Transcranial electrical stimulation individualized navigation positioning system based on visual guidance

Abstract

The invention discloses a transcranial electrical stimulation individualized navigation positioning system based on visual guidance, and the system comprises a head model reconstruction module which is used for constructing a magnetic resonance head model according to collected magnetic resonance data, and generating a leadfield matrix and coordinate data; the individualized navigation module is used for calculating an individualized electrode arrangement position according to the leadfield matrix and the coordinate data in combination with the coordinates of the target region; the depth camera system is used for sliding a circle along the sliding rail to collect a depth image of the tested head and constructing a point cloud head model according to the depth image; the individualized positioning module is used for registering the point cloud head model and the magnetic resonance head model to obtain real world space coordinates of electrode arrangement positions; and the laser indicator is used for indicating the real position of the electrode on the head through laser according to the real world space coordinates. The device can accurately guide the placement of the stimulating electrode according to the coordinates of the target region.

Description

Technical Field

[0001] The present invention belongs to the field of image vision technology, and in particular relates to a transcranial electrical stimulation individualized navigation and positioning system based on vision guidance. Background Art

[0002] With the development and continuous application of transcranial electrical stimulation technology, higher requirements have been put forward for transcranial electrical stimulation: first, many important regulatory targets are located deep in the brain, such as the hippocampus, striatum, thalamus, etc., requiring effective stimulation of transcranial electrical stimulation to have sufficient penetration to reach the deep brain areas; second, effective stimulation is required to be sufficiently focused, that is, only the target area is activated, and other target areas are not activated as much as possible (relevant studies have shown that the field strength required for neural regulation needs to be greater than 0.2V / m).

[0003] The electrode placement of traditional transcranial electrical stimulation is often selected based on a system such as 10-10. This means that if the electrodes are placed according to the division standards of the 10-10 system, they need to be manually measured and placed, and this process may cause errors. Similarly, placing electrodes at the positions marked by pre-made EEG caps can also introduce errors due to improper wearing of the EEG cap or individual differences. These errors may eventually lead to target deviation or decreased focus of transcranial electrical stimulation, making it difficult to achieve truly accurate and effective transcranial electrical stimulation. Therefore, a method that can accurately guide electrode positioning is needed. Summary of the Invention

[0004] The present invention proposes a transcranial electrical stimulation individualized navigation and positioning system based on vision guidance to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above objectives, the present invention provides a vision-guided transcranial electrical stimulation personalized navigation and positioning system, comprising:

[0006] A head model reconstruction module is used to construct a magnetic resonance head model based on the acquired magnetic resonance data and generate a leadfield matrix and coordinate data;

[0007] Individualized navigation module, used to calculate individualized electrode placement based on the leadfield matrix and coordinate data combined with target coordinates;

[0008] The depth camera system is used to slide along the slide rail to collect the depth image of the subject's head and construct a point cloud head model based on the depth image;

[0009] An individualized positioning module, configured to register the point cloud head model with the magnetic resonance head model to obtain the real-world spatial coordinates of the electrode arrangement positions;

[0010] A laser pointer is used to indicate the real position of the electrode on the head through a laser according to the real world space coordinates.

[0011] Preferably, the head model reconstruction module includes:

[0012] A magnetic resonance data acquisition unit, used for acquiring magnetic resonance imaging data of the subject's head;

[0013] A head model construction unit, configured to reconstruct a three-dimensional magnetic resonance head model based on the acquired magnetic resonance imaging data;

[0014] A leadfield calculation unit, configured to calculate a leadfield matrix based on the three-dimensional magnetic resonance head model;

[0015] The coordinate data generating unit is used to generate corresponding coordinate data for each electrode position.

[0016] Preferably, the personalized navigation module includes:

[0017] A data input unit, for receiving a magnetic resonance head model, a leadfield matrix, coordinate data and target area coordinates input by a user;

[0018] Electrode arrangement optimization unit, used to optimize the arrangement of electrodes using a multi-objective optimization algorithm based on the Leadfield matrix and target coordinates;

[0019] The solution screening unit is used to screen multiple sets of electrode arrangement-stimulation parameter pairing solutions generated by the multi-objective optimization algorithm.

