Spiking wave recognition model construction method, spiking wave recognition method, device and equipment

Abstract

The present application relates to the medical technical field, disclose a spike recognition model construction method, spike recognition method, device and equipment, the construction method includes: obtaining electroencephalogram signal;The electroencephalogram signal is segmented, and the spike electroencephalogram signal section is obtained;The spike electroencephalogram signal section is pulse coded, and the pulse sequence is obtained;Based on the pulse sequence and neuron dynamics mechanism, the initial neuron model is constructed;Based on the synaptic plasticity adjustment mechanism inside neuron, the initial neuron model is updated, and the adaptive neuron model is constructed;Based on the adaptive neuron model, the pulse neural network is constructed, and the pulse neural network is used for feature mapping of the pulse sequence;Based on the synaptic plasticity adjustment mechanism inside neuron, the pulse neural network is updated, and the target spike recognition model is obtained.The target spike recognition model, both excitatory neurons and inhibitory neurons, ensure the rationality of pulse neural network feature mapping, improve the accuracy of identifying non-invasive traceable type epilepsy-like spike.

Description

Technical Field

[0001] This invention relates to the field of medical technology, specifically to a method for constructing a spike wave recognition model, a spike wave recognition method, a device, and equipment. Background Technology

[0002] Epileptiform spikes are caused by the synchronous abnormal firing of a large number of pyramidal neurons in the intracranial cerebral cortex, which are then conducted through brain tissue such as the skull and recorded by extracranial electrodes. A spiking neural network (SNN) is a model that simulates the dynamics of brain neurons.

[0003] To date, neural networks have evolved to the third generation. Feature learning models built upon the first two generations of neural networks suffer from limitations in learning the inter-class differences in traceable epileptiform spikes due to a lack of understanding of the propagation mechanism of neuronal firing in the epileptogenic zone. Consequently, the recognition performance of these models falls far short of clinical requirements. While neuroscience-driven spiking neural networks aim to explain the dynamics of neuronal firing from a biophysical perspective, the development of traceable epileptiform spikes involves a balance of inputs between excitatory and inhibitory neurons fluctuating around the spike threshold.

[0004] Current spiking neuron models lack consideration for inhibitory neurons, making them incompletely biologically interpretable. This casts doubt on the rationality of the nonlinear mapping of spiking neural networks. Inaccurate biological interpretability inevitably affects the reliable learning of traceable epileptiform spike confidence features, resulting in inaccurate epileptiform spike identification. Summary of the Invention

[0005] In view of this, the present invention provides a method for constructing a spike recognition model, a spike recognition method, an apparatus and device, to solve the problem that the spike neural network constructed in the prior art is inaccurate in recognizing epileptiform spikes.

[0006] In a first aspect, the present invention provides a method for constructing a spike recognition model, the method comprising:

[0007] Acquiring EEG signals;

[0008] The EEG signal was segmented to obtain spike wave EEG signal segments;

[0009] Pulse coding was performed on the spike wave EEG signal segment to obtain a pulse sequence;

[0010] An initial neuron model was constructed based on pulse sequences and neuronal dynamics mechanisms.

[0011] Based on the plasticity regulation mechanism within the neuronal synapse, the initial neuron model is updated to construct an adaptive neuron model;

[0012] Based on the adaptive neuron model, a spiking neural network is constructed, which is used for feature mapping of pulse sequences;

[0013] Based on the plasticity regulation mechanism within the neuronal synapse, the spiking neural network is updated to obtain a target spike recognition model.

[0014] The constructed target spike recognition model based on adaptive spiking neural network takes into account both excitatory and inhibitory neurons, and realizes spike reception, processing and transmission in a more reasonable way that is more in line with the actual dynamic regulation of the nervous system. This ensures the rationality of the feature mapping of the spiking neural network and improves the accuracy of the spike recognition model in recognizing non-invasive traceable epileptiform spikes.

[0015] In one alternative implementation, an initial neuron model is constructed based on the pulse sequence and neuronal dynamics mechanisms, including:

[0016] The resting-state membrane potential, membrane capacitance, and membrane resistance of neurons are preset;

[0017] In the presence of membrane current input, the membrane potential of a neuron is determined, wherein the membrane current is obtained by weighting the pulse sequence of the presynaptic neuron;

[0018] Based on resting membrane potential, membrane potential, membrane capacitance, and membrane resistance, an initial neuron model is established. This initial neuron model is used to describe the dynamic chaotic system when endogenous signals in the brain are transmitted to the measurement space.

