CN110472733B - In-vivo neuron modeling method based on neuromorphics - Google Patents
In-vivo neuron modeling method based on neuromorphics Download PDFInfo
- Publication number
- CN110472733B CN110472733B CN201910660107.1A CN201910660107A CN110472733B CN 110472733 B CN110472733 B CN 110472733B CN 201910660107 A CN201910660107 A CN 201910660107A CN 110472733 B CN110472733 B CN 110472733B
- Authority
- CN
- China
- Prior art keywords
- neuron
- dendrite
- model
- compartment
- vivo
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- Molecular Biology (AREA)
- Neurology (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention relates to an in-vivo neuron modeling method based on neuromorphics, which comprises the following steps: firstly, constructing a dendrite single compartment model; secondly, constructing a multi-compartment cascade model: according to the specific morphological characteristics of the neuron dendrite, including the distribution area, the branch number and the topological structure of the dendrite, connecting the single compartment linear-nonlinear systems corresponding to N nodes according to the connection mode of the corresponding nodes, and establishing a multi-compartment cascade model for maintaining the original morphological characteristics of the neuron dendrite, so as to simulate the characteristics of the complete dendrite of the neuron. And thirdly, fitting the model parameters of the somatic neurons.
Description
Technical Field
The invention relates to the field of neuron modeling, in particular to an in-vivo neuron modeling method based on neuromorphics.
Background
The brain contains a large number of neurons which are connected with each other and play a very important role in the information processing and transmission process, and the neurons can receive information and output the information in a discharging mode. The neuron types are rich, and all have complex morphological structure characteristics. The neuron mainly comprises four parts of dendrite, cell body, axon and cell membrane, wherein the dendrite is used as an entrance for receiving input information of the neuron, has a complex morphological structure, can simultaneously receive a large amount of synaptic inputs from other neurons, and transmits integrated signals into the cell body after comprehensive integration, thereby generating neuron discharge. The dendrite can integrate synaptic input signals in different time and space, and the integration of the dendrite on input information plays a decisive role on the membrane potential of a cell body and finally influences the neuron discharge, namely the information transmission process, so that the construction of a neuron model is the basis for researching corresponding external stimulation.
At present, researchers have established various neuron models, such as a Hodgkin-Huxley model, an Izhikevich model, a FitzHugh-Nagulo model and the like, which solve a plurality of scientific problems to a certain extent, but ignore morphological and structural characteristics of neuron dendrites, and the application of the neuron models has certain defects in the process of researching the effect of neurons on information integration, so that research on an in-vivo neuron model based on neuromorphism has important significance.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an in-vivo neuron modeling method based on nerve morphology, which is characterized in that a single-compartment linear-nonlinear model of a neuron dendrite is established, a multi-compartment cascade system model of the dendrite whole is established according to the specific dendrite morphology of the neuron, and finally, a neuron discharge model is combined, and parameters of the neuron model are optimized by fitting by utilizing input signals and output discharge signals in a needling experiment. The key technical scheme adopted for solving the technical problems is as follows:
an in-vivo neuron modeling method based on neuromorphics, comprising the steps of:
first step, constructing a dendritic single compartment model
The dendrite of the neuron is divided into N nodes, that is, the dendrite is reduced to a cascade system formed by connecting N nodes, and a single-compartment linear-nonlinear system (X 1 、X 2 、……、X N ) Simulating a single node X i The single-chamber linear-nonlinear system can sequentially perform filtering treatment, linear superposition and nonlinear transformation on input signals, wherein the filtering treatment process sets a filtering threshold value, the linear superposition process needs to set weight distribution of each input stimulus, and the nonlinear transformation process needs to set membrane capacitance, axial conductance, input impedance, leakage current conductance and potassium-calcium ion channel.
Second step, constructing multi-compartment cascade model
According to the specific morphological characteristics of the neuron dendrite, including the distribution area, the branch number and the topological structure of the dendrite, a single compartment linear-nonlinear system (X 1 、X 2 、……、X N ) And connecting according to the connection mode of the corresponding nodes, and establishing a multi-compartment cascade model which keeps the original morphological characteristics of the neuron dendrites and is used for simulating the characteristics of the complete dendrites of the neurons.
Third step, fitting on the model parameters of the somatic neurons
According to the characteristics of neuron synapses obtained by in-vivo experiments, including the number of synapses, excitatory inhibitory distribution, synaptic conductance, attenuation rate, distance between synapses and cell bodies, spatial aggregation degree, spatial dispersion degree, time interval and time sequence distribution of synapses, a synapse input mode is designed, a pre-synaptic neuron discharge sequence measured by the experiments is input into a multi-compartment cascade model of dendrites, the processed signals are transmitted to a neuron cell body discharge model, action potentials are output, and compared with the post-synaptic discharge sequence obtained by the experiments, and parameters of the in-vivo neuron model are fitted and optimized to finally obtain the model of the in-vivo neuron.
