Dynamic interpolation brain topographic map drawing system and method

Abstract

The invention discloses a dynamic interpolation brain topographic map drawing system and method. An electroencephalogram signal collecting and preprocessing module obtains window data based on collected electroencephalogram signals; the feature extraction module is used for converting the electroencephalogram signals in a time domain to a frequency domain by utilizing fast Fourier transform based on the window data, respectively calculating a frequency band power spectrum and a total power spectrum of a total band-pass range for each target band on the frequency domain, and calculating an energy ratio of each target band according to each frequency band power spectrum and the total power spectrum; determining a standard deviation of the window data based on each piece of window data; the interpolation calculation module dynamically adjusts the smoothing coefficient of the interpolation method according to the energy ratio and / or the standard deviation of each target frequency band; performing spatial interpolation calculation on the normalized window data by using an interpolation function according to the smoothing coefficient to obtain an interpolation grid; the brain topographic map drawing module draws a two-dimensional brain topographic map. According to the method, the quality and practicability of the interpolation result are improved, and detail display is improved.

Description

Technical Field

[0001] The present application belongs to the interdisciplinary technical field of electroencephalographic signal processing and computer graphics, and specifically relates to a dynamic interpolation brain topography mapping system and method. Background Art

[0002] A brain-computer interface is a direct connection between the human or animal brain and an external device, enabling information exchange between the brain and the device. It directly reads neural signals from the brain and converts these signals into actionable commands, thereby controlling external devices or receiving information from external devices. Therefore, the analysis of EEG data is particularly important. Nowadays, there are many methods for visualizing EEG data, such as electroencephalogram (EEG), spectrogram, brain topography, and functional magnetic resonance imaging (fMRI), which can obtain brain activity data and perform visual analysis.

[0003] Compared to traditional EEG waveforms, brain topography maps can more intuitively show the activity of different brain regions. Brain topography maps use color coding to display the potential distribution of different electrode positions.

[0004] When mapping brain topography, interpolation is a key technology used to convert limited, discrete EEG data points into continuous EEG activity distribution images, allowing for a more intuitive and accurate presentation of the spatial characteristics of brain activity. However, the frequency components of EEG signals vary significantly across tasks and physiological states. Low-frequency energy dominates during states like relaxation and sleep, while high-frequency energy dominates during tasks like cognitive attention training. Traditional methods for mapping brain topography, which use fixed smoothing coefficients for interpolation calculations, are unable to adapt to these variations and struggle to accurately reflect brain activity. Summary of the Invention

[0005] The technical purpose of this application is to provide a dynamic interpolation brain topography mapping system and method to address the technical problem that the traditional brain topography mapping method using fixed smoothing coefficients for interpolation calculations cannot adapt to such changes and is difficult to accurately reflect brain activity, thereby reducing the impact of noise on the interpolation effect and improving the display of details.

[0006] In order to achieve the above technical objectives, this application adopts the following technical solutions.

[0007] In a first aspect, an embodiment of the present application provides a dynamic interpolation brain topography mapping system, comprising:

[0008] The EEG signal acquisition and preprocessing module is used to acquire EEG signals; perform sliding window processing based on the EEG signals to obtain sliding window data, perform channel superposition and window processing on the sliding window data of each channel, obtain each window data and transmit it to the feature extraction module;

[0009] a feature extraction module for converting the EEG signal in the time domain to the frequency domain using a fast Fourier transform based on the window data, calculating the frequency band power spectrum and the total power spectrum of the total bandpass range for each target band in the frequency domain, calculating the energy proportion of each target frequency band based on the frequency band power spectrum and the total power spectrum; and determining the standard deviation of the window data based on the window data;

[0010] an interpolation calculation module, connected to the feature extraction module, and dynamically adjusting a smoothing coefficient of the interpolation method according to the energy proportion and / or standard deviation of each target frequency band; performing spatial interpolation calculation on the normalized window data using an interpolation function according to the smoothing coefficient to obtain an interpolation grid;

[0011] The brain topography drawing module is used to connect with the interpolation calculation module and perform color value mapping according to the interpolation grid to draw a two-dimensional brain topography map.

