Chemometric Procedures Research Articles

Transfer of multivariate classification models applied to digital images and fluorescence spectroscopy data

This work evaluates the use of transfer of classification models for identifying adulteration of extra virgin olive oil (EVOO) samples involving, separately, two analytical techniques: fluorescence spectroscopy and digital imaging. The chemometric procedures, including development of classification models and application of classification transfer methods, were performed individually for each analytical technique. Methods of direct standardization (DS) and piecewise direct standardization (PDS) were applied to transfer samples sets in order to estimate an adjustment function and apply it to a samples set measured by the secondary instrument. For purposes of comparison, classification models were built based on linear discriminant analysis (LDA) with previous selection of variables by the successive projections algorithm (SPA), and partial least squares discriminant analysis (PLS-DA). The performance of the classification models was evaluated according to the number of errors and correct classification rate (CCR) for the prediction set measured by the secondary instrument. Before standardization, SPA-LDA and PLS-DA models achieved the same CCR using two analytical techniques: 54% for fluorescence emission spectra and 47% for histograms of digital images. After the standardization, a substantial increase of the CCR was observed. For the SPA-LDA models, a CCR of 88% was obtained for the fluorescence emission spectra and 82% for the histograms of the digital images. The PLS-DA classification models reached 85% and 76% of CCR for the fluorescence and imaging data, respectively, after standardization. These results demonstrate the efficiency of standardization procedures applied to multivariate classification models developed from fluorescence spectroscopy and digital images.

Attenuated total reflectance – Fourier transform infra-red spectral profiling of illicit heroin for forensic intelligence

An increasing source of illicit heroin, originating from the Golden Triangle or Afghanistan, has been shipped to other parts of Asia and to Oceania countries using Malaysian ports as transit hubs. Forensic drug profiling could provide useful information on the trafficking routes and the origin of drugs for intelligence purposes. This study is aimed at establishing the possible relationship among the heroin samples seized from different sources for predicting the trafficking network of illicit heroin in the northern region of Malaysia. One hundred and fifty-five seized heroin samples were analysed using Attenuated Total Reflectance-Fourier Transform Infra-red spectroscopy and the resulting spectral data points were subjected to chemometric procedures. Principal component analysis revealed the formation of four different clusters of illicit heroin: samples of high purity, samples cut with polysaccharide and caffeine, as well as samples of very low purity. Hierarchical cluster analysis grouped the samples into ten sub-clusters suggesting analytically distinguishable variability among the samples. In comparison, ten representative heroin samples seized from central regions of Malaysia gave significantly different profiles, separating them from those obtained in the northern region. This comparative information offers beneficial forensic intelligence to prove the existence of different clusters of trafficking networks.

Characterization of smokeless powders using multiplexed collision-induced dissociation mass spectrometry and chemometric procedures

This work demonstrates a non-targeted mass spectrometry approach for identification of organic compounds in smokeless powders. Unburned powders were removed from various commercial ammunitions of different brand, primer composition, caliber, and age. The unburned powders and corresponding fired residues were analyzed by liquid chromatography–atmospheric pressure chemical ionization-time-of-flight mass spectrometry (LC–APCI-TOFMS). Multiplexed collision-induced dissociation was performed at increasing collision potentials resulting in successive fragmentation that provided structural information for compound identification in a non-targeted manner. Nine compounds were identified in the powders, including akardite II, ethyl centralite, diphenylamine, N-nitrosodiphenylamine, and dibutyl phthalate.Multivariate statistical procedures were performed to first investigate association and discrimination of the unburned powders. Principal components analysis (PCA) of the chemical profiles suggested nine distinct groups of powders, according to the dominant organic compounds present. The clusters formed in hierarchical cluster analysis (HCA) were mostly in agreement with PCA groupings although HCA provided a metric to quantify the similarity. Finally, association of the fired residue to the corresponding unburned powder was possible although the success was highly dependent on the composition of the unburned powder and the extent of compound depletion as a result of firing.

