Authentication of an Italian PDO hazelnut ("Nocciola Romana") by NIR spectroscopy.

Common hazelnuts are widely present in human diet all over the world, and their beneficial effects on the health have been extensively investigated and demonstrated. Different in-depth researches have highlighted that the harvesting area can define small variations in the chemical composition of the fruits, affecting their quality. As a consequence, it has become relevant to develop methodologies which would allow authenticating and tracing hazelnuts. In the light of this, the present work aims to develop a non-destructive method for the authentication of a specific high-quality Italian hazelnut, "Nocciola Romana," registered with a protected designation of origin (PDO). Thus, different samples of this fruit have been analyzed by near-infrared (NIR) spectroscopy and then classification models have been built, in order to distinguish between the PDO fruits and the hazelnuts not coming from the designated region. In particular, two different classification approaches have been tested, a discriminant one, partial least squares-discriminant analysis, and a class-modeling one, soft independent modeling of class analogies. Both methods led to very high prediction capability in external validation on a test set (classification accuracy in one case, and sensitivity and specificity in the other, all higher than 92%), suggesting that the proposed methodologies are suitable for a rapid and non-destructive authentication of the product.

Simultaneous quantification of caffeine and chlorogenic acid in coffee green beans and varietal classification of the samples by HPLC-DAD coupled with chemometrics.

A chromatographic procedure (HPLC-DAD) using a relatively rapid gradient has been combined with a chemometric curve deconvolution method, multivariate curve resolution-alternating least squares (MCR-ALS), in order to quantify caffeine and chlorogenic acid in green coffee beans. Despite that the HPLC analysis (at these specific operating conditions) presents some coeluting peaks, MCR-ALS allowed their resolution and, consequently, the creation of a calibration curve to be used for the quantification of the analytes of interest; this procedure led to a high accuracy in the quantification of caffeine and chlorogenic acid present in the samples. In a second part of this study, the possibility of classifying the green coffee beans on the basis of their cultivar (Arabica or Robusta), by partial least squares discriminant analysis (PLS-DA) and soft independent modeling of class analogies (SIMCA), has been explored. SIMCA resulted in 100% of sensitivity and specificity for the Arabica class, while for the Robusta, it reached 66.7% of sensitivity and 100% of specificity, or 100% of sensitivity and 100% of specificity, depending on the extraction procedure followed prior to the chromatographic analysis; PLS-DA achieved 100% of correct classification independently of the procedure used for the extraction.

Data-fusion for multiplatform characterization of an Italian craft beer aimed at its authentication.

Five different instrumental techniques: thermogravimetry, mid-infrared, near-infrared, ultra-violet and visible spectroscopies, have been used to characterize a high quality beer (Reale) from an Italian craft brewery (Birra del Borgo) and to differentiate it from other competing and lower quality products. Chemometric classification models were built on the separate blocks using soft independent modeling of class analogies (SIMCA) and partial least squares-discriminant analysis (PLS-DA) obtaining good predictive ability on an external test set (75% or higher depending on the technique). The use of data fusion strategies - in particular, the mid-level one - to integrate the data from the different platforms allowed the correct classification of all the training and validation samples.

Chemometric classification techniques as a tool for solving problems in analytical chemistry.

Supervised pattern recognition (classification) techniques, i.e., the family of chemometric methods whose aim is the prediction of a qualitative response on a set of samples, represent a very important assortment of tools for solving problems in several areas of applied analytical chemistry. This paper describes the theory behind the chemometric classification techniques most frequently used in analytical chemistry together with some examples of their application to real-world problems.

Tracing the origin of extra virgin olive oils by infrared spectroscopy and chemometrics: a case study.

In this paper, the potential of coupling mid- and near-infrared spectroscopic fingerprinting techniques and chemometric classification methods for the traceability of extra virgin olive oil samples from the PDO Sabina was investigated. To this purpose, two different pattern recognition algorithm representative of the discriminant (PLS-DA) and modeling (SIMCA) approach to classification were employed. Results obtained after processing the spectroscopic data by PLS-DA evidenced a rather high classification accuracy, NIR providing better predictions than MIR (as evaluated both in cross-validation and on an external test set). SIMCA confirmed these results and showed how the category models for the class Sabina can be rather sensitive and highly specific. Lastly, as samples from two harvesting years (2009 and 2010) were investigated, it was possible to evidence that the different production year can have a relevant effect on the spectroscopic fingerprint. Notwithstanding this, it was still possible to build models that are transferable from one year to another with good accuracy.

Use of near infrared reflectance and transmittance coupled to robust calibration for the evaluation of nutritional value in naked oats.

A rapid accurate and precise method for simultaneous determination of ß-glucan and protein content in naked oat samples, based on the coupling of near-infrared spectroscopy and chemometrics, is presented. In particular, three different spectroscopic approaches [near infrared reflectance (NIR) and transmittance (NIT) on flour and NIT on whole grains] and various spectral pretreatments were considered. To account for the possibility of outlying samples, a robust version of the PLS algorithm (namely partial robust M-regression) was used. All the models resulted as accurate as the reference methods, reflectance spectroscopy being the technique providing the best outcomes. Variable reduction by inclusion of the most relevant predictors only (as evaluated by VIP scores) resulted in simpler and, in one case, more parsimonious models, without loss in accuracy.

