In vivo multi-channel neural electrical signal acquisition system and signal separator

Abstract

The present invention provides an in vivo multi-channel neural signal acquisition system and signal separator, comprising at least one 2#imgabs0#10 integrated interface, at least two electronic components, at least four conductive commutators, and at least four free electrode groups including free needles as electrodes. The 2#imgabs1#10 integrated interface is divided into Module I, Module II, Module III, and Module IV. A separator circuit separates the 32-to-16 electrode adapter configured with a preamplifier into four independent modules, enabling a single in vivo multi-channel recording system to simultaneously record neural signal signals from multiple regions of multiple freely moving mice and rats. The free needles used as electrodes enable flexible implantation in different brain regions, providing a foundation for constructing a multi-layer perception mechanism for stress response type recognition based on heterogeneous graph neural networks.

Description

Technical Field

[0001] The present invention relates to the field of signal acquisition, and in particular to an in-vivo multi-channel neural electrical signal acquisition system and a signal separator. Background Art

[0002] In vivo multi-channel neural signal recording is a method for studying neural electrical activity in living animals. Metal electrodes are implanted in a specific cortex or region of the animal's brain. The electrodes are connected to a high-sensitivity preamplifier and recording system, which converts the recorded analog signals into digital signals. Neuronal firing signals can then be observed using the accompanying recording software. This technology can simultaneously record the electrical activity of multiple neurons in multiple brain regions, displaying both high-frequency spike signals and low-frequency signals such as LFP, EEG, and EMG. It is a powerful tool for analyzing neural information encoding in the brain. Well-established systems and products for in vivo multi-channel recording are available, such as Plexon's OmniPlex-D system, Blackrock Microsystems' Cerebus system, Neuralynx's Digital Lynx system, Intan Technologies' Intan RHD system, and Ripple Neuro's Grapevine system. These systems offer high sampling and A / D conversion rates, ultra-low noise, and easy-to-use software, significantly enhancing neuroscience research.

[0003] However, most of these in vivo multi-channel neural signal recording systems are configured with preamplifiers that are configured as 32-channel modules. For example, a 32-channel recording system is configured with one preamplifier, a 64-channel recording system is configured with two preamplifiers, and so on. However, many animal experiments do not require such a large number of channels, so companies often include a 32-channel adapter to simplify electrode implantation. Despite this, most of these systems can only be used for in vivo neural signal recording in a single animal at a time. While they can simultaneously record from multiple brain regions in a single animal, they cannot simultaneously record from multiple brain regions in multiple animals. However, many animal experiments, such as those for screening epilepsy models in mice and rats and for monitoring EEG and myoelectric activity during sleep in mice, generally require no more than four channels per animal, yet long-term monitoring is required. Using a single system to complete these experiments would require a significant timeframe. Furthermore, neural signaling in freely moving animals is highly susceptible to environmental influences, animal behavior, and other factors. Therefore, consistency in the results of the same set of experiments conducted on different days is difficult to ensure. Of course, you can use multiple systems to start recording at the same time, but the cost of purchasing the system is very high, and installing multiple systems requires a large space. In addition, most channels of each system are not used, which will cause great waste.

[0004] Existing technology, such as invention patent application CN104524692A, discloses an implantable multi-channel neural electrical signal acquisition circuit. Figure 1 As shown, it includes a motor group for sensing weak neural electrical signals generated by cellular electrical activities in biological tissues; an amplifier group for amplifying the sensed weak electrical signals to obtain amplified neural electrical signals; a reference signal generator for generating a sinusoidal wave signal as a measurement reference; a modulator group for amplitude modulating the sinusoidal wave signal and the amplified neural electrical signal to output an amplitude modulated wave voltage signal; a signal adder for linearly adding the amplitude modulated wave voltage signals output by each modulator in the modulator group to obtain a summed electrical signal; an up-converter for up-converting the summed electrical signal to output a radio frequency signal; a radio frequency transmitter for power amplifying and antenna transmitting the radio frequency signal; a multi-channel carrier generator for providing a carrier signal for the modulator group and the up-converter; the output end of the motor group is connected to the input end of the radio frequency transmitter through the amplifier group, the modulator group, the signal adder and the up-converter in sequence, the output end of the reference signal generator is connected to the input end of the modulator group, and the output end of the multi-channel carrier generator is respectively connected to the input end of the modulator group and the input end of the up-converter. Each amplifier in the implantable multi-channel neural electrical signal acquisition circuit corresponds only to one biological tissue, and most of the amplifier's channels are not used, resulting in great waste and extremely high costs.

