertex set maintains sufficient vertex density to ensure path continuity, while being strictly limited to a narrow band region near the reference plane, ensuring the effectiveness of the anatomical constraints and thus achieving optimal positioning accuracy.

[0104] 2) Minimum connection threshold and maximum jump threshold The preferred value. When and At that time, electrode positioning achieves the highest accuracy. When the value is set too small (e.g., less than 18mm), in areas with high vertex density in the scalp mesh, the algorithm may select vertices too close to the current vertex as the next path vertex. This overly dense selection leads to redundant connections in the path within a local range, causing the path to oscillate back and forth within a small area, resulting in local oscillations and significantly increasing the overall path length, deviating from the ideal shortest path. When the setting is too large (e.g., greater than 24mm), it may be impossible to find any grids with a distance greater than or equal to the specified distance in areas with low scalp grid vertex density. The limited number of candidate vertices forces the algorithm to skip the normal step size and directly select vertices that are farther away. This long-distance jump causes the path to traverse large curved surface areas, making it unable to accurately follow the local curvature changes of the head model and resulting in geometric distortion. A threshold of 21mm strikes an optimal balance between avoiding localized oscillations and preventing long-distance jumps. This threshold matches the vertex distribution density and scalp curvature radius of a typical head model, ensuring that the selected vertex at each step is neither too close (avoiding oscillations) nor too far (ensuring candidate points are found). Experimental results show that once It is correctly calibrated; just make sure That's all.

[0105] In some specific embodiments, the technical effectiveness of the GPEL method provided in the embodiments of this application is verified through systematic experimental design, and compared with representative registration methods in related technologies, including FR (FieldTrip Registration) and PR (Philips Registration). The experiments comprehensively evaluate the technical advantages of the GPEL method provided in the embodiments of this application from multiple dimensions, such as electrode positioning accuracy, global geometric fidelity, EEG positive problem accuracy, simulated brain power imaging performance, and application to real EEG data.

[0106] This embodiment uses the following three MRI datasets to cover different application scenarios, from standard templates to individualized models and complex EEG signal scenarios: 1) Standard Head Model (SHM) data: Standard MRI data publicly available in the FieldTrip tutorial sample set. This data was acquired using a 1.5T scanner with T1-weighted MR scans, parameters: matrix size 256×256, slice thickness 1.0mm, and a total of 256 slices. This data is used to establish an unbiased unified reference benchmark for quantitative comparison of absolute accuracy. 2) Healthy Subject Head Model (HSHM) data: MRI data acquired from a healthy female volunteer. T1-weighted MR scans were acquired using a 3T scanner, parameters: FOV=250mm (FOV, Field of View), TR / TE=7.9 / 3.5ms (TR / TE, Repetition Time / Echo Time), matrix size 256×256, slice thickness 1.0mm, and a total of 160 slices. This data was used to evaluate the applicability of each method in real-world, individualized scenarios. 3) Clinical data: EEG data with epileptiform discharge annotations and corresponding individual head model MRI data were used from a publicly available dataset. This data was acquired using a 3T scanner with the following parameters: matrix size 256×256, slice thickness 1.0 mm, and a total of 256 slices. The dataset provides epileptogenic zone annotations determined based on pre- and post-operative images as a reference. This data was used to evaluate the source localization accuracy of each method in complex EEG signal scenarios.

[0107] Experiment 1: Electrode positioning accuracy comparison, used to evaluate the difference in electrode positioning accuracy between the GPEL method provided in the embodiments of this application and related registration methods (FR, PR).

[0108] For the standard head model (SHM), the standard electrode positions provided by FieldTrip are used as the gold standard. For the healthy individual head model (HSHM), since the actual electrode positions are unknown, the reliability of manually positioned electrodes is first verified on the SHM: experienced operators manually mark the electrode positions according to anatomical landmarks and compare them with the gold standard.

