基于iPLS和SiPLS算法的人体血清胆红素含量的可见-近红外光谱建模  

Visible-near infrared spectroscopy modeling on the contents of serum bilirubin based on iPLS and SiPLS

作　　者：郭慧娴 朱思祁[1] 黎远鹏 方涛[1] 黄富荣[1] 郑仕富[2] 陈振强[1] 

机构地区：[1]暨南大学光电工程系,广东广州510632 [2]暨南大学第一附属医院临床检验中心,广东广州510632

出　　处：《光电子．激光》2016年第10期1136-1144,共9页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(31371785);广东省自然科学基金(S2011040001850);广东省战略新兴产业核心技术攻关(2012A032300016);高等学校博士学科点专项科研基金(20124401120005);广东光阵光电科技有限公司光电信息工程院士工作站(2014B090905001)资助项目

摘　　要：为了建立血清胆红素（BR,bilirubin）样品总胆红素（TBIL）、直接胆红素（DBIL）和间接胆红素（IBIL）近红外（NIR）光谱分析最优模型,利用可见-NIR透射光谱技术与间隔偏最小二乘法（iPLS）及协合区间偏最小二乘法（SiPLS）算法相结合对建模区域进行优选,实现血清光谱特征波段选择,建立光谱与血清BR成分之间的定量预测模型,以均方根误差（RMSE）作为模型评价标准。结果表明：SiPLS模型效果更佳,TBIL、DBIL和IBIL的最优建模波长范围分别为400～536nm、1 366～1 502nm和2 324～2 460nm,400～502nm、608～710nm和1 644～1 746nm,400～502nm和1 746～1 848nm;3种BR最优预测模型的RMSE分别为0.598 9、0.207 2和0.386 2μmol/L;波段优选对提高预测结果的准确性有重要的意义;采用SiPLS建立TBIL、DBIL和IBIL定量分析模型,不仅可以提高模型的预测精度,而且克服了iPLS单一区间建模的缺点,优选出的特征谱区还可为设计小型专用光谱分析仪器提供依据。To establish the optimum visible-near infrared（NIR）spectroscopy analysis model of total bilirubin（TBIL）,direct bilirubin（DBIL）and indirect bilirubin（IBIL）in human serum,the modeling region is optimized by combining transmitted spectrum technology with interval partial least squares（iPLS）or synergy interval partial least squares（SiPLS）to choose characteristic bands of serum spectrum.Quantitative prediction models of the relationship between spectrum and three serum proteins are established accordingly.Root-mean-square error（RMSE）prediction was used as an evaluation standard of models.Results demonstrate that the SiPLS prediction model is superior to the iPLS prediction model.In the SiPLS prediction model,the optimum modeling wavelength ranges of TBIL,DBIL and IBIL in serum are400～536nm,1 366～1 502 nm,2 324～2 460 nm and 400～502nm,608～710nm,1 644～1 746 nm,as well as 1644～1746nm,400～502nm,1746～1848nm,respectively.RMSEs of these three proteins are0.598 9μmol/L,0.207 2μmol/L and 0.386 2μmol/L,respectively.Waveband optimization is of impor-tant significance to improve prediction accuracy.The SiPLS-based quantitative analysis model of TBIL,DBIL and IBIL in serum has high prediction accuracy and overcomes shortcomings of the iPLS-based model.Furthermore,the optimized characteristic spectrum region can provide references for designing small spectrophotometers for special purposes.

关 键 词：可见-近红外(NIR)光谱 间隔偏最小二乘法(iPLS) 协合区间偏最小二乘法(SIPLS) 波段优选 血清胆红素(BR) 

分 类 号：R454[医药卫生—治疗学]