A visual perception decision closed-loop neuromodulation system

Abstract

The application belongs to the technical field of neural regulation, and discloses a visual perception decision closed-loop neural regulation system, which comprises an individualized functional graph construction and target positioning module, an MRI compatible targeted light stimulation interface module and a behavior-brain network feedback driven closed-loop regulation module.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of neuromodulation, and particularly relates to a visual perception decision closed-loop neuromodulation system. BACKGROUND

[0002] Although the existing neuromodulation brain-computer interface system has made certain progress in target positioning and stimulation execution, there are still problems such as poor system integration, low target positioning accuracy, and lack of closed-loop feedback mechanism, which are difficult to meet the high-precision and systematic neuromodulation demand. The core of the neuromodulation brain-computer interface is to realize the precise coupling between external stimulation and brain function network, and this process not only depends on the matching of structure and function atlas, but also needs to have a clear input-output relationship and quantifiable feedback index as the basis for parameter optimization of the behavior paradigm.

[0003] However, most current systems still lack feedback mechanisms based on neural dynamics and behavioral performance under task states, resulting in non-quantifiable modulation effects and non-iterative parameter updates, which makes the neuromodulation process prone to non-specific intervention, significantly affecting the modulation effect.

[0004] In terms of positioning, traditional neuromodulation technology relies on standard anatomical atlas (such as MNI) or resting-state fMRI for target inference (for example, CN118903688A, CN119949772A), which is difficult to reflect the neural activation characteristics of individuals under task states, resulting in limited positioning accuracy of the modulation target. In addition, gadolinium-based markers are often used in traditional MRI-guided positioning (such as CN119113154B), but there are problems such as complex preparation and potential biological toxicity, which limit their long-term stable application in neuromodulation.

[0005] In terms of stimulation navigation, the target positioning and coordinate conversion in current neuromodulation systems rely on mechanical arm guidance or camera-assisted registration schemes (such as patent application CN116650114A). However, such devices often contain metal structural elements, which are prone to radio frequency artifacts and electromagnetic interference, seriously affecting image quality and positioning accuracy, and cannot work stably in high-field magnetic resonance environments.

[0006] To meet the magnetic compatibility requirements, existing solutions (such as patent application CN221730602U) attempt to use a propulsion mechanism to achieve precise positioning of electrodes or optical fibers. However, such structures are usually large in size, with a total height of several centimeters, making it difficult to adapt to the limited space within the nuclear magnetic resonance coil. If the magnetic resonance coil is too far from the cortex, it will seriously affect the imaging quality and modulation accuracy.

[0007] From the perspective of system integration, although some structures have attempted to introduce magnetically compatible materials such as nylon (PA), polycarbonate (PC) and other polymer materials to replace traditional metal elements. However, such materials generally have low rigidity, limited processing precision, especially in micro sliding rails, screws and other propulsion structures, mechanical deformation and wear are easy to occur, leading to the gradual accumulation of hole tolerance during long-term use, ultimately making it difficult to meet the millimeter-level target navigation accuracy and repeated stimulation stability requirements, limiting its long-term application in high-precision neural regulation experiments.

[0008] In terms of closed-loop feedback regulation, current closed-loop neural regulation systems are mainly used in pathological state electrical stimulation treatment scenarios, such as seizure suppression and Parkinson tremor control (for example, patent CN118059386A). Such systems usually rely on local field potentials (LFP) as feedback signals to achieve real-time closed-loop triggering based on neural events. Although preliminary results have been achieved in disease control, the design focuses on local neural response regulation of lesion activity, making it difficult to apply to complex task scenarios that require regulation of cognitive function networks.

[0009] Existing neural regulation systems such as CN119989823B, although combined with MRI structural image navigation and task feedback mechanisms, the core stimulation means is still electrical stimulation, which has extensive current diffusion and a centimeter-level spatial action range, making it difficult to achieve local fine regulation. Although EEG feedback evaluation is introduced, EEG records surface electric field signals, and the spatial resolution is also several centimeters, which cannot accurately reflect the more fine local neural dynamics.

[0010] However, the functional organization of the cerebral cortex has obvious millimeter-level fine partition characteristics, such as the functional domain of the visual cortex. Therefore, to meet the precise regulation requirements in perception tasks, the system must have millimeter-level spatial resolution for stimulation and feedback to achieve selective regulation of specific functional domains.

[0011] In addition, existing devices are limited to single functions and cannot achieve cross-module collaboration in the same system. Therefore, there is an urgent need to develop a neural regulation system with a "function positioning-precise navigation-closed-loop regulation" whole process. The system should support operation on a high-field magnetic resonance platform to promote the development of neural regulation technology towards systematization and refinement. SUMMARY

[0012] The present application provides a visual perception decision closed-loop neural regulation system, which is based on a perception-decision behavior paradigm to construct a task-induced functional atlas, combines an MRI-compatible optical stimulation interface with a behavior-brain network closed-loop feedback mechanism, and realizes high-precision, systematic and individualized neural regulation.

[0013] To achieve the above-mentioned purposes, the present application provides the following technical solutions:

[0014] A visual perception decision closed-loop neuromodulation system, the system comprises three modules:

[0015] An individualized functional map construction and target positioning module, based on a visual perception-decision task paradigm, an individualized functional map of a region is constructed, based on the functional map and a first structural imaging containing an MRI imaging marker, an implantable interface base is positioned and registered once, and based on the functional map and a second structural imaging containing an MRI imaging marker of the quick-release grid plugboard, a stimulation path is obtained;

[0016] An MRI-compatible targeted optical stimulation interface module, comprising an implantable interface base, a quick-release grid plugboard mounted on the implantable interface base, and a step-by-step optical fiber pin kit cooperating with the quick-release grid plugboard;

[0017] A behavior-brain network feedback-driven closed-loop control module, which controls the step-by-step optical fiber pin kit to act on the target functional domain to perform neural stimulation according to the stimulation path, collects behavior and brain network signals as feedback indicators, models the stimulation parameters and feedback indicators, and updates the stimulation parameters according to the modeling results.