[0019] The established initial neuron model can be used to describe the complex activities of neurons in the brain, providing a foundation for improving the accuracy of the target spike recognition model.

[0020] In one alternative implementation, an initial neuron model is updated based on the intrasynaptic plasticity regulation mechanism to construct an adaptive neuron model, including:

[0021] The initial neuron model is as follows:

[0022]

[0023] Among them, V t Let τ be the membrane potential at time t, and τ = C. m R m C is the time constant. m R m Let represent membrane capacitance and membrane resistance, respectively, and I(t) represent the total input current generated when the presynaptic neuron pulse arrives at the triggered synapse;

[0024] Discretize the initial neuron model:

[0025]

[0026] Among them, V t-1 Let I be the membrane potential at time t-1. t Let be the membrane current at time t;

[0027] When a neuron fires a pulse, the neuron's membrane potential is reset to the resting membrane potential, the membrane capacitance is set to a preset value, and the discretized initial neuron model is updated.

[0028] Determine an adaptive function to adaptively adjust the input to control the neuron's membrane potential, ensuring the output value remains between 0 and 1. The adaptive function is:

[0029] F t =σ(f(δ) t-1 ))

[0030] Among them, F t For the output value, δ t-1 This represents the pulse firing at time t-1, σ represents the sigmoid function, and f represents the convolution operator.

[0031] Based on the updated initial neuron model and the adaptive function, the adaptive neuron model is determined.

[0032] By simulating the simultaneous existence of excitatory and inhibitory neurons in the brain, the initial neuron model is updated to construct an adaptive neuron model. This model adaptively changes the neuron membrane potential, allowing the neuron model to adaptively adjust the input to control the membrane potential, thereby making pulse delivery more accurate.

[0033] In one alternative implementation, the adaptive neuron model is:

[0034]

[0035] Here, Θ represents element-wise multiplication.

[0036] In one alternative implementation, the spiking neural network is updated based on the intrasynaptic plasticity regulation mechanism of neurons to obtain a target spike recognition model, including:

[0037] Based on the Dirac function-based mechanism for regulating intrasynaptic plasticity, the parameter values ​​of neurons in a spiking neural network are updated. The intrasynaptic plasticity regulation mechanism is described as follows:

[0038]

[0039]

[0040] Where, τip The relative integration resolution of the initial neuron model and the intrasynaptic plasticity regulation mechanism, where β is the scale factor and r is the scale factor. C =1 / C m C m For film capacitance, r R =1 / R m R m Given the membrane resistance, the output response h of the neuron to input I is described as follows:

[0041]

[0042] Wherein, δ(tt) (f) ) is the Dirac function, t (f) Let f represent the pulse emitted at time t, and η represent the pulse intensity.

[0043] Secondly, the present invention provides a spike wave recognition method, the method comprising:

[0044] Acquire the EEG signal to be identified;

[0045] The EEG signal to be identified is input into the spike recognition model, and the traceable epileptiform spike signal segment of the EEG signal to be identified is obtained. The spike recognition model is constructed using the spike recognition model construction method of any one of the first aspects.

[0046] In this invention, a spike recognition model is used to identify traceable epileptiform spikes in the EEG signal to be identified, thereby improving the accuracy of identifying traceable epileptiform spikes in the EEG signal and providing support for subsequent diagnosis and treatment.

[0047] Thirdly, the present invention provides a spike recognition model construction device, the device comprising:

[0048] The acquisition module is used to acquire electroencephalogram (EEG) signals;

[0049] The segmentation module is used to segment the EEG signal to obtain spike EEG signal segments;

[0050] The encoding module is used to pulse encode the spike wave EEG signal segment to obtain a pulse sequence;

[0051] The neuron model building module is used to construct an initial neuron model based on pulse sequences and neuron dynamics mechanisms.

[0052] The adaptive neuron model building module is used to update the initial neuron model and build an adaptive neuron model based on the plasticity regulation mechanism inside the neuron synapse.

[0053] The spiking neural network building module is used to construct spiking neural networks based on the adaptive neuron model. The spiking neural network is used to perform feature mapping on the spiking sequence.

[0054] The network update module is used to update the spiking neural network based on the plasticity regulation mechanism inside the neuronal synapse to obtain the target spike recognition model.

[0055] Fourthly, the present invention provides a spike wave recognition device, the device comprising:

[0056] The module for acquiring EEG signals to be identified is used to acquire EEG signals to be identified.

