Model design for networks of heterogeneous Hodgkin-Huxley neurons
Model design for networks of heterogeneous Hodgkin-Huxley neurons
复制标题
DOI：
10.1016/j.neucom.2022.04.115
复制
发表时间：
2022-05-13
期刊：
影响因子：
6
通讯作者：
Astolfi, A.
中科院分区：
文献类型：
--
作者：
Giannari, A. G.;Astolfi, A.
我们提出了一种新颖的模块化、可扩展和适应性强的建模框架，可以基于受控制启发的反馈结构，准确地模拟由具有不同动态特性和不同放电模式的神经元组成的神经元网络。我们考虑三类重要的神经元：抑制性快速尖峰神经元、兴奋性规则尖峰适应神经元和兴奋性内在爆发神经元。我们还考虑了神经元互连的两种基本方式：电突触和化学突触。通过将神经元动力学与网络动力学分离，我们开发了一种完全灵活的反馈结构，可以进一步增强该结构以合并其他类型的神经元和/或突触。我们使用 Hodgkin-Huxley 模型的增强版本来描述单个神经元动力学，并使用图论来定义网络结构。我们提供了小型基本神经元基序以及更大的神经元网络的模拟结果，并验证了所提出方法的准确性、灵活性和可扩展性。因此，我们为全面的建模框架提供了基础，该框架能够模仿单个神经元和神经元网络的动力学，并能够复制基本的正常大脑功能。所提出的框架的结构非常适合控制和优化方法的应用，既可以模拟药理物质的作用，也可以模拟患病的神经元和网络条件。 (C) 2022 作者。由 Elsevier B.V. 出版
We present a novel modular, scalable and adaptable modelling framework to accurately model neuronal networks composed of neurons with different dynamic properties and distinct firing patterns based on a control-inspired feedback structure. We consider three important classes of neurons: inhibitory Fast spiking neurons, excitatory regular spiking with adaptations neurons, and excitatory intrinsic bursting neurons. We also take into consideration two basic means of neuronal interconnection: electrical and chemical synapses. By separating the neuronal dynamics from the network dynamics, we have developed a fully flexible feedback structure that can be further augmented to incorporate additional types of neurons and/or synapses. We use an augmented version of the Hodgkin-Huxley model to describe the individual neuron dynamics and graph theory to define the network structure. We provide simulation results for small fundamental neuron motifs as well as bigger neuronal networks and we verify the accuracy, flexibility and scalability of the proposed method. Therefore, we provide the basis for a comprehensive modelling framework that is able to imitate the dynamics of individual neurons and neuronal networks and is able to replicate basic normal brain function. The structure of the proposed framework is ideal for applications of control and optimization methods both for modelling the effect of pharmacological substances as well as for modelling diseased neuron and network conditions. (C) 2022 The Authors. Published by Elsevier B.V.