Classification and Quantitative Analysis of Azithromycin Tablets by Raman Spectroscopy and Chemometrics

Raman spectroscopy has been proven a noninvasive technique with high potential in pharmaceutical industry. In this study, micro Raman technique and chemometric tools were used for identification of azithromycin (AZM) tablets by different manufacturers and quantitative analysis of the active pharmaceutical ingredient (API) in the samples. Support vector machine (SVM), Bayes classifier and K-nearest neighbour (KNN) were employed for identification, partial least squares (PLS) regression was used for quantitative determination, and interval partial least squares (iPLS) and Monte Carlo based uninformative variable elimination (MC-UVE) methods were used to select informative variables for improving the models. The results show that all the samples can be classified into groups by manufacturers with high accuracy, and the correlation coefficient between the predicted API concentrations and reference values is as high as 0.96. Therefore, micro Raman spectroscopy coupled with chemometrics may be a fast and powerful tool for identification and quantitative determination of pharmaceutical tablets.

可解释深度学习在光谱和医学影像分析中的应用

深度学习是一种基于神经网络的建模方法,通过不同功能感知层的构建获得优化模型,提取大量数据的内在规律,实现端到端的建模。数据规模的增长和计算能力的提高促进了深度学习在光谱及医学影像分析中的应用,但深度学习模型可解释性的不足是阻碍其应用的关键因素。为克服深度学习可解释性的不足,研究者提出并发展了可解释性方法。根据解释原理的不同,可解释性方法划分为可视化方法、模型蒸馏及可解释模型,其中可视化方法及模型蒸馏属于外部解释算法,在不改变模型结构的前提下解释模型,而可解释模型旨在使模型结构可解释。本文从算法角度介绍了深度学习及三类可解释性方法的原理,综述了近三年深度学习及可解释性方法在光谱及医学影像分析中的应用。多数研究聚焦于可解释性方法的建立,通过外部算法揭示模型的预测机制并解释模型,但构建可解释模型方面的研究相对较少。此外,采用大量标记数据训练模型是目前的主流研究方式,但给数据的采集带来了巨大的负担。基于小规模数据的训练策略、增强模型可解释性的方法及可解释模型的构建仍是未来的发展趋势。

Titanium dioxide as an adsorbent to enhance the detection ability of near-infrared diffuse reflectance spectroscopy

A method for quantitative determination of fish sperm deoxyribonucleic acid(fsDNA)was developed by using titanium dioxide(TiO2)as an adsorbent and near-infrared diffuse reflectance spectroscopy(NIRDRS).The selective enrichment of fsDNA was proved by comparing the adsorption efficiency of bovine serum albumin,tyrosine and tryptophan,and the low adsorption background of TiO2 was illustrated by comparing the spectra of four commonly-used inorganic adsorbents(alkaline aluminium oxide,neutral aluminium oxide,nano-hydroxyapatite and silica).The spectral feature of fsDNA can be clearly observed in the spectrum of the sample.Partial least squares(PLS)model was built for quantitative determination of fsDNA using 28 solutions,and 13 solutions with interferences were used for validation of the model.The results showed that the correlation coefficient(R)between the predicted and the reference concentration is 0.9727 and the recoveries of the validation samples are in the range of 98.2%-100.7%.

A variable importance criterion for variable selection in near-infrared spectral analysis

Variable selection is a universal problem in building multivariate calibration models, such as quantitative structure-activity relationship(QSAR) and quantitative relationships between quantity or property and spectral data. Significant improvement in the prediction ability of the models can be achieved by reducing the bias induced by the uninformative variables. A new criterion,named as C, is proposed in this study to evaluate the importance of the variables in a model. The value of C is defined as the average contribution of a variable to the model, which is calculated by the statistics of the models built with different combinations of the variables. In the calculation, a large number of partial least squares(PLS) models are built using a subset of variables selected by randomly re-sampling. Then, a vector of the prediction errors, in terms of root mean squared error of cross validation(RMSECV), and a matrix composed of 1 and 0 indicating the selected and unselected variables can be obtained. If multiple linear regression(MLR) is employed to model the relationship between the RMSECVs and the matrix, the coefficients of the MLR model can be used as a criterion to evaluate the contribution of a variable to the RMSECV. To enhance the efficiency of the method, a multi-step shrinkage strategy was used. Comparison with Monte Carlo-uninformative variables elimination(MC-UVE), randomization test(RT) and competitive adaptive reweighted sampling(CARS) was conducted using three NIR benchmark datasets. The results show that the proposed criterion is effective for selecting the informative variables from the spectra to improve the prediction ability of models.