Use of different artificial neural networks to resolve binary blends of monocultivar Italian olive oils.

The problem of authenticating extra virgin olive oil varieties is particularly important from the standpoint of quality control. After having shown in our previous works the possibility of discriminating oils from a single variety using chemometrics, in this study a combination of two different neural networks architectures was employed for the resolution of simulated binary blends of oils from different cultivars. In particular, a Kohonen self-organizing map was used to select the samples to include in the training, test and validation sets, needed to operate the successive calibration stage, which has been carried out by means of several multilayer feed-forward neural networks. The optimal model resulted in a validation Q2 in the range 0.91-0.96 (10 data sets), corresponding to an average prediction error of about 5-7.5%, which appeared significantly better than in the case of random or Kennard-Stone selection.

UV-vis spectrophotometric method for the quantitation of all the components of Italian general denaturant and its application to check the conformity of alcohol samples.

A simple, fast and relatively inexpensive spectrophotometric method for the identification and the quantification of the individual components of the Italian general denaturant in alcohol samples is proposed. In particular, it is shown that bitrex (a quaternary ammonium salt), whose UV spectrum is completely masked by those of the other denaturant components, can be identified using its reaction with disulphine blue VN-150 (an anionic dye indicator), which leads to the formation of an intensely colored ion-association complex (mole ratio 1:1), easily extractable in chloroform. As far as the quantitative detection is involved, it is however necessary to shake the chloroform phase in the presence of 1 mol L(-1) NaClO(4) aqueous solution because of the fast adsorption of the ion pair on the walls of the glass cell. Perchlorate anion, due to mass action, substitutes the anionic dye indicator in the association complex: as a consequence, disulphine blue passes to the aqueous phase, where its absorbance at lambda=640 nm is measured. On the other hand, C.I. Reactive Red 24 dye is easily identifiable from the visible spectrum of the product without any further pretreatment: its concentration can be determined measuring the absorbance at lambda=542 nm. Thiophene, being significantly more concentrated than the other components, can be identified from the UV spectrum of a 1:100 diluted solution of the alcohol sample and quantitatively determined measuring the absorbance at lambda=230 nm. Lastly, methyl ethyl ketone (MEK) can be identified from the UV spectrum of a 1:5 diluted solution of the alcohol sample and quantitatively determined measuring the absorbance at lambda=273 nm. However, more accurate results can be obtained using a multiwavelength analysis in the range 220-250 and 250-310 nm for the determination of thiophene and MEK, respectively. Validation on standard denatured alcohol samples has proven the method to be both accurate and sufficiently precise (within- and between-days repeatability <5%) to be applied to the analysis of real commercial samples.

Comparison of three spectrophotometric methods for the determination of gamma-oryzanol in rice bran oil.

A comparison of the results obtained by applying three spectrophotometric methods (at fixed wavelength, second-derivative and multicomponent analysis) to the determination of gamma-oryzanol in rice bran oil is reported. At fixed wavelength the results are more accurate when using isopropyl alcohol, rather than n-heptane, to dilute the oil samples, because the absorption bands of gamma-oryzanol are red-shifted and the absorbance, measured at lambda(max)=327 nm, is less affected by the interference of the oil "matrix" (lambda(max)=314 nm in n-heptane).However, to obtain accurate results also in oils with a low content of gamma-oryzanol, it is necessary to perform the analysis using second-derivative ((2)D330.365) or multicomponent (lambda=310-360 nm) methods. The first one fully removes the interference of oil matrix whilst the second, which needs a specific computational program to process the spectrophotometric data, furnishes evidence the presence of some unexpected interference in the analysis and/or standards which are not representative of the analysed samples, from the square root of the sum of the squared differences at each point between the linear combination of the standards and the unknown spectra (RMS error).Finally, some aspects of the chemical, spectroscopic (UV, IR) and thermoanalytical (TG, DSC) behaviour of gamma-oryzanol and the values of the parameters which enable "computation" of its UV spectra are reported.

Chemical authentication of extra virgin olive oil varieties by supervised chemometric procedures.

This work has focused on discriminating extra virgin olive oils from Sabina (Lazio, Italy) by olive fruit variety (cultivar). A set of oils from five of the most widespread cultivars (Carboncella, Frantoio, Leccino, Moraiolo, and Pendolino) in this geographical area was analyzed for chemical composition using only the Official Analytical Methods, recognized for the quality control and commercial classification of this product. The obtained data set was converted into a computer-compatible format, and principal component analysis (PCA) and a method based on the Fisher F ratio were used to reduce the number of variables without a significant loss of chemical information. Then, to differentiate these samples, two supervised chemometric procedures were applied to process the experimental data: linear discriminant analysis (LDA) and artificial neural network (ANN) using the back-propagation algorithm. It was found that both of these techniques were able to generalize and correctly predict all of the samples in the test set. However, these results were obtained using 10 variables for LDA and 6 (the major fatty acid percentages, determined by a single gas chromatogram) for ANN, which, in this case, appears to provide a better prediction ability and a simpler chemical analysis. Finally, it is pointed out that, to achieve the correct authentication of all samples, the selected training set must be representative of the whole data set.