[0012] In a second aspect, an embodiment of the present application provides a dynamic interpolation brain topography drawing method of a dynamic interpolation brain topography drawing system provided by any possible implementation method of the first aspect, including: dynamically adjusting the smoothing coefficient of the interpolation method according to the energy proportion and standard deviation of each of the target frequency bands, including: establishing a functional relationship λ=f(σ,R) between the standard deviation and the energy proportion of each of the target frequency bands, and determining the smoothing coefficient λ through the functional relationship.

[0013] Furthermore, the modified interpolation function capability equation E(f) is used. The capability equation E(f) introduces multimodal data and is expressed as follows:

[0014]

[0015] Among them, f(x i ,y i ) is the interpolation function, (x i ,y i ) is the position of the i-th control point in the plane coordinates, f(x i ,y i ) is about the coordinate (x i ,y i ) function, which is used to estimate the values of other locations based on the known control point information, where n represents the number of control points, and z i is the function value of the i-th control point, λ is the smoothing coefficient, is the energy function, β is the weight coefficient, W(x, y) is the weight matrix of the new modality, and ∫∫W(x, y)·f(x, y)dxdy is the multimodal term;

[0016] The linear equations are modified synchronously to Where Dw is the diagonal matrix of W(x,y), K is the basis function matrix, I is the identity matrix, P is the polynomial matrix, ω is the weight vector, a is the polynomial coefficient vector, and z is a vector whose elements are the function values at the control points.

[0017] Furthermore, the target frequency band includes: α wave, whose frequency range is: 8~12Hz; β wave, whose frequency range is: 12~30Hz; γ wave, whose frequency range is: >30Hz; θ wave, whose frequency range is: 4~8Hz; δ wave, whose frequency range is: 0.5~4Hz.

[0018] Furthermore, dynamically adjusting the smoothing coefficient of the interpolation method according to the energy proportion of each target frequency band includes:

[0019] If R θ +R δ >0.5, reduce the smoothing coefficient λ;

[0020] If R γ >0.5, increase the smoothing coefficient λ;

[0021] If R β +R α >0.5, the preset smoothing coefficient λ is used;

[0022] where R θ is the energy proportion of theta wave, R δ is the energy proportion of δ wave, R β is the energy proportion of β wave, R α is the energy proportion of α wave, R γ is the energy proportion of γ waves.

[0023] Furthermore, the smoothing coefficient of the interpolation method is dynamically adjusted according to the standard deviation, including:

[0024] If the standard deviation is greater than or equal to a threshold, reducing the smoothing coefficient λ;

[0025] If the standard deviation is smaller than a threshold, the smoothing coefficient λ is increased.

[0026] Furthermore, the interpolation function is used to perform spatial interpolation calculation on the normalized window data to obtain an interpolation grid, including:

[0027] Before the task begins, a time window of resting-state data is collected as a baseline. The baseline data is divided according to the sliding window size, and each divided part overlaps half of the previous window to obtain sliding window data. The sliding window data of each channel are superimposed and windowed to obtain the data of each window.

[0028] A power spectrum set is obtained for each window data, and its mean and standard deviation are calculated. Then, the real-time task sliding window power spectrum is normalized according to the z-score formula to obtain normalized window data; interpolation calculation is performed based on the normalized window data to obtain the interpolation grid.

[0029] Furthermore, the method further comprises using the brain topography mapping module to perform the following operations:

[0030] Pre-generate the electrode UV point coordinates of each electrode based on the designed texture map, and perform alignment operations according to the electrode UV point coordinates in the 10-20 EEG specification to provide a coordinate basis for mapping two-dimensional data into three-dimensional space;

[0031] Using UV point coordinates as the interpolation grid coordinate system, the interpolation grid data of the two-dimensional brain topography map is linked to the three-dimensional space to achieve the conversion from two-dimensional to three-dimensional.

[0032] Furthermore, the method further includes: when performing interpolation calculations, the interpolation calculation module maps the interpolation grid to the GPU via CUDA for calculations; during the rendering process, the interpolated data is rendered in real time using OpenGL. To avoid video memory conflicts, the CUDA and OpenGL contexts are shared, and the drawing texture buffer is registered with CUDA using a shared video memory method. The graphics memory is attached exclusively during calculations and detached to OpenGL after the calculations are completed.