Класифікація різних видів питного молока та молока-сировини з використанням ультразвукового аналізу та хемометричних інструментів

Ultrasound analysis for more than 30 samples of different kinds of milk has been made: natural whole milk and diluted with water in the amount from 5 to 20 % of mass, drinking kinds of pasteurized, ultra-pasteurized and lactose-free milk and also reconstituted milk. Such indexes of milk content as fat, protein, lactose, non- fat milk solids, water, density and the temperature of freezing have been determined. According to the gained results, combination of ultrasound method of milk analysis with other instrumental methods and the use of chemometric procedures to develop simple and cheap way to determine the adulteration are envisaged in further research. Keywords: milk, fat, protein, lactose, density, ultrasonic method, chemometric method

Class-Modelling Approaches: Advantages over Discriminant Methods for Compliance Verifications

Our traditional Chemometric Space is undergoing some change at the moment. I have asked a number of different chemometric specialists to provide us with a series of tutorial type articles in which they will describe the theory and practice of a range of chemometric procedures as applied to (near infrared) NIR spectral data. This series of column articles will, I hope, form an invaluable resource for both the novice and experienced NIR spectroscopist given that some of the techniques to be covered are not as yet routinely applied in all laboratories. As always, we appreciate feedback on this new development. – Ed.

CNS Multiparameter Optimization Approach: Is it in Accordance with Occam's Razor Principle?

A detailed analysis of the possibility of using the Multiparameter Optimization approach (MPO) for CNS/non-CNS classification of drugs was carried out. This work has shown that MPO descriptors are able to describe only part of chemical transport in the CNS connected with transmembrane diffusion. Hence the "intuitive" CNS MPO approach with arbitrary selection of descriptors and calculations of score functions, search of thresholds of classification, and absence of any chemometric procedures, leads to rather modest accuracy of CNS/non-CNS classification models.

Authentication ofCoffea arabicaaccording to Triacylglycerol Stereospecific Composition

Stereospecific analysis is an important tool for the characterization of lipid fraction of food products. In the present research, an approach to characterize arabica and robusta varieties by structural analysis of the triacylglycerol (TAG) fraction is reported. The lipids were Soxhlet extracted from ground roasted coffee beans with petroleum ether, and the fatty acids (FA) were determined as their corresponding methyl esters. The results of a chemical-enzymatic-chromatographic method were elaborated by a chemometric procedure, Linear Discriminant Analysis (LDA). According to the total and intrapositional FA composition of TAG fraction, the obtained results were able to characterize roasted pure coffee samples and coffee mixtures with 10% robusta coffee added to arabica coffee. Totally correct classified samples were obtained when the TAG stereospecific results of the considered coffee mixture (90 : 10 arabica/robusta) were elaborated by LDA procedure.

Urban stream Osówka in Szczecin – how efficiently as possible should the variability of the stream water quality be studied along the road of runoff in time?

The water quality of the stream Osowka in the NW part of the Szczecin city was studied. Water samples were collected in four seasons of 2014 – January, April, July and October in 6 sampling stations, from station near the source of the stream – to inflow of Osowka stream water to the lake Rusalka. 20 selected water chemistry indices were measured. The aim of discernment variability of water quality along stream runoff path during studied period chemometric procedures were conducted: cluster analysis (CA), calculations of linear regression equation y i =y i +∑ 6 j=1 α j S j +∑ 4 k=1 s k T k +SEE for selected water quality indices (y i ), where S j and T k – are sampling stations and sampling terms numbers – respectively, and factor analysis (FA). It has been shown significant variability of Osowka water quality along the runoff path and established what parameters characterizing the variability of stream water chemistry should be examined in the places of indicated sampling stations – the same as in the studies – the quantities of terms of sampling in 4 consecutive seasons.

Targeted Metabolomics Approach To Detect the Misuse of Steroidal Aromatase Inhibitors in Equine Sports by Biomarker Profiling.

