Multi-channel transcranial magnetic stimulation coil array control method

Abstract

The invention relates to a multi-channel transcranial magnetic stimulation coil array control method, which comprises the following steps: solving a lead field matrix of each coil according to brain structure data of a subject and adopted multi-channel transcranial magnetic stimulation coil array arrangement; calculating the distribution of an induced electric field generated by each coil, and establishing an optimization function based on Elasticic-net regression by taking minimization of a difference value between the distribution of the induced electric field and the distribution of a target electric field as a target; respectively solving the optimization functions under different regularization intensities to obtain a plurality of alternative current time derivative vectors; and selecting the alternative current time derivative vector which enables the coil energy consumption to be lowest and the number of activated coils to be small, and controlling the current passing through each coil in the coil array according to a selection result. The method can accurately control the coil current to generate the required induction electric field, is widely applicable to a multichannel transcranial magnetic stimulation coil array, and has the advantages of high calculation speed, high flexibility and high usability.

Description

Technical Field

[0001] The present invention relates to the technical field of transcranial magnetic stimulation, and in particular to a multi-channel transcranial magnetic stimulation coil array control method. Background Art

[0002] Transcranial magnetic stimulation (TMS) is a non-invasive, safe, and highly effective neurostimulation technique. The time-varying magnetic field generated by the TMS coil can pass through the skull with near-zero loss, generating an induced electric field in the cerebral cortex, affecting brain metabolism and neural electrical activity. Clinically, TMS can be used to treat neurological conditions such as depression and sleep disorders. In scientific research, it can be used in conjunction with neuroimaging techniques such as electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) to study brain function.

[0003] Traditional transcranial magnetic stimulation devices use a single circular or figure-8 coil clamped by a human or robotic arm to focus the electric field at a specific location. This results in insufficient positioning accuracy and slow stimulation position switching speeds. Multi-channel transcranial magnetic stimulation devices, on the other hand, use an array of multiple coils to achieve electric field focusing at a specific location by superimposing the induced electric fields generated by each coil. Circuit control replaces the mechanical control of traditional devices, maintaining sufficient positioning accuracy while providing extremely fast stimulation position switching speeds. However, how to specifically control the current pulses passing through each coil to achieve the ideal electric field focusing effect, and how to implement it in an actual system, remains an urgent problem to be solved. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a multi-channel transcranial magnetic stimulation coil array control method, which can accurately control the coil current to generate the required induced electric field and is widely applicable to multi-channel transcranial magnetic stimulation coil arrays.

[0005] The technical solution adopted by the present invention to solve the technical problem is to provide a multi-channel transcranial magnetic stimulation coil array control method, comprising the following steps:

[0006] Based on the subject's brain structure data and the multi-channel transcranial magnetic stimulation coil array used, the lead field matrix of each coil is solved;

[0007] Calculating the induced electric field distribution generated by each coil, wherein the induced electric field distribution is the product of the coil's current time derivative vector and the lead field matrix;

[0008] An optimization function based on Elastic-net regression is established with the goal of minimizing the difference between the induced electric field distribution and the target electric field distribution;

[0009] Solving the optimization function under different regularization strengths respectively to obtain multiple candidate current time derivative vectors;

[0010] The candidate current time derivative vector that minimizes coil energy consumption and activates a smaller number of coils is selected, and the current passing through each coil in the coil array is controlled according to the selection result.

[0011] Furthermore, the selecting of the candidate current time derivative vector that minimizes coil energy consumption and activates a smaller number of coils includes:

[0012] determining a maximum element in each of the candidate current time derivative vectors;

[0013] The candidate current time derivative vector corresponding to the smallest maximum element is selected.

[0014] Furthermore, the method of solving the lead field matrix of each coil based on the subject's brain structure data and the multi-channel transcranial magnetic stimulation coil array arrangement includes:

[0015] A volume conductor model of the subject's head is established based on the subject's brain structure data;

[0016] According to the arrangement of the multi-channel transcranial magnetic stimulation coil array, a spatial position model of the coil array is established;

[0017] Based on the volume conductor model and the spatial position model, the lead field matrix of each coil is calculated.

