正交信号校正在人体血糖近红外光谱分析中的研究  

Study of Orthogonal Signal Correction Analyze in Human Serum Glucose with Near Infrared Spectroscopy

作　　者：罗运平[1] 张军[1] 郑仁富[2] 曹燕[2] 温旺荣[2] 陈星旦[1] 

机构地区：[1]暨南大学光电工程系,广州市黄埔大道西601号510632 [2]暨南大学附属第一医院临床学检验中心,广州市510632

出　　处：《光谱实验室》2007年第5期773-777,共5页Chinese Journal of Spectroscopy Laboratory

摘　　要：本文以血清的近红外光谱为研究对象,首先运用载荷向量分析确定最佳定标谱区为4192—4943cm-1与5294—7200cm-1两个波段,然后分别用平滑、一阶导数、正交信号校正(OSC)预处理方法,结合偏最小二乘回归方法建立了血清中血糖的模型。用平滑、一阶导数所建立模型的预测标准偏差(RMSEP)分别是0.545、0.568,用OSC校正后所建立模型的RMSEP是0.390,结果表明使用OSC校正对光谱进行合理的校正,能够滤除光谱矩阵与浓度矩阵无关的信号,降低模型的因子数,从而降低模型的复杂性,提高模型的稳健性。Two optimal spectral region 4192-4943cm^-1 and 5294-7200cm^-1 are selected based on human serum near-infrared spectroscopy for human glucose scale by loading vector analysis method. The spectra are preprocessed by smooth, first derivative and orthogonal signal correction （OSC） data preprocessing methods. The calibration models of serum glucose are established with partial least squares regression （PI.SR）. The root mean squared errors of prediction （RMSEP） are 0.545 and 0. 568 for spectra pretreated by smooth and first derivative preprocessing methods. But the RMSEP is 0. 390 for OSC spectra preprocessing methods. The OSC preprocessing method can correct the spectral matrix,and remove the information that is orthogonal to the concentration matrix and can reduce the factors number of calibration. Thereby the complexity is diminished and the stability is improved to the serum glucose model by OSC method.

关 键 词：正交信号校正 偏最小二乘回归 血糖 近红外光谱 

分 类 号：O657.33[理学—分析化学]