[0005] Furthermore, existing technologies use pin headers to achieve multi-channel signal acquisition. However, because different parts of the animal brain represent different neuronal activity functions, fixed pin header layouts cannot flexibly achieve spatially distributed placement of pins for each channel in the brain. The signal loss associated with multiple pin headers in a second-order system is also significant.

[0006] Finally, the electronic components and integrated interfaces are uniformly wrapped with epoxy resin, which cannot guarantee the screen connection between the lines and the risk of short circuit during the wrapping process. Summary of the Invention

[0007] In order to solve the above problems, the present invention provides a signal separator for an in vivo multi-channel neural electrical signal acquisition system, which separates the 32-to-16 electrode adapter configured by the preamplifier into four independent modules, thereby achieving the purpose of a set of in vivo multi-channel recording system to simultaneously record neural electrical signals in multiple areas of multiple freely moving mice.

[0008] A signal separator for an in vivo multi-channel neural electrical signal acquisition system, comprising at least one 2 10 integrated interfaces, at least 2 electronic components, at least 4 conductive commutators and at least 4 free electrode groups;

[0009] 1 of which 2 10 The integrated interface is divided into Module I, Module II, Module III and Module IV;

[0010] 2 10 modules of integrated interface 1, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form the first group of separation circuits;

[0011] 2 Module II of 10 integrated interfaces, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a second group of separation circuits;

[0012] 2 Module III of 10 integrated interfaces, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a third group of separation circuits;

[0013] 2 10 integrated interface module IV, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a fourth group of separation circuits, wherein at least one 2 10 integrated interfaces and at least 2 electronic components are independently and separately packaged as a whole by a plastic cover having relatively independent chambers.

[0014] Preferably, at least 1 2 Each of the 10 integrated interfaces includes 20 pins, which are divided into 6 groups. Two groups are named ① and ②, each with 2 pins, and four groups are named module I, module II, module III and module IV, each with 4 pins.

[0015] Preferably, each of the at least two electronic components has five pins, a pin pitch of 2.54 mm, and a resistance of 5.1 KΩ.

[0016] Preferably, each of the at least four conductive commutators comprises six conductive wires, a housing, a 2 3 pin headers and 1 threaded sleeve.

[0017] Preferably, the housing comprises a stator side and a rotor side.

[0018] Preferably, each of the at least four free electrode groups comprises six wires, one 2 3 female rows, 6 free pins, 1 spring sleeve and 1 threaded cap, each free pin is set at the end of a corresponding wire, the 2 3 female headers can be used with the 2 3 pin headers are detachable.

[0019] More preferably, the six wires are wrapped with insulating jackets of different colors or with insulating jackets of the same color and provided with markers for distinguishing them from each other.

[0020] The present invention also provides an in vivo multi-channel neural electrical signal acquisition system, comprising the signal separator of the in vivo multi-channel neural electrical signal acquisition system as described above.

[0021] Preferably, it also includes a preamplifier, an analog-to-digital converter and a data processing unit, and the data processing unit is provided with a storage medium capable of executing artificial intelligence model processing, which is used to establish a heterogeneous graph neural network model based on signals from different brain regions, and to construct animal stress response recognition using an attention mechanism.

[0022] Optionally, a method for establishing a heterogeneous graph neural network model based on signals from different brain regions and constructing animal stress response recognition using an attention mechanism includes:

[0023] S1 regards each free needle as a node and establishes a heterogeneous graph H (V, E, X), where V is the node set, E is the relationship set, and X is the bioelectric information set. A corresponding label is set for each type of stress response;

[0024] S2 builds a multi-node path L, aggregates the nodes, and calculates the node relationships are the representations of node i and any neighbor node j on the multi-node path L of node i, respectively. To implement the neural network with node attention mechanism, ,right Normalized to , the aggregation representation of the aggregation node i on the node path L , It refers to the sum of the number of all neighbor nodes on the multi-node path L;

[0025] S3 continues to build the aggregation of node paths and calculates the path relationship of the specific path containing node i are attention vector, weight matrix, bias vector, is to sum all nodes on V, then Normalized to , and get the aggregate representation , is a specific path representation containing node i, which is calculated by the following path neural network A neural network to implement the path attention mechanism;

[0026] S4 establishes a multi-layer perception mechanism for stress response recognition, trains the accuracy of stress response types, and establishes a loss function to optimize A NODE (·)、A ROUTE (·),Multi-layer perception mechanism network parameters.

