基于主成分分析和反向传播神经网络的肝癌细胞后向散射显微.PDF

? 246 ? Journal of Biomedical Engineering, Apr. 2017, Vol. 34, No.2 ? ·论 著· 基于主成分分析和反向传播神经网络的肝癌 细胞后向散射显微光谱判别 杨静1，王成1，谢成颖2，翁小阜2，魏勋斌2 1. 上海理工大学 医疗器械与食品学院 生物医学光学研究所（上海 ?200093） 2. 上海交通大学 生物医学工程学院 Med-X研究院（上海 ?200030） 【摘要】 ? 为了实现血液中肝癌细胞的自动识别，本文基于主成分分析（PCA）和反向传播（BP）神经网络算 法对三种细胞（小鼠的白细胞、红细胞和人体肝癌细胞 HepG2）进行了识别研究。利用光纤共聚焦后向散射 （FCBS）光谱仪获取光谱数据后进行 PCA，选取前两个主成分作为光谱的特征，建立一个具有 2 个输入层节点、11 个隐层节点、3 个输出节点的神经网络模式识别模型。选取 195 例对象数据训练???模型，随机抽取 150 组数据作 为训练集，45 组数据作为测试集，验证模型给出的细胞是否识别准确。结果显示三种细胞的整体识别准确率在 90% 以上，平均相对偏差只有 4.36%。实验结果预示采用 PCA+BP 算法能够从红细胞和白细胞中自动识别肝癌细 胞，这将为研究肝癌的转移与肝癌的生物代谢特性提供有利的工具。 【关键词】 ?显微光谱；肝癌细胞；主成分分析法；反向传播神经网络；细胞识别 Backscatter micro-spectra discrimination of liver cancer cell based on principal component analysis arithmetic and back propagation neural network YANG Jing1, WANG Cheng1, XIE Chengying2, WENG Xiaofu2, WEI Xunbin2 1. Laboratory of Biomedical Optics Optometry, School of Medical Instrument and Food Engineering, University of Shanghai for Science and Technology, Shanghai 200093, P.R.China 2. Institute of Med-X, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai 200030, P.R.China Corresponding author: WANG Cheng, Email: shhwangcheng@163.com 【Abstract】 In order to achieve the automatic identification of liver cancer cells in the blood, the present study adopted a principal component analysis (PCA) and back propagation (BP) algorithm of feedforward neural networks to identify white blood cells and red blood cells in mice and human liver cancer cells, HepG2. The present paper shows the process in which PCA was carried out after obtaining spectral data by fiber confocal back-scattering spectrograph, selecting the first two principal components as spectral features, and establishing a neural network pattern recognition model with two input layer nodes, eleven hidden layer nodes and three output nodes. In order to verify whether the model would give accura