神经元形态分类和识别方法的研究

Research on classification and identification of neuronal morphology

以数学建模的方式对神经元的形态分类进行了研究。首先通过神经元的几何模型数据,得到一系列刻画神经元空间形态的特征参数;然后根据特征参数建立特征参数的主成分分析数学模型,提取了神经元的综合形态特征;最后依据综合形态特征,提出了一种基于系统聚类分析法和贝叶斯判别法的神经元空间形态分类算法。实验结果验证了该分类算法的正确性。

It investigates morphological classification of neurons by means of mathematical modeling. A series of characteristic parameters is firstly set up by geometric model data of neurons. Then establish the geometric parameters of neurons for principal component analysis, and extract general features of neurons. Finally, in the light of general morphological characteristics of neurons, present a classification algorithm which uses neurons space form based on Hierarchical cluster analysis and Bayesian Criterion. The result verifies the correctness of the classification algorithm.