[0033] Furthermore, the method further includes: if the two-dimensional brain topography map has a boundary, performing an interpolation 0 constraint on the boundary.

[0034] Compared with the prior art, the dynamic interpolation brain topography mapping system and method provided in the embodiments of the present application achieve the following beneficial technical effects: EEG signals are extremely susceptible to noise, and noise can interfere with the analysis of the brain's real activity signals. Dynamically changing the smoothing coefficient can take signal noise evaluation (such as standard deviation) into consideration, and increase the smoothing coefficient when the noise is large. During the EEG data acquisition process, if the signal standard deviation increases due to external electromagnetic interference, increasing the smoothing coefficient at this time can suppress the noise during the interpolation process, reduce the manifestation of noise in the interpolation results, and enable subsequent analysis to focus more on the effective signals generated by brain activity, thereby improving the accuracy of EEG data processing. Dynamically changing the smoothing coefficient for interpolation provides a personalized processing method for different EEG signal characteristics and task requirements, which can make the interpolation results better adapt to these differences, more accurately reflect the characteristics of brain activity in different scenarios, improve the quality and practicality of the interpolation results, and improve the display of details. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings described herein are for illustrative purposes only and are not intended to limit the scope of the present application in any way. In addition, the shapes and proportional dimensions of the components in the drawings are only schematic and are used to help understand the present application. They do not specifically limit the shapes and proportional dimensions of the components of the present application. Those skilled in the art can select various possible shapes and proportional dimensions to implement the present application according to the specific circumstances under the guidance of the present application. In the drawings:

[0036] Figure 1 A schematic diagram of the structure of a dynamic interpolation brain topography mapping system provided in an embodiment;

[0037] Figure 2 A schematic diagram of the principle of a method for drawing a brain topography map based on dynamic interpolation provided in an embodiment;

[0038] Figure 3 Schematic diagram of the three-dimensional head model and texture in the embodiment;

[0039] Figure 4 Schematic diagram of the texture information of the three-dimensional head model in the embodiment;

[0040] Figure 5 This is an example of a three-dimensional textured brain topography map in the embodiment;

[0041] Figure 6 This is the GPU parallel computing mode in the embodiment. DETAILED DESCRIPTION

[0042] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0043] See Figure 1 The embodiment of the present application provides a dynamic interpolation brain topography mapping system, including an EEG signal acquisition and preprocessing module, a feature extraction module, an interpolation calculation module and a brain topography mapping module.

[0044] Among them, the EEG signal acquisition module is used to collect EEG signals, perform sliding window processing based on the EEG signals to obtain sliding window data, superimpose the sliding window data of each channel and then perform window processing to obtain each window data and transmit it to the feature extraction module.

[0045] The feature extraction module is used to convert the EEG signal in the time domain to the frequency domain based on the data of each window using fast Fourier transform, calculate the frequency band power spectrum and the total power spectrum of the total bandpass range for each target band in the frequency domain, calculate the energy proportion of each target frequency band based on the power spectrum of each frequency band and the total power spectrum; and determine the standard deviation of the window data based on the data of each window.

[0046] The interpolation calculation module is used to connect with the feature extraction module and dynamically adjust the smoothing coefficient of the interpolation method according to the energy proportion and / or standard deviation of each target frequency band; based on the smoothing coefficient, the interpolation function is used to perform spatial interpolation calculation on the normalized window data to obtain the interpolation grid.

[0047] The brain topography drawing module is used to connect with the interpolation calculation module, and perform color value mapping according to the interpolation grid to draw a two-dimensional brain topography map.

[0048] This embodiment also provides a dynamic interpolation brain topography drawing method based on the above dynamic interpolation brain topography drawing system, such as Figure 2 As shown, the following steps are included:

[0049] Step 1: Real-time acquisition and preprocessing of EEG data, band proportion and power spectrum, and noise level assessment:

[0050] By selecting a 10-20 standard electrode arrangement EEG cap device, wearing and connecting the EEG device, and then collecting real-time data.

