Classification of binary mass spectra of toxic compounds with an inductive expert system and comparison with SIMCA class modeling

The performance of an inexpensive, inductive rule-building expert shell system, based on the ID3 algorithm, was compared to that of SIMCA class modeling in classifying the binary mass spectra of 78 toxic and related compounds. The compressed mass spectra consisted of 17 masses chosen by using information theory. The expert rules verified the six main classes and two subclasses found with SIMCA class modeling. These classes were: all benzenes and all alkanes/ alkenes (alka(e)nes); nonhalobenzenes, chlorobenzenes, bromoalka(e)nes, and chloroalka(e)nes; and mono-, dichloroalka(e)nes and polychloroalka(e)nes. Training set classification accuracies obtained with the expert system were 93–100% as opposed to 62–98% for SIMCA. For 73 compounds, the expert rules gave a classification accuracy of 97–100% vs. 79–96% for SIMCA. Predictive accuracy for the four main classes was 78%. In general, fewer masses were involved with the rules than with the SIMCA models, and the rules are normally optimized with regard to minimum number of steps in the rule, not minimum number of variables. The expert rules work best with closed sets of objects where all possibilities can be included in the training sets. The expert rules represent planes partitioning the multidimensional measurement space (hypercube) into a subvolume nearest the SIMCA cylinders for an appropriate class. Overall, the performance of the expert system was very good.     

Discrimination of Osteonecrosis and Normal Tissues by Near-Infrared Spectroscopy and Successive Projections Algorithm-Linear Discriminant Analysis

Osteonecrosis of femoral head (ONFH) is a disease characterized by an impaired blood flow in the bone. The pathogenesis is still unknown, which makes an exact diagnosis troublesome and heavily dependent on experience. Exploring the information of molecular level by modern spectroscopy may help to discover the underlying pathogenesis and find its diagnostic application in clinical medicine. The study focuses on the combination of near-infrared (NIR) spectroscopy and classification models for discriminating ONFH and normal tissues. A total of 128 surgical specimens was prepared and NIR spectra were recorded by an integrating sphere. The experiment data set was divided into three subsets, i.e., the training set, validation set, and test set. Successive projection algorithm-linear discriminant analysis (SPA-LDA) was used to compress variables and build the diagnostic model. Partial least square-discriminant analysis (PLS-DA) was used as the reference. Principal component analysis (PCA) was used for exploratory analysis. The results showed that compared to PLS-DA, SPA-LDA provided a more parsimonious model using only seven variables and achieved better performance, i.e., sensitivity of 90.5 and 85%, and specificity of 100 and 95.5% for the validation and test sets, respectively. It indicated that NIR spectroscopy combined with SPA-LDA algorithm was a feasible aid tool for discriminating ONFH from normal tissue.     

Adulteration of the anthocyanin content of red wines: perspectives for authentication by Fourier transform-near infrared and 1H NMR spectroscopies

In the Italian oenological industry, the regular practice used to naturally increase the colour of red wines consists in blending them with a wine very rich in anthocyanins, namely Rossissimo. In the Asian market, on the other hand, anthocyanins extracted by black rice are frequently used as correctors for wine colour. This practice does not produce negative effects on health; however, in many countries, it is considered as a food adulteration. The present study is therefore aimed to discriminate wines containing anthocyanins originated from black rice and grapevine by using reliable spectroscopic techniques requiring minimum sample preparation. Two series of samples have been prepared from five original wines, that were added with different amounts of Rossissimo or of black rice anthocyanins solution, until the desired Colour Index was reached. The samples have been analysed by FT-NIR and (1)H NMR spectroscopies and the resulting spectra matrices were subjected to multivariate classification. Initially, PLS-DA was used as classification method, then also variable selection/classification methods were applied, i.e. iPLS-DA and WILMA-D. The classification with variable selection of NIR spectra permitted to classify the test set samples with an efficiency of about 70%. Probably these not excellent performances are due to the matrix effect, together with the lack of sensitivity of NIR with respect to minor compounds. On the contrary, very satisfactory results were obtained on NMR spectra in the aromatic region between 6.5 and 9.5 ppm. The classification method based on wavelet-based variables selection, permitted to reach an efficiency in validation greater than 95%. Finally, 2D correlation analysis was applied to FT-NIR and (1)H NMR matrices, in order to recognise the spectral zones bringing the same chemical information.     

