Mouse motor training system for invasive brain-computer interface research

Abstract

The application belongs to the field of brain-computer interface, and particularly relates to a mouse motion training system for invasive brain-computer interface research, which comprises a training platform, a main control module, a power supply system, a reward module, a visual induction module, a physiological information acquisition module and a monitoring and processing module; the power supply module and the main control module are integrated in a box as a support platform structure in the training platform, a small animal fixing device and a traction component are arranged on the top of the box, and the structure is more compact. The main control module uniformly dispatches the visual induction, the stepping motor traction, the capacitive touch detection, the water reward release, the multi-axis mechanical measurement and the neural signal acquisition and other multifunctional modules, so that the synchronous acquisition, real-time processing and closed-loop interaction of cross-modal data are realized. The application significantly improves the integrity, quantification level and repeatability of brain-computer interface experiments.

Description

Technical Field

[0001] This invention belongs to the field of brain-computer interfaces, specifically relating to a mouse motor training system for invasive brain-computer interface research. Background Technology

[0002] Neuroscience and brain-computer interface research typically requires systematic behavioral training of laboratory mice to establish the correspondence between neural signals, motor outputs and task feedback, thereby gaining a deeper understanding of motor intentions, neural regulation and behavioral mechanisms.

[0003] In existing technologies, some mouse behavior training platforms are mainly used for judging mouse behavior in free-movement scenarios. For example, invention patent CN120713073A discloses a mouse behavior judgment platform based on auditory stimulation induction, which focuses on guiding animal behavior through sound signals under free-movement conditions. However, it does not integrate active interaction modules such as mechanical traction and tactile feedback, nor does it construct a closed-loop control system synchronized with invasive neural signal acquisition. Another type of technical solution introduces visual stimulation and task switching mechanisms to enrich behavioral paradigms. For instance, utility model patent CN221749314U provides a mouse behavior training device with a visual stimulation and channel switching structure. Although it introduces visual guidance and path selection mechanisms, it still mainly focuses on discrete behavioral selection tasks, lacking integration of mechanical traction control, real-time mechanical detection, capacitive tactile feedback, and signal acquisition circuits deeply coupled with brain-computer interfaces.

[0004] In invasive brain-computer interface (BCI) experiments, researchers not only need to collect neural signals from the cortex or deep brain regions, but also simultaneously record information such as mouse movement, traction force changes, tactile feedback, and behavioral videos, and achieve multi-module collaborative operation and unified scheduling. However, existing devices generally suffer from low integration and lack of data detection.

[0005] Therefore, it is necessary to provide an invasive mouse training platform that integrates visual induction, mechanical traction, tactile feedback, water reward, power supply and drive circuitry, and supports multi-source behavioral monitoring, in order to meet the needs of refined research in brain-computer interface experiments. Summary of the Invention

[0006] The purpose of this invention is to provide a mouse motor training system for invasive brain-computer interface research, addressing the shortcomings of the prior art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A mouse motor training system for invasive brain-computer interface research, characterized in that it includes: a training platform, a main control module, a power supply system, a reward module, a visual induction module, a physiological information acquisition module, and a monitoring and processing module;

[0009] The training platform includes a housing; the top of the housing is a support platform, and a fixing device for securing the animal to be tested is installed in the middle of the support platform; a water tank and a traction module are located outside the housing; the water tank serves as a liquid supply, counterweight, or working condition simulation auxiliary component; the traction module includes a stepper motor, an elastic traction structure, and a force sensor; wherein:

[0010] The stepper motor's control terminal is connected to the main control module to receive the excitation timing control signal provided by the main control module, and generates traction force based on the excitation timing control signal to transmit to the elastic traction structure;

[0011] One end of the elastic traction structure is connected to the output end of the stepper motor, and the other end is fixed to the forelimb of the animal under test. It is used to transmit traction force and provide flexible cushioning.

[0012] A force sensor is placed on the traction force transmission path between the stepper motor and the forelimb of the animal under test to detect the magnitude of the traction force;

[0013] The main control module is located inside the housing. Its stepper motor output is connected to the stepper motor control terminal, its touch signal input is connected to the signal output of the capacitive touch sensor in the reward module, its water pump control output is connected to the water pump control terminal of the reward module, its communication terminal is connected to the host computer in the monitoring and processing module, and its synchronization signal input is connected to the synchronization signal output of the visual guidance module. The main control module is configured as follows:

[0014] It receives touch signals sent by a capacitive touch sensor. When the touch signal is a high-level signal, it uses it as a water pump control signal and sends it to the water pump.

[0015] It receives the synchronization signal sent by the visual guidance module and generates an excitation timing control signal to drive the stepper motor to perform traction action based on the synchronization signal.

[0016] Receive training instructions from the host computer;

[0017] The reward module includes a capacitive touch sensor and a water pump; the capacitive touch sensor is fixedly installed in a position within the training platform that the animal's forelimbs can reach, and is used to generate a touch signal when the animal touches it and send it to the main control module; the water pump is used to power on and output liquid rewards according to the received water pump control signal.

