Department of Mechanical Engineering, National Cheng Kung University, Tainan, Taiwan.
Academy of Innovative Semiconductor and Sustainable Manufacturing, National Cheng Kung University, Tainan, Taiwan.
J Biophotonics. 2024 Sep;17(9):e202400052. doi: 10.1002/jbio.202400052. Epub 2024 Jul 2.
A Mueller matrix polarimetry system at 532 nm wavelength is developed for noninvasive glucose sensing in turbid media such as human's fingertip. The system extracts mean absorbance and anisotropic properties, demonstrated numerically and experimentally with phantom glucose samples. It is found that mean absorbance ( ), depolarization index (Δ), and linear dichroism (LD) show linear variation with glucose concentration 100-500 mg/dL. In addition, LightTools simulations indicate proportional scaling of scattering effects with , Δ, and LD. Real-world tests on fingertip show a strong correlation between these properties and blood glucose levels with a mean absolute relative deviation (MARD) of 12.56% and a correlation coefficient (R) of 0.875 in prediction by a neural network (NN) model, highlighting the advantages of Mueller matrix in extracting more parameters related to blood glucose.
我们研发了一种工作在 532nm 波长的 Mueller 矩阵偏振光系统,用于探测混浊介质(如人体指尖)中的血糖浓度。系统通过数值模拟和实验,提取了平均吸光度和各向异性参数。实验结果表明,平均吸光度( )、退偏指数(Δ)和线性二色性(LD)与 100-500mg/dL 浓度范围内的葡萄糖浓度呈线性变化。此外,LightTools 模拟表明散射效应与 、Δ 和 LD 呈比例缩放关系。在指尖的实际测试中,这些特性与血糖水平之间具有很强的相关性,神经网络(NN)模型预测的平均绝对相对偏差(MARD)为 12.56%,相关系数(R)为 0.875,突出了 Mueller 矩阵在提取更多与血糖相关参数方面的优势。