The use of anabolic androgenic steroids (AAS) is prohibited in both human and equine sports. The conventional approach in doping control testing for AAS (as well as other prohibited substances) is accomplished by the direct detection of target AAS or their characteristic metabolites in biological samples using hyphenated techniques such as gas chromatography or liquid chromatography coupled with mass spectrometry. Such an approach, however, falls short when dealing with unknown designer steroids where reference materials and their pharmacokinetics are not available. In addition, AASs with fast elimination times render the direct detection approach ineffective as the detection window is short. A targeted metabolomics approach is a plausible alternative to the conventional direct detection approach for controlling the misuse of AAS in sports. Because the administration of AAS of the same class may trigger similar physiological responses or effects in the body, it may be possible to detect such administrations by monitoring changes in the endogenous steroidal expression profile. This study attempts to evaluate the viability of using the targeted metabolomics approach to detect the administration of steroidal aromatase inhibitors, namely androst-4-ene-3,6,17-trione (6-OXO) and androsta-1,4,6-triene-3,17-dione (ATD), in horses. Total (free and conjugated) urinary concentrations of 31 endogenous steroids were determined by gas chromatography-tandem mass spectrometry for a group of 2 resting and 2 in-training thoroughbred geldings treated with either 6-OXO or ATD. Similar data were also obtained from a control (untreated) group of in-training thoroughbred geldings (n = 28). Statistical processing and chemometric procedures using principle component analysis and orthogonal projection to latent structures-discriminant analysis (OPLS-DA) have highlighted 7 potential biomarkers that could be used to differentiate urine samples obtained from the control and the treated groups. On the basis of this targeted metabolomic approach, the administration of 6-OXO and ATD could be detected for much longer relative to that of the conventional direct detection approach.

Chemometric analysis of the volatile fraction evolution of Portuguese beer under shelf storage conditions

In this work we present a multivariate statistical analysis of the evolution of the volatile fraction of Portuguese beer over an extended period of 1year under standard shelf storage conditions, using gas chromatography coupled with mass spectrometry (GC–MS). A systematic methodology is proposed for detecting the onset of meaningful changes in chemical composition during shelf storage and to monitor its evolution along time. We also put forward and discuss chemometric procedures for analyzing the contributions of different chemical components in the definition of dynamic ageing trends. In summary, the chemometric analysis reveals that the chemical composition of beer presents a statistically meaningful deviation from the reference scenario after a period of 7months, although the deviation trend has its onset during the 6th month. The analysis performed also underlines the limitations of current variable contribution methods, and an alternative procedure was proposed based on the analysis in the original domain which finally led to a consistent and interpretable clustering structure of the volatile fraction compounds. Esters and higher alcohol compounds stand up on a cluster arrangement suggesting that their strict control can effectively point out meaningful changes on beer aroma. Organic acids, namely caprylic, capric and acetic acids can also be very helpful in that sense.

Classification of wine distillates using multivariate statistical methods based on their direct GC-MS analysis

AbstractThis work describes a novel methodology for the recognition of brandies based on direct injection of a raw sample followed by GC-MS analysis. Direct injection was chosen for its simplicity and the fact that the composition of the samples analysed remains unchanged compared to original brandy. The repeatability of the analytical procedure was evaluated by a comparison of the peak areas for randomly selected compounds obtained from 10 parallel measurements. A novel chemometric procedure was investigated in order to separate the samples studied on the basis of their geographical origin, processing technology or maturation time. In this procedure, a principal component analysis was applied to full chromatograms to select the time interval that shows the significant differences between the samples studied. It was shown that the chromatogram recorded at 36-39 min bore the maximal differences, hence it could be used to classify the brandy samples. The chromatographic peaks found within this time interval were identified and their peak areas determined. These compounds could be used as specific markers for determining geographical origin or processing technology.

TRACEABILITY OF OLIVE OIL BY CARBON STABLE ISOTOPES RATIO AND FATTY ACIDS COMPOSITION

Olive oil is a commodity for which high value is attached to the region of its origin. In Italy, higher prices can be asked for extra-virgin olive oils depending on the region of origin. Recently, the European law (EC Reg. 182/2009) requires the indication of the geographical origin for all extra-virgin olive oils, irrespective the protected denomination of origin (PDO) and/or protected geographic indication (PGI) of origin. Therefore, in the last few years attention has been focused on authentication methods for the geographical origin of olive oils. In this work some preliminary results on the estimation of olive oils origin by carbon isotope ratio measurements and fatty acids composition (UE Reg. 61/2011) have been obtained. Particularly, results from 80 extra virgin olive oil samples, produced during the crop year 2011, collected from Chile, Israel, Italy, Mexico, Portugal and Spain are presented and discussed. The aim of the study was to investigate if selected olive oils produced in areas with different climatic and geological characteristics could be discriminated on the basis of carbon isotopic ratio and fatty acids composition. All data obtained were treated by means of proper chemometric procedures. Finally, as preliminary results, a good discrimination between the different selected sites was achieved.

