Low temperature cleanup combined with magnetic nanoparticle extraction to determine pyrethroids residue in vegetables oils

To quantify trace pesticide residue in vegetable oil rapidly, low temperature cleanup combined with magnetic nanoparticle based solid phase extraction was developed to determine eight pyrethroids in vegetable oils, including tetramethrin, fenpropathrin, cypermethrin, decamethrin, fenvalerate, acrinathrin, permethrin and bifenthrin. Polystyrene coated magnetic nanoparticles were synthesised by a modified chemical coprecipitation combined with emulsion polymerisation method. The nanoparticles were afterwards characterised by Fourier transform-infrared spectroscopy, X-ray diffraction, transmission electron microscopy as well as vibrating sample magnetometer, and successfully employed as adsorbents for the magnetic solid phase extraction of pyrethroids which were cleaned up using low temperature approach in advance. Critical impact factors on the efficiency of the extraction method such as the mass of adsorbents used, volume and type of eluent solvent, extraction time as well as elution time were optimised subsequently. Regression analysis of the calibration curves of the eight pyrethroids yielded satisfactory correlation coefficients within the range of 0.980–0.998. Limit of detection and limit of quantification were calculated to be between 0.0290-0.0658 and 0.0890–0.1994 ng g⁻¹, respectively. Intra-day and inter-day reproducibility at different concentration levels also produced satisfactory recovery rates of 83.18–112.79% with relative standard deviations not exceeding 10.84% and 12.01%, respectively, suggesting desirable stability of the proposed method.     

基于高光谱成像技术的青花椒产地识别研究

目的 基于高光谱成像技术结合机器学习建立一种青花椒产地的快速识别方法,实现四川、贵州、云南、重庆等10个青花椒主要产地样品的快速无损鉴别。方法 本研究利用全平皿法、五点平均法和中心点法3种不同的兴趣区域(region of interest, ROI)提取方式,获得平行光谱数据,分别采用5种预处理方法消除数据噪声提升模型性能,并比较了偏最小二乘判别分析(partial least square-discriminant analysis, PLS-DA)、随机森林(random forests, RF)和支持向量机(support vector machine, SVM) 3种模型的产地识别效果。结果 采用全平皿法提取兴趣区域,通过二阶导(second derivative, D2)预处理后建立的RF模型分类效果最佳,训练集和测试集的准确率均可达到100%。进一步采用连续投影算法(successive projections algorithm, SPA)选择27个特征波长建模,结果表明多元散射校正(multiplicative scatter correction, MSC)-RF模...     

Viscosity transition of ZnO-containing rosin

Eutectic alloy Sn–9Zn is regarded as one of the potential substitutes for Pb-containing solders, although it has some drawbacks yet to overcome. One of the problems involved is drastic viscosity increase of rosin-based flux during soldering or storage. In the present work we investigated the role of ZnO in the viscosity transition of rosin. The results show that the viscosity transition is directly caused by the presence of ZnO in molten rosin and, at a certain temperature, there exists a critical ZnO concentration for its occurrence. The critical ZnO concentration increases with temperature. A graph is presented to show the conditions for the viscosity transition, in a ZnO concentration–temperature space. The transition is found to be reversible when shifting temperature or ZnO concentration across the boundary and to assume the features of a thermally activated phase transformation.     

基于深度学习的小样本中药材粉末显微图像识别

针对中药材种类繁多、数据量稀少以及难以对其导管进行分类的问题,提出一种基于多通道颜色空间与注意力机制模型的卷积神经网络改进方法。首先,采用多通道颜色空间将RGB颜色空间与其他颜色空间合并为6通道作为网络输入,使网络学习亮度、色调和饱和度等特征信息,弥补数据量的不足;其次,在网络中加入注意力机制模型,其中通道注意力模型将两个池化层紧密连接到一起,空间注意力模型将多尺度空洞卷积结合到一起,使网络将注意力聚焦于小样本中关键的特征信息。实验结果表明,针对34种中药材样本的8774张导管图像,采用多通道颜色空间和注意力机制模型的方法,与原始ResNet网络相比,准确率分别提升了1.8个百分点和3.1个百分点,将二者结合后准确率提升了4.1个百分点,说明所提方法对小样本分类的准确率有着大幅度的提升。     

Chemical constituents and biological research on plants in the genus Curcuma

Curcuma, a valuable genus in the family Zingiberaceae, includes approximately 110 species. These plants are native to Southeast Asia and are extensively cultivated in India, China, Sri Lanka, Indonesia, Peru, Australia, and the West Indies. The plants have long been used in folk medicine to treat stomach ailments, stimulate digestion, and protect the digestive organs, including the intestines, stomach, and liver. In recent years, substantial progress has been achieved in investigations regarding the chemical and pharmacological properties, as well as in clinical trials of certain Curcuma species. This review comprehensively summarizes the current knowledge on the chemistry and briefly discusses the biological activities of Curcuma species. A total of 720 compounds, including 102 diphenylalkanoids, 19 phenylpropene derivatives, 529 terpenoids, 15 flavonoids, 7 steroids, 3 alkaloids, and 44 compounds of other types isolated or identified from 32 species, have been phytochemically investigated. The biological activities of plant extracts and pure compounds are classified into 15 groups in detail, with emphasis on anti-inflammatory and antitumor activities.