[0018] The visual perception decision closed-loop neuromodulation system provided by the application is based on a perception-decision task paradigm to construct an individualized functional map, combined with an MRI-compatible targeted optical stimulation interface and a behavior-brain network feedback-driven closed-loop control mechanism, and is realized at a millimeter-level spatial resolution.

[0019] The individualized functional map construction and target positioning module comprises:

[0020] Based on the visual perception-decision task paradigm, a BOLD (Blood Oxygen Level-Dependent) response is induced, functional magnetic resonance imaging signals are collected, and an individualized functional map across regions is constructed;

[0021] By laying out vitamin E capsule imaging markers on the surface of the skull and collecting T1 high-resolution structural images, a first structural imaging is obtained, and a spatially registered three-dimensional reference system is established;

[0022] Based on the functional map and the first structural imaging, the installation position and angle of the implantable interface base on the craniofacial surface are simulated and calculated;

[0023] The quick-release grid plugboard filled with MRI imaging material is inserted into the implantable interface base, a second structural imaging is performed, and a secondary positioning and registration with the functional map is performed, a spatial mapping relationship between the jack array position in the quick-release grid plugboard and the functional activation area is established, the stimulation path is fine-tuned based on the spatial mapping relationship, and an optimal stimulation path is obtained.

[0024] Wherein, the specific implementation method of constructing the individualized functional atlas across regions can be:

[0025] The individualized functional atlas construction and target positioning module introduces a visual perception-decision task paradigm with clear structure and quantifiable response as the driving condition for functional atlas construction. The task paradigm requires the subject to distinguish and make a selection response within a limited time to different visual stimuli presented on both sides. The visual stimulus parameters include contrast, color, motion direction, and space-time frequency, and the task paradigm has high controllability of task parameters and behavior relevance.

[0026] During task execution, the system collects functional magnetic resonance imaging (fMRI) signals, extracts task-induced BOLD response patterns, and constructs an individualized functional atlas covering multiple cortical regions, including visual-related cortex, parietal cortex, and prefrontal cortex. Compared with resting-state functional atlas or anatomical template, this method has the following advantages: (1) controllable visual stimulus input parameters; (2) functional response related to behavior performance; (3) dynamic quantifiable capture of neural coordination across cortical regions. The above characteristics effectively improve the task relevance and spatial precision of functional target identification, providing key basis for precise targeted stimulation in subsequent modules.

[0027] Thanks to the resolution capability of 7T ultra-high field functional magnetic resonance imaging (fMRI) superior to conventional imaging methods in spatial dimension, the voxel size can reach 1 mm or even sub-millimeter level. This system can accurately locate the millimeter-scale functional domain within the cortex, especially suitable for regions with highly differentiated functional columnar structures such as the visual cortex. In contrast, traditional electroencephalogram (EEG) technology is limited by the scalp volume conductance effect in spatial resolution, and the actual functional positioning capability is usually at the centimeter level, making it difficult to accurately analyze the local neural spatial topology.

[0028] Wherein, the specific implementation method of the first positioning registration can be:

[0029] By establishing the spatial coordinate conversion relationship between the five geometric reference points on the implantable interface base and the imaging markers, the relative coordinate parameters of the implantable interface base on the craniofacial surface are obtained, and the installation position and angle of the implantable interface base on the craniofacial surface are obtained.

[0030] Wherein, the specific implementation method of the second positioning registration can be:

[0031] The second structure imaging is used for acquiring an actual three-dimensional position distribution (spatial coordinate information) of the grid array in the magnetic resonance space, rigidly registering and performing affine transformation on the second structure imaging and the functional atlas, and establishing a spatial mapping relationship of the jack array; based on the spatial mapping relationship, an Euclidean distance from each jack center to a centroid of a target functional domain is calculated, and an optimal stimulation path is determined according to the Euclidean distance, so as to ensure that a subsequent stimulation operation is accurately applied to the functional target area.

[0032] The individualized functional atlas construction and target positioning module provided by the application constructs an individualized functional atlas by using task-state fMRI signals induced under a perception-decision task paradigm and combining high-resolution structure images, significantly improves the spatial accuracy and behavior correlation of functional positioning, and provides quantitative basis and spatial coordinate guarantee for the millimeter-level target intervention of the subsequent light stimulation module.

[0033] The MRI-compatible targeted light stimulation interface module comprises:

[0034] The implantable interface base seat is made of polyether ether ketone (PEEK), is provided with a reinforcing rib structure on an outer wall, and is internally provided with a limiting boss, and is used in cooperation with bone cement and a quick-release grid plugboard.

[0035] The quick-release grid plugboard is made of PEEK, has a regularly arranged jack array and an asymmetric limiting groove.

[0036] The step-by-step optical fiber insertion needle kit comprises a plurality of standardized insertion needles with a spacing of 0.5 mm, the insertion needles are provided with ceramic limiting structures at tail ends, and the insertion depth is 2-20 mm, and the insertion needles are matched with the jack array.

[0037] The protective cover is closed matched with the implantable interface base seat by using an internal thread connection structure.

[0038] The jack in the quick-release grid plugboard is filled with an MRI developing agent (such as the content of a vitamin E capsule), and T1 magnetic resonance structure image acquisition is combined to realize secondary spatial registration of the jack array position and the task-state functional atlas; based on the registration result, the jack plane coordinates and the insertion angle can be customized as required to meet the spatial intervention requirements of different cortical target points; the assembly supports multiple insertion needle operations and imaging calibration, has good reusability and spatial stability, and can effectively guarantee the spatial accuracy and individualized adaptation ability of the stimulation path.

[0039] The top of the quick-release grid plug-in board is provided with a threading through hole and cooperates with a nylon line lifting structure; the outer edge of the quick-release grid plug-in board is provided with three limiting concave table structures, and mechanical cooperation is realized with the limiting convex table of the inner wall of the implantable interface base; the threading through hole provided at the top of the quick-release grid plug-in board is used for plug-in operation in cooperation with the nylon line lifting structure; the mechanical cooperation between the convex tables ensures that the plug-in board is installed in the same direction and has good spatial consistency for multiple times.