[0027] The beneficial effects of the present invention include: using a separation circuit to separate the preamplifier's 32-to-16-electrode adapter into four independent modules, each of which does not interfere with each other. This allows for a single in vivo multi-channel recording system to simultaneously record neural electrical signals from multiple animals, greatly improving experimental efficiency, ensuring the reliability of experimental results, and significantly reducing experimental costs. Furthermore, the free needle electrodes can be flexibly implanted in different brain regions, providing an instrumental foundation for constructing a multi-layer perception mechanism for stress response type recognition based on heterogeneous graph neural networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a structural framework diagram of an implantable multi-channel neural electrical signal acquisition circuit in the prior art;

[0029] Figure 2 This is a physical diagram of a signal separator of an in vivo multi-channel neural electrical signal acquisition system of the present invention;

[0030] Figure 3 yes Figure 2 The circuit structure diagram of the signal separator of the in vivo multi-channel neural electrical signal acquisition system shown;

[0031] Figure 4 yes Figure 22. The signal separator of the in vivo multi-channel neural electrical signal acquisition system shown in FIG. 10. Schematic diagram of the integrated interface;

[0032] Figure 5 yes Figure 2 Schematic diagram of electronic components of a signal separator of an in vivo multi-channel neural electrical signal acquisition system;

[0033] Figure 6 yes Figure 2 The schematic diagram of the structure of the conductive commutator of the signal separator of the in vivo multi-channel neural electrical signal acquisition system is shown;

[0034] Figure 7 yes Figure 2 Schematic diagram of a free electrode group of a signal separator of an in vivo multi-channel neural electrical signal acquisition system;

[0035] Figure 8 This is a schematic diagram of four types of free needles implanted in different areas of the left and right brain hemispheres of animals, forming specific node pathways. DETAILED DESCRIPTION

[0036] The embodiments of the present invention are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only illustrative and not restrictive.

[0037] Figure 2 This is a physical diagram of a signal separator of an in vivo multi-channel neural electrical signal acquisition system of the present invention. As shown in the figure, a signal separator of an in vivo multi-channel neural electrical signal acquisition system of an embodiment of the present invention includes 1 2 10 integrated interfaces (11), 2 electronic components, 4 conductive commutators (12) and 4 free electrode groups (13). 10 integrated interfaces (11) are packaged together with 2 electronic components.

[0038] As shown in the figure, the plastic cover (14) is made of two pieces that are buckled together. The two pieces have three cavities a, b, and c after being buckled together. The figure shows a cross-section of a cavity on one piece. Cavity a is used to place the integrated interface, and cavities b and c are used to place one electronic component each. The welding wires between the two are connected through the wiring channel (20). In this way, the three device units are separated, and the mutual interference and short circuit risks during assembly are avoided after the arrangement of the devices is completed.

[0039] Figure 3 yes Figure 2 The circuit structure diagram of the signal separator of the in vivo multi-channel neural electrical signal acquisition system is shown in the figure. 10 The integrated interface (11) is divided into module I, module II, module III and module IV. 2 10 integrated interface module 1, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form the first group of separation circuits; 2 10 modules II of the integrated interface, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form the second group of separation circuits; 2 10 modules III of the integrated interface, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form the third group of separation circuits; 2 Module IV of the 10-integrated interface, two electronic components, a conductive commutator, and a free electrode group constitute the fourth separation circuit. The two electronic components in the first, second, third, and fourth separation circuits are identical, the conductive commutator is a different one, and the free electrode group is a different one. These four separation circuits separate the 32-to-16 electrode adapter configured in the preamplifier into four independent modules, enabling a single in vivo multi-channel recording system to simultaneously record neural electrical signals from multiple regions of multiple freely moving mice.

[0040] Figure 4 yes Figure 2 2. The signal separator of the in vivo multi-channel neural electrical signal acquisition system shown in FIG. 10 Schematic diagram of the integrated interface. As shown in the figure, at least 1 2 Each of the 10 integrated interfaces consists of 20 pins with a pin pitch of 1.27 mm. The 20 pins are divided into six groups: two groups, named ① and ②, each with two pins, and four groups, named Module I, Module II, Module III, and Module IV, each with four pins.