Fast determination of ethylene glycol, 1,2-propylene glycol and glycolic acid in blood serum and urine for emergency and clinical toxicology by GC-FID

A simple, cost effective, and fast gas chromatography method with flame ionization detection (GC-FID) for simultaneous measurement of ethylene glycol, 1,2-propylene glycol and glycolic acid was developed and validated for clinical toxicology purposes. This new method employs a relatively less used class of derivatization agents – alkyl chloroformates, allowing the efficient and rapid derivatization of carboxylic acids within seconds while glycols are simultaneously derivatized by phenylboronic acid. The entire sample preparation procedure is completed within 10min. To avoid possible interference from naturally occurring endogenous acids and quantitation errors 3-(4-chlorophenyl) propionic acid was chosen as an internal standard. The significant parameters of the derivatization have been found using chemometric procedures and these parameters were optimized using the face-centered central composite design. The calibration dependence of the method was proved to be quadratic in the range of 50–5000mgmL−1, with adequate accuracy (92.4–108.7%) and precision (9.4%). The method was successfully applied to quantify the selected compounds in serum of patients from emergency units.

The effect of the water on the curcumin tautomerism: A quantitative approach

The tautomerism of curcumin has been investigated in ethanol/water binary mixtures by using UV–Vis spectroscopy and advanced quantum-chemical calculations. The spectral changes were processed by using advanced chemometric procedure, based on resolution of overlapping bands technique. As a result, molar fractions of the tautomers and their individual spectra have been estimated. It has been shown that in ethanol the enol–keto tautomer only is presented. The addition of water leads to appearance of a new spectral band, which was assigned to the diketo tautomeric form. The results show that in 90% water/10% ethanol the diketo form is dominating. The observed shift in the equilibrium is explained by the quantum chemical calculations, which show that water molecules stabilize diketo tautomer through formation of stable complexes. To our best knowledge we report for the first time quantitative data for the tautomerism of curcumin and the effect of the water.

Non-Targeted Electrospray Mass Spectrometry-Based Metabolic Fingerprinting and PLS-DA-Based Classification of Brazilian "Malvas"

Malva sylvestris, popularly known as malva in Brazil, is a medicinal plant used for its antiinflammatory effects. However, multiple plants are known as malvas in Brazil; these species include Sida cordifolia (malva-branca) and Pelargonium graveolens (malva-cheirosa). In this work, electrospray mass spectrometry metabolic fingerprints from these species were obtained by direct infusion of extracts of commercial samples, purchased as crushed dried leaves. The spectral data were used to classify these species using a partial least squares discriminant analysis method. The identities of all samples were confirmed by morphological analyses, emphasizing the trichomes morphology. While over 80% of the samples were labeled as M. sylvestris or malva, in the present study only 37% of these samples were classified as M. sylvestris species. Approximately 50% of these samples were classified as S. cordifolia, whose consumption may be dangerous for patients with cardiovascular diseases. Thus, these analytical and chemometric procedures worked as a fast and simple method for classifying species of malvas.

Hyperspectral imaging of intact bell peppers

Agricultural engineering technologies have successfully addressed certain challenges by the use of advanced sensors and machine vision technologies. The objective of this study was to develop a non-destructive method to evaluate and to map quality indices in bell pepper. Three cultivars of bell pepper (‘Ever Green’, ‘No. 117’ and ‘Celica’) were studied during maturation by using hyperspectral imaging in the visible and near-infrared (550–850 nm) region. Peppers were marked in the flowering stage and 20 samples from each variety were collected weekly, along a growing period of seven weeks, until full growth. Quality parameters like total soluble solids, total chlorophyll, carotenoid and ascorbic acid content were determined and correlated with the spectral data. Images of intact peppers were collected by an acousto-optic-tuneable-filter (AOTF) hyperspectral charged-coupled-device (CCD) camera, in spectral resolution of 5 nm. Spectral information of the hyper cubes was analysed by chemometric procedures. Partial least squares regression was used for model development. Comparisons were made between the PLS regression analysis of the reflectance spectra (R), and the pre-processed spectra such as the first derivative (D1R), log(1/R), D1(log(1/R)) and D2(log(1/R)). Models were established to predict the quality attributes creating the basis for multiple sampling of a particular fruit or individual peppers from many fruits in the same time. High correlations were obtained by the established models with coefficients of determination of 0.95, 0.95, 0.97, and 0.72 for total soluble solids, total chlorophyll, carotenoid and ascorbic acid content, respectively.