[0040] The step-by-step optical fiber insertion needle kit can meet the stimulation requirements of different depths of the skin functional domain, and each insertion needle tail end is provided with a ceramic limiting structure to realize an operation accuracy of better than ±0.5 mm. The design avoids the interference of the traditional push structure on the MRI environment, takes into account the magnetic compatibility and fine control ability, and is suitable for directional intervention on different depths of the skin functional domain.

[0041] The step-by-step optical fiber insertion needle kit is used for optical stimulation on the target functional domain, and the laser can be conducted through the extended optical fiber system. The extended section can be accurately connected with the optical fiber insertion needle by using a ceramic sleeve, so as to accurately guide the laser energy to the target functional domain, realizing millimeter-level spatial resolution of neural regulation operation.

[0042] The protective cover is used for long-term implantation protection: after the stimulation experiment is completed, a non-metal protective cover is installed to close the interface to reduce the risk of infection and ensure the tissue compatibility and stability of long-term implantation of the interface.

[0043] The MRI compatible targeted optical stimulation interface module is used to perform precise neural stimulation operation in a high-field magnetic resonance environment, and has excellent spatial adaptability, magnetic compatibility and stimulation targeting accuracy. It is a key component for supporting the conversion of target positioning results of the system into an implementable stimulation path. It works cooperatively with the individualized functional atlas construction and target positioning module to accurately map the identified cortical target points in the individualized functional atlas to the operation space, and then realize spatial targeted intervention on the target region.

[0044] To adapt to the neural regulation experiment requirements in a high-field magnetic resonance environment, the total height of the MRI compatible targeted optical stimulation interface module is controlled to be within 8 mm. The flattened design can make the radio frequency coil close to the cerebral cortex, effectively improving the signal-to-noise ratio of fMRI imaging; and through compact structure design and fine control, it ensures that the millimeter-level targeted intervention on the target skin functional domain is realized under the imaging conditions with strict magnetic compatibility requirements.

[0045] The behavior-brain network feedback driven closed-loop regulation module comprises:

[0046] A neural stimulation control executor is configured to set stimulation parameters, and the output laser is optically stimulated on the target functional domain according to the stimulation path by the step-by-step optical fiber insertion needle kit.

[0047] The behavioral and brain network signal acquisition unit is used to synchronously record feedback indicators (including the amplitude of BOLD signals in task-related brain regions and the functional connectivity between regions): behavioral indicators and fMRI brain network data. The behavioral indicators include selection accuracy, reaction time and selection bias, and the fMRI brain network data includes activation amplitude and connectivity strength.

[0048] The feedback index modeling unit is used to build a regression model for the mapping relationship between stimulus parameters and feedback indices, and to train and update it round by round.

[0049] The parameter update unit is used to adjust the stimulation parameters based on the output of the regression model and update them to the neural stimulation control actuator to achieve closed-loop optimization control.

[0050] In the neural stimulation control actuator, optical stimulation of the target cortical region is performed according to set parameters, including power, frequency, and target coordinates. The laser source is a laser with a center wavelength of 1870±10nm, which is introduced into the brain region through a polymer optical fiber with a core diameter of 200μm and NA=0.22. In the feedback index modeling unit, the stimulation parameters include power, frequency, and target coordinates. The model is trained and updated round by round using multiple linear regression, Ridge regression, or XGBoost ensemble learning method based on gradient boosting. In the training and round by round updates, the fitting evaluation index includes R² value and standardized regression coefficient.

[0051] Specifically, the feedback index modeling unit is used to establish the mapping relationship between stimulus parameters and behavioral-neural feedback. This modeling unit introduces a multivariate regression analysis method, using the parameters set for each round of stimulation, including stimulus power, frequency, and target coordinates, as input independent variables; and using synchronously collected behavioral indicators, including selection accuracy, reaction time, and selection bias, and fMRI neural feedback indicators, including activation amplitude of task-related brain regions and cross-regional functional connectivity strength, as the dependent variable set to construct a regression model. The model can adopt linear or nonlinear regression, supports data updates round by round, and outputs an assessment of the impact of stimulus setting parameters on behavior and brain networks, thereby dynamically adjusting subsequent stimulus parameters.

[0052] The regulation parameter update unit includes:

[0053] The current stimulus parameter efficacy evaluation mechanism judges the effectiveness of the stimulus based on whether the change in the feedback indicator exceeds a set threshold; if the stimulus is judged to be ineffective, a parameter search and iteration mechanism is adopted.

[0054] The parameter search and iteration mechanism uses a minimum mean square error optimization algorithm to adjust the combination of stimulus parameters.

[0055] A safety threshold setting mechanism is used to limit the maximum stimulation energy and frequency to ensure the safety of the subjects; if the stimulation is judged to be invalid, a parameter search and iteration mechanism is used, and the adjusted stimulation parameter combination is used for the next round of light stimulation.

[0056] The control parameter updating unit is used to automatically update the stimulation parameters according to the modeling results, and send the new parameters to the neural stimulation control executor, to realize closed-loop control. The system can adjust the key parameters including stimulation energy and frequency according to the regression model output, to improve the individualized adaptability and control effect of the stimulation. This mechanism significantly enhances the targeting, interpretability and closed-loop regulation ability of the control strategy.

[0057] The system runs on a 7T ultra-high field magnetic resonance imaging platform, and supports millimeter-level resolution BOLD signal acquisition.

[0058] The application constructs a visual perception decision closed-loop neural regulation system, and the overall structure is composed of the following three functional modules: an individualized functional atlas construction and target positioning module, which is used to induce functional magnetic resonance imaging signals under the driving of a perception-decision task, obtain an individualized task state functional atlas, and realize millimeter-level functional domain identification and target positioning in combination with an MRI imaging marker and an image registration algorithm; an MRI compatible targeted light stimulation interface module, which is suitable for neural stimulation operation in a high field strength magnetic resonance environment, has millimeter-level targeting precision and good radio frequency coil spatial adaptability, has a compact structure, and supports high-precision optical fibers to perform targeted stimulation on a specific functional area of the cortex on a millimeter scale; and a behavior-brain network driven closed-loop regulation module, which is used to record behavior indicators and fMRI brain network responses synchronously, construct a feedback model and update stimulation parameters, and realize closed-loop optimization of the regulation parameters. The logical association between the modules is close, and the modules cooperate to realize the whole process regulation from functional positioning to accurate stimulation to closed-loop feedback under the millimeter-level spatial resolution.