[0018] The visual induction module is used to present visual stimulation signals to small animals and generate a synchronization signal when the visual stimulation begins and send it to the main control module.

[0019] The output end of the physiological information acquisition module is connected to the physiological signal input end of the host computer in the monitoring and processing module, and is used to acquire the brain nerve electrical signals of the animal under test and transmit them to the monitoring and processing module.

[0020] The monitoring and processing module includes a host computer and a camera, used for behavior recording and data synchronous analysis;

[0021] The power module is also located inside the enclosure, and its power output terminal is connected to the power input terminals of the training platform, main control module, reward module, visual guidance module, physiological information acquisition module and monitoring and processing module, respectively, to supply power to each module.

[0022] Furthermore, the elastic traction structure includes a spring and a flexible traction element; one end of the spring is connected to the output end of the stepper motor, and the other end is connected to the forelimb of the small animal through the flexible traction element.

[0023] Furthermore, the stepper motor is a 28BYJ-48 type stepper motor, which is connected to the main control module through a ULN2003 driver board and driven in a four-phase eight-beat control mode; the spring has a length of 300 mm, a wire diameter of 0.3 mm, and an outer diameter of 2 mm; the force sensor is a ZNZL-V-10 type tension / compression sensor with a range of 10N and a sensitivity of 2.0mV / V; the tension / compression sensor is used to generate a differential voltage signal proportional to the magnitude of the traction force during traction and send it to the main control module to generate traction force monitoring data.

[0024] Furthermore, the main control module adopts an STM32F103VET6 microcontroller;

[0025] The main control module establishes bidirectional data communication with the host computer via the JDY31 Bluetooth module, based on the Bluetooth 3.0 protocol and serial port transparent transmission mode. The transmitted monitoring data includes: stepper motor running status, traction monitoring data, capacitive touch trigger information, and water output time of the reward module.

[0026] 5. The training system according to claim 4, characterized in that the main control module uses an external interrupt method to detect the touch signal sent by the capacitive touch sensor; the main control module is equipped with a timer controller, which controls the high-level duration of the water pump control signal through a timer to adjust the water output.

[0027] Furthermore, the capacitive touch sensor adopts a double-sided TTP223 type; the water pump drive circuit adopts an NCE4080K type N-channel enhancement-mode MOSFET to form a low-end switching drive structure, the gate of the MOSFET is connected to the water pump control output terminal through a current-limiting resistor, the source is connected to system ground, and the drain is connected to the negative terminal of the water pump; the main control module controls the water pump output by controlling the high-level duration of the MOSFET gate signal or by using PWM modulation.

[0028] Furthermore, the visual induction module includes a ceramic display screen and visual stimulus generation software;

[0029] The visual stimulus generation software is used to parameterize parameters such as stimulus intensity, flashing frequency, stimulus duration, display position, and motion direction, and to generate various visual stimulus signals based on the parameter configuration; the visual stimulus signals include light spot stimuli that flash at a fixed frequency, geometric shapes that move in a preset direction and speed, and target stimulus shapes that periodically change color or brightness.

[0030] The ceramic crystal display screen is used to receive and play visual stimulus signals;

[0031] The stimulation start and termination times of the visual induction module are synchronized with the timing of the main control module, and the traction module and reward module are triggered according to the preset timing sequence to construct a closed-loop training control process of visual induction, behavioral response, neural signal acquisition, traction and reward feedback.

[0032] Furthermore, the physiological information acquisition module includes an acquisition device, an operational amplifier, and a filter processor;

[0033] The acquisition device is an invasive microelectrode array implanted in the sensory or motor cortex of the animal under test, used to acquire the firing signals of a single neuron or a cluster of neurons, i.e., neural electrical signals.

[0034] The input terminal of the operational amplifier is connected to the output terminal of the acquisition device, and is used to receive neural electrical signals and amplify them.

[0035] The input terminal of the filter processor is connected to the output terminal of the operational amplifier, and the output terminal is connected to the physiological signal input terminal of the host computer; it is used to filter the amplified neural electrical signal; the filtering process includes high-pass filtering and low-pass filtering, wherein the cutoff frequency of the high-pass filter is set between 10Hz and 100Hz, and the cutoff frequency of the low-pass filter is set between 100Hz and 1000Hz.

[0036] Furthermore, the camera is connected to a host computer via a USB or serial video output interface for collaborative recording of behavioral video; the host computer is configured as follows:

[0037] Receive data information from the main control module and send training instructions to the main control module;

[0038] It receives neural electrical signals, identifies and decodes them, and displays in real time the subcortical EEG data of small animals and the changes in surface muscle potential in specific brain regions;

[0039] Receive traction monitoring data and display and store it in real time;

[0040] The system displays the execution time of each module's tasks and system event logs in a timeline format.