[0041] Figure 5 yes Figure 2 The schematic diagram of the electronic components of the signal separator of the in vivo multi-channel neural electrical signal acquisition system is shown. As shown in the figure, each of the at least two electronic components has 5 pins, a pin pitch of 2.54mm, and a resistance of 5.1KΩ. The 5 pins of electronic component 1 are named ① ’ , A, B, C, D (from left to right), the 5 pins of electronic component 2 are named ② ’ 、A ’ 、B ’ 、C ’ 、D ’ (From left to right). 2 10. Solder pin ① of the integrated interface to pin ①' of the electronic component 1. Solder pin ② of the integrated interface 10 to pin ②' of the electronic component 2.

[0042] Figure 6 yes Figure 2 The schematic diagram of the structure of the conductive commutator of the signal separator of the in vivo multi-channel neural electrical signal acquisition system is shown. As shown in the figure, at least four conductive commutators are named a, b, c, and d respectively. Each conductive commutator includes 6 wires, a shell (40), a 2 3 pin headers (50) and 1 threaded sleeve (60). The wire specification is a 250mm long multi-core 28AWG Teflon insulated wire with a maximum output circuit of 2A. The housing (40) includes a stator side (41) and a rotor side (42). The contact material of the stator side and the rotor side is gold-gold, and the housing (40) is made of ABS plastic. The maximum speed reaches 250RPM, the contact resistance is less than 150mΩ, and the electrical noise is less than 10mΩ. The outer diameter of the stator side is 12.5mm, and the outer diameter of the rotor side is 5mm. The stator side outlet (43) and the rotor side outlet (44) are 6-way wires. The 6-way wires of the rotor side outlet (44) are welded to 2 On the 6 pins of the 3-row pin (50), the 6-way wires of the stator side outlet (43) are connected to the 2 10. Solder the pins of the integrated interface module and the pins of the electronic components; 1. Sleeve a threaded sleeve (60) onto the outside of the 6-way wires of the rotor side outlet (44), with the threaded opening facing 2. 3 rows of pins (50). The conductive commutator prevents the connection wires from getting tangled when the animal is free to move.

[0043] Figure 7 yes Figure 2 The schematic diagram of the free electrode group of the signal separator of the in vivo multi-channel neural electrical signal acquisition system is shown. As shown in the figure, each of the at least 4 free electrode groups includes 6 wires, 1 2 3 rows of female (70), 6 free pins (30) led by the wire, 1 spring sleeve (80) and 1 threaded cap (90). The wire is a single-strand Teflon silver-plated high-temperature wire with an inner diameter of outer diameter Length=0.15 0.30 450mm; Spring sleeve (80) is 304 stainless steel, its wire diameter outer diameter inner diameter Length is 0.4 4 3.2 420mm. The specific connection method is: 6 wires are put into the spring sleeve (80) as a whole, and the outside of the wires are uniformly covered with red insulation sleeves, and then one end of the wires is connected to the 2 3 female row (70) welding, and the 2 3 female rows (70) with Figure 6 2 of the conductive commutator 3 rows of pins are connected, and the other end is welded to the free pin (30), and then the threaded cap (90) is put on the outside of the spring sleeve (80), with the cap opening of the threaded cap (90) facing 2 3. The female end (70) of the 6 free pin welding positions are provided with markers (31) that are distinguished by different colors.

[0044] Figure 8 is a distribution diagram of four types of electrodes implanted in the animal brain using free needles (30). Figure 8 In the figure, a is the distribution of free needle implantation points in the left hemisphere, and its color is Figure 2 and Figure 7 The colors of the markers (31) in Figure 8 In the figure, b represents the right hemisphere distribution, c and d represent the left and right hemisphere distributions, respectively. Ad shows examples of specific paths, where node i is the study node. a represents a specific path with a branch (defined in this invention as the merger of two specific paths), and c and d represent examples of specific path distributions, respectively, with one asymmetric and three symmetric paths.