[0109] After verifying the reliability of the manual positioning method, the manually positioned electrode positions were used as the gold standard for HSHM. Electrode positioning was performed using the GPEL, FR, and PR methods provided in the embodiments of this application under two head molds and two electrode configurations (21-channel and 64-channel arrangement), and the AELE values ​​of each method were calculated.

[0110] The experimental results are shown in Table 1 below. On the SHM, the mean AELE of the manually positioned electrodes is less than 1 mm (21 channels: 0.75±0.39 mm; 64 channels: 0.83±0.72 mm), verifying the reliability of the manual positioning method as the gold standard. The mean AELE of the GPEL method provided in this application embodiment is 5.16 mm (21 channels) and 5.29 mm (64 channels), significantly better than the FR method's 9.41 mm (21 channels) and 10.38 mm (64 channels). On the HSHM, the GPEL method provided in this application embodiment also shows the best positioning accuracy, with mean AELE of 3.35 mm (21 channels) and 4.38 mm (64 channels), significantly lower than the FR method's 11.10 mm (21 channels) and 11.31 mm (64 channels). The PR method performs well in the 64-channel configuration (7.84 mm), but the error increases significantly in the 21-channel configuration (15.39 mm), indicating that it is more sensitive to electrode density.

[0111] Table 1 The above test results show that the GPEL method provided in this application can achieve better electrode positioning accuracy than related registration methods under different head molds and different electrode configurations, and can maintain stable positioning performance even under sparse electrode configuration.

[0112] Experiment 2: Global geometric fidelity comparison, used to evaluate the performance of each method in maintaining electrode array shape consistency and topological fidelity.

[0113] Using the gold standard electrode array as a reference, the AGD (Absolute Geometric Deviation) and AMD (Adjacency Matrix Difference) indices of the electrode arrays obtained by each method were calculated. Statistical significance analysis was performed using the two-tailed Wilcoxon signed-rank test, and the Bonferroni correction method was used to eliminate statistical bias caused by multiple comparisons.

[0114] Experimental results show that, on standard head molds, the manually positioned electrodes achieved the lowest AGD (0.09-0.88 mm) and AMD (2.67-9.70 mm). The GPEL method, in a 21-channel configuration, achieved an AGD of 5.87 mm and an AMD of 36.53 mm, representing reductions of 65.1% and 44.8% respectively compared to the FR method's 16.80 mm and 66.16 mm. In a 64-channel configuration, the GPEL method achieved an AGD of 6.94 mm and an AMD of 50.90 mm, representing reductions of 62.6% and 13.2% respectively compared to the FR method's 18.51 mm and 58.66 mm. On healthy individual head molds, the GPEL method achieved AGDs of 3.80 mm (21 channels) and 4.90 mm (64 channels), and AMDs of 24.65 mm (21 channels) and 38.62 mm (64 channels), all significantly outperforming the FR and PR methods. The PR method performs relatively well in a 64-channel configuration, but the geometric error increases significantly in a 21-channel configuration.

[0115] The above experimental results show that the GPEL method provided in this application embodiment is significantly better than related registration methods in maintaining electrode array shape adaptability and topological fidelity, and achieves global geometric fidelity of the electrode array.

[0116] Experiment 3: Accuracy Comparison of EEG Positive Problems. This experiment was used to evaluate the impact of different electrode localization methods on the accuracy of EEG positive problem models, and indirectly reflect the geometric fidelity of the electrode array.

[0117] Forward models were constructed based on electrode arrays obtained using the gold standard and the GPEL, FR, and PR methods, respectively. All simulated source dipoles were activated to generate scalp potential distributions. The RDM (Relative Difference Measure) and lnMAG (logarithm of Magnitude ratio) indices were used to compare the differences between the potential distributions generated by each method and the gold standard.