[0041] Furthermore, the power supply module consists of a main power supply and an LM2596 step-down regulator module connected after the main power supply; the main power supply uses an 18650 6S lithium battery pack with a nominal output voltage of 22.2 V; the LM2596 step-down regulator module receives the 22.2 V voltage and converts it into a stable 5 V DC voltage, which powers the stepper motor, camera, main control module, and other peripheral modules; the output current range of the LM2596 step-down regulator module is 2 A to 3 A, and the conversion efficiency is 85%–90%.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The exercise training system of the present invention uses a box as a support platform structure, integrates the power module and the main control module inside the box, and sets the small animal fixing device and traction component on the top of the box, making the structure more compact.

[0044] 2. This invention utilizes a main control module to uniformly schedule multiple modules, including visual induction, stepper motor-driven traction, capacitive touch detection, water reward release, multi-axis mechanical measurement, and neural signal acquisition. All modules operate on the same main control clock, avoiding the drift of independent module clocks and achieving synchronous acquisition of cross-modal data. The interrupt handling mechanism ensures timely triggering of reward behavior, while researchers can observe the mouse's state in real time, enabling real-time processing and closed-loop interaction. All data in the feedback loop can be quantitatively measured, and the standardized experimental procedure reduces variations caused by human intervention. This significantly improves the completeness, quantification level, and reproducibility of brain-computer interface experiments.

[0045] 3. This invention integrates a pattern recognition unit based on neural electrical signal matching into the system to identify "pseudo-learning" behaviors in animals online. This solves the problem of traditional training systems that only reward or punish based on the correctness of behavioral responses, failing to distinguish whether the response stems from the animal's true understanding of the rules or is accidental, guesswork, or unconscious, thus improving training efficiency and success rate. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the overall structure of the sports training system of the present invention;

[0047] Figure 2 This is a diagram showing the connection structure between the main control module and each module of the present invention;

[0048] Figure 3 This is a schematic diagram of the driving principle of the stepper motor traction module of the present invention;

[0049] Figure 4 This is a schematic diagram of the wiring between the power supply and the voltage regulator module in an embodiment.

[0050] Figure 5 This is a schematic diagram of the mouse training process in an example. Detailed Implementation

[0051] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0052] like Figure 1 As shown in this embodiment, a mouse motor training system for invasive brain-computer interface research includes: a training platform, a traction module, a main control module, a power supply system, a reward module, a visual induction module, a physiological information acquisition module, and a monitoring and processing module. The reward module includes a capacitive touch sensor and a water pump, and the monitoring and processing module includes a host computer and a camera. These modules cooperate to complete the mouse motor training.

[0053] The training platform includes a box; the top of the box is a support platform, and a fixation device for securing the mouse is located in the middle of the support platform. A water tank and a traction module are located outside the box; the water tank serves as a liquid supply, counterweight, or working condition simulation auxiliary component. The traction module includes a stepper motor, an elastic traction structure, and a force sensor. Wherein:

[0054] The stepper motor is a 28BYJ-48 type, with a rated operating voltage of 5V and a starting torque of not less than 34 mN·m. The basic step angle of a single motor body is 5.625°. Combined with an internal 1:64 reduction gear structure, the equivalent step angle of its output shaft is approximately 0.088°, thus achieving high-precision angle control. The stepper motor is connected to the main control module via a ULN2003 driver board. The ULN2003 integrates a multi-channel Darlington transistor array, providing sufficient current amplification and reverse freewheeling protection to the stepper motor coils, ensuring stable motor operation.

[0055] like Figure 3As shown, in this embodiment, the 28BYJ-48 stepper motor adopts a 4-1-8 step control mode, also known as half-step drive control, to control the stepper motor. The 4-1-8 step control refers to a driving method that combines single-phase excitation (4 steps) with dual-phase excitation (8 steps), with the driving sequence alternating between single-phase energization and simultaneous energization of two adjacent phases. Compared with traditional full-step drive, 4-1-8 step control halves the actual step angle of the motor, improving the smoothness of rotation and position resolution while ensuring torque continuity, effectively reducing vibration and noise of the motor at low speeds. By changing the order of the excitation sequence, the motor can rotate forward and backward; by adjusting the beat frequency, precise control of the motor speed can be achieved.

[0056] The elastic traction structure includes a spring and a flexible traction element. One end of the spring is connected to the output end of the stepper motor, and the other end is connected to the forelimb of the small animal via the flexible traction element. In this embodiment, the spring is 300 mm long, has a wire diameter of 0.3 mm, and an outer diameter of 2 mm. The spring mainly serves as a flexible buffer and provides restoring force in the traction structure. When the stepper motor rotates forward, it pulls and stretches the spring. The tension of the spring acts on the mouse's forelimb via the flexible traction element, causing the mouse's forelimb to move until it touches the touch module. When the stepper motor rotates forward, it pulls and stretches the spring, applying traction force to the mouse's forelimb through the spring and the flexible traction element, causing the mouse's forelimb to move and touch the touch module. When the stepper motor rotates in the reverse direction to reset, the pulling effect is removed, the spring releases its stored elastic potential energy, and retracts under its own elastic restoring force, causing the flexible traction element and the mouse's forelimb to reset synchronously, allowing it to smoothly return to its initial position. By introducing a spring, the impact of rigid traction can be avoided, significantly improving the safety and smoothness of the traction action.