[0059] The visual perception decision closed-loop neural regulation system provided by the application is suitable for neural function regulation experiments in a high field strength magnetic resonance imaging environment, and is especially suitable for millimeter-level functional domain positioning and regulation based on individualized task states. The application realizes the systematic integration of "functional positioning-accurate navigation-closed-loop regulation" under millimeter-level spatial resolution, and promotes the development of neural regulation technology in the direction of high precision, systematization and individualization.

[0060] The visual perception decision closed-loop neural regulation system provided by the application solves the following problems in existing neural regulation technology: 1) the spatial resolution of the functional positioning means is insufficient, and it is difficult to accurately identify and target the millimeter-scale functional domain of the cortex; 2) the current diffusion exists in the electrical stimulation means; and 3) there is a lack of individualized behavior-brain network feedback mechanism. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 A schematic diagram of a visual perception decision-making closed-loop neuromodulation system provided for an embodiment;

[0062] Figure 2 A schematic diagram of individualized functional domain mapping construction in an embodiment;

[0063] Figure 3 A schematic diagram of an MRI-compatible targeted optical stimulation interface module in an embodiment;

[0064] Among them, in Figure 3 : 1 is an implantable base seat, 2 is a quick-release grid plugboard, 3 is a stepped depth optical fiber pin kit, 4 is a protective cover; 21 is an asymmetric groove mark, 22 is a threading through hole, 23 is a jack array, 24 is a limiting recess; 11 is a limiting boss, 12 is a raised reinforcing rib;

[0065] Figure 4 A schematic diagram of a closed-loop feedback index. DETAILED DESCRIPTION

[0066] The visual perception decision-making closed-loop neuromodulation system provided by the embodiment includes three modules:

[0067] An individualized functional map construction and target positioning module is used to complete functional map construction and target positioning.

[0068] An MRI-compatible targeted optical stimulation interface module is used to implement precise implementation of the physical stimulation path.

[0069] A behavior-brain network feedback-driven closed-loop control module is used to dynamically optimize and adjust the stimulation parameters.

[0070] The specific implementation process of each module will be described one by one in combination with Figure 1 .

[0071] Module One: Individualized Functional Map Construction and Stimulation Target Positioning Module

[0072] This module aims to achieve millimeter-precision functional target identification and targeted stimulation path planning under high-field magnetic resonance imaging conditions: based on the visual perception-decision-making task paradigm, an individualized functional map of the region is constructed, based on the functional map and the first structural imaging containing the MRI imaging marker, the implantable interface base seat is positioned and registered once, and based on the functional map and the second structural imaging containing the MRI imaging marker of the quick-release grid plugboard, the stimulation path is obtained.

[0073] This module specifically includes the following four steps:

[0074] Step 1-1): Functional map construction based on perception-decision-making tasks

[0075] In this embodiment, the visual contrast discrimination task is used as a neural functional activation paradigm to construct the subject's individual spatiotemporal specific functional atlas in the process of perception and decision-making.

[0076] In this embodiment, the bilateral visual perception and decision-making task is used as a paradigm for constructing a functional atlas. In each test, after the subject fixates on the central point, a sinusoidal wave drifting grating is presented in the left and right visual fields. The standard stimulus on one side has a fixed contrast of 50%, and the test stimulus on the other side has a contrast gradient set between 10% and 90%. The stimulus is presented in the lower quadrant of both sides, with an eccentricity of about 4°, and the stimulus area is a circular grating with a diameter of 2°. Each contrast condition is presented in a pseudo-random order, and the conditions near the perception threshold have a higher sampling density to improve the discrimination and behavioral measurement sensitivity of the task. The subject selects the stimulus direction with higher contrast through eye movement, and the behavioral monitoring program provides rewards based on the selection results.

[0077] In this embodiment, the functional magnetic resonance data is collected using a T2* weighted gradient echo planar imaging (EPI) sequence, combined with a 7T high-field magnetic resonance imaging system and a custom coil, to achieve high temporal and spatial resolution BOLD signal recording. The key acquisition parameters include: repetition time (TR) of 2000ms, echo time (TE) of 25ms, flip angle of 90°, each scan layer thickness of 1.0mm, spatial resolution of 1.0×1.0×1.0mm³, covering 38 coronal slices to achieve whole brain coverage. The image matrix size is 96×96, the field of view (FOV) is 96mm, and the GRAPPA acceleration mode (acceleration factor 2) is used to shorten the acquisition time and improve the image signal-to-noise ratio.

[0078] As shown in a of Figure 2 Under the driving of the visual perception and decision-making task, this embodiment can stably induce blood oxygen level dependent (BOLD) signal responses in multiple brain regions, covering visual-related cortex, parietal cortex and prefrontal cortex. The task-related functional atlas constructed thereby has good spatial resolution and task specificity, and can be used as a key reference for subsequent individualized functional domain target stimulation positioning. Further, the centroid coordinates of each activated functional domain can be extracted as the spatial coordinates of the control target, to achieve millimeter-level precise positioning of the functional target and subsequent stimulation path planning.

[0079] Step 1-2): MRI imaging marker positioning and structural image acquisition

[0080] As shown in a of Figure 2As shown in b, to achieve spatial mapping between functional maps and physical stimulation pathways, three MRI imaging markers are pre-placed on the surface of the individual skull to form a stable spatial positioning reference system. In this embodiment, vitamin E soft capsules are used as imaging markers. Their contents consist of a mixture of dl-α-tocopherol acetate and vegetable oil, rich in lipid-soluble hydrocarbon chain structures (-CH2-CH3), exhibiting high proton density and short T1 relaxation characteristics. Therefore, they appear as high-signal bright spots in T1-weighted imaging sequences, providing clear imaging and demonstrating good magnetic resonance compatibility and biosafety.

[0081] The radiopaque markers are adhered to pre-designed locations on the skull using PMMA dental bone cement. The contact surfaces at the adhesion points can be marked with carbon ink for surgical reference. The spacing between the markers is designed to ensure that the three points are not collinear, thus establishing a stable three-dimensional coordinate system.