[0051] In the embodiment, the real-time EEG signal can be differentially filtered by using a 0.5-45 Hz Butterworth bandpass filter to ensure the real-time filtering effect of the data. The expression is as follows:

[0052]

[0053] y[n] represents the output signal value of the filter at discrete time point n, x[n] represents the input signal value of the filter at discrete time point n, b k is the input coefficient, a k is the output coefficient, M is the maximum delay number when the input signal x[n] is weighted and summed; N is the maximum delay number when the filter's past output signal y[nk] is weighted and summed.

[0054] In the embodiment, a 48-52 Hz band-stop filter is further performed to eliminate power frequency interference.

[0055] As an example, a 4-second time window is used to perform FIFO (first-in-first-out) operation on the real-time filtered signal, and each sliding window is updated for 0.5 seconds. The updated sliding window data is first channel superimposed and then a window function (such as Hanning window) is added to reduce spectrum leakage X. w[n] = X[n]·W[n], where the Hanning window function

[0056] Through Fast Fourier Transform (FFT) Get the frequency domain data of the sampling interval, where X is the converted value corresponding to the frequency domain resolution at the sampling rate, X w It is the EEG signal window data after superimposing the signals and adding windows.

[0057] Then the power spectrum of different bands and the power spectrum of the total bandpass range are calculated for subsequent energy proportion preparation.

[0058]

[0059] Where U is the window function normalization factor, and P represents the power value in each frequency domain range.

[0060] The target bands commonly used in EEG analysis are α band (8-12Hz), β band (12-30Hz), γ band (>30Hz), θ band (4-8Hz), and δ band (0.5-4Hz). The energy of the above bands is calculated and then the energy ratio is calculated with the total band energy to obtain the energy ratio R. The energy ratio R of the α band is 0. α For example, the expression is as follows:

[0061]

[0062] Among them E total Represents the energy sum of each band, and R represents the ratio of the corresponding frequency band to the total energy.

[0063] For EEG noise analysis, the standard deviation σ is calculated from the superimposed window data Xw. Where N represents the number of data points in the sliding window, Z i For each time series point, zˉ is the time-averaged mean of the superimposed sliding window, and the expression is as follows:

[0064]

[0065] Step 2: Dynamically adjust the smoothing coefficient according to the task mode:

[0066] In some embodiments, the standard deviation σ shown above is used as a single-channel signal noise evaluation. When it is large (such as greater than or equal to a threshold), the smoothing coefficient λ is increased to reduce noise interference. Conversely, if the standard deviation is less than the threshold, the smoothing coefficient λ is reduced to retain more details.

[0067] In some embodiments, the energy proportion shown above is used to dynamically adjust λ based on the actual task situation. For example, during tasks such as relaxation and sleep, the proportion of low-frequency energy is high (>0.5), and the signal is in a stable state, so λ is reduced to increase details; in tasks that require concentration, such as cognitive attention training, the proportion of high-frequency energy is high (>0.5), and there may be noise, so λ is increased to reduce the impact of noise; if the mid-frequency part is dominant (>0.5), the default λ value is used.

[0068] In some embodiments, the range of λ is dynamically adjusted based on the relationship between the signal-to-noise evaluation σ and the energy ratio R, and the relationship λ = f(σ, R) is established, where a simple linear combination can be used to implement constraints to reduce computational complexity. Specifically, the parameter settings can be dynamically adjusted based on actual test results and task requirements.

[0069] Step 3: Use the smoothing coefficient and multimodal data to perform interpolation calculations.

[0070] As an example, thin plate spline interpolation (TPS) can be used. The TPS method has unique advantages in drawing brain topography, mainly in terms of its ability to accurately fit data, adapt to complex EEG distributions, maintain smoothness, and be flexible.

[0071] First, define the grid matrix G[x,y,z] which represents the position of all pixels (point coordinates) in the drawing area, and P[x,y,z] as the control point information. Then construct the following matrix, where K is the basis function matrix Radial basis function U(r)=r 2 log(r). P is the polynomial matrix, λ is the smoothing coefficient, and I is the identity matrix. The polynomial coefficients a1, a2, a3 and weight ω are solved using the conjugate gradient method: Then generate the interpolation function f(x,y)

[0072]

[0073] Just get the energy value of each point.