[0057] The force sensor is positioned along the traction force transmission path between the stepper motor and the forelimb of the small animal under test, and is used to detect the magnitude of the traction force. This embodiment uses a ZNZL-V-10 tension / compression sensor with a range of 10 N and a sensitivity of 2.0 mV / V, which can meet the high-precision measurement requirements of the traction force of the small animal's forelimb. The ZNZL-V-10 tension / compression sensor operates based on the resistance strain effect: when the sensor is subjected to tension or compression, the internal strain gauge undergoes micron-level elastic deformation with the load, thus changing its resistance value. The strain gauges are typically connected in a Wheatstone bridge configuration. When an external force causes the bridge circuit to become unbalanced, a differential voltage signal proportional to the magnitude of the external force is output. This differential voltage signal is amplified and converted from analog to digital before being transmitted to the main control board for acquisition and processing, generating traction force monitoring data which is then sent to the host computer, thereby realizing real-time monitoring and data uploading of the force on the mouse's forelimb during traction.

[0058] like Figure 2As shown, the main control module, as the core component of the sports training system, is configured as follows:

[0059] (1) Receive the synchronization signal provided by the visual guidance module, generate the excitation timing control signal according to the synchronization signal, and control the stepper motor to perform mechanical traction action;

[0060] (2) Receive the touch signal sent by the capacitive touch sensor, generate a water pump control signal based on the touch signal and send it to the water pump; thereby completing the output of the reward liquid;

[0061] (3) Establish a data communication link with the host computer to realize bidirectional transmission of system monitoring information.

[0062] The main control module in this embodiment uses a 72MHz STM32F103VET6 microcontroller. The STM32F103VET6 microcontroller integrates multiple general-purpose input / output (GPIO) interfaces, four timers, three USART communication interfaces, one USB interface, and multiple ADC input channels, fully meeting all the functional requirements of this system in terms of actuator control, peripheral trigger detection, and data communication. Specifically:

[0063] The first to fourth output GPIO interfaces of the main control module are connected one-to-one to the four-phase control terminals of the stepper motor. The main control module outputs high and low level signals to each GPIO pin in sequence according to the preset four-phase eight-step excitation timing sequence, controlling the stepper motor windings to switch on and off in the preset order, thereby enabling the stepper motor to produce stable rotational motion and achieve precise mechanical traction.

[0064] The first input GPIO interface of the main control module is connected to the output of the capacitive touch sensor. It detects the touch signal received by the first input GPIO interface in real time through an external interrupt to obtain the level state of the touch signal and determine whether a reward trigger event has occurred.

[0065] When the main control module detects a change in the touch signal's level, it uses the changed level signal as the water pump control signal and outputs it to the water pump control terminal via the fifth output GPIO interface. Specifically, in this embodiment, the MOSFET driver switch serves as the water pump control terminal. When the main control module detects a high-level touch signal, it amplifies and drives the high-level signal through the MOSFET driver switch, providing the rated operating voltage to the water pump and starting it to achieve the output of the excitation liquid.

[0066] The main control module uses an internal timer to precisely time and control the duration of the high-level signal output from the fifth output GPIO interface. By adjusting the pump's on-time, the amount of water dispensed can be controlled. When the timer expires, the main control module pulls the level of the fifth output GPIO interface low, the MOSFET driver switch is turned off, and the pump stops working.

[0067] The main control module and the host computer establish a wireless data communication connection via a JDY-31 Bluetooth module for transmitting various monitoring data during system operation. This monitoring data specifically includes the stepper motor's operating status, traction parameters, capacitive touch trigger information, and the water output time of the reward module. The data communication is based on the Bluetooth 3.0 protocol and uses a Serial Port Profile (SPP) working mode to achieve reliable, low-latency bidirectional data transmission between the main control module and the host computer. The JDY-31 Bluetooth module operates in the 2.4GHz band, with a maximum transmission distance of 30 meters and a maximum throughput of 16K bytes / s, which meets the data transmission requirements of the experimental equipment.

[0068] The SPP transparent transmission mode used in this embodiment is a virtual serial communication method based on the Bluetooth protocol. In this mode, the JDY-31 Bluetooth module transparently maps the Bluetooth wireless link to a serial communication interface: the main control module sends monitoring data to the JDY-31 Bluetooth module through its USART interface. The JDY-31 Bluetooth module automatically encapsulates and processes the monitoring data before transmitting it wirelessly to the host computer. After receiving the monitoring data, the host computer directly parses it according to the serial communication protocol, eliminating the need for an additional protocol conversion module and simplifying the system communication structure. The SPP transparent transmission mode has the advantages of simple communication protocol, strong compatibility, and convenient development, making it suitable for command issuance and data feedback scenarios between experimental equipment and the host computer in this system. It should be noted that the monitoring data includes execution time points and system event logs, including: stepper motor operating status, traction monitoring data, capacitive touch trigger information, and the water output time of the reward module.