Determination of detergent and dispensant additives in gasoline by ring-oven and near infrared hypespectral imaging

A method using the ring-oven technique for pre-concentration in filter paper discs and near infrared hyperspectral imaging is proposed to identify four detergent and dispersant additives, and to determine their concentration in gasoline. Different approaches were used to select the best image data processing in order to gather the relevant spectral information. This was attained by selecting the pixels of the region of interest (ROI), using a pre-calculated threshold value of the PCA scores arranged as histograms, to select the spectra set; summing up the selected spectra to achieve representativeness; and compensating for the superimposed filter paper spectral information, also supported by scores histograms for each individual sample. The best classification model was achieved using linear discriminant analysis and genetic algorithm (LDA/GA), whose correct classification rate in the external validation set was 92%. Previous classification of the type of additive present in the gasoline is necessary to define the PLS model required for its quantitative determination. Considering that two of the additives studied present high spectral similarity, a PLS regression model was constructed to predict their content in gasoline, while two additional models were used for the remaining additives. The results for the external validation of these regression models showed a mean percentage error of prediction varying from 5 to 15%.     

Coffee varietal differentiation based on near infrared spectroscopy

Near infrared spectroscopy (NIRS) was used to discriminate between arabica and robusta pure coffee varieties and blends of varied varietal composition. Direct orthogonal signal correction (DOSC) pre-processing method was applied on a set of 191 roasted coffee NIR spectra from both pure varieties and blends varying the final robusta content from 0 to 60% (w/w) in order to remove information unrelated to the actual varietal composition of samples. The corrected NIR spectra, as well as raw NIR spectra, were used to develop separate classification models using the potential functions method as class-modelling technique, exploring several options more or less restrictive according to the final number of considered categories. All constructed classification models were compared to evaluate their respective qualities and to show the suitability of applying DOSC method as pre-processing step for developing improved classification models for coffee varietal identification purposes.     

Computer-aided structural analysis of organic compounds by an artificial intelligence system

The computer program SEAC is designed to elucidate the structure of organic compounds comprising C (up to 40 C atoms in a molecule), H, N, O, S, Cl, Br and/or I atoms. Artificial intelligence techniques are exploited for empirical interpretation of spectralstructural correlations obtained directly from i.r., ¹H-n.m.r. and u.v. spectra. SEAC works on five levels depending on the entropy of the information input on tested molecules. The computer prints out a set of alternative substructures (levels I–IV) or a set of complete structures which are consonant with the input data.     

A program for generating empirical spectrum classification schemes

Programs are described which derive a set of compound classification tests from data specifying a required classification hierarchy and example low resolution mass spectra. The generated tests are expressed in terms of conventional operations such as searching for a molecular ion and checking for fragment ions. These tests can be used in a laboratory computer to give real-time classification of spectra collected by gas chromatography mass spectrometry.     

Development of a new file search system for nuclear magnetic resonance spectra: Production of an Enlarged Data Base and Search Test

A practical search system for proton n.m.r. spectra is reported. The coding rules and search algorithms are described in detail. Data for 8000 spectra have been converted into a computer-readable file from printed charts. Several search tests are used to evaluate the usefulness of the search system, and various effects of experimental conditions such as different instruments, frequencies and solvents on recall efficiency are described. The results presented indicate that the system should be applicable to routine analytical work.     

Systematization method for distinguishing plastic groups by using NIR spectroscopy.

A systematic classification method for polymers is not yet available in case of using near infrared spectra (NIR). That is why we have been searching for a systematic method. Because raw NIR spectra usually have few obvious peaks, NIR spectra have been pretreated by 2nd derivation for taking well modulated spectra. After the pretreatment, we applied classification and regression trees (CART) to the discrimination between the spectra and the species of polymers. As a result, we obtained a relatively simple classification tree. Judging from the obtained splitting conditions and the classified polymers, we concluded that obtained knowledge on the chemical function groups estimated by the important wavelength regions is not always applicable to this classification tree. However, we clarified the splitting rules for polymer species from the NIR spectral point of view.     

Orthogonal signal correction applied to the classification of wine and molasses vinegar samples by near-infrared spectroscopy. Feasibility study for the detection and quantification of adulterated vinegar samples

The most common fraudulent practice in the vinegar industry is the addition of alcohol of different origins to the base wine used to produce wine vinegar with the objective of reducing manufacturing costs. The mixture is then sold commercially as genuine wine vinegar, thus constituting a fraud to consumers and an unfair practice with respect to the rest of the vinegar sector. A method based on near-infrared spectroscopy has been developed to discriminate between white wine vinegar and alcohol or molasses vinegar. Orthogonal signal correction (OSC) was applied to a set of 96 vinegar NIR spectra from both original and artificial blends made in the laboratory, to remove information unrelated to a specific response. The specific response used to correct the spectra was the extent of adulteration of the vinegar samples. Both raw and corrected NIR spectra were used to develop separate classification models using the potential functions method as a class-modeling technique. The previous models were compared to evaluate the suitability of near-infrared spectroscopy as a rapid method for discrimination between vinegar origin. The transformation of vinegar NIR spectra by means of an orthogonal signal-correction method resulted in notable improvement of the specificity of the constructed classification models. The same orthogonal correction approach was also used to perform a calibration model able to detect and quantify the amount of exogenous alcohol added to the commercial product. This regression model can be used to quantify the extent of adulteration of new vinegar samples.     