[0074] Since traditional interpolation methods are only based on a single data modality and cannot fully utilize multi-faceted information, the interpolation results cannot well reflect the physiological characteristics of the brain.

[0075] If it is desired to add multimodal data to improve physiological plausibility and display effects, in some embodiments, this is achieved by modifying the interpolation function capability equation E. The modified interpolation function capability equation E(f) introduces multimodal data and is expressed as:

[0076]

[0077] The first two terms on the right side of the equation are the energy functions of the traditional TPS interpolation, and the third term is the multimodal term W(x,y) which is the weight matrix of the new modality. i ,y i ) is the interpolation function, (x i ,y i ) is the position of the i-th control point in the plane coordinates, f(x i ,y i ) is about the coordinate (x i ,y i ) function, which is used to estimate the values of other locations based on the known control point information, where n represents the number of control points, and z i is the function value of the i-th control point, λ is the smoothing coefficient, is the energy function, β is the weight coefficient, W(x, y) is the weight matrix of the new modality, and ∫∫W(x, y)·f(x, y)dxdy is the multimodal term;

[0078] At this time, the linear equations are synchronously modified to Among them D w is the diagonal matrix W(x,y).

[0079] In this embodiment, the smoothing coefficient λ still plays a role and affects the interpolation result together with the multimodal term, balancing the relationship between data fitting and function smoothness, ensuring that when fusing multimodal data, the interpolation result not only conforms to multiple data characteristics but also has appropriate smoothness.

[0080] In the embodiment, before the task starts, a time window of resting state data is collected as the baseline X base , divide the baseline data into sliding window sizes, and each divided part overlaps half of the previous window to obtain sliding window data, and perform channel superposition and windowing on the sliding window data of each channel to obtain the data of each window;

[0081] For each window data, the power spectrum set E is obtained C , calculate the power spectrum set E C mean and standard deviation Then normalize the real-time task sliding window power spectrum according to the z-score formula Normalized window data is obtained; interpolation calculation is performed based on the normalized window data to obtain an interpolation grid.

[0082] In the embodiment, the brain topography drawing module is further used to perform the following operations: pre-generate the electrode UV point coordinates of each electrode according to the designed texture map, such as Figure 3 As shown, Figure 3(a) is a schematic diagram of the 3D head model, and (b) is a schematic diagram of the texture. The alignment operation is performed according to the electrode UV point coordinates in the 10-20 EEG specification to provide a coordinate basis for mapping the 2D data into the 3D space. The UV point coordinates are used as the interpolation grid coordinate system to establish a connection between the interpolation grid data of the 2D brain topography map and the 3D space, thus realizing the conversion from 2D to 3D. Figure 5 shown.

[0083] Figure 4 shows a schematic diagram of the three-dimensional head model texture information, Figure 4 (a) is the 3D head model texture information. Figure 4 (b) is the 3D head model texture information in the grid coordinate system. Figure 4 As shown, if the texture has a boundary, the boundary can be interpolated with 0 constraints according to the above TPS interpolation, which can basically adjust it to the appropriate position. In addition, the contour coordinates can be obtained and the edge blur effect of the contour can be performed through edge detection to achieve the final terrain map.

[0084] Brain mapping is available in both two-dimensional and three-dimensional forms. Compared to two-dimensional brain mapping, three-dimensional mapping can be applied to complex brain patterns or the surface of the scalp, making it more intuitive. However, current three-dimensional brain mapping visualization solutions are primarily based on Python or MATLAB offline tools such as mne and eeglab, which lack cross-platform real-time rendering capabilities. Traditional data interpolation algorithms for brain mapping, such as thin plate spline (TPS) and kriging interpolation, require significant computing resources and cannot guarantee good real-time performance if calculated using a CPU. Furthermore, existing brain mapping solutions all use a single EEG signal as a baseline for preprocessing calculations, without multimodal fusion, resulting in blurry displays.