[0118] Experimental results show that, on standard head models, the potential distribution generated by the GPEL method is more than 50% lower than the gold standard RDM value of the FR method, and has a smaller interquartile range, indicating that the reconstructed potential distribution has higher accuracy and stability. Regarding the lnMAG index, although the means of all methods are close to zero, the interquartile range of GPEL is significantly smaller than that of FR. On healthy individual head models, the average RDM and interquartile range of GPEL are significantly lower than those of FR and PR. The PR method exhibits a large upper quartile and many outliers in the 21-channel configuration, indicating its poor stability.

[0119] The above experimental results show that the GPEL method provided in this application embodiment can construct a more accurate EEG positive problem model, providing a more reliable physical basis for subsequent brain power imaging.

[0120] Experiment 4: Comparison of simulated brain power source imaging, used to quantitatively evaluate the impact of different electrode localization methods on the accuracy of brain power source imaging under the condition that the actual source location is known.

[0121] Synthetic EEG signals were generated from randomly selected source locations throughout the brain to simulate the frequency components of typical EEG rhythms. A total of 2000 simulations were conducted, each generating a 2-second source activity at a sampling rate of 250Hz, with measurement noise of varying signal-to-noise ratios (5-25dB) added. Dipole fitting was used for source localization, with position error (PE) as the evaluation metric.

[0122] The experimental results are shown in Table 2 below. The GPEL method provided in this application embodiment exhibits the lowest source localization error under all experimental conditions: except for the 21-channel SHM where the average PE is approximately 4mm, the average PE under other conditions is around 3mm, and the standard deviation is within 5mm, indicating robust and reliable localization accuracy. The FR method produces an average PE of approximately 12-13mm under all conditions, with a standard deviation exceeding 6.1mm. The PR method achieves a lower average PE (approximately 5.87-6.67mm) with a 64-channel configuration, but under a sparse 21-channel configuration, the average PE increases sharply to 15.89-16.37mm due to topological distortion introduced by nonlinear deformation, even exceeding that of the FR method.

[0123] Table 2 The above experimental results show that the GPEL method provided in this application embodiment can significantly improve the source localization accuracy of brain power imaging, and maintain stable performance under different signal-to-noise ratios and electrode configurations.

[0124] Experiment 5: Comparison of real motor imagery EEG data, used to verify the accuracy of the GPEL method in locating task-related brain regions in real motor imagery EEG data.

[0125] We used a publicly available motor imagery EEG dataset, which contains 64-channel EEG recordings of healthy subjects performing left-hand and right-hand motor imagery tasks at a sampling rate of 512 Hz. Mean scalp potentials within a time window of 1.5–2.5 seconds after the start of motor imagery were selected for source localization. Using the contralateral motor cortex region as a reference, dipole fitting was employed to compare the localization results of the GPEL, FR, and PR methods.

[0126] The experimental results showed that the GPEL method accurately localized the activity to the contralateral motor cortex in both the left-hand and right-hand motor imagery tasks. The FR method successfully identified the contralateral motor cortex only in the left-hand task, and its localization was off in the right-hand task. The PR method failed to localize the activity to the contralateral motor cortex in both tasks.

[0127] The above experimental results show that the GPEL method provided in this application embodiment can accurately locate task-related brain regions in real motor imagery EEG data, demonstrating superior source localization ability compared to related registration methods.

[0128] Experiment 6: Comparison of real EEG data containing epileptiform discharges, used to verify the accuracy of the GPEL method in locating source activity regions in real EEG data containing epileptiform discharges.

[0129] We used publicly available datasets containing EEG data annotated with epileptiform discharges and their corresponding individual head models. For each annotated epileptiform discharge event, EEG segments before and after the discharge peak were extracted and preprocessed. Average potentials at multiple time points near the discharge peak were selected for dipole fitting to obtain source localization results. Using the epileptogenic zone annotations based on clinical imaging provided in the dataset as a reference, the localization error (PE) of each method was calculated.

[0130] Experimental results show that the average distance between the source location and the center of the labeled area located by the GPEL method is 5.63 mm, which is significantly better than the PR method's distance of over 16 mm and the FR method's distance of over 8 mm. The GPEL positioning results have the highest spatial consistency with the labeled area.