[0018] Furthermore, the volume conductor model of the subject's head is established by assigning corresponding conductivity values ​​to each brain structure of the subject.

[0019] Furthermore, the spatial position model of the coil array is established by taking the center position of the subject's head as the origin and calculating the three-dimensional coordinates and coil orientation of each coil in the coil array during transcranial magnetic stimulation.

[0020] Furthermore, the calculating of the three-dimensional coordinates and coil orientation of each coil in the coil array during transcranial magnetic stimulation includes:

[0021] Abstract the ring elements of the coils of each coil unit into current loops;

[0022] A ring element is uniquely determined by the ring radius, the center position of the ring, and the normal direction of the ring;

[0023] The data of all the ring elements are saved as a matrix to obtain the spatial position model.

[0024] Furthermore, the optimization function is expressed as

[0025]

[0026] Among them, ||·||1 represents the L1 norm, ||·||2 represents the L2 norm, represents the target electric field distribution, J represents the current time derivative vector, represents the lead field matrix, α represents the regularization weight, and λ represents the regularization strength.

[0027] Furthermore, the optimal parameters of the optimization function are calculated by using a coordinate descent method and an alternating direction multiplier method.

[0028] Beneficial effects

[0029] Due to the adoption of the above-mentioned technical solution, the present invention has the following advantages and positive effects compared with the prior art: Based on the volume conductor model of the subject's head and the spatial position model of the coil array, the present invention calculates the lead field matrix of each coil, and then solves the current time derivative vector that minimizes the distance between the actual induced electric field and the target induced electric field. There are no special requirements for the structural parameters and arrangement of the coil array, and it is suitable for most multi-channel transcranial magnetic stimulation coil array designs, and has the advantages of fast calculation speed, strong flexibility and high ease of use; Based on optimization theory, the present invention uses Elastic-net regression to calculate the current in each coil required to generate the target induced electric field distribution. At the same time, by controlling the regularization parameter of the optimization function, the current intensity and focusing effect can be adjusted, and coils that do not need to be started can be effectively excluded, thereby reducing the control burden of the multi-channel transcranial magnetic stimulation system; By selecting the solution that minimizes the maximum current time derivative from multiple groups of solutions under different regularization strengths, the present invention enables the control scheme to meet the lowest energy consumption while using fewer coils. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of an embodiment of the present invention;

[0031] Figure 2 It is an effect diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0032] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall equally within the scope limited by the appended claims of the application.

[0033] The embodiment of the present invention relates to a multi-channel transcranial magnetic stimulation coil array control method based on optimization theory, such as Figure 1 As shown, the following steps are included:

[0034] S1 solves the lead field matrix of each coil based on the subject's brain structure data and the multi-channel transcranial magnetic stimulation coil array used;

[0035] S2 calculates the induced electric field distribution generated by each coil. The induced electric field distribution is the product of the coil's current time derivative vector and the lead field matrix;

[0036] S3 aims to minimize the difference between the induced electric field distribution and the target electric field distribution and establishes an optimization function based on Elastic-net regression;

[0037] S4 solves the optimal parameters of the above optimization function under different regularization strengths respectively to obtain multiple candidate current time derivative vectors;

[0038] S5 selects an alternative current time derivative vector that minimizes coil energy consumption and activates a smaller number of coils, and controls the current passing through each coil in the coil array according to the selected result.

[0039] The lead field matrix is ​​obtained by calculating the coil's lead-field matrix based on the volume conductor model of the subject's head and the spatial position model of the coil array. Specifically, the following method can be used for modeling:

[0040] S101 collects brain structural data of the subject. In this embodiment, magnetic resonance imaging (MRI) is used to scan the subject's head to obtain structural data of various parts of the brain, including distribution data of various tissues such as cerebrospinal fluid, gray matter, and white matter.