Nondestructive Discrimination of Pharmaceutical Preparations Using Near-Infrared Spectroscopy and Partial Least-Squares Discriminant Analysis

Metronidazole is a widely used antibacterial and amoebicide drug. The feasibility of the classification of metronidazole samples with respect to their brands was investigated by near-infrared (NIR) spectroscopy along with chemometrics. A total of 92 samples of different lots and four brands were collected for measurements. First, principal component analysis was conducted to visualize the difference between metronidazole samples of different brands. Then, based on an effective classifier-independent method, i.e., joint mutual information, only the 30 most important variables were selected for modeling. From the independent test set, the partial least-squares discriminant analysis model based on the reduced variable set was compared with the corresponding full-spectrum model using all variables, which indicates the model based on the reduced variable set outperforms the full-spectrum model. It appears that the combination of NIR spectroscopy, joint mutual information, and partial least-squares discriminant analysis is a potential method for the classification of metronidazole from different brands and can, therefore, be used in the screening of counterfeit pharmaceutical products.     

An evaluation of orthogonal signal correction methods for the characterisation of arabica and robusta coffee varieties by NIRS

Two orthogonal signal correction methods (OSC and DOSC) were applied on a set of 83 roasted coffee NIR spectra from varied origins and varieties in order to remove information unrelated to a specific chemical response (caffeine), which was selected due to its high discriminant ability to differentiate between arabica and robusta coffee varieties. These corrected NIR spectra, as well as raw NIR spectra and three chemical quantities (caffeine, chlorogenic acids and total acidity), were used to develop separate classification models accordingly using the potential functions method as a class-modelling technique in order to evaluate their respective capacities to discriminate between coffee varieties and the influence of these pre-processing methods on the classification of the coffee samples into their corresponding variety class. The transformation of roasted coffee NIR spectra by means of an orthogonal signal correction method, taking into account in this correction a chemical response closely related to the sample origin, prompted a notable improvement in the specificity of the constructed classification models.     

Evaluation of SARs for the prediction of skin irritation/corrosion potential: structural inclusion rules in the BfR decision support system

The German Federal Institute for Risk Assessment (BfR) has developed a Decision Support System (DSS) to assess certain hazardous properties of pure chemicals, including skin and eye irritation/corrosion. The BfR-DSS is a rule-based system that could be used for the regulatory classification of chemicals in the European Union. The system is based on the combined use of two predictive approaches: exclusion rules based on physicochemical cut-off values to identify chemicals that do not exhibit a certain hazard (e.g., skin irritation/corrosion), and inclusion rules based on structural alerts to identify chemicals that do show a particular toxic potential. The aim of the present study was to evaluate the structural inclusion rules implemented in the BfR-DSS for the prediction of skin irritation and corrosion. The following assessments were performed: (a) a confirmation of the structural rules by rederiving them from the original training set (1358 substances), and (b) an external validation by using a test set of 200 chemicals not used in the derivation of the rules. It was found as a result that the test data set did not match the training set relative to the inclusion of structural alerts associated with skin irritation/corrosion, albeit some skin irritants were in the test set.     

Evaluation of SARs for the prediction of skin irritation/corrosion potential–structural inclusion rules in the BfR decision support system†

The German Federal Institute for Risk Assessment (BfR) has developed a Decision Support System (DSS) to assess certain hazardous properties of pure chemicals, including skin and eye irritation/corrosion. The BfR–DSS is a rule-based system that could be used for the regulatory classification of chemicals in the European Union. The system is based on the combined use of two predictive approaches: exclusion rules based on physicochemical cut-off values to identify chemicals that do not exhibit a certain hazard (e.g., skin irritation/corrosion), and inclusion rules based on structural alerts to identify chemicals that do show a particular toxic potential. The aim of the present study was to evaluate the structural inclusion rules implemented in the BfR–DSS for the prediction of skin irritation and corrosion. The following assessments were performed: (a) a confirmation of the structural rules by rederiving them from the original training set (1358 substances), and (b) an external validation by using a test set of 200 chemicals not used in the derivation of the rules. It was found as a result that the test data set did not match the training set relative to the inclusion of structural alerts associated with skin irritation/corrosion, albeit some skin irritants were in the test set.     

