• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文学位 >Optimization methods for quantitative measurement of glucose based on near-infrared spectroscopy.
【24h】

Optimization methods for quantitative measurement of glucose based on near-infrared spectroscopy.

机译：基于近红外光谱定量测量葡萄糖的优化方法。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The application of near-infrared (NW) spectroscopy in a dedicated glucose sensor requires portable, rugged, and low-cost instrumentation. Filter-based instruments have the potential to be used in this application, although a drawback to the use of these instruments is the lack of wavelength regions selective to spectral features of glucose. Therefore, a good filter design strategy is needed to tailor the filter set to this specific application.;In the first part of the research, a general filter design strategy is developed through a simulation approach. Gaussian or square-shaped optical filters were simulated to generate filter responses from single-beam spectra of six-component sample matrixes, which were collected by a Fourier transform infrared (FT-IR) spectrometer. The filter bandpass specifications, as well as the number of filters, are optimized by use of a genetic algorithm. It is found that a multiple linear regression (MLR) model using 11 simulated filters is successful for prediction of glucose over a long period of time, indicating the robustness of the model and stability of the filter-based method.;This filter-based strategy is further applied to measure glucose concentrations in rats using in vivo NW spectra of rat skin tissue collected during glucose clamp experiments. Results show that the predicted glucose concentrations are consistent with the reference values for all of the rats with the calibration models based on the optimal simulated filters when the prediction set is chosen randomly from all in vivo spectra for each rat. The filter-based method is also applied to glucose measurements in four-component matrixes with spectra collected by an acousto-optic tunable filter (AOTF) - based spectrometer. Both short-term and long-term data sets are used and the results demonstrate that MLR models with the optimal filters have good prediction performance.;The second part of the research involves employing particle swarm optimization (PSO) in an automated wavelength selection procedure for building multivariate calibration models based on partial least-squares regression. Two different fitness functions are studied, showing that both guide the PSO to find the best variable settings with similar prediction performance.
机译：近红外（NW）光谱在专用葡萄糖传感器中的应用需要便携式，坚固且低成本的仪器。基于过滤器的仪器具有在该应用中使用的潜力，尽管使用这些仪器的缺点是缺乏对葡萄糖的光谱特征具有选择性的波长区域。因此，需要一种良好的滤波器设计策略来使滤波器组适应该特定应用。在研究的第一部分中，通过一种仿真方法来开发一种通用的滤波器设计策略。模拟高斯或方形滤光片，以从六组分样品矩阵的单束光谱中生成滤光片响应，这些光谱由傅立叶变换红外（FT-IR）光谱仪收集。滤波器的带通规格以及滤波器的数量通过使用遗传算法进行了优化。发现使用11个模拟滤波器的多元线性回归（MLR）模型在长时间内成功地预测了葡萄糖，这表明该模型的鲁棒性和基于滤波器的方法的稳定性。使用在葡萄糖钳制实验期间收集的大鼠皮肤组织的体内NW光谱，将其进一步用于测量大鼠中的葡萄糖浓度。结果表明，当从每只大鼠的所有体内光谱中随机选择预测集时，基于基于最佳模拟滤波器的校准模型，预测的葡萄糖浓度与所有大鼠的参考值一致。基于过滤器的方法还适用于四分量矩阵中的葡萄糖测量，其光谱由基于声光可调过滤器（AOTF）的光谱仪收集。短期和长期数据集都被使用，结果表明具有最佳滤波器的MLR模型具有良好的预测性能。;第二部分研究涉及在自动波长选择过程中采用粒子群优化（PSO）基于偏最小二乘回归建立多元校准模型。对两个不同的适应度函数进行了研究，结果表明两者均指导PSO查找具有相似预测性能的最佳变量设置。

著录项

  • 作者

    Wu, Yuping.;

  • 作者单位

    The University of Iowa.;

  • 授予单位 The University of Iowa.;
  • 学科 Chemistry Analytical.
  • 学位 Ph.D.
  • 年度 2008
  • 页码 237 p.
  • 总页数 237
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 化学;
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号