[0131] The above experimental results show that the GPEL method provided in this application demonstrates superior source localization accuracy in real EEG data containing epileptiform discharges, and can provide a more accurate localization basis for subsequent clinical interpretation.

[0132] Experiment 7: Ablation experiment, used to evaluate the contribution of sub-vertex interpolation and scalp vertex mapping modules in the GEL algorithm to electrode positioning accuracy.

[0133] Using the full GPEL method, which includes sub-vertex interpolation and scalp vertex mapping, as a baseline, and comparing it with an ablation method that directly selects geodesic path vertices as electrode locations based on distance ratios, the AELE index was used to evaluate the positioning accuracy.

[0134] Experimental results show that the electrode placement of the baseline method is closer to the gold standard than that of the ablation method. In terms of the AELE index, the baseline method outperforms all ablation methods in both the median and upper quartile, with an average localization accuracy improvement of at least 68.92%.

[0135] The above experimental results show that sub-vertex interpolation and scalp vertex mapping are the core components of the GPEL method provided in this application embodiment, and play a key role in achieving high-precision electrode positioning.

[0136] Accordingly, please refer to Figure 5 , Figure 5 A schematic diagram of a brain power imaging electrode localization system provided in this application embodiment is shown below. Figure 5 As shown, the system includes: an acquisition module for acquiring a set of scalp vertices in an individual head model, the set of scalp vertices including multiple anatomical landmarks; a generation module for determining, based on the multiple anatomical landmarks, the anatomical constraint points and two endpoints of the electrode geodesic to be constructed; generating a reference plane based on the anatomical constraint points and the two endpoints, the reference plane providing anatomical constraints for the orientation of the electrode geodesic; a construction module for extracting candidate vertices adapted to the reference plane from the set of scalp vertices, and constructing the electrode geodesic matching the individual head model based on the candidate vertices; and a determination module for determining the actual position of each electrode based on the total length of the electrode geodesic, a preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices.

[0137] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0138] In this embodiment, the brain power imaging electrode localization system is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0139] Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of a brain-computer interface device provided in an embodiment of this application, such as... Figure 6As shown, the brain-computer interface device includes one or more processors 10, a memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the brain-computer interface device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple brain-computer interface devices can be connected, each providing some of the necessary operations (e.g., as a server array, a set of blade servers, or a multiprocessor system). Figure 6 Take a processor 10 as an example.

[0140] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0141] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0142] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the brain-computer interface device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transient memory, such as at least one disk storage device, flash memory device, or other non-transient solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, which can be connected to the brain-computer interface device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0143] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0144] The brain-computer interface device also includes a communication interface 30 for communicating with other devices or communication networks.

[0145] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the methods shown in the above embodiments are implemented.

[0146] This application provides a computer program product including computer instructions stored in a computer-readable storage medium. A processor of a brain-computer interface device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the brain-computer interface device to perform the method of any embodiment of this application.

[0147] The systems and modules described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0148] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0149] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0150] This application is described with reference to flowchart illustrations and / or block diagrams of methods, systems, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0151] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0152] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0153] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0154] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0155] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

[0156] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for locating electrodes in brain power imaging, characterized in that, The method includes: Obtain the set of scalp vertices in the individual head model, the set of scalp vertices including multiple anatomical landmarks; For the electrode geodesic to be constructed, based on the multiple anatomical landmarks, the anatomical constraint points and two endpoints of the electrode geodesic are determined; based on the anatomical constraint points and the two endpoints, a reference plane is generated, which is used to provide anatomical constraints for the orientation of the electrode geodesic. Candidate vertices that fit the reference plane are extracted from the set of scalp vertices, and the electrode geodesics that match the individual head model are constructed based on the candidate vertices. The actual position of each electrode is determined based on the total length of the electrode geodesic, the preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices.