[0041] S102 determines the multi-channel transcranial magnetic stimulation coil array to be used, specifically including the coil orientation, coil radius, number of turns, and other necessary spatial position distribution parameters; the coil material, shape, winding method, number of turns and other structural parameters, as well as the arrangement of the coil array, can be selected according to actual needs. In this embodiment, the preferred coil array is a coil composed of three ring elements with a diameter of 50 mm and 20 turns wound with copper Litz wire with an outer diameter of 2 mm, arranged perpendicularly to each other as the basic unit, and arranged at the electrode positions of the standard EEG 10-10 system;

[0042] S103 physical model building, including establishing a volume conductor model of the subject's head and a spatial position model of the coil array;

[0043] The volume conductor model includes conductivity data for various locations in the conductor to determine the relationship between the current density at any point in the conductor and the electric field strength at that point. Specifically, the structural data of the subject's brain obtained in step S101 can be used to construct a brain structure model, and corresponding conductivity values ​​are assigned to various parts of the subject's brain. More specifically, the conductivity of the scalp is 0.33 S / m, the conductivity of the skull is 0.0042 S / m, the conductivity of the cerebrospinal fluid is 1.79 S / m, the conductivity of the gray matter is 0.33 S / m, and the conductivity of the white matter is 0.14 S / m.

[0044] When establishing the spatial position model of the coil array, the spatial position, coil radius, and coil orientation of each coil in the coil array are imported as data. Specifically, the center of the subject's brain is used as the coordinate origin, and the annular elements of the coils of each coil unit are abstracted as current loops. A ring element is uniquely determined by the ring radius, the center position of the ring, and the normal direction of the ring. The data of all the ring elements are stored in a matrix to obtain the spatial position model.

[0045] This embodiment uses a method of calculating the lead-field of the coil to calculate the distribution of the induced electric field in the subject's brain when a current of unit intensity passes through each coil in the coil array.

[0046] The definition formula of Lead-field is as follows:

[0047]

[0048] Among them, L i (r) represents the lead-field of coil i at position r, e k represents the unit vector along the three-dimensional Cartesian coordinates x, y, and z, Φ represents the magnetic flux, Q(r) represents the unit current dipole source at position r, S i represents the area enclosed by coil i, B Q represents the magnetic field generated by a unit current dipole source.

[0049] The magnetic field generated by a unit current dipole source is calculated using Biotsavart's law:

[0050]

[0051] Among them, B Q (r′) represents the magnetic field of the current dipole source Q at position r′, r represents the position of the current dipole source, μ0 represents the vacuum permeability, J tot represents the current of the dipole source plus the body current it produces in the brain, and V represents the location contained in the brain volume.

[0052] The formula for calculating the induced electric field generated by the coil through the lead field of the coil is as follows:

[0053] E i =-J i L i

[0054] Among them, E i represents the induced electric field generated by coil i, J i =dI i / dt represents the time derivative of the current through coil i, L i represents the lead-field matrix of coil i.

[0055] The lead-field matrix of all coils is calculated and stored as data. This data is used to establish an optimization function based on Elastic-net regression and solve the current passing through each coil, so that an induced electric field distribution closest to the required induced electric field can be generated.

[0056] The optimization function based on Elastic-net regression is expressed as follows:

[0057]

[0058] Among them, ||·||1 represents the L1 norm, which is used to achieve sparse activation of coils (i.e., exclude coils that do not need to be started), and ||·||2 represents the L2 norm, which is used to limit the current (current time derivative) size (i.e., reduce energy consumption). The induced electric field distribution matrix representing the demand, represents the lead-field matrix of the coil array, J represents the time derivative vector of the current passing through the coil array, α represents the regularization weight, which is used to adjust the intensity of the two norms, and λ represents the regularization strength, which is used to adjust the focusing effect, satisfying 0<α<1,λ≥0.

[0059] The optimization function can be solved by existing technical means, such as using the alternating direction method of multipliers (ADMM) to calculate the solution of the optimization function to obtain the vector J that minimizes the value of the optimization function.