[0020] Preferably, the multi-objective optimization algorithm adopts a multi-objective differential evolution algorithm, including: initializing the population, mutation, crossover, selection, calculating fitness and determining whether to terminate.

[0021] Preferably, the depth camera system comprises:

[0022] Depth camera unit, used to collect RGB images and depth images of the subject's head;

[0023] The slide motion control unit is used to control the depth camera to slide along the slide;

[0024] A processing unit is used to process the RGB image and depth image collected by the depth camera, extract feature points and generate point cloud data;

[0025] The point cloud head model construction unit is used to construct a three-dimensional point cloud head model based on multiple sets of point cloud data collected by the depth camera.

[0026] Preferably, the individualized positioning module includes:

[0027] A feature point extraction unit, used for extracting feature points from the point cloud head model and the magnetic resonance head model;

[0028] a registration unit, configured to register the point cloud head model with the magnetic resonance head model and calculate the spatial transformation relationship between the cloud head model and the magnetic resonance head model;

[0029] The space conversion unit is used to convert the electrode arrangement position from the magnetic resonance head model space to the real world space according to the registration result.

[0030] Preferably, the registration unit comprises:

[0031] a coarse registration unit, configured to perform preliminary alignment of the point cloud head model and the magnetic resonance head model using a coarse registration algorithm;

[0032] The fine registration unit is used to accurately align the point cloud head model and the magnetic resonance head model through a fine registration algorithm.

[0033] Preferably, the registration unit performs the following steps:

[0034] In the coarse registration stage, the four-point consistency set algorithm is used to quickly match the feature points in the point cloud head model and the magnetic resonance head model to obtain an initial solution; in the fine registration stage, the iterative closest point algorithm is used to gradually optimize the spatial transformation relationship between the point cloud head model and the magnetic resonance head model through continuous iterative calculations until the preset registration accuracy requirements are met.

[0035] Compared with the prior art, the present invention has the following advantages and technical effects:

[0036] The present invention discloses a transcranial electrical stimulation personalized navigation and positioning system based on vision guidance, comprising: a head model reconstruction module for constructing a magnetic resonance head model based on collected magnetic resonance data and generating a leadfield matrix and coordinate data; a personalized navigation module for calculating personalized electrode arrangement positions based on the leadfield matrix and coordinate data combined with target area coordinates; a depth camera system for collecting a depth image of the subject's head by sliding along a slide rail for one circle and constructing a point cloud head model based on the depth image; a personalized positioning module for aligning the point cloud head model with the magnetic resonance head model to obtain the real-world spatial coordinates of the electrode arrangement positions; and a laser pointer for indicating the real position of the electrodes on the head using a laser according to the real-world spatial coordinates. The present invention can accurately point out the electrode installation position and avoid installation errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0038] Figure 1 This is a schematic diagram of the overall process of the device according to an embodiment of the present invention;

[0039] Figure 2 Schematic diagram of the device structure of an embodiment of the present invention;

[0040] Figure 3 This is a flowchart of the navigation module of an embodiment of the present invention;

[0041] Figure 4 A flowchart of individualized stimulation electrode arrangement and stimulation parameter generation according to an embodiment of the present invention;

[0042] Figure 5 This is a flowchart of the construction and registration of a point cloud head model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0044] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0045] Example 1

[0046] like Figure 1-2 As shown, this embodiment provides a transcranial electrical stimulation personalized navigation and positioning system based on vision guidance, including:

[0047] A head model reconstruction module is used to construct a magnetic resonance head model based on the acquired magnetic resonance data and generate a leadfield matrix and coordinate data;

[0048] Individualized navigation module, used to calculate individualized electrode placement based on the leadfield matrix and coordinate data combined with target coordinates;

[0049] The depth camera system is used to slide along the slide rail to collect the depth image of the subject's head and construct a point cloud head model based on the depth image;

[0050] An individualized positioning module, configured to register the point cloud head model with the magnetic resonance head model to obtain the real-world spatial coordinates of the electrode arrangement positions;

[0051] A laser pointer is used to indicate the real position of the electrode on the head through a laser according to the real world space coordinates.