[0082] T1-weighted high-resolution structural imaging was then performed to clearly visualize the biomarkers. Specifically, a 7T ultra-high field magnetic resonance imaging system with a custom head coil was used, and the acquisition sequence parameters were as follows: MPRAGE sequence was used, with a spatial resolution of 0.5×0.5×0.5mm³, TR = 2300ms, TE = 3.61ms, TI = 1100ms, a flip angle of 7°, based on the GRAPPA parallel acceleration mode (acceleration factor 2), an acquisition matrix of 192×192, 120 layers, covering a 96mm cranial region.

[0083] like Figure 2 As shown in b, the acquired structural images can clearly display the locations of imaging markers in three-dimensional space. This imaging signal is stable and has high contrast, making it suitable as a non-invasive MRI spatial localization reference point, demonstrating good practicality in experiments. Furthermore, a stable three-dimensional spatial reference system is constructed and used as the geometric benchmark for subsequent image registration and interface installation position calculations.

[0084] Steps 1-3) - Simulation calculation and surgical installation of the implantable stimulation interface base

[0085] This embodiment aims to accurately map the target coordinates identified in the functional map to the actual surgical operation coordinates, and to perform simulation design and surgical implantation of the implantable stimulation interface base based on the spatial registration results, thereby achieving precise targeted coverage of the functional activation domain.

[0086] Spatial registration was performed between the functional atlas acquired in step 1 and the structural images with radiopaque markers acquired in step 2. Simulation software was then used to calculate the optimal installation angle and planar position of the stimulation interface on the cranial surface. Specifically:

[0087] Firstly, the fMRI functional maps obtained in Step 1 and the high-resolution T1 structural images collected in Step 2 are rigidly transformed and affinely registered using AFNI and FreeSurfer as image processing software.

[0088] After completing the image registration, the functional activation maps are projected onto the structural images with the visual markers using the Planner simulation software to calculate the optimal installation plane and angle of the implantable stimulation interface base, as shown in c of Figure 2 The simulation design needs to meet the following constraints: (1) there is no large blood vessel covering the installation area of the implantable stimulation interface base; (2) other implants are avoided; (3) the implantable stimulation interface base can completely cover the stimulation target, and a space redundancy of ≥0.5 mm is reserved to compensate for the operation error.

[0089] Further, by establishing the spatial coordinate conversion relationship between the five geometric reference points on the implantable stimulation interface base and the visual markers, a parameter set is obtained to guide the installation of the interface surgery, and the precise surgical implantation of the interface base is realized, specifically:

[0090] The key geometric points on the implantable stimulation interface base, including the center and the four-direction edges, constitute two independent spatial reference systems with the three visual markers, as shown in d of Figure 2 , where P1-P3 are three visual marker diagrams, and the red circle in the lower right corner represents the interface base, and the blue dot on the red circle represents the key geometric point of the interface base. By extracting its three-dimensional coordinates in the T1 structure image and calculating the transformation matrix between the two coordinate systems, the relative coordinate parameters of the accurate installation of the interface on the individual cranial surface can be derived, and the parameters are derived in CSV format for intraoperative reference.

[0091] Finally, according to the simulation results described above, the surgical installation of the implantable stimulation interface base is completed in a sterile environment. PMMA bone cement MRI compatible fixation material is used to adhere the implantable stimulation interface base during the operation, providing a stable installation platform for the subsequent interface module.

[0092] Step 1-4) - Grid visualization imaging for secondary positioning calibration

[0093] In this embodiment, after completing the surgical installation of the implantable stimulation interface base, a quick-release grid plug-in board filled with vitamin E capsule contents as MRI visualization material is inserted into the interface base. The grid plug-in board has a regular array structure, and the spatial geometric relationship of all plug-in hole positions has been calibrated. The visualization material inside the plug-in hole can present a clear high-intensity signal in the T1 weighted structure image, forming a recognizable bright spot array. A new structure image is collected to obtain the actual three-dimensional position distribution of the grid array in the magnetic resonance space (see Figure 2(e) Then, the structural diagram is rigidly registered and affine transformed with the task-related functional map obtained in step 1 to establish the spatial correspondence between the mesh hole coordinates and the functional activation region (see e). Figure 2 (f) Based on this mapping relationship, the system automatically calculates the Euclidean distance from the center of each jack to the centroid of the target functional domain, and selects the nearest jack or jack combination as the actual stimulation path, thereby achieving individualized fine-tuning and precise coverage of the functional target area.

[0094] The secondary positioning and calibration step in this embodiment optimizes the registration accuracy between the aperture array and the functional map in the grid plate by spatially arranging the imaging markers and the stimulation path. This is a key step in achieving millimeter-level stimulation targeting and positioning.

[0095] Module 2: MRI-compatible Targeted Photostimulation Interface Module

[0096] This embodiment provides a targeted neurostimulation interface module suitable for high-field magnetic resonance environments. It features a compact structure, good magnetic compatibility, and the ability to perform millimeter-level neurostimulation manipulation on specific functional domains of the cortex. This module combines the individualized functional map and stimulation target coordinate information obtained in Module 1 to convert the map space into the actual operational space, achieving precise fiber optic insertion and stimulation path control. Figure 3 As shown in 'a', it mainly consists of the following four structures:

[0097] 2-1) Implantable interface base 1

[0098] like Figure 3 As shown in Figure c, the implantable interface base 1 is the core support platform of the interface system. Its design goal is to achieve long-term implantation and precise spatial positioning while ensuring MRI compatibility and mechanical stability. Its main structural features include:

[0099] Materials and processing techniques: Polyetheretherketone (PEEK) is used, and the material is formed by high-precision CNC machining. This material has high strength, low magnetic susceptibility, and excellent biocompatibility, making it suitable for long-term cranial implantation.

[0100] The mechanically stable structure features four evenly distributed raised reinforcing ribs 12 on the outer wall of the implantable interface base 1, which can form a mechanical interlocking structure with dental bone cement during surgery, significantly improving adhesion and preventing dislocation.