C-H stretching vibrations of methyl, methylene and methine groups at the vapor/alcohol (N = 1-8) interfaces

In IR and Raman spectral studies, the congestion of the vibrational modes in the C-H stretching region between 2800 and 3000 cm(-1) has complicated spectral assignment, conformational analysis, and structural and dynamics studies, even with quite a few of the simplest molecules. To resolve these issues, polarized spectra measurement on a well aligned sample is generally required. Because the liquid interface is generally ordered and molecularly thin, and sum frequency generation vibrational spectroscopy (SFG-VS) is an intrinsically coherent polarization spectroscopy, SFG-VS can be used for discerning details in vibrational spectra of the interfacial molecules. Here we show that, from systematic molecular symmetry and SFG-VS polarization analysis, a set of polarization selection rules could be developed for explicit assignment of the SFG vibrational spectra of the C-H stretching modes. These polarization selection rules helped assignment of the SFG-VS spectra of vapor/alcohol (n = 1-8) interfaces with unprecedented details. Previous approach on assignment of these spectra relied on IR and Raman spectral assignment, and they were not able to give such detailed assignment of the SFG vibrational spectra. Sometimes inappropriate assignment was made, and consequently misleading conclusions on interfacial structure, conformation and even dynamics were reached. With these polarization rules in addition to knowledge from IR and Raman studies, new structural information and understanding of the molecular interactions at these interfaces were obtained, and some new spectral features for the C-H stretching modes were also identified. Generally speaking, these new features can be applied to IR and Raman spectroscopic studies in the condensed phase. Therefore, the advancement on vibrational spectra assignment may find broad applications in the related fields using IR and Raman as vibrational spectroscopic tools.     

A neuro-fuzzy system for X-ray spectra interpretation

A decision scheme for the interpretation of spectra from wavelength dispersive X-ray fluorescence spectrometry is described that encompasses elements from three areas of artificial intelligence: fuzzy logic, rule based expert systems and neural net technology.After transforming the recorded spectra to line spectra by appropriate background correction a reasoning scheme is applied that takes into account not only the observed spectra, but also the recording conditions and prior spectroscopic information regarding the relative emission probabilities and the usefulness of the different lines for the purpose of element identification. The latter is done on the basis of a previously described scheme to compute conditional a posteriori Bayes probabilities for a mean matrix. These different pieces of information are then assembled into a battery of fuzzy rules. The importance of the rules as well as the importance of the X-ray lines is determined in a training process, similar to the one in a feedforward back-propagation network.To further stabilize the results this network is pruned in a second training cycle. This, however, had little effect on the quality of interpretation.The advantages of this approach to the interpretation of X-ray spectra over older ones are numerous: the system adapts itself to better interpret spectra that are of greater importance to a laboratory as these are better represented in the training set; the fuzzy logic is capable of working with incomplete and uncertain knowledge, and the neural network results based on these fuzzy rules is readily interpretable by the X-ray spectroscopist as every rule can be expressed also in natural language as in any classical rule based system.On leave from Silesian University, Katowice, Poland     

Classification for high-throughput data with an optimal subset of principal components

High-throughput data have been widely used in biological and medical studies to discover gene and protein functions. Due to the high dimensionality, principal component analysis (PCA) is often involved for data dimension reduction. However, when a few principal components (PCs) are selected for dimension reduction or considered for dimension determination, they are typically ranked by their variances, eigenvalues. However, this approach is not always effective in subsequent multivariate analysis, particularly classification. To maximize information from data with a subset of the components, we apply a different ranking criterion, canonical variate criterion, which considers within- and between-group variance rather than total variance in the classical criterion. Four prevalent classification methods are considered and compared using leave-one-out cross-validation. These methods are illustrated with three real high-throughput data sets, two microarray data sets and a nuclear magnetic resonance spectra data set.     

Quick method for analysis of anisotropic ESR spectra of particles with a g factor having rhombic symmetry in partially ordered and unordered systems

The exact integral expression describing the form of the anisotropic ESR spectrum of particles with a g factor having rhombic symmetry and a spin S=1/2 has been expanded into a rapidly converging series. This has made it possible to formulate simple rules for the quick analysis of experimental spectra in the case of individual absorption lines with Gaussian and Lorentzian shapes and to determine the corrections associated with the approximate nature of the approach. The rules obtained are universal, i.e., they are equally valid for the ESR spectra of both unordered and partially ordered systems. They can be used to determine the components of the g factor, the width of an individual absorption line, and the orientational distribution functions of the particles in a sample from spectra with sufficient accuracy. The applicability of the expressions obtained has been demonstrated by comparing them with exact computer-simulated ESR spectra.Translated from Teoreticheskaya i Éksperimental'naya Khimiya, Vol. 24, No. 6, pp. 738–742, November–December, 1988.