Compared with the prior art, the invention has the beneficial effects that:
(1) The modeling method provided by the invention uses a single-compartment linear-nonlinear system to simulate a single node obtained by simplifying a neuron dendrite, and fully considers the active characteristic and the passive characteristic of the dendrite, so that the simulation rate is improved, and the integration effect of the neuron dendrite on synaptic inputs is better simulated;
(2) The modeling method of the invention connects single compartment linear-nonlinear systems according to the original topological structure of the neuron dendrite during modeling to form a multi-compartment cascade model of the dendrite, and fully considers the influence of the morphology structure of the neuron dendrite on the neuron model.
(3) The method utilizes the pre-synaptic neuron discharge signals measured in the needling experiments and the discharge signals output by the post-synaptic neurons to fit and optimize various parameters in the neuron model, so that the constructed neuron model is more similar to a real neuron.
Drawings
FIG. 1 is a schematic diagram of an in vivo neuron modeling method based on neuromorphics of the present invention
FIG. 2 is a schematic diagram of a single-chamber-based linear-nonlinear system according to the present invention
FIG. 3 is a schematic diagram of a parameter fitting of a neuron model constructed in accordance with the present invention
Detailed Description
The following description and drawings are illustrative of the embodiments of the present invention, but are not intended to limit the scope of the present invention.
An in vivo neuron modeling method (as shown in fig. 1-3) based on neuromorphics, the steps of the method are:
first step, constructing a dendritic single compartment model
Dividing the dendrite of the neuron into N nodes, namely, looking at a cascade system formed by connecting the N nodes, and further dividing each node X i Linear-nonlinear system (X) 1 、X 2 、……、X N ) The method comprises the steps of sequentially carrying out filtering treatment, linear superposition and nonlinear transformation on synaptic input signals in a single-compartment linear-nonlinear system, wherein a filtering threshold value is set in the filtering treatment process, weight distribution of each input stimulus needs to be set in the linear superposition process, and membrane capacitance, axial conductance, input impedance, leakage current conductance and a potassium-calcium ion channel need to be set in the nonlinear transformation process.
Second step, constructing multi-compartment cascade model
According to the dendritic morphological structure (distribution, branch number and topological structure) of different types of neurons, a single compartment linear-nonlinear system (X 1 、X 2 、……、X N ) And the nodes are connected according to the connection mode of the corresponding nodes, so that a multi-compartment cascade model which retains the morphological characteristics of the neuron dendrites is established and is used for simulating the whole dendrites.
Third step, fitting on the model parameters of the somatic neurons
According to the characteristics of neuron synapses (synapse number, excitatory inhibitory distribution, synaptic conductance, attenuation rate, distance between synapses and cell bodies, space aggregation degree, space dispersion degree, time interval and time sequence distribution) obtained by in-vivo experiments, a synapse input mode is designed, a pre-synaptic neuron discharge sequence measured by the experiments is input into a multi-compartment cascade model of dendrites, the processed signals are transmitted to a neuron cell body discharge model, action potentials are output, compared with the post-synaptic discharge sequence obtained by the experiments, and parameters of the in-vivo neuron model are fitted and optimized, so that the model of the in-vivo neuron is finally obtained.
Claims (1)
1. An in-vivo neuron modeling method based on neuromorphics, comprising the steps of:
first step, constructing a dendritic single compartment model
The dendrite of the neuron is divided into N nodes, that is, the dendrite is reduced to a cascade system formed by connecting N nodes, and a single-compartment linear-nonlinear system (X 1 、X 2 、……、X N ) Simulating a single node X i The method comprises the steps of sequentially carrying out filtering treatment, linear superposition and nonlinear transformation on input signals by a single-chamber linear-nonlinear system, wherein a filtering threshold value is set in the filtering treatment process, weight distribution of each input stimulus is required to be set in the linear superposition process, and a membrane capacitor, axial conductance, input impedance, leakage current conductance and a potassium-calcium ion channel are required to be set in the nonlinear transformation process;
second step, constructing multi-compartment cascade model
According to the specific morphological characteristics of the neuron dendrite, including the distribution area, the branch number and the topological structure of the dendrite, a single compartment linear-nonlinear system (X 1 、X 2 、……、X N ) Connecting according to the connection mode of the corresponding nodes, and establishing a multi-compartment cascade model which keeps the original morphological characteristics of the neuron dendrite and is used for simulating the characteristics of the complete dendrite of the neuron;
third step, fitting on the model parameters of the somatic neurons