[0057] The spike recognition module is used to input the EEG signal to be recognized into the spike recognition model and recognize the traceable epileptiform spike signal segment of the EEG signal to be recognized. The spike recognition model is constructed by any of the spike recognition model construction devices of the third aspect.

[0058] Fifthly, the present invention provides a computer device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the spike recognition model construction method of the first aspect and any corresponding embodiment described above, or to perform the spike recognition method of the second aspect.

[0059] In a sixth aspect, the present invention provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the spike recognition model construction method of the first aspect and any corresponding embodiment thereof, or to execute the spike recognition method of the second aspect. Attached Figure Description

[0060] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0061] Figure 1 This is a flowchart illustrating the method for constructing a spike recognition model according to an embodiment of the present invention.

[0062] Figure 2 This is a flowchart illustrating another method for constructing a spike recognition model according to an embodiment of the present invention.

[0063] Figure 3 This is a flowchart of some steps in the spike recognition model construction method according to an embodiment of the present invention.

[0064] Figure 4 This is a flowchart illustrating the spike recognition method according to an embodiment of the present invention.

[0065] Figure 5 This is a structural block diagram of a spike recognition model construction device according to an embodiment of the present invention.

[0066] Figure 6 This is a structural block diagram of a spike recognition device according to an embodiment of the present invention.

[0067] Figure 7 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Neuroscience-guided spiking neural networks aim to explain the dynamics of neuronal firing from a biophysical perspective. However, the unique neuronal spiking interactions during the development of traceable epileptiform spikes limit the reliability of spiking neural networks in learning interclass differences.

[0070] A mathematical computational model based on the pulse firing process and electrophysiological characteristics of biological neurons—the spiking neuron model—is used to construct a large-scale spiking neural network, simulating the reception, processing, and transmission of signals in biological nervous systems. Therefore, constructing a spiking neural network based on a neuron model that conforms to the dynamic regulation of actual nervous systems can achieve nonlinear output mapping of traceable epileptiform spikes. Since the pulse firing of the Leaky Integrate-and-Fire neuron model depends on the input intensity of any neuron of interest, the voltage balance of neurons near the pulse threshold is ensured by regulating strong and weak synaptic inputs. Thus, the Leaky Integrate-and-Fire neuron model is currently the most biologically interpretable nonlinear neuron model for constructing spiking neural networks. However, due to the spontaneous electrical activity of neurons in traceable epileptiform spikes and the release of neurotransmitters caused by synchronous firing of neuronal groups, traceable epileptiform spikes are primarily activated by excitatory neurons in the epileptogenic zone and terminated by inhibitory neurons, exhibiting an overall asymmetrical waveform variability. Furthermore, related studies have shown that the peak wave component of traceable epileptiform spikes within a duration of 20–200 ms originates solely from the synchronous firing of inhibitory neurons. Clearly, the peak preceding wave component of traceable epileptiform spikes ( Δ RP is dominated by the firing of excitatory neurons in the epileptogenic zone, while the after-peak component ( Δ PF is dominated by inhibitory neurons.

[0071] The development and progression of traceable epileptiform spikes involve the dynamic regulation of the nervous system, resulting from the balanced input of excitatory and inhibitory neurons fluctuating around the spike threshold. However, the aforementioned studies, based on the Leaky Integrate-and-Fire neuron model for nonlinear input-output mapping, missed the opportunity to construct a more biologically interpretable neuron model due to a lack of consideration for the inhibitory neuron model. This casts doubt on the rationality of the nonlinear mapping of the spiking neural network, as inaccurate biological interpretation inevitably affects the reliable learning of the confidence features of traceable epileptiform spikes.

[0072] Based on the above, and building upon the well-interpretable Leaky Integrate-and-Fire neuron model, a neuron model capable of adaptively altering neuronal membrane potential is constructed. This model aims to ensure the rationality of spiking neural network feature mapping and realize pulse reception, processing, and transmission in a more reasonable and realistic manner, in accordance with the dynamic regulation of the actual nervous system. This is worthy of further investigation.

[0073] Therefore, this invention provides a method for constructing a spike recognition model, a spike recognition method, an apparatus, and a device for use in a computer device. It should be noted that the executing entity can be a spike recognition model construction device, which can be implemented as part or all of the computer device through software, hardware, or a combination of both. The computer device can be a terminal, client, or server. The server can be a single server or a server cluster composed of multiple servers. In this embodiment, the terminal can be a smartphone, personal computer, tablet computer, or other smart hardware device. The following method embodiments all use a computer device as the executing entity for illustration.