[0069] The reward module consists of a touch detection section and a water delivery execution section, used to output a water reward after the subject completes a specified behavior. The touch detection section uses a double-sided TTP223 capacitive touch sensor. This sensor features short response time, stable triggering, and sensitivity to light touch behavior in mice, making it suitable for trigger detection in behavioral experiments. The output signal of the TTP223 touch module is directly connected to the first input GPIO interface of the main control module. By reading the level state of this pin, the touch event can be determined.

[0070] In practical implementation: The TTP223 capacitive touch sensor includes three pins: GND, VCC, and SIG. The GND pin is connected to system ground, the VCC pin is connected to the power supply voltage, and the SIG pin is the output terminal used to input touch signals.

[0071] The water supply actuator consists of a miniature water pump operating at 5V and a MOSFET driver switch group. The positive terminal of the miniature water pump is connected to the 5V power supply, and the negative terminal is connected to system ground through the MOSFET driver switch. The on / off control of the water pump is achieved by the MOSFET driver switch. The MOSFET used is the NCE4080K model, which is an N-channel enhancement-mode MOSFET with low on-resistance, typically about 8mΩ, enabling stable operation of the miniature water pump with a small drive current. The gate (G) of the NCE4080K is connected to the GPIO output pin of the main control module through a current-limiting resistor, the source (S) is connected to system ground, and the drain (D) is connected to the negative terminal of the water pump, forming a low-side switching drive structure.

[0072] In operation, the main control module precisely regulates the water pump's operating status by timing or PWM modulation of the MOSFET gate control signal. After detecting a high-level output from the TTP223 touch module, the main control module outputs a high-level signal for a specific duration to the gate of the NCE4080K, thereby controlling the water pump's operating time. Adjusting the duration of this high-level signal allows for precise control of the water output, meeting the needs of different experimental conditions.

[0073] The visual induction module is used to output controllable visual stimuli to mice during experimental training, inducing them to generate specific attentional orientations and behavioral intentions. This module consists of a ceramic display screen and visual stimulus generation software. The software allows for parameterized configuration of stimulus intensity, flashing frequency, stimulus duration, display position, and movement direction, and generates various types of visual stimuli, including fixed-frequency flashing light spots, geometric shapes moving at preset directions and speeds, and periodically changing color / brightness target graphics. Researchers can flexibly adjust parameters according to the training stage and experimental needs to achieve precise control of visual input conditions. The ceramic display screen receives and plays the visual stimulus signals.

[0074] During training, this module presents preset visual stimuli to guide the mouse's attention to a specific spatial area or target, leading to a clear behavioral intention. This, in turn, triggers changes in the neural activity patterns of relevant brain regions, providing behavioral triggering conditions for the brain-computer interface module to collect target neural signals. Simultaneously, the activation and deactivation times of the visual induction module are strictly synchronized with the main control module, triggering the traction and reward modules in a preset sequence, forming a closed-loop training control process of visual induction—behavioral response—neural signal acquisition—traction and reward feedback.

[0075] The physiological information acquisition module is a key module in this exercise training system for acquiring neural information. It is used to collect electrical signals from the brains of small animals and transmit them to the monitoring and processing module. Its acquisition principle and implementation method are as follows:

[0076] We used WE-LINKING technology from Microlink Medical to collect neural signals through invasive microelectrode array implantation. Invasive microelectrode array implantation refers to the precise surgical implantation of an electrode array consisting of multiple micro-scale electrodes into specific brain regions (such as the sensory cortex and motor cortex) of mice, maintaining a micrometer-level distance between the electrode tips and the neuronal cell bodies or axons.

[0077] Brain neural signals originate from the electrical activity of neuronal action potentials and postsynaptic potentials. When neurons are stimulated by external stimuli (such as visual stimuli presented by a visual induction module), the cell membrane potential changes rapidly, generating short-duration, high-amplitude action potential signals. When multiple neurons fire synchronously or nearly synchronously, extracellular potential changes that can be detected by electrodes are formed in local brain regions. Invasive microelectrode arrays sense these weak voltage changes and convert them into electrical signals for output. Compared to non-invasive EEG acquisition methods, invasive acquisition has a higher signal-to-noise ratio and finer temporal resolution, making it suitable for precise analysis of single neuron firing or neuronal cluster activity.