Monitoring the polymorphic transformation on the surface of carbamazepine tablets generated by heating using near-infrared chemical imaging and chemometric methodologies

This work presents the monitoring of the polymorphic transformation of carbamazepine (form III to form I) in pharmaceutical formulations as tablets, generated by heating at a temperature of 140°C for 7h, with data acquisition in each hour. Near infrared chemical imaging in conjunction with chemometric procedures was used to evaluate the change on the distribution of each polymorph in the tablet surface. The concentration distribution maps of polymorphic forms I and III of carbamazepine in the tablet were estimated by matrix-augmented multivariate curve resolution (MCR) and parallel factor analysis (PARAFAC). The concentration profile as a function of time under heating allowed a dynamic study of the system. The results show that MCR was able to provide information about the polymorphic transformation of carbamazepine tablets in the dynamic process, generating distribution maps for each acquisition time. PARAFAC provided global information about the system, but it was not able to completely resolve the spectral components.

A fast chemometric procedure based on NIR data for authentication of honey with protected geographical indication

In this work, information contained in near infrared (NIR) spectra of honeys with protected geographical indication (PGI) “Mel de Galicia” was processed by means of different chemometric techniques to develop an authentication system for this high quality food product. Honey spectra were obtained in a fast and single way, and they were pretreated by means of standard normal variate transformation in order to remove the influence of particle size, scattering and other factors, and prior to their use as input data. As the first step in chemometric study, display techniques such as principal component analysis and cluster analysis were applied in order to demonstrate that the NIR data contained useful information to develop a pattern recognition classification system to authenticate honeys with PGI. The second step consisted in the application of different pattern recognition techniques (such as D-PLS: Discriminant partial least squares regression; SIMCA: Soft independent modelling of class analogy; KNN: K-nearest neighbours; and MLF-NN: Multilayer feedforward neural networks) to derive diverse models for PGI-honey class with the objective of detecting possible falsification of these high-quality honeys. Amongst all the classification chemometric procedures, SIMCA achieved to be the best PGI-model with 93.3% of sensitivity and 100% of specificity. Therefore, the combination of NIR information data with SIMCA developed a single and fast method in order to differentiate between genuine PGI-Galician honey samples and other commercial honey samples from other origins that, due to their lower price, could be used as substrates for falsification of genuine PGI ones.

Quantitative analysis of Cistanches Herba using high‐performance liquid chromatography coupled with diode array detection and high‐resolution mass spectrometry combined with chemometric methods

We aim to determine the chemical constituents of three species of Cistanches Herba using HPLC coupled with diode array detection and high-resolution MS. Ten phenylethanoid glycosides were identified and further quantified as marker substances by HPLC coupled with diode array detection method. The separation was conducted using an Agilent TC-C18 column with 0.1% formic acid and methanol as the mobile phases under gradient elution. The analytical method was fully validated in terms of linearity, sensitivity, precision, repeatability as well as recovery, and subsequently applied to evaluate the quality of 36 batches of Cistanche plants. The chemometric procedures (i.e., hierarchical clustering analysis and principal component analysis) were used to compare different species of Cistanches Herba, leading to successful classification of the Cistanche samples in accordance with their origins. In conclusion, this study provides a chemical basis for quality control of Cistanches Herba.

Harmonizing a chemometric procedure for the classification of illicit heroin using manufacturing impurities: preliminary investigation of heroin simulated links

This paper describes the steps involved in harmonizing a statistical method for the classification of illicit heroin using manufacturing impurity data. Sixteen linked samples were prepared from each of the five unrelated heroin seizures. Five links, totaling 80 samples were subjected to two sample weight approaches for liquid–liquid extraction. A wide range of intra-batch variations were observed in the dataset and this was probably due to the poor extraction efficiency and weight differences. Notwithstanding this, this paper aims to find the best clustering technique (pretreatment and clustering tool) that is able to minimize the intra-batch variation without undermining the inter-batch variation for sample classification. Three pretreatment methods were first evaluated by principle component analysis (PCA) and normalization coupled with standardization was found to be able to show promising results. Subsequently, hierarchical cluster analysis (HCA) was chosen to evaluate the pretreated data as this technique can clearly depict the relationships between samples on a dendogram. In the HCA, Ward-Squared Euclidean and Ward-Squared Pearson were found to be the successful linkage-distance combinations for clustering all the linked samples under five distinct groups.