血糖浓度的短波近红外定量分析  

The quantitative analysis of glucose in human serum by short-wave NIR spectra

作　　者：王伟[1,2] 李振庆[1,2] 倪一[1,2] 刘晨晨[1,2] 赵志强[3] 

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093 [2]上海理工大学上海市现代光学系统重点实验室,上海200093 [3]河南省电子产品质量监督检验所,河南郑州450003

出　　处：《光学仪器》2013年第3期20-24,共5页Optical Instruments

基　　金：国家自然科学基金资助项目(21205078);上海市科学发展基金资助项目(10540500700);上海市重点学科项目第三期资助项目(S30502)

摘　　要：采用间隔偏最小二乘法(IPLS)和移动窗口偏最小二乘方法(MWPLS),在640～1 100nm范围内建立血糖短波近红外的优化模型。使用马氏距离对人血清样品中的奇异样品进行筛选,将检测光谱分别等分为2～15份进行IPLS分析,对比建立预测模型。设窗宽为151nm,成分数范围(1～20),全谱进行MWPLS,对预测模型进行优化。结果显示,依据马氏距离采用最小半球体积法能有效筛选所采集光谱中的奇异光谱,IPLS可以有效地找到葡萄糖分子官能团对应的近红外特征谱段,MWPLS能够找到适合建模的精确起止波长点,通过偏最小二乘法建立血糖浓度的预测模型,相关系数R=0.982 2,预测均方差RMSEP=0.163 5 mmol/L,偏差Bias=-0.087 3mmol/L。The optimal model for the short wave near-infrared （640～1 100 nm） spectroscopic analysis of serum glucose was established by interval partial least square （IPLS） and moving window partial least square （MWPLS） methods. The outlier samples were eliminated based on mahalanobis distance, and then the spectra was equally divided into 2～15 parts for IPLS model. The spectra window for MWPLS was 151 nm, and the principal component was between 1 and 20. Results demonstrated that mahalanobis distance method can effectively eliminate the outlier samples. The optimal spectra domain by IPLS corresponded well with the functional groups of the glucose molecules. By MWPLS, the relative coefficient R, root meant square error for prediction RMSEP, and Bias error reached 0. 982 2, 0. 163 5 mmol/L, and -0. 087 3 mmol/L, respectively.

关 键 词：短波近红外光谱 血糖 间隔偏最小二乘法 移动窗口偏最小二乘法 

分 类 号：O657.33[理学—分析化学]