According to the characteristics of neuron synapses obtained by in-vivo experiments, including the number of synapses, excitatory inhibitory distribution, synaptic conductance, attenuation rate, distance between synapses and cell bodies, spatial aggregation degree, spatial dispersion degree, time interval and time sequence distribution of synapses, a synapse input mode is designed, a pre-synaptic neuron discharge sequence measured by the experiments is input into a multi-compartment cascade model of dendrites, the processed signals are transmitted to a neuron cell body discharge model, action potentials are output, and compared with the post-synaptic discharge sequence obtained by the experiments, and parameters of the in-vivo neuron model are fitted and optimized to finally obtain the model of the in-vivo neuron.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910660107.1A CN110472733B (en) | 2019-07-22 | 2019-07-22 | In-vivo neuron modeling method based on neuromorphics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910660107.1A CN110472733B (en) | 2019-07-22 | 2019-07-22 | In-vivo neuron modeling method based on neuromorphics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110472733A CN110472733A (en) | 2019-11-19 |
| CN110472733B true CN110472733B (en) | 2023-05-02 |
Family
ID=68508768
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910660107.1A Active CN110472733B (en) | 2019-07-22 | 2019-07-22 | In-vivo neuron modeling method based on neuromorphics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110472733B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110956259B (en) * | 2019-11-22 | 2023-05-12 | 联合微电子中心有限责任公司 | Photon neural network training method based on forward propagation |
| CN112529166A (en) * | 2020-12-25 | 2021-03-19 | 中国科学院西安光学精密机械研究所 | Fusion neuron model, neural network structure, training and reasoning method, storage medium and device |
| CN114861864A (en) * | 2022-02-24 | 2022-08-05 | 天津大学 | Neuron network modeling method and device with dendritic morphology |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104798088A (en) * | 2012-11-20 | 2015-07-22 | 高通股份有限公司 | Piecewise linear neuron modeling |
| CN105631208A (en) * | 2015-12-25 | 2016-06-01 | 天津大学 | Data-driven acupuncture neural discharge reconfiguration platform |
| CN109325596A (en) * | 2018-09-17 | 2019-02-12 | 中国传媒大学 | A kind of Synaptic plasticity calculation method based on calcium concentration |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8914315B2 (en) * | 2012-01-27 | 2014-12-16 | International Business Machines Corporation | Multi-compartment neuron suitable for implementation in a distributed hardware model by reducing communication bandwidth |
| KR101593224B1 (en) * | 2014-06-10 | 2016-02-12 | 재단법인 대구경북과학기술원 | Reduced modelling method for neurons |
-
2019
- 2019-07-22 CN CN201910660107.1A patent/CN110472733B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104798088A (en) * | 2012-11-20 | 2015-07-22 | 高通股份有限公司 | Piecewise linear neuron modeling |
| CN105631208A (en) * | 2015-12-25 | 2016-06-01 | 天津大学 | Data-driven acupuncture neural discharge reconfiguration platform |
| CN109325596A (en) * | 2018-09-17 | 2019-02-12 | 中国传媒大学 | A kind of Synaptic plasticity calculation method based on calcium concentration |
Non-Patent Citations (1)
| Title |
|---|
| 金淇涛等.经颅磁刺激感应外电场作用下最小神经元模型放电起始动态机理分析.物理学报.2011,第61卷(第11期),517-526. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110472733A (en) | 2019-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110472733B (en) | In-vivo neuron modeling method based on neuromorphics | |
| Yu et al. | Biophysical neural spiking, bursting, and excitability dynamics in reconfigurable analog VLSI | |
| US11429857B2 (en) | Secure voice signature communications system using local and remote neural network devices | |
| JP2019517085A (en) | Artificial neuron | |
| CN107480597B (en) | Robot obstacle avoidance method based on neural network model | |
| CN111976733B (en) | Method and system for continuously predicting steering intention of driver | |
| Hussain et al. | Delay learning architectures for memory and classification | |
| Rachmuth et al. | Transistor analogs of emergent iono‐neuronal dynamics | |
| CN113792863A (en) | Impulse neural network modeling method, system and application thereof | |
| CN101916393B (en) | Realized circuit of pulse coupled neural network with function of image segmentation | |
| Zeng et al. | Temporal learning with biologically fitted SNN models | |
| Chatzipaschalis et al. | Parkinson's treatment emulation using asynchronous cellular neural networks | |
| Uwate et al. | Nonlinear Time Series Analysis of Spike Data of Izhikevich Neuron Model | |
| Navarro-Guerrero et al. | A neurocomputational amygdala model of auditory fear conditioning: A hybrid system approach | |
| Chauvet et al. | A class of functions for the adaptive control of the cerebellar cortex | |
| Kulkarni et al. | Scalable digital CMOS architecture for spike based supervised learning | |
| Kaminski et al. | Liquid state machine built of Hodgkin–Huxley neurons | |
| Hui et al. | A plausible method for assembling a neural circuit for decision-making | |
| Zhao et al. | A hippocampus ca3 model with autoassociative and heteroassociative memory functions | |
| Valenza et al. | Stochastic modeling of spontaneous bursting activity to simulate neural responses of in-vitro networks on multielectrode arrays | |
| Li et al. | Multi-delay dynamic neural group network global asymptotic stability and synchronization state analysis | |
| AHUJA | Implementing Spike-Timing Dependent Plasticity on the Izhikevich model of a Spiking Neuron | |
| Kapoor et al. | Effect of event sequence on intragroup dynamics: Simulation using modified Hopfield neural network | |
| CN116774601A (en) | Neuron simulation device and simulation system | |
| Slavova et al. | Nano Computing in Bioinspired Systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |