丘脑皮层神经群模型仿真及现场可编程门阵列实现研究  

Thalamocortical Neural Mass Model Simulation and Study Based on Field Programmable Gate Array

作　　者：梁振虎[1] 周敬亮 李小俚[1,2] 

机构地区：[1]燕山大学电气工程学院,秦皇岛066004 [2]北京师范大学认知神经科学与学习国家重点实验室,北京100875

出　　处：《生物医学工程学杂志》2016年第4期616-625,共10页Journal of Biomedical Engineering

基　　金：国家自然科学基金资助项目(61304247;81230023);中国博士后科学基金资助项目(2014M551051;2015T80231);河北省自然科学基金资助项目(F2014203127)

摘　　要：利用计算机模拟大脑的神经振荡对于分析大脑的功能机制具有重要的意义。丘脑皮层神经群模型(TNMM)是一种通过建立丘脑和皮层耦合关系来反映神经活动机制的模型,有助于我们理解大脑特定认知功能及各个节律脑电信号的神经振荡活动。随着神经群模型的复杂性和规模的不断增加,常规的计算机系统性能无法实现快速和大规模的模型仿真。本文提出了以FPGA为基础的硬件计算仿真方法,利用Altera公司的计算机辅助设计工具DSP Builder,同时结合MATLAB/Simulink实现复杂神经群算法模型的搭建和硬件平台移植。此方法充分利用FPGA的并行计算能力来实现大规模复杂神经群模型的快速仿真,为神经群模型的计算机实现提供了新的方案和思路。Using the computer to imitate the neural oscillations of the brain is of great significance for the analysis of brain functions.Thalamocortical neural mass model（TNMM）reflects the mechanisms of neural activities by establishing the relationships between the thalamus and the cortex,which contributes to the understanding of some specific cognitive functions of the brain and the neural oscillations of electroencephalogram（EEG）rhythms.With the increasing complexity and scale of neural mass model,the performance of conventional computer system can not achieve rapid and large-scale model simulation.In order to solve this problem,we proposed a computing method based on Field Programmable Gate Array（FPGA）hardware in this study.The Altera＇s DSP Builder module combined with MATLAB/Simulink was used to achieve the construction of complex neural mass model algorithm,which is transplanted to the FPGA hardware platform.This method takes full advantage of the ability of parallel computing of FPGA to realize fast simulation of large-scale and complex neural mass models,which provides new solutions and ideas for computer implementation of neural mass models.

关 键 词：丘脑皮层神经群模型 神经振荡 DSP BUILDER 现场可编程门阵列 脑电图 

分 类 号：R338[医药卫生—人体生理学] TP391.9[医药卫生—基础医学] TN791[自动化与计算机技术—计算机应用技术]