[0060] In some preferred embodiments, a coil array control scheme can be selected from several sets of coil currents that successfully focus the induced electric field, minimizing energy consumption while using fewer coils. During calculations, coordinate descent can be used to simultaneously solve optimization problems for multiple sets of different λ parameters. The ADMM algorithm is also employed to achieve better performance and improve solution speed for tall matrices with unusual shapes (i.e., those with significantly more rows than columns). During the solution process, the values ​​of λ are arranged from small to large, so that the number of nonzero elements in the corresponding solution vector J decreases from high to low.

[0061] From the multiple calculated solutions, the solution that minimizes the maximum time derivative of the current is selected as the control scheme for the coil array. The value of each element in vector J corresponds to the time derivative of the current passing through each coil in the coil array. For example, if the first element in vector J is J1, the time derivative of the current passing through coil 1 should be J1 / N, where N is the number of turns in the coil. A corresponding solution vector is calculated for each desired induced electric field. The result is the control scheme for the current when the coil array is focused at the corresponding position. By controlling the time derivative of the current passing through each coil, the coil array can produce a distribution that most closely matches the desired induced electric field.

[0062] When actually performing transcranial magnetic stimulation, it is only necessary to control the current pulses passing through each coil through the circuit according to the scheme given in this method to achieve rapid alternating stimulation of the dorsolateral prefrontal cortex (DLPFC) and the M1 area. Figure 2 As shown, the circles around the brain represent the coils at the corresponding positions. The colored circles represent that the coils need to be activated, and the black circles represent that they are not activated.

Claims

1. A multi-channel transcranial magnetic stimulation coil array control method, characterized in that: The following steps are involved: Based on the subject's brain structure data and the multi-channel transcranial magnetic stimulation coil array used, the lead field matrix of each coil is solved; Calculating the induced electric field distribution generated by each coil, wherein the induced electric field distribution is the product of the coil's current time derivative vector and the lead field matrix; An optimization function based on Elastic-net regression is established with the goal of minimizing the difference between the induced electric field distribution and the target electric field distribution; Solving the optimization function under different regularization strengths respectively to obtain multiple candidate current time derivative vectors; The candidate current time derivative vector that minimizes coil energy consumption and activates a smaller number of coils is selected, and the current passing through each coil in the coil array is controlled according to the selection result.

2. The method according to claim 1, characterized in that The selecting the candidate current time derivative vector that minimizes coil energy consumption and activates a smaller number of coils includes: determining a maximum element in each of the candidate current time derivative vectors; The candidate current time derivative vector corresponding to the smallest maximum element is selected.

3. The method according to claim 1, characterized in that The method of solving the lead field matrix of each coil based on the subject's brain structure data and the multi-channel transcranial magnetic stimulation coil array arrangement includes: A volume conductor model of the subject's head is established based on the subject's brain structure data; According to the arrangement of the multi-channel transcranial magnetic stimulation coil array, a spatial position model of the coil array is established; Based on the volume conductor model and the spatial position model, the lead field matrix of each coil is calculated.

4. The method according to claim 3, characterized in that The volume conductor model of the subject's head is established by assigning corresponding conductivity values ​​to each brain structure of the subject.

5. The method according to claim 3, characterized in that The spatial position model of the coil array is established by taking the center position of the subject's head as the origin and calculating the three-dimensional coordinates and coil orientation of each coil in the coil array during transcranial magnetic stimulation.

6. The method according to claim 5, characterized in that The calculating of the three-dimensional coordinates and coil orientation of each coil in the coil array during transcranial magnetic stimulation includes: Abstract the ring elements of the coils of each coil unit into current loops; A ring element is uniquely determined by the ring radius, the center position of the ring, and the normal direction of the ring; The data of all the ring elements are saved as a matrix to obtain the spatial position model.

7. The method according to claim 1, characterized in that The optimization function is expressed as Among them, ||·||1 represents the L1 norm, ||·||2 represents the L2 norm, represents the target electric field distribution, J represents the current time derivative vector, represents the lead field matrix, α represents the regularization weight, and λ represents the regularization strength.

8. The method according to claim 1, characterized in that The optimal parameters of the optimization function are obtained by calculating the coordinate descent method and the alternating direction multiplier method.

Images