中红外光谱无创血糖测量基本问题研究

发布时间：2018-08-22 16:13

【摘要】：糖尿病是一种严重危害人类健康的重大疾病，目前还没有根治的手段。为了有效地降低并发症的发病率，需要对血糖进行频繁测量。实现血糖水平的无创伤、连续测量具有重要的现实意义。中红外（Mid-Infrared，MIR）光谱对分子的选择性好，不同分子之间的光谱重叠和干扰不严重，信息丰富。中红外光谱无创血糖测量方法研究已经开展了很多年，但一直没有得到突破性的研究成果。该方法所面临的主要问题在于：a)缺乏对中红外光谱无创血糖测量方法中存在的光与生物组织的相互作用、葡萄糖的来源、光谱与血糖之间的关系等问题的细致研究，因此，也缺乏提高分析结果精度的有效手段。b)人体皮肤是一复杂、时变的系统，其光谱易受到一些物理、化学因素的干扰，会直接影响定标模型的可靠性与稳健性。c)与血糖有关的吸收信号非常微弱，给信息的提取带来了困难。本文对中红外光谱无创血糖测量方法中存在的上述问题进行了研究，主要内容如下： 1）对中红外光谱无创血糖测量方法中存在的光与生物组织的相互作用、皮肤光谱中与血糖有关的吸收信息来源、组织液中葡萄糖含量与血糖水平的关系等问题进行了分析，提出利用中红外光谱测量表皮组织液中葡萄糖含量来实现血糖无创伤测量的物理模型。与分析血液信息相比，组织液光谱中不存在血红蛋白等生化成分的干扰，较容易实现无创伤血糖测量。 2）对中红外光谱分析组织液中葡萄糖的定标方法进行了研究。中红外光谱分析技术中的化学计量学信息提取方法直接影响测量结果的精度与可靠性。本文研究结果表明，分析组织液中葡萄糖含量时，偏最小二乘回归（Partial Least Squares Regression，PLSR）方法能获得较好的测量结果。 3）利用胶带粘贴、水冲洗及打磨等方式分别对皮肤进行预处理，研究皮肤预处理方式对其衰减全反射(Attenuated Total Reflection,ATR)红外光谱的影响，并分析造成这种影响的机制。结果表明，光程变化是导致皮肤光谱变化的主要原因。当光程变化信息与葡萄糖的浓度信息之间存在偶然相关时，所得PLSR模型的分析精度与稳健性都会受到影响；利用附加散射校正（Multiplicative Scatter Correction，MSC）、变量标准化（StandardNorrmalized Variate，SNV）方法对可以对光谱中的光程信息进行校正。 4）根据PLSR算法原理，提出了评估光谱干扰因素对定标结果影响的方法；结合具体实验，对中红外光谱无创血糖测量方法中存在的光谱测量时机、角质层的吸收变化、接触压力等干扰因素影响进行研究。并通过口服葡萄糖耐量实验（Oral Glucose Tolerance Test，，OGTT）实验，发现表皮中的组织液很难渗透到皮肤角质层中进而被倏逝波所探测到。 5）提出了中红外光谱无创血糖测量系统的物理设计方案，对提高光谱信噪比的技术手段进行了分析。通用的光谱仪不能满足中红外无创血糖测量的要求，基于可调谐量子级联光器激(Quantum Cascade Laser,QCL)光源的光谱测量系统物理设计方案能有效提高所得皮肤光谱的信噪比，有助于实现血糖的无创伤测量。
[Abstract]:Diabetes mellitus is a major disease that seriously endangers human health. There is no radical cure at present. In order to effectively reduce the incidence of complications, it is necessary to measure blood sugar frequently.
Mid-Infrared (MIR) spectroscopy has good selectivity to molecules, and the overlap and interference between different molecules are not serious, and the information is abundant. There are many problems in the non-invasive measurement of blood glucose by external spectroscopy, such as the interaction between light and biological tissue, the source of glucose, the relationship between spectrum and blood glucose. Therefore, there is no effective means to improve the accuracy of the analysis results. b) Human skin is a complex, time-varying system, and its spectrum is susceptible to some physical and chemical factors. Interference will directly affect the reliability and robustness of the calibration model. c) The absorption signal related to blood glucose is very weak, which makes it difficult to extract information.
In this paper, the problems in the non-invasive measurement of blood glucose by mid-infrared spectroscopy are studied.
1) The interaction between light and biological tissues, the source of absorption information related to blood glucose in skin spectra, the relationship between glucose content in tissue fluid and blood glucose level were analyzed. The method of measuring glucose content in epidermal tissue fluid by mid-infrared spectroscopy was proposed. Physical model of non-invasive measurement. Compared with the analysis of blood information, there is no interference of hemoglobin and other biochemical components in tissue-fluid spectra, so it is easier to achieve non-invasive blood glucose measurement.
2) The calibration method of glucose in tissue fluids by mid-infrared spectroscopy is studied. The stoichiometric information extraction method in mid-infrared spectroscopy directly affects the accuracy and reliability of measurement results. Ression (PLSR) method can get better results.
3) The skin was pretreated by adhesive tape, water washing and polishing, and the effect of skin pretreatment on ATR infrared spectra was studied, and the mechanism of the effect was analyzed. When there is an accidental correlation between chemical information and glucose concentration information, the analytical accuracy and robustness of the PLSR model will be affected, and the optical path information in the spectrum can be corrected by using the method of Multiplicative Scatter Correction (MSC) and Standard Norrmalized Variate (SNV).
4) According to the principle of PLSR algorithm, a method for evaluating the influence of spectral interference factors on the calibration results is proposed. Combined with specific experiments, the influence of spectral measurement timing, cuticle absorption changes, contact pressure and other interference factors in the non-invasive mid-infrared blood glucose measurement method is studied. Coste Tolerance Test (OGTT) showed that it was difficult for tissue fluids in the epidermis to penetrate into the cuticle and be detected by evanescent waves.
5) The physical design scheme of the mid-infrared noninvasive blood glucose measurement system is proposed, and the technical means to improve the signal-to-noise ratio of the spectrum are analyzed. The case can effectively improve the signal to noise ratio of the obtained skin spectrum, and help to achieve the non-invasive measurement of blood sugar.
【学位授予单位】：暨南大学
【学位级别】：博士
【学位授予年份】：2012
【分类号】：R318.0

【参考文献】