[0078] Since the amplitude of neural discharge signals is typically in the range of μV to mV and is easily affected by environmental noise, power supply interference, and other factors, the acquired raw neural electrical signals need to be pre-amplified and filtered. Specifically:

[0079] The pre-processing method employs a high-input-impedance, low-noise operational amplifier structure to perform primary amplification of weak neural electrical signals, thereby increasing the signal amplitude and reducing the load effect of subsequent circuits. The amplification factor is set according to the electrode characteristics and signal amplitude range to ensure that the signal is within the dynamic range of subsequent processing circuits.

[0080] The filtering process consists of two parts: high-pass filtering and low-pass filtering. High-pass filtering removes DC drift and low-frequency baseline fluctuations, with a cutoff frequency set around tens of Hz. Low-pass filtering suppresses high-frequency noise and electromagnetic interference, with a cutoff frequency set below 1000 Hz. Through this bandpass filtering process, the main frequency band characteristics of the neural discharge signal can be effectively preserved, while simultaneously improving the signal-to-noise ratio.

[0081] In this embodiment, the physiological information acquisition module consists of three parts: an acquisition device, an operational amplifier, and a filter processor; wherein:

[0082] The acquisition device uses WE-LINKING technology from Microlink Medical, which is an invasive microelectrode array implanted in the sensory or motor cortex of small animals. It is mainly used to acquire the discharge signals of a single neuron or a cluster of neurons, i.e., neural electrical signals.

[0083] The operational amplifier receives the neural electrical signals transmitted by the acquisition device and amplifies them;

[0084] The filter processor is used to filter the amplified neural electrical signals. This filtering process includes high-pass filtering and low-pass filtering, with the high-pass filter cutoff frequency set at tens of Hz and the low-pass filter cutoff frequency set below 1000 Hz. The filtered neural electrical signals are then transmitted to a host computer for recognition and decoding.

[0085] The monitoring and processing module includes a host computer and a camera. Specifically:

[0086] The camera used is an HSX63 model with a resolution of 1280×720 and a frame rate of 25fps. It collaborates with a host computer via USB or serial video output to record behavioral video. The camera's high resolution and stable frame rate can clearly capture the mouse's behavior and posture changes, providing a reliable image data foundation for subsequent behavioral analysis, event tagging, and training progress evaluation.

[0087] The host computer possesses comprehensive functions including multi-source signal synchronous acquisition, real-time visualization analysis, closed-loop control, and data management. Specifically, the host computer is configured as follows:

[0088] It receives data information from the main control module and sends training instructions to the main control module to achieve bidirectional control of data transmission.

[0089] It receives neural electrical signals, identifies and decodes them, displays real-time subcortical EEG data and surface muscle potential changes in specific brain regions of small animals, and draws relevant analytical images to help researchers obtain information on muscle-brain coupling.

[0090] It receives traction force monitoring data, displays and stores the force changes of the force sensor on the mouse's upper limb in real time, and provides biomechanical behavior information to assess the mouse's force output and postural stability during training.

[0091] The system displays the execution time points of each module's tasks and system event logs in a timeline format, including the arrival of input signals, the start and completion of the data processing phase, and the time when output commands are sent.

[0092] Through the above functions, the monitoring and processing module can simultaneously monitor and analyze behavior, EEG, EMG and mechanical state during the training process, ensuring the reliability and repeatability of the training loop.

[0093] The power module is also located inside the enclosure and is used to supply power to each module. For example... Figure 4 As shown, the power supply module includes a main power supply and an LM2596 step-down regulator module, wherein:

[0094] The main power supply uses an 18650 6S lithium battery pack, with each cell having a nominal voltage of 3.7V, resulting in a nominal output voltage of 22.2V for the battery pack. By connecting the batteries in series, a higher voltage level suitable for centralized power supply can be achieved while maintaining high energy density, reducing current loss and line voltage drop during transmission, making it suitable for applications requiring long-term, stable power supply.

[0095] The LM2596 buck regulator module is a switching buck regulator chip used to convert the 22.2V output voltage from the 18650-6S lithium battery pack into a stable 5V DC voltage, which is then supplied to the stepper motor, camera, main control module, visual screen, and water pump. It integrates a power switch, error amplifier, and control logic circuitry, achieving high-efficiency voltage conversion through high-frequency switching modulation. Its typical output current range is 2A to 3A, with a conversion efficiency of approximately 85%–90%, maintaining output voltage stability under large load variations. The LM2596 buck regulator module has an output current range of 2A to 3A and a conversion efficiency of 85%–90%.

[0096] The procedure for training mice using the above training system is as follows: Figure 5 As shown, it includes the following steps:

[0097] Step 1. Configure the visual stimulation software parameters according to the requirements, generate visual stimulation signals and play them on the visual display to visually induce stimulation in mice; and generate a synchronization signal to send to the main control module when the stimulation begins.

[0098] Step 2. The main control module drives the stepper motor to pull the mouse's forelimb to touch the capacitive touch sensor to generate a touch signal.