[0101] The embedded limiting structure has three limiting protrusions 11 inside the base, which are used to form a stable fit with the quick-release mesh insert 2 to prevent the quick-release mesh insert 2 from rotating or tilting during use and improve the repeatability of the insertion hole positioning.

[0102] Implantation, the size and geometry of the implantable interface base 1 support individual customization, and can be combined with ceramic micro bone nails and bone cement to fix in the craniotomy area, to ensure the long-term stability of the stimulation interface and the consistency of the spatial reference.

[0103] 2-2), quick-release grid plugboard 2

[0104] As shown in b in the embodiment of the quick-release grid plugboard 2 is designed to be installed on the implantable base 1, as a detachable component of the stimulation interface, with the functions of developing secondary calibration and stimulation targeting, with the advantages of precise positioning and quick replacement. Figure 3

[0105] The top of the plugboard is provided with two symmetrical through holes 22 on both sides, which can be used to pass nylon wire as a lifting structure, so as to quickly extract the quick-release grid plugboard 2 in the space-limited area, and improve the operation convenience and module reuse efficiency.

[0106] In order to ensure that the quick-release grid plugboard 2 is installed in the same direction every time, an asymmetric groove structure 21 is provided on the outer edge of the quick-release grid plugboard 2 in the vertical direction, which is used as a direction identification mark and cooperates with the registration boss structure of the base to ensure the spatial consistency of the quick-release grid plugboard 2 during multiple installations, and avoid stimulation coordinate error caused by wrong direction.

[0107] The quick-release grid plugboard 2 is provided with a regular array of jack array 23 in the center, and each jack is provided with a limiting structure inside for fixing the inserted ceramic optical fiber pin, preventing the pin from slipping axially during the experiment, and improving the spatial stability and repeatability of the targeted stimulation.

[0108] After the functional atlas registration and target identification of module one are completed, MRI developing materials such as vitamin E capsule contents can be injected into the jack to recalibrate the plugboard position through T1 structural image, so as to realize the secondary spatial registration of the grid array and the functional atlas, and accurately guide the stimulation path. In the formal stimulation stage, the selected jack angle and coordinate can also be customized according to the individual target to match the required stimulation path, so as to realize the millimeter-level personalized targeted intervention.

[0109] In addition, the outer edge of the quick-release grid plugboard 2 is provided with three limiting concave platforms 24 structures, which form mechanical cooperation with the limiting boss 11 in the inner wall of the base. During installation, this structure can realize the quick alignment and positioning of the quick-release grid plugboard 2, avoid rotation deviation and axial misplacement, and effectively improve the assembly efficiency and overall structural stability of the modular interface.

[0110] 2-3), step-by-step optical fiber pin kit 3

[0111] ​To meet the demand for precise stimulation of different depth cortical functional areas, the embodiment is equipped with a set of step-by-step optical fiber needle kit 3, which contains multiple step-by-step needles with a standardized pitch of 0.5 mm, covering a depth range of 2-20 mm. Each needle is precisely machined and assembled with a ceramic limiting structure at the tail end. The insertion depth is determined by the length of the needle itself, avoiding magnetic compatibility interference caused by traditional push mechanisms, ensuring stable operation in a limited magnetic resonance coil area, and making the insertion error less than ±0.5 mm;

[0112] The needle body is composed of polymer-coated quartz optical fiber, which has excellent light guiding performance and structural rigidity. The head end is optically polished, with uniform light output, which can efficiently conduct laser light to achieve optical stimulation of the cortical target. The tail end ceramic limiting structure is precise in size and can be matched with the jack of the quick-release grid plugboard to form stable fixation.

[0113] In this embodiment, the optical stimulation is based on laser output, which is coupled into the needle through the optical fiber path, and finally focuses the laser energy on the millimeter-scale functional domain, realizing high spatial precision of neural regulation without current diffusion risk.

[0114] 2-4), protective cover 4

[0115] The protective cover 4 in this embodiment is used to close the interface during the non-stimulation stage to prevent contamination and foreign matter from entering, thereby improving the stability and biological safety of long-term implantation. The protective cover 4 is made of polyether ether ketone (PEEK) material and is designed with internal threads that can be tightly screwed with the implantable base 1 to form a good seal. The protective cover 4 has a vertical groove on the top of the cover body, which can be easily screwed on or off with a plastic ruler tool, suitable for operation under space-limited conditions. The overall structure is compact, with good magnetic resonance compatibility and reusability.

[0116] All interface components can be sterilized with ethylene oxide to meet the use and safety requirements under long-term experimental conditions.

[0117] The MRI-compatible targeted optical stimulation interface module structure provided by the embodiment can correspond to the interface-functional area registration result established in module one, thereby realizing complete mapping and execution from individualized atlas to spatial targeting path.

[0118] Module three: behavior-brain network feedback driven closed-loop regulation module

[0119] The module described in this embodiment aims to, under the visual perception-decision-making task paradigm, construct a stimulation parameter regulation model by real-time acquisition of behavior and brain network feedback signals, and realize a closed-loop control mechanism of stimulation-feedback-optimization. This module is composed of the following four functional units:

[0120] Unit 1: neural stimulation control executor

[0121] The nerve stimulation control executor in the embodiment is used to set parameter control instructions and perform nerve stimulation operation on the target cortical area according to the parameter control instructions. Taking near-infrared light stimulation as an example, the system configuration center is configured with a laser with a center wavelength of 1870±1 nm as a stimulation source. The laser is conducted through a flexible polymer optical fiber with a core diameter of 200 μm and a numerical aperture (NA) of 0.22, and is accurately introduced into the target functional domain through the MRI-compatible targeting light stimulation interface of module 2, so as to realize the spatial alignment of the stimulation site and the target point in the task-induced functional atlas. The laser output parameters can be flexibly modulated according to the preset power, frequency and pulse width, and are accurately controlled by the control software through a TTL signal or a digital communication interface, so as to ensure that the stimulation waveform is highly synchronized with the experimental task.

[0122] Unit 2: Behavior and brain network signal acquisition

[0123] As shown in Figure 4 , the behavior and brain network signal acquisition unit in the embodiment is designed to synchronously acquire multi-modal feedback data during the execution of the visual perception-decision task, so as to provide quantitative basis for subsequent feedback modeling and stimulation parameter updating.

[0124] In terms of behavior feedback acquisition, the behavioral feedback includes quantifiable behavior indicators such as selection accuracy, reaction time and selection bias. The embodiment is equipped with an MRI-compatible infrared eye tracking device (Eyelink 1000 Plus, SRResearch) with a sampling frequency of 250 Hz, which is used to record the eye movement behavior trajectory of the subject during the task. The embodiment sets a fixed fixation point and a selection response window to calculate the selection accuracy, reaction time and left-right selection bias of each trial, and the results are recorded and output in units of trials.

[0125] In terms of brain network feedback acquisition, based on functional magnetic resonance imaging (fMRI) signals, the activation amplitude and cross-regional functional connection strength of the task-related brain area are extracted. The embodiment synchronously acquires whole brain BOLD signals on a 7T high-field magnetic resonance platform, uses a T2* weighted gradient echo EPI sequence (TR=2000 ms, TE=25 ms, flip angle=90°, voxel resolution 1 mm isometric, 38 layers cover the whole brain), to ensure good temporal and spatial resolution for the task-related brain area.

[0126] The acquired behavior and brain network data will be aligned according to the trial timestamp and bound with each round of stimulation setting parameters, and stored in CSV standardized format for calling by the feedback modeling unit.

[0127] The embodiment can automatically detect abnormal reactions or low-quality signals, set data cleaning rules, and exclude trials with more than a threshold of lost fixation, to improve the reliability of subsequent model construction.

[0128] The synchronous data acquisition capability of the embodiment ensures that multi-dimensional and quantifiable behavior-brain network feedback indicators are obtained after each round of stimulation, constituring a key data support path in the system closed-loop regulation mechanism.

[0129] Unit 3: Feedback indicator modeling unit

[0130] The feedback indicator modeling unit in the embodiment is used to establish a functional relationship between the neural stimulation parameters and the behavior performance and brain network feedback obtained synchronously, so as to realize quantitative evaluation and adjustment of the stimulation effect. The unit is a key calculation module for realizing the core logic of the closed-loop regulation, and its main functions include data preprocessing, feature extraction, regression modeling and model updating.

[0131] The embodiment first standardizes the data of each round of stimulation experiment. The stimulation parameter set of each experiment includes: light stimulation power, stimulation frequency, target point three-dimensional coordinates and stimulation execution time as additional meta information, which are used as independent variables of the model.

[0132] At the same time, the embodiment synchronously reads the feedback indicator set corresponding to each round from the behavior and brain network signal acquisition unit as the dependent variable of the model. The set includes: behavioral indicator dimensions, correct selection rate, reaction time, selection bias probability; brain network feedback dimensions, BOLD signal amplitude of activated regions under the behavior paradigm, and functional connectivity index of each region to other regions.

[0133] The above variables are standardized and input into the modeling process. The modeling unit can use multiple regression algorithms to construct the mapping relationship between the stimulation parameters and the feedback indicators. The basic form is a multiple linear regression model. When there is a nonlinear relationship in the feedback structure, it can be switched to Ridge regression or XGBoost integrated learning method based on gradient boosting.

[0134] The regression model training is performed in a round-by-round updating manner: after each round of stimulation experiment, the system appends the new stimulation-feedback data pair to the historical data to refit the model, and calculates the goodness of fit of each predicted dependent variable, such as R², and the parameter importance coefficient, such as the standardized β value, to evaluate the regulation effect of different stimulation parameters on the behavior and brain network state.

[0135] In addition, the embodiment also provides a regulation threshold and indicator weight setting mechanism. For example, if a specific feedback indicator, such as the activation intensity of the visual area, is the primary regulation target, the system can increase the weight of the indicator in the model evaluation function, so that the parameter optimization strategy converges in this direction.

[0136] The final output model not only evaluates the regulatory effect of the current stimulation setting, but also provides directional suggestions for stimulation parameter updates to unit 4, such as "whether increasing the stimulation power will help improve the correct selection rate", so as to realize model-driven closed-loop regulation.

[0137] In summary, by introducing statistical modeling and round-by-round updating mechanisms, the embodiment structures, quantifies and predicts the effects of neural stimulation, providing technical support for efficient and individualized neural intervention strategies.

[0138] Unit 4: Regulation parameter update

[0139] The regulation parameter update unit in the embodiment is used to dynamically adjust the neural stimulation parameters according to the output results of the behavior-neural feedback regression model established in unit 3, so as to realize a closed-loop neural regulation strategy for individual states.

[0140] The regulation parameter update unit receives the modeling estimation results and performs the following two types of operations:

[0141] 1. Current parameter effectiveness evaluation: the system first judges whether the current stimulation achieves the expected regulatory effect according to the comparison between the model prediction value and the actual feedback data. The evaluation criteria that can be set include: change of behavior index: change of correct selection rate by more than 2%; change of neural index: significant change (p<0.05) of BOLD amplitude in the target functional domain; and enhancement of functional connection strength between functional domains: significantly higher than the baseline round;

[0142] 2. Parameter search and update: if the expected effect is not achieved, the system will perform parameter iterative search within the current parameter neighborhood. For example, based on the least mean square error optimization algorithm, the stimulation power, frequency, or target point coordinates are adjusted to form the recommended stimulation parameter configuration for the next round.

[0143] The updated stimulation parameters are transmitted to the laser control interface in unit 1 through TTL or digital communication by the control software after user evaluation, for setting the stimulation scheme for the next round of experiment. This process forms a complete closed-loop regulation path, supporting feedback-based multi-round parameter optimization iteration.

[0144] During the parameter update process, the embodiment also supports setting safety range limit conditions (such as maximum stimulation energy not exceeding 1.0 J / cm² and frequency not higher than 400 Hz) to ensure the safety of the subjects and avoid stimulation overload or cumulative risks.

