基于晶状体中微量元素与白内障关系的广义回归神经网络模式识别

The Pattern of General Regression Neural Network for Recognition of Relation between the Microamount of Elements in Crystalline Lens and Cataract

选择微量元素Sr,Mg,Na,K,Mn,Cu,Fe和Zn在晶状体中的含量作为识别白内障患者的指标,建立了广义回归神经网络(GRNN)模式识别。选择20个样本为训练集,5个样本为预测集。结果表明,与BP神经网络相比,该种网络具有设计简单与收敛快的优点,对给定的数据能完全识别,预示着通过对晶状体中的微量元素含量的分析,可能作为白内障患者诊断的一种辅助手段。

The contents of eight microamount elements （Sr, Mg, Na, K, Mn, Cu and Fe） in crystalline lens were chosen as recognition index of cataract disease patients and normal people. The pattern recognition of general regression neural network （GRNN） was established with twenty samples as a training group and five samples as a test group. The design of GRNN is simple and the calculation time needed by GRNN is significantly shorter compared with the BP neural network. The given data could be completely identifed,which indicates the method could be a supplementary tool to diagnose this kind of disease with the determined contents of microamount elements in crystalline lens.