[0074] According to an embodiment of the present invention, a method for constructing a spike recognition model is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0075] This embodiment provides a method for constructing a spike recognition model, which can be used in the aforementioned computer equipment. Figure 1 This is a flowchart of a spike recognition model construction method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0076] Step S101: Acquire EEG signals.

[0077] In one example, the EEG signal was a non-invasive scalp EEG signal, rather than an intracranial EEG signal, which can reduce the patient's pain and burden.

[0078] Step S102: Segment the EEG signal to obtain spike wave EEG signal segments.

[0079] In one example, the EEG signal segment is an epileptiform spike signal segment, which can be obtained by segmenting a time window of a certain length.

[0080] Specifically, the segmentation can be performed according to the following formula:

[0081] X(t i And t i =(A a -L / 2)×f s , ..., (A a ×f s -L / 2)

[0082] Among them, A a This indicates the time at which the expert marker point is located, where L is the length of the time window, and f is the value of f. sThe frequency of epileptic EEG-like spikes is represented by a = 1, 2, ..., Q, where Q represents the total amount of epileptic spikes marked by experts. Specifically, multi-lead signal segments within 100 milliseconds before and after the points marked by clinicians are extracted to obtain the database of samples to be analyzed.

[0083] Step S103: Pulse coding is performed on the spike wave EEG signal segment to obtain a pulse sequence.

[0084] In one example, by simulating the sparsity of pulses fired by a single neuron, a sparse pulse coding method is proposed to encode the EEG signal segment sparsely to obtain a sparse pulse sequence, thus solving the problem that the sparsity of firing leads to information coding redundancy and affects coding accuracy.

[0085] Step S104: Construct an initial neuron model based on the pulse sequence and neuronal dynamics mechanism;

[0086] Neurons are the basic functional units of the nervous system and fundamental components of neural network models. The dynamics of neurons can be explained as the processes of neuronal potential changes and the generation and transmission of membrane potential.

[0087] An initial neuron model constructed based on pulse sequences and neuronal dynamics can be a mathematical model used to simulate the interactions between neurons in the brain. It is based on the connections and electrical activity transmission between neurons, as well as their dynamic behavior.

[0088] Step S105: Based on the internal plasticity regulation mechanism of neuronal synapses, the initial neuron model is updated to construct an adaptive neuron model.

[0089] In this embodiment, the phenomenon of simultaneous excitatory and inhibitory neurons in the brain can be simulated to update the initial neuron model, thereby constructing an adaptive neuron model. This adaptively changes the neuron membrane potential, allowing the neuron model to adaptively adjust the input to control the membrane potential input, thus making the pulse delivery more accurate.

[0090] Step S106: Based on the adaptive neuron model, construct a spiking neural network, which is used to perform feature mapping on the pulse sequence.

[0091] In this embodiment, an adaptive neuron model is used to design a spiking neural network that uses pulses for information transmission or processing. The adaptive neuron model is built upon the initial neuron model and is also a mathematical model used to simulate the interaction between neurons in the brain, that is, a mathematical description of the behavior of biological neurons. The neuron model is the basic computational unit of the spiking neural network, and establishing a suitable adaptive spiking neuron model is of great significance for simulating neural network functions and solving practical problems. When constructing a spiking neural network, the adaptive neuron model can be selected as the basic unit in the network. Each neuron processes the input pulse by simulating changes in membrane potential and generates an output pulse under specific conditions. Then, by designing the topology and determining the hierarchy, neurons can be connected into a spiking neural network through synapses. The constructed spiking neural network can be used to achieve nonlinear feature mapping of pulse sequences.

[0092] Step S107: Based on the plasticity regulation mechanism inside the neuronal synapse, the spiking neural network is updated to obtain the target spike recognition model.

[0093] Because the firing of the peak-precursor component of traceable epileptiform spikes is dominated by a large number of excitatory neurons in the epileptogenic zone, while the peak-postcursor component is dominated by inhibitory neurons, it indicates that the propagation regulation process of traceable epileptiform spikes involves both excitatory and inhibitory neurons. However, in some techniques, when performing nonlinear input-output mapping based on the Leaky Integrated-and-Fire neuron model, the lack of consideration for the inhibitory neuron model leads to the omission of constructing a more biologically interpretable neuron model, thus casting doubt on the rationality of the nonlinear mapping of spiking neural networks.