[0052] Furthermore, the head model reconstruction module includes:

[0053] A magnetic resonance data acquisition unit, used for acquiring magnetic resonance imaging data of the subject's head;

[0054] A head model construction unit, configured to reconstruct a three-dimensional magnetic resonance head model based on the acquired magnetic resonance imaging data;

[0055] A leadfield calculation unit, configured to calculate a leadfield matrix based on the three-dimensional magnetic resonance head model;

[0056] The coordinate data generating unit is used to generate corresponding coordinate data for each electrode position.

[0057] Furthermore, the personalized navigation module includes:

[0058] A data input unit, for receiving a magnetic resonance head model, a leadfield matrix, coordinate data and target area coordinates input by a user;

[0059] Electrode arrangement optimization unit, used to optimize the arrangement of electrodes through a multi-objective optimization algorithm based on the leadfield matrix and target area coordinates;

[0060] The solution screening unit is used to screen multiple sets of electrode arrangement-stimulation parameter pairing solutions generated by the multi-objective optimization algorithm.

[0061] Furthermore, the multi-objective optimization algorithm adopts a multi-objective differential evolution algorithm, including: initializing the population, mutation, crossover, selection, calculating fitness and determining whether to terminate.

[0062] Furthermore, the depth camera system includes:

[0063] Depth camera unit, used to collect RGB images and depth images of the subject's head;

[0064] The slide motion control unit is used to control the depth camera to slide along the slide;

[0065] A processing unit is used to process the RGB image and depth image collected by the depth camera, extract feature points and generate point cloud data;

[0066] The point cloud head model construction unit is used to construct a three-dimensional point cloud head model based on multiple sets of point cloud data collected by the depth camera.

[0067] Furthermore, the individualized positioning module includes:

[0068] A feature point extraction unit, used for extracting feature points from the point cloud head model and the magnetic resonance head model;

[0069] A registration unit, configured to register the point cloud head model with the magnetic resonance head model and calculate the spatial transformation relationship between the point cloud head model and the magnetic resonance head model;

[0070] The space conversion unit is used to convert the electrode arrangement position from the magnetic resonance head model space to the real world space according to the registration result.

[0071] Furthermore, the registration unit includes:

[0072] a coarse registration unit, configured to perform preliminary alignment of the point cloud head model and the magnetic resonance head model using a coarse registration algorithm;

[0073] The fine registration unit is used to accurately align the point cloud head model and the magnetic resonance head model through a fine registration algorithm.

[0074] The implementation process is as follows:

[0075] S1: Reconstructs the head model based on the previously acquired MRI data and generates the leadfield matrix and coordinate data. This data is input into the personalized navigation module to calculate the personalized electrode arrangement and stimulation parameters;

[0076] S2: Use the depth camera to slide along the slide rail for one circle to collect images, reconstruct the point cloud based on the collected depth image, align it with the MRI reconstructed head model, and then input it into the individual positioning module;

[0077] S3: Use the personalized positioning module to convert the electrode positions obtained by personalized navigation into real-world positions. The laser pointer will sequentially indicate the placement of the electrodes in the real world, and then the electrodes are put on according to the instructions. After putting on the electrodes, the stimulation parameters are configured according to the personalized navigation results and stimulation begins.

[0078] S4: During stimulation, the infrared camera will monitor the electrode temperature in real time. If the electrode temperature is too high, the terminal will alarm.

[0079] S5: After the stimulation is completed, the terminal will prompt that the stimulation is completed. The stimulation process data and stimulation feedback are saved;

[0080] In order to obtain the basic prior information required for subsequent simulation and personalized navigation, in step S1, it is necessary to collect magnetic resonance images of the subject's head in advance, and use the magnetic resonance data to obtain personalized leadfield data and corresponding position data. The leadfield is essentially a linear mapping matrix used to quantify the physical transmission relationship between neural electrical activity in the brain and scalp or external electrode signals. In TI stimulation, the leadfield is usually represented by a matrix L, which satisfies the following linear relationship:

[0081] E=LI

[0082] Where E represents the brain's electric field distribution, and I represents the electrode current vector. Under artificial constraints, the row or column index of the leadfield corresponds to a specific stimulation electrode. The position data stores the coordinates of the stimulation electrode corresponding to the same row or column index. This data, along with the target coordinates, is fed into the personalized navigation module to determine the stimulation electrode layout and stimulation current parameters.

[0083] Step S2 describes the construction of the individualized positioning module before formal stimulation: the subject takes his place next to the transcranial electrical stimulation device and uses Figure 1 The head fixation device in the terminal assists in securing the subject's head, reducing data noise caused by head movement. After the terminal selects a series of feature points, depth camera data acquisition begins: the depth camera slides along the rails for one full rotation, collecting RGB and depth images. The terminal then automatically calculates and constructs a point cloud model of the subject's head and displays it. Once the depth camera data is confirmed to be correct, the point cloud head model and the MRI head model are aligned. Once the alignment is complete, the results are displayed on the terminal. Once the alignment is confirmed to be correct, the construction of the personalized positioning module is completed.

[0084] Step S3 describes the use of the personalized positioning module to convert the electrode position obtained by personalized navigation into the real space position and the formal pre-stimulation preparations such as wearing the stimulation electrodes: the personalized stimulation electrode arrangement coordinates obtained in step S1 are input into the personalized positioning module constructed in step S2, and the personalized stimulation electrode arrangement in the real world space coordinates are obtained. Then, the laser pointer will indicate the position of the stimulation electrode on the subject's head in the real world according to these coordinates. Then, it is only necessary to put on the saline electrodes for the subjects in turn according to the points indicated by the laser pointer. Finally, the stimulation parameters are configured according to the personalized navigation results and the stimulation begins.

[0085] Step S4 describes real-time monitoring of electrode temperature during stimulation. After the stimulation begins, the infrared camera monitors the electrode temperature in real time. If the electrode temperature exceeds a set threshold, a warning is issued, and stimulation is terminated to ensure safety.

[0086] After the stimulation is completed, the terminal will prompt that the stimulation is completed. At the same time, the terminal will save the stimulation process data (including basic information of the subject, stimulation parameters, electrode arrangement, etc.) and stimulation feedback.

[0087] Further optimization scheme, such as Figure 3-4 As shown, navigation includes the following aspects:

[0088] In terms of input, the personalized navigation system needs to provide three basic data: one is the personalized leadfield data, the second is the position data corresponding to the personalized leadfield, and the third is the basic information of the target area of this stimulation.

[0089] In terms of output, the personalized navigation system outputs a set of corresponding stimulation electrode arrangements and stimulation parameters.

[0090] The overall navigation optimization process involves obtaining leadfield and corresponding location data and target area information, encoding electrode information and current parameters, and feeding these into a multi-objective differential evolution algorithm for parallel optimization. Because the multi-objective optimization approach utilizes a Pareto multi-objective optimization scheme, and the differential evolution algorithm employs a population optimization approach, the output is multiple paired solutions for electrode placement and stimulation parameters. Through a combination of automated and manual screening, the desired electrode placement and current parameters are determined.

[0091] Individualized leadfield data and corresponding position data require the collection of magnetic resonance imaging data of the subjects in advance and are obtained through calculation. Leadfield reflects the conduction characteristics of electric current in complex biological tissues, and its row or column index corresponds to a certain stimulation electrode. The position data stores the coordinates of a stimulation electrode corresponding to the same row or column index. In practice, it can be artificially stipulated that the leadfield is a tensor of size n×m×3, and the position data is an n×3 matrix. Among them, n represents the total number of electrodes in the electrode arrangement template, m represents the total number of grids divided by the head model, and 3 represents the three directions of x, y, and z in space.

[0092] In practice, the total number of electrodes n in an electrode layout template can be strictly defined according to the 10-10 system, or it can be 64, 128, 256, or other values. In fact, as long as the leadfield corresponds to the position data, the electrode layout template can be customized without having to follow the universal 10-10 system. This customizable electrode layout template helps expand the solution space for subsequent navigation optimization and achieve better results in guided electrical stimulation.

[0093] To implement the navigation optimization function, the temporal interference stimulation navigation optimization problem is a non-convex problem that requires simultaneous consideration of both target stimulation effectiveness and focus. Therefore, this personalized navigation system uses a multi-objective differential evolution (DE) algorithm as its core optimization algorithm. The differential evolution algorithm optimization process consists of five main steps: population initialization, mutation, crossover, selection, fitness calculation, and termination determination. Initialization can simply use a random initialization method. The mutation process is as follows:

[0094] Vi (g+1)=X r1 (g)+F(X r2 (g)-X r3 (g))

[0095] Among them, r1, r2, r3 are three random numbers, X r1 (g) is the solution extracted by random number r1 in the g-th generation solution, V i (g+1) is the g+1th generation solution compiled from the gth generation solution, and F is the mutation factor.

[0096] The purpose of crossover is to randomly select individuals:

[0097]

[0098] Among them, U i,j (g+1) is the offspring solution after the i-th and j-th g+1-generation solutions cross over, and CR is the crossover probability.

[0099] After crossover, the offspring solution is selected, generally using a greedy strategy:

[0100]

[0101] Among them, X i (g+1) is the ith g+1 generation solution, f(X i (g)) is produced by X i (g) Calculate the objective function result. Repeat this process until the calculated fitness (i.e. the value of the objective function) meets the termination requirements.

[0102] Further optimization of the plan includes the following aspects:

[0103] In terms of input, the personalized positioning system requires three parts of data as input: one is the magnetic resonance head model constructed based on the magnetic resonance imaging data collected from the subject in the early stage; the second is the depth data collected by the depth camera before stimulation; and the third is a set of manually specified feature points.

[0104] In terms of output, the personalized positioning system provides a conversion relationship for the input magnetic resonance head model space coordinates to the real-world space coordinates of the point cloud head model, thereby outputting the stimulation electrode arrangement coordinates of the real-world space of the point cloud head model.

[0105] After the subject is in place, before the formal stimulation, the MRI head model constructed by the subject's previously collected MRI imaging data should have been saved and input into the individual positioning system. At this time, the depth camera is controlled to slide along the slide rail for one week to collect RGB and depth images, and then confirm the manually selected feature points. Figure 5As shown in the figure, a point cloud head model is constructed. The point cloud head model is mainly constructed by using a registration algorithm to obtain multiple sets of point cloud data captured by a depth camera at different positions.

[0106] After obtaining the point cloud head model, a registration algorithm is used to align the point cloud head model with the MRI head model. This registration yields a conversion relationship between the MRI head model's spatial coordinates and the point cloud head model's real-world spatial coordinates. This conversion relationship is used to convert the stimulation electrode layout coordinates, output by the personalized navigation module and based on the MRI head model's spatial coordinates, into the point cloud head model's real-world spatial coordinates.

[0107] The registration algorithm is divided into two steps: coarse registration and fine registration. Coarse registration uses the 4-Point Consongruent Sets (4PCS) algorithm, and fine registration uses the Iterative Closest Point (ICP) algorithm. The specific registration process is as follows:

[0108] Based on the manually specified feature points, a four-point consensus set algorithm is first used for coarse registration. This step provides a relatively good initial solution for subsequent fine registration while being relatively fast. After coarse registration, an iterative closest point algorithm is used for iterative calculations of fine registration. Considering that, in addition to translation, spatial transformations are mainly represented by a set of affine relations in linear algebra, we can set:

[0109] Q=AP+B

[0110] Among them, Q is the real-world space coordinate matrix of the point cloud head model; P is the space coordinate matrix of the magnetic resonance head model; A and B are space conversion factors, covering space conversion information such as translation, rotation, and scaling.