[0045] By setting up multiple specific pathways, assigning corresponding labels based on stress response types, and using a bioelectric information matrix, we can train a multi-layer perception mechanism model based on a heterogeneous graph neural network composed of these pathways to identify stress types. Specifically, the steps include:

[0046] S1 regards each free needle as a node and establishes a heterogeneous graph H (V, E, X), where V is the node set, E is the relationship set (including the node relationships between nodes), and X is the bioelectric information set (including the bioelectric information matrix). A corresponding label is set for each type of stress response;

[0047] S2 builds a multi-node path L, aggregates the nodes, and calculates the node relationships They are node i and any neighbor node j on the multi-node path L of node i (such as Figure 8 a) in the representation, To implement the neural network with node attention mechanism, ,right Normalized to , the aggregation representation of the aggregation node i on the node path L ,Σ refers to the sum of the number of neighbor nodes on the multi-node path L;

[0048] S3 continues to build the aggregation of node paths and calculates the path relationship of the specific path (∈L) containing node i are attention vector, weight matrix, bias vector, is to sum all nodes on V, then Normalized to , and get the aggregate representation , is a specific path representation containing node i, which is calculated by the following path neural network A neural network to implement the path attention mechanism;

[0049] S4 establishes a multi-layer perception mechanism for stress response recognition, trains the accuracy of stress response types, and establishes a loss function to optimize A NODE (·)、A ROUTE (·),Multi-layer perception mechanism network parameters.

[0050] As mentioned above, 2 10 modules of integrated interface 1, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form the first group of separation circuits;

[0051] 2 Module II of 10 integrated interfaces, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a second group of separation circuits;

[0052] 2 Module III of 10 integrated interfaces, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a third group of separation circuits;

[0053] 2 The module IV with 10 integrated interfaces, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form the fourth separation circuit.

[0054] The specific connection method of the first group of separation circuits, the second group of separation circuits, the third group of separation circuits and the fourth group of separation circuits is as follows:

[0055] The first group of separation circuits: Among the 6 wires of the stator side of the conductive commutator, 4 of them are connected to 2 10. Solder the 4 pins of module 1 of the integrated interface, solder one wire to pin A of electronic component 1, and solder one wire to pin A' of electronic component 2;

[0056] The second group of separation circuits: Among the 6 wires of the stator side of the conductive commutator b, 4 of them are connected to 2 Solder 4 pins of module II of the 10 integrated interface, solder 1 wire to pin B of electronic component 1, and solder 1 wire to pin B' of electronic component 2;

[0057] The third group of separation circuits: Among the 6 wires of the stator side of the conductive commutator c, 4 of them are connected to 2 Solder 4 pins of module III of the 10 integrated interface, solder 1 wire to pin C of electronic component 1, and solder 1 wire to pin C' of electronic component 2;

[0058] The fourth group of separation circuits: Among the six wires of the stator side of the conductive commutator d, four of them are connected to the two Solder the 4 pins of module IV of the 10 integrated interface, solder one wire to pin D of electronic component 1, and solder one wire to pin D' of electronic component 2.

[0059] Finally, use a plastic cover such as Figure 3 As shown, 2 10 integrated interfaces and 2 electronic components are arranged and isolated in each chamber, and the whole is wrapped together. Each conductive commutator has 2 outlets on the rotor side. 3 rows of pins with 1 free electrode set of 2 3 rows of female connections, then screw the threaded sleeve of the conductive commutator and the threaded cap of the free electrode group together.

[0060] The functions of the 6 wires in each split circuit are as follows: The four wires soldered to the four pins of the 10 integrated interface module serve as four free movable electrodes (32), the one wire soldered to the one pin (A / B / C / D) of the electronic component serves as a free ground electrode (33), and the one wire soldered to the two pins (A' / B' / C' / D') of the electronic component serves as a free reference electrode (34).

[0061] Four channels are used for free-moving electrodes (32), which can directly capture the electrical activity of the brain area where the electrodes are located. One channel is used for a free reference electrode (34), which can provide a stable comparison baseline and is a reference point for measuring the potential difference of the free-moving electrodes (32). One channel is used for a free ground electrode (33), which can help stabilize the system from electrical noise and other external electrical interference.

[0062] According to the present invention, there is also provided an in vivo multi-channel neural electrical signal acquisition system, comprising the signal separator of the in vivo multi-channel neural electrical signal acquisition system as described above.

[0063] The in vivo multi-channel neural electrical signal acquisition system also includes a preamplifier, an analog-to-digital converter and a data processing unit, and the data processing unit is provided with a storage medium capable of executing artificial intelligence model processing, which is used to establish a heterogeneous graph neural network model based on signals from different brain regions, and to construct animal stress response recognition using an attention mechanism.

[0064] The connection relationship between these components is well known in the relevant technical field and will not be described in detail here.