Determination of triglycerides in human serum by near-infrared diffuse reflectance spectroscopy using silver mirror as a substrate

Near infrared diffuse reflectance spectroscopy(NIRDRS) has gained wide attention due to its convenience for rapid quantitative analysis of complex samples. A method for rapid analysis of triglycerides in human serum using NIRDRS with silver mirror as the substrate is developed. Due to the even and high reflectance of the silver mirror, the spectral response is enhanced and the background interference is reduced.Furthermore, both linear and nonlinear modeling strategies were investigated adopting the partial least squares(PLS) and least squares support vector regression(LS-SVR), continuous wavelet transform(CWT)was used for spectral preprocessing, and variable selection was tried using Monte Carlo uninformative variable elimination(MC-UVE), randomization test(RT) and competitive adaptive reweighted sampling(CARS) for optimization the models. The results show that the determination coefficient(R) between the predicted and reference concentration is 0.9624 and the root mean squared error of prediction(RMSEP) is 0.21. The maximum deviation of the prediction results is as low as 0.473 mmol/L. The proposed method may provide an alternative method for routine analysis of serum triglycerides in clinical applications.

Near-infrared spectroscopy and chemometric modelling for rapid diagnosis of kidney disease

Rapid diagnosis is important for efficient treatment in clinical medicine.This study aimed at development of a method for rapid and reliable diagnosis using near-infrared(NIR)spectra of human serum samples with the help of chemometric modelling.The NIR spectra of sera from 48 healthy individuals and 16 patients with suspected kidney disease were analyzed.Discrete wavelet transform(DWT)and variable selection were adopted to extract the useful information from the spectra.Principal component analysis(PCA),linear discriminant analysis(LDA)and partial least squares discriminant analysis(PLSDA)were used for discrimination of the samples.Classification of the two-class sera was obtained using LDA and PLSDA with the help of DWT and variable selection.DWT-LDA produced 93.8%and 83.3%of the recognition rates for the validation samples of the two classes,and 100%recognition rates were obtained using DWT-PLSDA.The results demonstrated that the tiny differences between the spectra of the sera were effectively explored using DWT and variable selection,and the differences can be used for discrimination of the sera from healthy and possible patients.NIR spectroscopy and chemometrics may be a potential technique for fast diagnosis of kidney disease.

Chemometric methods for extracting information from temperature-dependent near-infrared spectra

Temperature-dependent near-infrared(NIR) spectroscopy is a new technique for measuring the NIR spectra of a sample at different temperatures. Taking the advantage of the temperature effect, the technique has shown its potential in both quantitative and qualitative analysis. The technique has been proved to be powerful in determination of the analytes in complex samples,particularly in studying the functions of water in aqueous systems due to the significant effect of temperature on the NIR spectra of water. Because of the complicated interactions in the samples and the overlapping of the broad peaks in NIR spectra, it is difficult to extract the temperature-dependent information from the spectra. Chemometric methods, therefore, have been developed for improving the spectral resolution and extracting the temperature-induced spectral information. In this review, recent advances in the studies of chemometric methods and the applications in resolution, quantitative and structural analysis of temperature-dependent NIR spectra were summarized.

Deciphering the Mechanism Involved in the Switch On/Off of Molecular Pistons

生产在分子的水平把精力变换成机械工作的机器是一条重要小径探索显微镜的世界。一种合用的分子的引擎， β-cyclodextrin (β-CD ) 镇静，芳基，链烯和酰胺一半用模拟与免费精力的计算相结合的分子的动力学被调查。理解怎么综合链烯工作在引擎上执行了的两倍契约控制，原型的二链烯异构体被看作二台分子的引擎。描出的免费精力的侧面酰胺(Z) 的有约束力的过程 - 并且为有 1-adamantanol 的每链烯异构体的(E) 异构体为明显的工作在酰胺契约上执行了的链烯(E)-isomer, 显示那是 1.6 kcal/mol，当链烯(Z) 异构体是不能的表现时，工作。因此，链烯契约的异构化引起的引擎的直接开关开／关被见证，与试验性的大小一致。进不同部件和结构的分析的免费精力的侧面的分解建议链烯契约的异构化相对 CD 的洞控制芳基单位的位置，在免费精力的侧面和工作的鲜明对比之中导致差别表现在引擎上。