[0099] Step 3. The main control module detects the level of the touch signal output by the capacitive touch sensor; when the signal output by the capacitive touch sensor is detected to be a high level signal, the force sensor traction is stopped or the high level signal is sent to the water pump as a water pump control signal and the water pump is controlled to release liquid to reward the mouse; at the same time, the force sensing data is recorded.

[0100] Step 4. After the reward period ends, the motor returns to its initial position;

[0101] Step 5. End the training or wait for the next round of exercise training.

[0102] In addition, to improve the accuracy of training behavior and training data, a pattern recognition unit is also provided in the host computer in practical applications. The pattern recognition unit includes a feature extraction unit, a matching unit, and a judgment unit.

[0103] The feature extraction unit is used to extract signal segments from the neural electrical signals reported by the physiological information acquisition module within a time window of 200ms to 500ms before the response after receiving the behavior correctness flag sent by the main control module, and extract its time domain or frequency domain features.

[0104] The matching unit is used to calculate the similarity between the extracted time-domain or frequency-domain features and the pre-trained effective cognitive response template;

[0105] The determination unit is used to output a true learned label when the similarity is greater than a preset threshold, and output a false learned label otherwise.

[0106] Based on the genuine learning label, the main control module triggers the reward module to provide the first water volume; otherwise, a fake learning judgment signal is output, triggering the reward module to perform the second water volume and marking the trial as invalid; wherein, the first water volume is 0.05ml to 0.10ml, the second water volume is 0.01ml to 0.02ml, and the second water volume is less than 30% of the first water volume.

[0107] In summary, the training system of this embodiment, integrating an elastic traction structure, a main controller, a reward module, a visual stimulation module, and a host computer, is applied to the training of test animals. It regulates the activity range and trajectory of the test animals through traction, reducing interference from ineffective behaviors. Events such as the onset of visual stimulation, stepper motor action, touch triggering, water pump activation, and neural signal sampling all use the same main control clock as a reference, avoiding drift from independent module clocks and ensuring that the collected neural signals, mechanical signals, and behavioral monitoring are aligned in time. Its interrupt handling method ensures that reward behavior is triggered and output in a timely manner (<1ms). Simultaneously, wireless transmission allows personnel to observe the mouse's state in real time; neural signals → real-time decoding → reward / punishment adjustment → influencing subsequent behavior → re-collection of neural signals, forming a closed "neural-behavior-feedback" training loop. Firstly, it improves the completeness of training and achieves quantitative training indicators: the closed-loop feedback data chain allows experimental data to fully describe the animal's perception, decision-making, execution, and learning process. Secondly, the training operation is repeatable: the experimental process is solidified into repeatable code logic, reducing human error.

Claims

1. A mouse motor training system for invasive brain-computer interface research, characterized in that, include: Training platform, main control module, power system, reward module, visual guidance module, physiological information acquisition module, and monitoring and processing module; The training platform includes a housing; the top of the housing is a support platform, and a fixing device for securing the animal to be tested is installed in the middle of the support platform; a water tank and a traction module are located outside the housing; the water tank serves as a liquid supply, counterweight, or working condition simulation auxiliary component; the traction module includes a stepper motor, an elastic traction structure, and a force sensor; wherein: The control terminal of the stepper motor is connected to the main control module to receive the excitation timing control signal provided by the main control module, and generate traction force based on the excitation timing control signal to transmit to the elastic traction structure. One end of the elastic traction structure is connected to the output end of the stepper motor, and the other end is fixed to the forelimb of the animal under test. It is used to transmit traction force and provide flexible cushioning. A force sensor is placed on the traction force transmission path between the stepper motor and the forelimb of the animal under test to detect the magnitude of the traction force; The main control module is located inside the housing. Its stepper motor output is connected to the stepper motor control terminal, its touch signal input is connected to the signal output of the capacitive touch sensor in the reward module, its water pump control output is connected to the water pump control terminal of the reward module, its communication terminal is connected to the host computer in the monitoring and processing module, and its synchronization signal input is connected to the synchronization signal output of the visual guidance module. The main control module is configured as follows: It receives touch signals sent by a capacitive touch sensor. When the touch signal is a high-level signal, it uses it as a water pump control signal and sends it to the water pump. It receives the synchronization signal sent by the visual guidance module and generates an excitation timing control signal to drive the stepper motor to perform traction action based on the synchronization signal; Receive training instructions from the host computer; The reward module includes a capacitive touch sensor and a water pump; the capacitive touch sensor is fixedly installed in a position within the training platform that the animal's forelimbs can reach, and is used to generate a touch signal when the animal touches it and send it to the main control module; the water pump is used to power on and output liquid rewards according to the received water pump control signal. The visual induction module is used to present visual stimulation signals to small animals and generate a synchronization signal when the visual stimulation begins and send it to the main control module. The output end of the physiological information acquisition module is connected to the physiological signal input end of the host computer in the monitoring and processing module, and is used to acquire the brain nerve electrical signals of the animal under test and transmit them to the monitoring and processing module. The monitoring and processing module includes a host computer and a camera, used for behavior recording and data synchronous analysis; The power module is also located inside the enclosure, and its power output terminal is connected to the power input terminals of the training platform, main control module, reward module, visual guidance module, physiological information acquisition module and monitoring and processing module, respectively, to supply power to each module.