[0065] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A signal separator for an in vivo multi-channel neural electrical signal acquisition system, characterized by: Include at least 1 2 10 integrated interfaces, at least 2 electronic components, at least 4 conductive commutators and at least 4 free electrode groups; 1 of which 2 10 The integrated interface is divided into Module I, Module II, Module III and Module IV; 2 10 modules of integrated interface 1, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form the first group of separation circuits; 2 Module II of 10 integrated interfaces, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a second group of separation circuits; 2 Module III of 10 integrated interfaces, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a third group of separation circuits; 2 10 integrated interface module IV, 2 electronic components, 1 conductive commutator and 1 free electrode group are connected to form a fourth group of separation circuits, wherein at least one 2 10 integrated interfaces and at least 2 electronic components are independently and separately packaged as a whole by a plastic cover having relatively independent chambers.

2. The signal separator of the in vivo multi-channel neural electrical signal acquisition system according to claim 1, characterized in that: The at least one 2 Each of the 10 integrated interfaces includes 20 pins, and the 20 pins are divided into six groups, two of which are named ① and ②, each with 2 pins; and four of which are named module I, module II, module III and module IV, each with 4 pins.

3. The signal separator of the in vivo multi-channel neural electrical signal acquisition system according to claim 2, characterized in that: Each of the two electronic components has five pins, a pin pitch of 2.54 mm, and a resistance of 5.1 kΩ.

4. The signal separator of the in-vivo multi-channel neural electrical signal acquisition system according to claim 3, characterized in that: Each of the at least four conductive commutators includes six conductive wires, a housing, a 2 3 pin headers and 1 threaded sleeve.

5. The signal separator of the in vivo multi-channel neural electrical signal acquisition system according to claim 4, characterized in that: The housing includes a stator side and a rotor side.

6. The signal separator of the in vivo multi-channel neural electrical signal acquisition system according to claim 5, characterized in that: Each of the at least four free electrode groups includes six wires, one 2 3 female headers, 6 free pins, 1 spring sleeve and 1 threaded cap.

7. The signal separator of the in vivo multi-channel neural electrical signal acquisition system according to claim 5, characterized in that: Four of the six free needles are free active electrodes that directly capture the electrical activity of the brain area where the electrodes are located. The remaining two are free ground electrodes that help stabilize the electrical noise and external electrical interference of the system and free reference electrodes that provide a reference point for measuring the potential difference of the free active electrodes. The four wires soldered to the four pins of the 10 integrated interface module serve as four free active electrodes, the one wire soldered to the pin of the first electronic component of the two electronic components serves as a free ground electrode, and the one wire soldered to the pin of the second electronic component of the two electronic components serves as a free reference electrode.

8. An in vivo multi-channel neural electrical signal acquisition system, characterized by: A signal separator for an in-vivo multi-channel neural electrical signal acquisition system comprising the device described in any one of claims 1 to 7.

9. The in vivo multi-channel neural electrical signal acquisition system according to claim 8, characterized in that: It also includes a preamplifier, an analog-to-digital converter and a data processing unit, and the data processing unit is provided with a storage medium capable of executing artificial intelligence model processing, which is used to establish a heterogeneous graph neural network model based on signals from different brain regions, and to construct animal stress response recognition using an attention mechanism.

10. The in vivo multi-channel neural electrical signal acquisition system according to claim 9, characterized in that: Methods for establishing a heterogeneous graph neural network model based on signals from different brain regions and using the attention mechanism to construct animal stress response recognition include: S1: Treat the free pin of each free electrode group as a node and build a heterogeneous graph , is a node set, is a relation set, For the bioelectric information set, set a corresponding label for each type of stress response; S2: Building a multi-node path , and aggregate the nodes to calculate the node relationships Node and nodes The representation of node For nodes The multi-node path to which it belongs Go to any neighbor node; To implement the neural network with node attention mechanism, ,right Normalized to , aggregation node In the node path Aggregate representation on , Refers to multi-node paths Sum the number of all neighbor nodes; S3: Continue to build the aggregation of node paths and calculate the nodes Path relationship of a specific path are attention vector, weight matrix, bias vector, It is for all Sum the nodes on Normalized to , and get the aggregate representation , For nodes The specific path representation of is calculated by the following path neural network A neural network to implement the path attention mechanism; S4: Establish a multi-layer perception mechanism for stress response recognition, train the accuracy of stress response types, and establish a loss function to optimize Multi-layer perception mechanism network parameters.

Images