[0111] Therefore, the optimization objective can be set as:

[0112]

[0113] The solution is then obtained through an iterative method. In fact, there are currently a variety of iterative solutions for such optimization problems, including but not limited to the singular value decomposition (SVD) method. Here, the singular value decomposition method can be simply used to complete the iterative registration.

[0114] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A vision-guided transcranial electrical stimulation personalized navigation and positioning system, characterized in that: include: A head model reconstruction module is used to construct a magnetic resonance head model based on the acquired magnetic resonance data and generate a leadfield matrix and coordinate data; Individualized navigation module, used to calculate individualized electrode placement based on the leadfield matrix and coordinate data combined with target coordinates; The depth camera system is used to slide along the slide rail to collect the depth image of the subject's head and construct a point cloud head model based on the depth image; An individualized positioning module, configured to register the point cloud head model with the magnetic resonance head model to obtain the real-world spatial coordinates of the electrode arrangement positions; A laser pointer is used to indicate the real position of the electrode on the head through a laser according to the real world space coordinates.

2. The device according to claim 1, characterized in that The head model reconstruction module includes: A magnetic resonance data acquisition unit, used for acquiring magnetic resonance imaging data of the subject's head; A head model construction unit, configured to reconstruct a three-dimensional magnetic resonance head model based on the acquired magnetic resonance imaging data; A leadfield calculation unit, configured to calculate a leadfield matrix based on the three-dimensional magnetic resonance head model; The coordinate data generating unit is used to generate corresponding coordinate data for each electrode position.

3. The device according to claim 1, characterized in that The personalized navigation module includes: A data input unit, for receiving a magnetic resonance head model, a leadfield matrix, coordinate data and target area coordinates input by a user; Electrode arrangement optimization unit, used to optimize the arrangement of electrodes through a multi-objective optimization algorithm based on the leadfield matrix and target area coordinates; The solution screening unit is used to screen multiple sets of electrode arrangement-stimulation parameter pairing solutions generated by the multi-objective optimization algorithm.

4. The device according to claim 3, characterized in that The multi-objective optimization algorithm adopts a multi-objective differential evolution algorithm, including: initializing population, mutation, crossover, selection, calculating fitness and judging whether to terminate.

5. The device according to claim 1, characterized in that The depth camera system comprises: Depth camera unit, used to collect RGB images and depth images of the subject's head; The slide motion control unit is used to control the depth camera to slide along the slide; A processing unit is used to process the RGB image and depth image collected by the depth camera, extract feature points and generate point cloud data; The point cloud head model construction unit is used to construct a three-dimensional point cloud head model based on multiple sets of point cloud data collected by the depth camera.

6. The device according to claim 1, characterized in that The individualized positioning module includes: A feature point extraction unit, used for extracting feature points from the point cloud head model and the magnetic resonance head model; a registration unit, configured to register the point cloud head model with the magnetic resonance head model and calculate the spatial transformation relationship between the cloud head model and the magnetic resonance head model; The space conversion unit is used to convert the electrode arrangement position from the magnetic resonance head model space to the real world space according to the registration result.

7. The device according to claim 6, characterized in that The registration unit comprises: a coarse registration unit, configured to perform preliminary alignment of the point cloud head model and the magnetic resonance head model using a coarse registration algorithm; The fine registration unit is used to accurately align the point cloud head model and the magnetic resonance head model through a fine registration algorithm.

8. The device according to claim 6, characterized in that The working steps of the registration unit are as follows: In the coarse registration stage, the four-point consistency set algorithm is used to quickly match the feature points in the point cloud head model and the magnetic resonance head model to obtain an initial solution; in the fine registration stage, the iterative closest point algorithm is used to gradually optimize the spatial transformation relationship between the point cloud head model and the magnetic resonance head model through continuous iterative calculations until the preset registration accuracy requirements are met.

Images