[0085] In some embodiments, the dynamic interpolation brain topography mapping method further includes step 5: using a GPU to perform parallel acceleration on the real-time drawing process: during the TPS interpolation calculation process, the interpolated grid can be mapped to the GPU through CUDA and then calculated using GPU resources, and during the drawing process, the interpolated data can be drawn in real time through OpenGL.

[0086] like Figure 6 As shown in the figure, in order to avoid the memory conflict between OpenGL drawing and CUDA calculation, the context of CUDA and OpenGL is shared. The drawing texture buffer is registered to CUDA using the shared memory method. By attaching it, it exclusively uses the memory when calculation is required. After the calculation is completed, it is detached and handed over to OpenGL for use. This reduces the memory transfer cost of copying from GPU to CPU and then to GPU, as well as the allocation and release of memory resources. Figure 6As shown. For example, for the calculation of the radial basis function matrix K. It is preferred to apply for a piece of video memory to store the data for TPS calculation, and divide the data block into multiple sub-blocks for parallel operation. Create a thread pool with a maximum number of 6 threads to perform parallel calculations on each data block. If it is not assigned to a thread, it enters the task queue and the flag is changed to undo. If it is assigned to a thread but has not yet ended, it is set to doing. If a task is completed, it is set to done and the thread resources are returned to the task queue for use. When all blocks are set, it means that a matrix operation is completed. At this time, the video memory resources are demapped to opengl through detach and marked with None. Other data calculation operations follow the same logic until an interpolation calculation is completely completed.

[0087] This step utilizes CUDA / OpenCL to call the GPU for parallel computing to perform thin-plate spline interpolation. Pre-generated electrode position-dependent basis function matrices limit real-time operations to matrix operations, accelerating the rendering process and improving rendering efficiency, enabling real-time rendering. GPU parallel computing reduces CPU burden and significantly increases speed by requiring only matrix operations during dynamic basis function calculations.

[0088] In some embodiments, real-time rendering capabilities across platforms and languages can be provided through dynamic link libraries or microservices. For example, a dynamic link library based on C / C++ can be used to separate the computing module from the display module to achieve cross-language calls and the efficiency of C / C++. Alternatively, the microservice model can be used to implement the above functions through process communication.

[0089] The above is a detailed introduction to the dynamic interpolation brain topography mapping system and method provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the concept of the present application and should not be understood as limiting the scope of protection of the present application.

Claims

1. A dynamic interpolation brain topography mapping system, characterized in that: include: The EEG signal acquisition and preprocessing module is used to acquire EEG signals, perform sliding window processing based on the EEG signals to obtain sliding window data, perform channel superposition and window processing on the sliding window data of each channel, obtain each window data and transmit it to the feature extraction module; a feature extraction module for converting the EEG signal in the time domain to the frequency domain using a fast Fourier transform based on the window data, calculating the frequency band power spectrum and the total power spectrum of the total bandpass range for each target band in the frequency domain, calculating the energy proportion of each target frequency band based on the frequency band power spectrum and the total power spectrum; and determining the standard deviation of the window data based on the window data; an interpolation calculation module, connected to the feature extraction module, and dynamically adjusting a smoothing coefficient of the interpolation method according to the energy proportion and / or standard deviation of each target frequency band; performing spatial interpolation calculation on the normalized window data using an interpolation function according to the smoothing coefficient to obtain an interpolation grid; The brain topography drawing module is used to connect with the interpolation calculation module and perform color value mapping according to the interpolation grid to draw a two-dimensional brain topography map.

2. A method for drawing a dynamic interpolation brain topography map according to the system of claim 1, characterized in that: Dynamically adjust the smoothing coefficient of the interpolation method according to the energy proportion and standard deviation of each target frequency band, including: establishing a functional relationship λ=f(σ,R) between the standard deviation and the energy proportion of each target frequency band, and determining the smoothing coefficient λ through the functional relationship.

