College of Integrated Traditional Chinese and Western Medicine, Binzhou Medical University, Yantai 264003, China.
Shandong Runzhong Pharmaceutical Co., Ltd., Yantai 256603, China.
Molecules. 2023 Feb 13;28(4):1778. doi: 10.3390/molecules28041778.
A synergetic strategy was proposed to address the critical issue in the brand characterization of (Ejiao, CCA), a precious traditional Chinese medicine (TCM). In all brands of CCA, Dong'e Ejiao (DEEJ) is an intangible cultural heritage resource. Seventy-eight CCA samples (including forty DEEJ samples and thirty-eight samples from other different manufacturers) were detected by laser-induced breakdown spectroscopy (LIBS) and near-infrared spectroscopy (NIR). Partial least squares discriminant analysis (PLS-DA) models were built first considering individual techniques separately, and then fusing LIBS and NIR data at low-level. The statistical parameters including classification accuracy, sensitivity, and specificity were calculated to evaluate the PLS-DA model performance. The results demonstrated that two individual techniques show good classification performance, especially the NIR. The PLS-DA model with single NIR spectra pretreated by the multiplicative scatter correction (MSC) method was preferred as excellent discrimination. Though individual spectroscopic data obtained good classification performance. A data fusion strategy was also attempted to merge atomic and molecular information of CCA. Compared to a single data block, data fusion models with SNV and MSC pretreatment exhibited good predictive power with no misclassification. This study may provide a novel perspective to employ a comprehensive analytical approach to brand discrimination of CCA. The synergetic strategy based on LIBS together with NIR offers atomic and molecular information of CCA, which could be exemplary for future research on the rapid discrimination of TCM.
针对（阿胶，CCA）品牌特征的关键问题，提出了一种协同策略。在所有 CCA 品牌中，东阿阿胶（DEEJ）是一种非物质文化遗产资源。通过激光诱导击穿光谱（LIBS）和近红外光谱（NIR）检测了 78 个 CCA 样本（包括 40 个 DEEJ 样本和 38 个来自其他不同制造商的样本）。首先分别考虑单个技术构建了偏最小二乘判别分析（PLS-DA）模型，然后在低水平融合了 LIBS 和 NIR 数据。计算了分类准确率、灵敏度和特异性等统计参数来评估 PLS-DA 模型性能。结果表明，两种单独的技术都具有良好的分类性能，尤其是 NIR。经多重散射校正（MSC）预处理的单 NIR 光谱的 PLS-DA 模型表现出色。尽管单独的光谱数据获得了良好的分类性能，但也尝试了数据融合策略来融合 CCA 的原子和分子信息。与单个数据块相比，经 SNV 和 MSC 预处理的数据融合模型具有良好的预测能力，没有误分类。本研究可能为采用综合分析方法对 CCA 的品牌鉴别提供新视角。基于 LIBS 与 NIR 的协同策略提供了 CCA 的原子和分子信息，可为未来对中药的快速鉴别研究提供范例。