2. The method according to claim 1, characterized in that, The plurality of anatomical landmarks include the nasal root point, the external occipital protuberance point, the left preauricular point, and the right preauricular point. Based on the plurality of anatomical landmarks, the anatomical constraint points and two endpoints of the electrode geodesic are determined, including: For the sagittal electrode geodesic to be constructed, the nasal root point and the external occipital protuberance point are determined as the two endpoints of the sagittal electrode geodesic. Calculate the midpoint between the left and right preauricular points, and determine the midpoint as the anatomical constraint point of the sagittal electrode geodesic.

3. The method according to claim 1 or 2, characterized in that, The candidate vertices include the two endpoints. Based on the candidate vertices, the electrode geodesics matching the individual head model are constructed, including: The two endpoints are respectively used as the starting path vertex and the ending path vertex of the electrode geodesic line. The path vertices that constitute the electrode geodesic line are determined sequentially from the candidate vertices according to the direction from the starting path vertex to the ending path vertex. Based on the vertices of each path, the electrode geodesic is constructed.

4. The method according to claim 3, characterized in that, The path vertices constituting the electrode geodesic are determined sequentially from the candidate vertices, including: For any vertex on the current path, determine the candidate vertices whose Euclidean distance to the current path vertex is greater than or equal to a first threshold and less than or equal to a second threshold. Extract the candidate vertex with the smallest Euclidean distance to the current path vertex from the candidate vertices, use the extracted candidate vertex as the next path vertex, and update the next path vertex to the current path vertex.

5. The method according to claim 4, characterized in that, Based on the total length of the electrode geodesic, the preset relative distance ratio of each electrode along the electrode geodesic, and the set of scalp vertices, the actual position of each electrode is determined, including: For any electrode, the product of the preset relative distance ratio and the total length of the electrode geodesic is calculated to obtain the target cumulative distance of the electrode along the electrode geodesic. Based on the target cumulative distance, the target path vertex among the various path vertices is determined; Based on the target path vertices, determine the theoretical location of the electrode; The actual position of the electrode is determined based on the theoretical position of the electrode and the set of scalp apex points.

6. The method according to claim 5, characterized in that, Based on the target cumulative distance, the target path vertex among the various path vertices is determined, including: For any adjacent first path vertex and second path vertex among the path vertices, if the first cumulative distance of the first path vertex is less than the target cumulative distance, and the second cumulative distance of the second path vertex is greater than or equal to the target cumulative distance, then the first path vertex is determined as the first target path vertex, and the second path vertex is determined as the second target path vertex.

7. The method according to claim 6, characterized in that, For any path vertex other than the starting path vertex, the cumulative distance of the path vertex is determined as follows: For any pair of adjacent path vertices from the starting path vertex to the path vertex, calculate the path segment length between the adjacent path vertices; The cumulative distance between the path vertices is obtained by summing the path segment lengths between each pair of adjacent path vertices.

8. The method according to claim 7, characterized in that, The method further includes: The cumulative distance of the terminating path vertex is determined as the total length of the electrode geodesic.

9. The method according to claim 6, characterized in that, Based on the target path vertices, the theoretical location of the electrode is determined, including: Calculate the difference between the cumulative distance to the target and the cumulative distance to the first target path vertex; Calculate the length of the target path segment between the second target path vertex and the first target path vertex; Calculate the ratio between the difference and the length of the target path segment, and determine the ratio as the interpolation ratio; Based on the interpolation ratio, linear interpolation is performed between the first target path vertex and the second target path vertex to obtain the theoretical position of the electrode.

10. The method according to claim 5, characterized in that, Based on the theoretical location of the electrodes and the set of scalp apex points, the actual location of the electrodes is determined, including: For any scalp vertex in the set of scalp vertices, calculate the spatial distance between the theoretical position and the scalp vertex; Identify the minimum value among various spatial distances, and determine the scalp vertex corresponding to the minimum value as the actual position of the electrode.

Images