[0145] This module realizes the closed-loop regulation and iterative optimization of neural stimulation parameters in the state of visual perception-decision-making tasks, and is an important part of the system to achieve precise neuromodulation. Based on the completion of individualized functional atlas construction and target positioning in module one and the execution of targeted stimulation operation in module two, the system synchronously collects behavioral indicators and functional magnetic resonance (fMRI) brain network data through module three, constructs a quantitative stimulation effect evaluation model, and dynamically updates the stimulation parameters based on this, forming a closed-loop regulation path of stimulation-feedback-optimization.

Claims

1. A visual perception decision-making closed-loop neuromodulation system, characterized in that, The system comprises three modules: The individualized functional map construction and target positioning module is based on a visual perception-decision task paradigm to construct an individualized functional map of a region, based on the functional map and first structural imaging containing an MRI imaging marker to perform a primary positioning registration of an implantable interface base, and based on the functional map and second structural imaging containing an MRI imaging marker of a quick-release grid plug-in board to perform a secondary positioning registration to obtain a stimulation path; The MRI-compatible targeted optical stimulation interface module comprises an implantable interface base, a quick-release grid plug-in board mounted on the implantable interface base, and a step-by-step optical fiber pin kit cooperating with the quick-release grid plug-in board; The behavior-brain network feedback-driven closed-loop regulation module controls the step-by-step optical fiber pin kit to act on a target functional domain to perform neural stimulation according to the stimulation path, collects behavior and brain network signals as feedback indicators, models the stimulation parameters and the feedback indicators, and updates the stimulation parameters according to the modeling results; The individualized functional map construction and target positioning module comprises: Based on the visual perception-decision task paradigm, a BOLD response is induced, functional magnetic resonance imaging signals are collected, and an individualized functional map across regions is constructed; By arranging vitamin E capsule imaging markers on the surface of the skull and collecting T1 high-resolution structural images, first structural imaging is obtained, and a spatially registered three-dimensional reference system is established; Based on the primary positioning registration of the functional map and the first structural imaging, the installation position and angle of the implantable interface base on the craniofacial surface are simulated and calculated; The quick-release grid plug-in board filled with MRI imaging material is inserted into the implantable interface base, second structural imaging is performed, and secondary positioning registration is performed with the functional map to establish a spatial mapping relationship between the jack array position in the quick-release grid plug-in board and the functional activation area, and the stimulation path is fine-tuned based on the spatial mapping relationship to obtain an optimal stimulation path.

2. The visual perception decision loop neuromodulation system of claim 1, wherein, In the primary positioning registration, the spatial coordinate conversion relationship between the five geometric reference points on the interface implantable interface base and the imaging markers is established to obtain the relative coordinate parameters of the implantable interface base on the craniofacial surface.

3. The visual perception decision loop neuromodulation system of claim 1, wherein, In the secondary positioning registration, rigid body registration and affine transformation are performed between the second structural imaging and the functional map to establish the spatial mapping relationship of the jack array; Based on the spatial mapping relationship, the Euclidean distance from each jack center to the centroid of the target functional domain is calculated, and the optimal stimulation path is determined accordingly.

4. The visual perception decision loop neuromodulation system of claim 1, wherein, The MRI-compatible targeted optical stimulation interface module comprises: The implantable interface base is made of polyether ether ketone (PEEK), has a reinforcing rib structure on the outer wall, and has a limiting boss inside, which is used in cooperation with bone cement and a quick-release grid plug-in board; The quick-release grid plug-in board is made of PEEK and has a regular arrangement of jack array and an asymmetric limiting groove; The step-by-step optical fiber pin kit contains multiple standardized pins with a spacing of 0.5 mm, and the tail end of the pin is provided with a ceramic limiting structure with an insertion depth of 2-20 mm, which cooperates with the jack array; The protective cover is closed with the implantable interface base by using an internal thread connection structure.

5. The visual perception decision loop neuromodulation system of claim 4, wherein, The top of the quick-release grid plug-in board is provided with a threading through hole and cooperates with a nylon line lifting structure; the outer edge of the quick-release grid plug-in board is provided with three limiting concave table structures, and the limiting convex table structures in the inner wall of the implantable interface base realize mechanical cooperation.

6. The visual perception decision loop neuromodulation system of claim 1, wherein, The behavior-brain network feedback driven closed-loop regulation module comprises: A nerve stimulation control executor is configured to set stimulation parameters, and output laser is optically stimulated to a target functional domain according to a stimulation path through a stepping optical fiber needle kit. A behavior and brain network signal acquisition unit is configured to synchronously record feedback indicators, including behavioral indicators and fMRI brain network data, wherein the behavioral indicators include correct selection rate, response time and selection bias, and the fMRI brain network data includes activation amplitude and connection strength. A feedback indicator modeling unit is configured to construct a regression model for a mapping relationship between stimulation parameters and feedback indicators, and to train and update the regression model round by round. A regulation parameter updating unit is configured to adjust stimulation parameters according to an output result of the regression model and update the stimulation parameters to the nerve stimulation control executor, so as to realize closed-loop optimal control.

7. The visual perception decision closed-loop neural regulation system according to claim 6, wherein In the nerve stimulation control executor, the stimulation parameters include power, frequency and target point coordinates; in the feedback indicator modeling unit, a multiple linear regression, a Ridge regression or an XGBoost integrated learning method based on gradient boosting is used for model training and round-by-round updating, and in the training and round-by-round updating, fitting evaluation indicators include R² value and standardized regression coefficient.

8. The visual perception decision loop neuromodulation system of claim 7, wherein, The regulation parameter updating unit comprises: A current stimulation parameter effectiveness evaluation mechanism is configured to judge stimulation effectiveness according to whether a feedback indicator change amplitude exceeds a set threshold value; A parameter search and iteration mechanism is configured to adjust a stimulation parameter combination by using a least mean square error optimization algorithm; A safety threshold setting mechanism is configured to limit maximum stimulation power and frequency, and to guarantee subject safety; If it is judged that the stimulation is invalid, the parameter search and iteration mechanism is used, and the adjusted stimulation parameter combination is used for next round of optical stimulation.

9. The visual perception decision loop neuromodulation system of any of claims 1-8, wherein, The system is operated on a 7T ultra-high field magnetic resonance imaging platform, and supports millimeter-level resolution BOLD signal acquisition.

Images