2. The mouse exercise training system according to claim 1, characterized in that, The elastic traction structure includes a spring and a flexible traction component; one end of the spring is connected to the output end of the stepper motor, and the other end is connected to the forelimb of the small animal through the flexible traction component.

3. The training system according to claim 2, characterized in that, The stepper motor is a 28BYJ-48 type, connected to the main control module via a ULN2003 driver board, and driven using a four-phase eight-step control mode; the spring has a length of 300 mm, a wire diameter of 0.3 mm, and an outer diameter of 2 mm; the force sensor is a ZNZL-V-10 type tension / compression sensor with a range of 10N and a sensitivity of 2.0mV / V; the tension / compression sensor is used to generate a differential voltage signal proportional to the magnitude of the traction force during traction and send it to the main control module to generate traction force monitoring data.

4. The training system according to claim 1, characterized in that, The main control module uses an STM32F103VET6 microcontroller. The main control module establishes bidirectional data communication with the host computer via the JDY31 Bluetooth module, based on the Bluetooth 3.0 protocol and serial port transparent transmission mode. The transmitted monitoring data includes: stepper motor running status, traction monitoring data, capacitive touch trigger information, and water output time of the reward module.

5. The training system according to claim 4, characterized in that, The main control module uses an external interrupt method to detect the touch signal sent by the capacitive touch sensor; the main control module has a timer controller inside, which controls the high level duration of the water pump control signal through a timer to adjust the water output.

6. The training system according to claim 1, characterized in that, The capacitive touch sensor is a double-sided TTP223 type; the water pump drive circuit uses an NCE4080K type N-channel enhancement-mode MOSFET to form a low-end switching drive structure. The gate of the MOSFET is connected to the water pump control output terminal through a current-limiting resistor, the source is connected to system ground, and the drain is connected to the negative terminal of the water pump; the main control module controls the water pump output by controlling the high-level duration of the MOSFET gate signal or by using PWM modulation.

7. The training system according to claim 1, characterized in that, The visual induction module includes a ceramic display screen and visual stimulus generation software. The visual stimulus generation software is used to parameterize parameters such as stimulus intensity, flashing frequency, stimulus duration, display position, and motion direction, and to generate various visual stimulus signals based on the parameter configuration; the visual stimulus signals include light spot stimuli that flash at a fixed frequency, geometric shapes that move in a preset direction and speed, and target stimulus shapes that periodically change color or brightness. The ceramic crystal display screen is used to receive and play visual stimulus signals; The stimulation start and termination times of the visual induction module are synchronized with the timing of the main control module, and the traction module and reward module are triggered according to the preset timing sequence to construct a closed-loop training control process of visual induction, behavioral response, neural signal acquisition, traction and reward feedback.

8. The training system according to claim 1, characterized in that, The physiological information acquisition module includes an acquisition device, an operational amplifier, and a filter processor; The acquisition device is an invasive microelectrode array implanted in the sensory or motor cortex of the animal under test, used to acquire the firing signals of a single neuron or a cluster of neurons, i.e., neural electrical signals. The input terminal of the operational amplifier is connected to the output terminal of the acquisition device, and is used to receive neural electrical signals and amplify them. The input terminal of the filter processor is connected to the output terminal of the operational amplifier, and the output terminal is connected to the physiological signal input terminal of the host computer; it is used to filter the amplified neural electrical signal; the filtering process includes high-pass filtering and low-pass filtering, wherein the cutoff frequency of the high-pass filter is set between 10Hz and 100Hz, and the cutoff frequency of the low-pass filter is set between 100Hz and 1000Hz.

9. The training system according to claim 8, characterized in that, The camera is connected to the host computer via a USB or serial video output interface for collaborative recording of behavioral video; the host computer is configured as follows: Receive data information from the main control module and send training instructions to the main control module; It receives neural electrical signals, identifies and decodes them, and displays in real time the subcortical EEG data of small animals and the changes in surface muscle potential in specific brain regions; Receive traction monitoring data and display and store it in real time; The system displays the execution time of each module's tasks and system event logs in a timeline format.

10. The training system according to claim 1, characterized in that, The power supply module consists of a main power supply and an LM2596 step-down regulator module connected after the main power supply. The main power supply uses an 18650 6S lithium battery pack with a nominal output voltage of 22.2 V. The LM2596 step-down regulator module receives the 22.2 V voltage and converts it into a stable 5 V DC voltage, which powers the stepper motor, camera, main control module, and other peripheral modules. The output current range of the LM2596 step-down regulator module is 2 A to 3 A, and the conversion efficiency is 85%–90%.

Images