State Key Laboratory of Precision Measuring Technology and Instruments, Tianjin University, Tianjin, China.
Appl Spectrosc. 2017 Sep;71(9):2076-2082. doi: 10.1177/0003702817712227. Epub 2017 Jun 1.
Noninvasive sensing of blood glucose based on near-infrared (NIR) spectroscopy is a research hotspot in the biomedical field. However, its accuracy is severely limited by the weak specific signal of glucose and the strong background variations caused by other constituents in the blood, the measuring instrument, and the environment. In this paper, special source-detector distances, defined as the floating reference position, are used to conduct relative measurements and correct for background variations. These floating reference positions are chosen so that the diffuse reflectance is not sensitive to the change in glucose concentration due to the combined effects of absorption and scattering. Nine 10% intralipid samples with glucose concentrations in the range of 1000-5000 mg dL at an interval of 500 mg dL were prepared. Using a custom-built, continuously moving, spatially resolving, double-fiber measurement system with a superluminescent diode (SLD) as the light source, the diffuse reflectance of intralipid samples containing glucose under different source-detector distances (0.2-5 mm, with intervals of 0.2 mm) were collected. Then, a correlation analysis between the spectra and the glucose concentration was carried out to determine the floating reference position and the optimal measuring position. The signal in the floating reference position was used to correct the background variation because it contains the same systematic drift and interference as the signal in the optimal measuring position. The results showed that the correlation between the diffuse reflectance and the glucose concentration was increased significantly compared with traditional correction by subtracting the nearest spectrum of pure 10% intralipid solution. The correlation between the diffuse reflectance and the concentration of glucose is significantly increased, which indicated that the combination of the correlation analysis and a floating reference is able to eliminate the influence of background variations.
基于近红外（NIR）光谱的无创血糖传感是生物医学领域的一个研究热点。然而，由于葡萄糖的特异性信号较弱，以及血液、测量仪器和环境中的其他成分的背景变化较强，其准确性受到严重限制。在本文中，使用特殊的源-探测器距离（定义为浮动参考位置）进行相对测量，并校正背景变化。这些浮动参考位置的选择使得漫反射对于由于吸收和散射的综合作用而导致的葡萄糖浓度变化不敏感。制备了 9 个 10%中链甘油三酯（Intralipid）样品，葡萄糖浓度范围为 1000-5000mg/dL，间隔为 500mg/dL。使用具有超辐射二极管（SLD）作为光源的定制、连续移动、空间分辨、双光纤测量系统，收集不同源-探测器距离（0.2-5mm，间隔 0.2mm）下含有葡萄糖的 Intralipid 样品的漫反射率。然后，对光谱与葡萄糖浓度之间的相关性进行分析，以确定浮动参考位置和最佳测量位置。由于浮动参考位置中的信号包含与最佳测量位置中的信号相同的系统漂移和干扰，因此可以使用该信号来校正背景变化。结果表明，与传统的用纯 10%中链甘油三酯溶液的最近光谱相减的校正方法相比，漫反射率与葡萄糖浓度之间的相关性显著增加。漫反射率与葡萄糖浓度之间的相关性显著增加，这表明相关分析和浮动参考的结合能够消除背景变化的影响。
