摘要： 生物神经细胞具有复杂多样的空间几何形态,而这种形态结构是研究单个神经细胞信息处理和整个神经系统连接的基础。为此,提出一种针对三维神经元几何形态的发育生成方法。采用人工基因组对基因调控网络进行编码,用基因表达的动态特性来表示神经元树突树的发育过程。实验结果表明,生成的虚拟神经元与实际神经元具有相似的几何形态结构。与已有基于发育机制的生长方法相比,更真实地表现了生物神经元的发育过程。与基于统计分析的重建方法相比,能够更方便地得到几何形态参数的统计分布函数。

关键词: 虚拟神经元, 神经元形态, 基因调控网络, 发育机制, 参数分析

Abstract: Biological neural cells have complex and diverse spatial geometry morphologies,and the geometry morphologies are the basis of single neuron cell information processing and whole nervous system connections.This paper proposes a novel developmental method for generating the 3D neuron morphology.Using the artificial genome to encode the genetic regulatory network,the dynamic features of gene expression are used to express as the developmental process of dendritic morphologies.Experimental results show that the generated virtual neurons have similar geometric structure with the actual neurons.Compared with the existing growth methods based on developmental mechanism,the proposed method shows the development of biological neurons more realistically.Compared with the reconstruction method based on statistical analysis,the statistical distribution function of geometric morphological parameters can be obtained more conveniently.

Key words: virtual neuron, neuron morphology, genetic regulatory network, developmental mechanism, parameter analysis

中图分类号: 

• TP183