[0094] In this embodiment, the target spike recognition model based on adaptive spiking neural network takes into account both excitatory and inhibitory neurons, and realizes the reception, processing and transmission of spikes in a more reasonable way that is more in line with the actual dynamic regulation of the nervous system. This ensures the rationality of the feature mapping of the spiking neural network and improves the accuracy of the spike recognition model in recognizing non-invasive traceable epileptiform spikes.

[0095] This embodiment provides a method for constructing a spike recognition model, which can be used in the aforementioned computer equipment. Figure 2 This is a flowchart of another spike recognition model construction method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0096] Step S401: Acquire EEG signals; for details, please refer to step S101 of the above embodiment, which will not be repeated here.

[0097] Step S402: Segment the EEG signal to obtain spike wave EEG signal segments; for details, please refer to step S102 of the above embodiment, which will not be repeated here.

[0098] Step S403: Pulse coding is performed on the spike wave EEG signal segment to obtain a pulse sequence; for details, please refer to step S103 of the above embodiment, which will not be repeated here.

[0099] Step S404: Preset the resting-state membrane potential, membrane capacitance, and membrane resistance of the neuron. The resting-state membrane potential refers to the membrane potential of the neuron when there is no external stimulus. The resting-state membrane potential, membrane capacitance, and membrane resistance of the neuron can be preset in advance.

[0100] Step S405: In the presence of membrane current input, determine the membrane potential of the neuron, wherein the membrane current is obtained by weighting the pulse sequence of the presynaptic neuron.

[0101] Membrane potential refers to the electrical potential difference across a neuron, which can change over time. In the presence of membrane current input, the change in membrane potential depends on membrane capacitance and membrane resistance.

[0102] In one example, the membrane current can be obtained according to the following formula:

[0103]

[0104] Where i = 1, 2, ..., N is the number of neurons, and O(t) is the encoded pulse sequence. φ represents the weights of the feedforward spiking neural network, j = 1, 2, ..., and φ-1 represents the number of layers in the feedforward spiking neural network.

[0105] Step S406: Based on resting-state membrane potential, membrane potential, membrane capacitance, and membrane resistance, an initial neuron model is established. This initial neuron model is used to describe the dynamic chaotic system when endogenous signals in the brain are transmitted to the measurement space. The measurement space refers to the region used to measure or observe brain activity.

[0106] In one example, a Leaky Integrated-and-Fire neuron model is constructed using mathematical descriptions of biological neurons to describe the dynamical chaotic system as endogenous signals are transmitted from the brain to the measurement space:

[0107]

[0108] Where V(t) is the membrane potential of a single neuron at time t, V rest The resting membrane potential has a time constant τ. m For film capacitance C m and film resistance R mThe product of and , I(t) represents the total input current generated by the presynaptic neuron pulse arriving at the triggered synapse.

[0109] In some alternative implementations, a φ-1 layer fully connected feedforward neural network is constructed to perform nonlinear feature mapping on the input pulse sequence, thereby obtaining pulse features:

[0110]

[0111] Where j = 1, 2, ..., φ-1, and W represents the weight values ​​of the neural network.

[0112] In this embodiment, the established initial neuron model can be used to describe the complex activities of neurons in the brain, providing a foundation for improving the accuracy of the target spike recognition model.

[0113] Step S407: Based on the internal plasticity regulation mechanism of neuronal synapses, the initial neuron model is updated to construct an adaptive neuron model. For details, please refer to step S105 of the above embodiment, which will not be repeated here.

[0114] Step S408: Based on the adaptive neuron model, a spiking neural network is constructed. The spiking neural network is used for feature mapping of the pulse sequence. For details, please refer to step S106 of the above embodiment, which will not be repeated here.

[0115] Step S409: Based on the plasticity regulation mechanism within the neuronal synapse, the spiking neural network is updated to obtain the target spike recognition model. For details, please refer to step S107 of the above embodiment, which will not be repeated here.

[0116] like Figure 3 As shown, in some optional embodiments, step S105 above updates the initial neuron model based on the neuronal synaptic plasticity regulation mechanism to construct an adaptive neuron model. This process includes the following steps:

[0117] The initial neuron model is as follows:

[0118]

[0119] Among them, V t Let τ be the membrane potential at time t, and τ = C. m R m C is the time constant. m R m Let represent the membrane capacitance and the membrane resistance, respectively, and let I(t) represent the total input current generated when the presynaptic neuron pulse arrives at the triggered synapse;

[0120] Step S501, Discretize the initial neuron model. The discretized initial neuron model is as follows:

[0121]

[0122] Among them, V t-1 Let I be the membrane potential at time t-1. t Let be the membrane current at time t;

[0123] Step S502: After a neuron fires a pulse, the neuron's membrane potential is reset to the resting membrane potential V. rest =0, setting the membrane capacitance to a preset value, and updating the discretized initial neuron model. We can assume the membrane capacitance C is... m If = 1, then the model is updated as follows:

[0124]

[0125] Step S503: Determine the adaptive function. The adaptive function can be composed of convolution operations and the sigmoid function to adaptively adjust the input to control the neuron's membrane potential, ensuring that the output value is between 0 and 1.

[0126] F t =σ(f(δ) t-1 ))

[0127] Among them, F t For the output value, δ t-1 This represents the pulse firing at time t-1, σ represents the sigmoid function, and f represents the convolution operator.

[0128] Step S504: Determine the adaptive neuron model based on the updated initial neuron model and the adaptive function.

[0129] In this embodiment, the simultaneous existence of excitatory and inhibitory neurons in the brain can be simulated to update the initial neuron model, thereby constructing an adaptive neuron model. This model adaptively changes the neuron membrane potential, allowing the neuron model to adaptively adjust its input to control the membrane potential, thus making pulse delivery more accurate. The target spike recognition model, based on an adaptive spiking neural network, considers both excitatory and inhibitory neurons, achieving pulse reception, processing, and transmission in a more reasonable and realistic manner that aligns with the dynamic regulation of the actual nervous system. This ensures the rationality of the spiking neural network feature mapping and improves the accuracy of the spike recognition model in identifying non-invasive, traceable epileptiform spikes.

[0130] In some alternative implementations, the adaptive neuron model is as follows:

[0131]

[0132] Among them, V tV is the membrane potential at time t. t-1 Let τ be the membrane potential at time t-1, and τ = C. m R m C is the time constant. m R m These represent membrane capacitance and membrane resistance, respectively.

[0133] In some optional implementations, step S107 above, which involves updating the spiking neural network based on the intrasynaptic plasticity regulation mechanism of neurons to obtain the target spike recognition model, includes:

[0134] Based on the intrasynaptic regulatory mechanism of the Dirac function, the parameter values ​​of neurons in a spiking neural network are updated. The intrasynaptic regulatory mechanism can be described as follows:

[0135]

[0136]

[0137] Where, τ ip The relative integration resolution of the initial neuron model and the intrasynaptic plasticity regulation mechanism, where β is the scale factor and r is the scale factor. C =1 / C m C m For film capacitance, r R =1 / R m R m Given the membrane resistance, the output response h of the neuron to input I can be described as:

[0138]

[0139] Wherein, δ(tt) (f) ) is the Dirac function, t (f) Let f represent the pulse emitted at time t, and η represent the pulse intensity.

[0140] In one example, by utilizing the intrasynaptic plasticity regulation mechanism of neurons, a feedforward spiking neural network is updated to obtain a target spike recognition model. This model can achieve high-precision recognition of traceable epileptiform spike signal segments. Due to the synaptic regulation mechanism based on the Dirac function, the network parameters can be efficiently optimized by updating the parameter values ​​of neurons in the spiking neural network at time t.

[0141] The epileptiform spike S input is defined based on the network output expectation. n Attribute tags

[0142]

[0143] Where W represents the learnable weights of the network.

[0144] In this embodiment, a self-feedback time-delayed synaptic regulation mechanism was explored that can maintain the dynamic balance between excitatory and inhibitory neurons and take into account the millisecond-level delay of synaptic neurotransmitter release. This mechanism accelerates the convergence of spiking neural networks and accurately identifies non-invasive traceable epileptiform spikes.

[0145] This embodiment provides a spike wave recognition method, which can be used in the aforementioned computer equipment. Figure 4 This is a flowchart of a spike recognition method according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps:

[0146] Step S601: Acquire the EEG signal to be identified.

[0147] Step S602: Input the EEG signal to be identified into the spike recognition model to identify the traceable epileptiform spike signal segment of the EEG signal to be identified.

[0148] In this embodiment of the invention, the spike recognition model is constructed using the spike recognition model construction method of any one of the above embodiments.

[0149] The spike recognition method provided in this embodiment uses a spike recognition model to identify traceable epileptiform spikes in the EEG signal to be identified, thereby improving the accuracy of identifying traceable epileptiform spikes in EEG signals and providing support for subsequent diagnosis and treatment.

[0150] This embodiment also provides a spike recognition model construction device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0151] This embodiment provides a spike recognition model construction device, such as... Figure 5 As shown, it includes:

[0152] Acquisition module 701 is used to acquire electroencephalogram (EEG) signals;

[0153] The segmentation module 702 is used to segment the electroencephalogram (EEG) signal to obtain spike wave EEG signal segments;

[0154] Encoding module 703 is used to pulse encode spike wave EEG signal segments to obtain pulse sequences;

[0155] Neuron model building module 704 is used to build an initial neuron model based on pulse sequences and neuron dynamics mechanisms;

[0156] The adaptive neuron model building module 705 is used to update the initial neuron model and build an adaptive neuron model based on the internal plasticity regulation mechanism of neuronal synapses.

[0157] The spiking neural network construction module 706 is used to construct a spiking neural network based on an adaptive neuron model. The spiking neural network is used to perform feature mapping on the spiking sequence.

[0158] The network update module 707 is used to update the spiking neural network based on the plasticity regulation mechanism inside the neuronal synapse to obtain the target spike recognition model.

[0159] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0160] In this embodiment, the spike recognition model construction device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0161] This embodiment provides a spike wave recognition device, such as... Figure 6 As shown, it includes:

[0162] The EEG signal acquisition module 801 is used to acquire the EEG signal to be identified.

[0163] The spike recognition module 802 is used to input the EEG signal to be recognized into the spike recognition model and identify the traceable epileptiform spike signal segment of the EEG signal to be recognized.

[0164] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0165] In this embodiment, the spike recognition device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0166] This invention also provides a computer device having the above-described features. Figure 5 The spike recognition model construction device shown is the same as the one described above. Figure 6 The spike wave recognition device shown.

[0167] Please see Figure 7 , Figure 7 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 7 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 7 Take a processor 10 as an example.

[0168] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0169] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0170] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0171] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0172] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0173] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0174] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for constructing a spike recognition model, characterized in that, The method includes: Acquiring EEG signals; The EEG signal is segmented to obtain spike wave EEG signal segments; The spike wave EEG signal segment is pulse-coded to obtain a pulse sequence; An initial neuron model is constructed based on the pulse sequence and neuronal dynamics mechanism. This includes: pre-setting the resting-state membrane potential, membrane capacitance, and membrane resistance of the neuron; determining the membrane potential of the neuron in the presence of membrane current input, wherein the membrane current is obtained by weighting the pulse sequence of the presynaptic neuron; and establishing the initial neuron model based on the resting-state membrane potential, the membrane potential, the membrane capacitance, and the membrane resistance. This initial neuron model is used to describe the dynamic chaotic system when endogenous signals are transmitted from the brain to the measurement space. Based on the internal plasticity regulation mechanism of neuronal synapses, the initial neuron model is updated to construct an adaptive neuron model; wherein, the initial neuron model is: ;in, for Membrane potential at time t, It is a time constant. These represent the membrane capacitance and the membrane resistance, respectively. This represents the total input current generated by the presynaptic neuron impulse arriving at the triggered synapse; the initial neuron model is discretized: ;in, for Membrane potential at time t, for The membrane current at any given time; when a neuron fires a pulse, the neuron's membrane potential is reset to the resting-state membrane potential, the membrane capacitance is set to a preset value, and the discretized initial neuron model is updated; an adaptive function is determined to adaptively adjust the input to control the neuron's membrane potential, ensuring the output value is between 0 and 1, the adaptive function being: ;in, The output value is... Indicates in The pulse is emitted at a specific time. This represents the sigmoid function. Represents the convolution operator; determines the adaptive neuron model based on the updated initial neuron model and the adaptive function; Based on the aforementioned adaptive neuron model, a spiking neural network is constructed; wherein, the process includes: selecting the adaptive neuron model as the basic unit in the network, each neuron processing the input pulse by simulating changes in membrane potential; the spiking neural network is used to perform feature mapping on the pulse sequence. Based on the intrasynaptic plasticity regulation mechanism of neurons, the spiking neural network is updated to obtain a target spike recognition model; wherein, it includes: updating the parameter values ​​of neurons in the spiking neural network based on the intrasynaptic plasticity regulation mechanism of the Dirac function; the intrasynaptic plasticity regulation mechanism of neurons is described as follows: ; ;in, This represents the relative integration resolution of the initial neuron model and the intrasynaptic plasticity regulation mechanism of the neuron. As a scale factor, , , For film capacitors, For membrane resistance, neuronal input Output response Described as: ;in, For the Dirac function, Indicates in Pulse emitted at any time , Indicates pulse intensity.

2. The method according to claim 1, characterized in that, The adaptive neuron model is as follows: in, This indicates element-wise multiplication.

3. A spike wave recognition method, characterized in that, The method includes: Acquire the EEG signal to be identified; The EEG signal to be identified is input into the spike recognition model to identify the traceable epileptiform spike signal segment of the EEG signal to be identified, wherein the spike recognition model is constructed using the spike recognition model construction method of claim 1 or 2.

4. A spike wave recognition model construction device, characterized in that, The device includes: The acquisition module is used to acquire electroencephalogram (EEG) signals; The segmentation module is used to segment the electroencephalogram (EEG) signal to obtain spike wave EEG signal segments; The encoding module is used to perform pulse encoding on the spike EEG signal segment to obtain a pulse sequence; A neuron model construction module is used to construct an initial neuron model based on the pulse sequence and neuronal dynamics mechanism; wherein, it includes: pre-setting the resting-state membrane potential, membrane capacitance, and membrane resistance of the neuron; determining the membrane potential of the neuron in the presence of membrane current input, wherein the membrane current is obtained by weighting the pulse sequence of the presynaptic neuron; and establishing the initial neuron model based on the resting-state membrane potential, the membrane potential, the membrane capacitance, and the membrane resistance, wherein the initial neuron model is used to describe the dynamic chaotic system when endogenous signals in the brain are transmitted to the measurement space; An adaptive neuron model construction module is used to update the initial neuron model based on the intrasynaptic plasticity regulation mechanism to construct an adaptive neuron model; wherein, the initial neuron model is: ;in, for Membrane potential at time t, It is a time constant. These represent the membrane capacitance and the membrane resistance, respectively. This represents the total input current generated by the presynaptic neuron impulse arriving at the triggered synapse; the initial neuron model is discretized: ;in, for Membrane potential at time t, for The membrane current at any given time; when a neuron fires a pulse, the neuron's membrane potential is reset to the resting-state membrane potential, the membrane capacitance is set to a preset value, and the discretized initial neuron model is updated; an adaptive function is determined to adaptively adjust the input to control the neuron's membrane potential, ensuring the output value is between 0 and 1, the adaptive function being: ;in, The output value is... Indicates in The pulse is emitted at a specific time. This represents the sigmoid function. Represents the convolution operator; determines the adaptive neuron model based on the updated initial neuron model and the adaptive function; A spiking neural network construction module is used to construct a spiking neural network based on the adaptive neuron model; wherein, it includes: selecting the adaptive neuron model as the basic unit in the network, each neuron processing the input pulse by simulating changes in membrane potential; the spiking neural network is used to perform feature mapping on the pulse sequence; The network update module is used to update the spiking neural network based on the intrasynaptic plasticity regulation mechanism of neurons to obtain a target spike recognition model; wherein, it includes: updating the parameter values ​​of neurons in the spiking neural network based on the intrasynaptic plasticity regulation mechanism of the Dirac function; the intrasynaptic plasticity regulation mechanism of neurons is described as follows: ; ;in, This represents the relative integration resolution of the initial neuron model and the intrasynaptic plasticity regulation mechanism of the neuron. As a scale factor, , , For film capacitors, For membrane resistance, neuronal input Output response Described as: ;in, For the Dirac function, Indicates in Pulse emitted at any time , Indicates pulse intensity.

5. A spike wave recognition device, characterized in that, The device includes: The module for acquiring EEG signals to be identified is used to acquire EEG signals to be identified. The spike recognition module is used to input the EEG signal to be recognized into the spike recognition model and recognize the traceable epileptiform spike signal segment of the EEG signal to be recognized, wherein the spike recognition model is constructed using the spike recognition model construction method of claim 1 or 2.

6. A computer device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the spike recognition model construction method of claim 1 or 2, or the spike recognition method of claim 3.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the spike recognition model construction method of claim 1 or 2, or the spike recognition method of claim 3.

Images