3. The method for drawing a dynamic interpolation brain topography map according to claim 2, wherein: The modified interpolation function capability equation E(f) is adopted. The capability equation E(f) introduces multimodal data and is expressed as follows: Among them, f(x i ,y i ) is the interpolation function, (x i ,y i ) is the position of the i-th control point in the plane coordinates, f(x i ,y i ) is about the coordinate (x i ,y i ) function, which is used to estimate the values of other locations based on the known control point information, where n represents the number of control points, and z i is the function value of the i-th control point, λ is the smoothing coefficient, is the energy function, β is the weight coefficient, W(x, y) is the weight matrix of the new modality, and ∫∫W(x, y)·f(x, y)dxdy is the multimodal term; The linear equations are modified synchronously to Where Dw is the diagonal matrix of W(x,y), K is the basis function matrix, I is the identity matrix, P is the polynomial matrix, ω is the weight vector, a is the polynomial coefficient vector, and z is a vector whose elements are the function values at the control points.

4. The method for drawing a dynamic interpolation brain topography map according to claim 2, wherein: The target frequency bands include: α wave, whose frequency range is: 8-12 Hz; β wave, whose frequency range is: 12-30 Hz; γ wave, whose frequency range is: >30 Hz; θ wave, whose frequency range is: 4-8 Hz; δ wave, whose frequency range is: 0.5-4 Hz.

5. The method for drawing a dynamic interpolation brain topography map according to claim 2, characterized in that: Dynamically adjusting the smoothing coefficient of the interpolation method according to the energy proportion of each target frequency band includes: If R θ +R δ >0.5, then reduce the smoothing coefficient λ; If R γ >0.5, then increase the smoothing coefficient λ; If R β +R α >0.5, the preset smoothing coefficient λ is used; where R θ is the energy proportion of theta wave, R δ is the energy proportion of δ wave, R β is the energy proportion of β wave, R α is the energy proportion of α wave, R γ is the energy proportion of γ waves.

6. The method for drawing a dynamic interpolation brain topography map according to claim 2, characterized in that: Dynamically adjust the smoothing coefficient of the interpolation method based on the standard deviation, including: If the standard deviation is greater than or equal to a threshold, reducing the smoothing coefficient λ; If the standard deviation is smaller than a threshold, the smoothing coefficient λ is increased.

7. The method for drawing a dynamic interpolation brain topography map according to claim 2, characterized in that: Use the interpolation function to perform spatial interpolation calculation on the normalized window data to obtain the interpolation grid, including: Before the task begins, a time window of resting-state data is collected as a baseline. The baseline data is divided according to the sliding window size, and each divided part overlaps half of the previous window to obtain sliding window data. The sliding window data of each channel are superimposed and windowed to obtain the data of each window. A power spectrum set is obtained for each window data, and its mean and standard deviation are calculated. Then, the real-time task sliding window power spectrum is normalized according to the z-score formula to obtain normalized window data; interpolation calculation is performed based on the normalized window data to obtain the interpolation grid.

8. The method for drawing a dynamic interpolation brain topography map according to claim 2, characterized in that: The method further includes performing the following operations using the brain topography mapping module: Pre-generate the electrode UV point coordinates of each electrode based on the designed texture map, and perform alignment operations according to the electrode UV point coordinates in the 10-20 EEG specification to provide a coordinate basis for mapping two-dimensional data into three-dimensional space; Using UV point coordinates as the interpolation grid coordinate system, the interpolation grid data of the two-dimensional brain topography map is linked to the three-dimensional space to achieve the conversion from two-dimensional to three-dimensional.

9. The method for drawing a dynamic interpolation brain topography map according to claim 2, wherein: The method further includes: when performing interpolation calculations, the interpolation calculation module maps the interpolation grid to the GPU via CUDA for calculations; during the rendering process, the interpolated data is rendered in real time using OpenGL. To avoid video memory conflicts, the CUDA and OpenGL contexts are shared, and a drawing texture buffer is registered with CUDA using a shared video memory method. The graphics memory is attached exclusively during calculations and detached to OpenGL after the calculations are completed.

10. The method for drawing a dynamic interpolation brain topography map according to claim 2, characterized in that: The method further includes: if the two-dimensional brain topography map has a boundary, performing zero-interpolation constraint on the boundary.

Images