近红外无创血糖测量—位置浮动基准的光谱修正方法及仿体实验

发布时间：2018-10-21 13:10

【摘要】：近红外光谱无创血糖测量技术是至今尚未实现临床应用的国际前沿课题。活体自身背景变化是目前影响近红外无创血糖测量的重要干扰因素,位置基准参考测量方法将有助于解决该问题。论文围绕位置浮动基准点展开,从模拟、理论和实验三方面对位置浮动基准点进行研究,重点分析了基准光谱修正方法。基于蒙特卡罗模拟方法,论文研究了三层皮肤模型的位置浮动基准点的存在特性,结果验证了位置浮动基准点在皮肤模型中存在。分析了皮肤组织光学特性(吸收系数和散射系数)和皮肤组织结构特性(表皮层厚度和真皮层厚度)与位置浮动基准点的关系,得出位置浮动基准点随皮肤组织特征而变化的范围,模拟结果为光学检测探头结构设计提供依据。论文分析了人体无创血糖检测中的各种干扰因素及其消除方法,介绍了近红外无创血糖检测中常用的数学建模方法,及光谱数据处理方法,并分析了这些方法的优点和不足。在此基础上,理论推导了基准光谱修正方法,分析了利用该方法消除人体背景噪声干扰的优点。根据蒙特卡罗模拟结果,论文设计了用于位置浮动基准点实验的光纤测头,并基于5%的Intralipid葡萄糖溶液,完成了葡萄糖浓度梯度实验。根据实验得到的基准光谱和测量光谱,论文研究了基准光谱修正方法,结果表明:与应用原始的测量光谱的模型相比,该方法建立的偏最小二乘模型的预测误差平均减小了35%。该方法能够提高校正模型的稳定性和葡萄糖的预测精度。
[Abstract]:Non-invasive near-infrared spectroscopy (NIR) measurement of blood glucose is an international frontier subject that has not yet been applied in clinical practice. Background change in vivo is an important interference factor affecting NIR noninvasive blood glucose measurement. The reference method of position reference will be helpful to solve this problem. This paper focuses on the position floating datum, and studies the position floating datum from three aspects: simulation, theory and experiment. Based on the Monte Carlo simulation method, the existence characteristics of the position floating reference point of the three-layer skin model are studied, and the results verify the existence of the position floating reference point in the skin model. The relationship between the optical properties of skin tissue (absorption coefficient and scattering coefficient) and skin tissue structure (epidermis thickness and dermis thickness) and the position floating reference point are analyzed. The range of the position floating reference points varying with the skin tissue characteristics is obtained, and the simulation results provide the basis for the structural design of the optical detection probe. In this paper, the interference factors and their elimination methods in noninvasive blood glucose detection of human body are analyzed. The mathematical modeling methods and spectral data processing methods commonly used in NIR noninvasive blood glucose detection are introduced. The advantages and disadvantages of these methods are also analyzed. On this basis, the reference spectral correction method is derived theoretically, and the advantages of using this method to eliminate human background noise interference are analyzed. Based on the Monte Carlo simulation results, the optical fiber probe used in the position floating reference point experiment is designed, and the glucose concentration gradient experiment is completed based on the 5% Intralipid glucose solution. According to the experimental datum spectrum and measurement spectrum, this paper studies the correction method of datum spectrum. The results show that compared with the original model, the prediction error of the partial least square model is reduced by 35% on average. This method can improve the stability of calibration model and the precision of glucose prediction.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.51

【参考文献】