Factorial Discriminant Analysis Research Articles

Solid‐phase microextraction gas chromatography/mass spectrometry: a new method for species identification of truffles

This study describes a rapid method to identify different truffle species by analysis of their volatile compound fraction using static headspace solid-phase microextraction gas chromatography/mass spectrometry. The volatile organic compounds (VOCs) were extracted using a new 2-cm 50/30 microm DVB/CAR/PDMS fiber placed for 10 min in the headspace of the truffle sample with the vial maintained at 20 degrees C (in a thermostatically controlled analysis room). The mass spectra of the VOC chromatograms were represented as 'fingerprints' of the analysed samples. Next, stepwise factorial discriminant analysis afforded a limited number of characteristic fragment ions that allowed a classification of the truffle species studied. This new method provides an effective approach to rapid quality control and identification of truffle species by analysis of their volatile fraction. Moreover, this method offers the advantage of minimizing thermal, mechanical, and chemical modifications of the truffles, thereby reducing the risk of analytical artifacts.

Corticobasal degeneration and Parkinson's disease assessed by HmPaO SPECT: The utility of factorial discriminant analysis

The diagnosis of corticobasal degeneration (CBD) is difficult despite the existence of some typical clinical features. Single photon emission computerized tomography (SPECT) in CBD presents an original pattern (with asymmetric hypoperfusion in pre- and retrorolandic regions) that could facilitate the differential diagnosis of CBD relative to the other degenerative parkinsonian syndromes. The objective of our study was to compare the regional cerebral blood flow measurements studied by SPECT in both CBD and Parkinson's disease (PD) using a multivariate procedure. Twenty-one patients with probable CBD and 20 patients with probable PD underwent brain (99m)Tc HmPaO SPECT. We used factorial discriminant analysis (FDA) to study the relative fixation of 26 regions of interest (ROIs) drawn on two transverse slices, together with the asymmetry indexes of 13 pairs of ROIs. FDA performed using the full set of parameters classified all the patients correctly. In order to classify the patients more easily, a predictive score using a selection of parameters was established. The most discriminating ROIs were the temporoinsular, temporoparietal, and frontal medial regions. We believe that this semiautomatic classification may be a precious tool for reinforcing the current clinical differential diagnosis of CBD and PD.

COMPLEMENTARY MEASUREMENTS FOR APPLE TEXTURE DISCRIMINATION: MECHANICAL TESTS

Force-Displacement and Force-Time curves loaded from penetrometric and compression tests, receptively, were analyzed for their ability to discriminate the apple texture. The parameters extracted from the curves of each measurement, were processed in a single factorial discriminant analysis (FDA) in order to identified five apple progenitors. The correct identification of each apple progenitor, using both penetrometric and compression data curve, were realized with a discrimination accuracy of 100% for X4956, X6908, X2033, 6064 and 95% for X2888. This study showed that the identification close to 100% was due to the complementary information given by these two kinds of measurements.

STATISTICAL MULTIVARIATE ANALYSIS OF MELON SHAPE: A CASE STUDY USING NEAR ISOGENIC LINES

Melon fruit shape is a complex character of high interest for the agrifood chain due to it represents, a critical factor to reduce mechanical damage, transportation costs and for designing new melon varieties to satisfy consumer's future needs. The aim of our melon project is to increase the knowledge on the genetics of fruit market and fruit quality traits. An Introgression Line population consisting of Near Isogenic Lines (NILs) derived from a cross between the Spanish cultivar Piel de Sapo' (PS) and an exotic Koran accession (PI 161375), each of them having a single introgresslon from PI 161375 into the PS background, was used to detect alleles of genes or QTLs controlling fruit traits. The morphological fruit traits that better discriminated NILs were length, diameter, and fruit shape (the ratio length divided by diameter). Useful indices derived from these variables were obtained by discriminant analysis. The practical discriminant power of principal component analysis or discriminant analysis was not improved by using additional variables as fruit weight, the rates between seed cavity or epidermal thickness and diameter, or fruit density. Round shaped or small-sized fruit were good characterized, particularly by discriminant or classical factor analysis. Melon fruit with these quantitative traits loci can be designed in order to satisfy consumer's needs and reduce mechanical damages during the agrifood chain.

Using Image Analysis to Look into the Effect of Impurity Concentration in NaCl Crystallization

This paper presents and describes a methodology for morphology assessment, crystal particle classification and quantification of the complexity level of a crystal or a population of crystals. It further describes the application of such methodology to the study of the influence of MgCl 2 in the crystallization of NaCl, namely on crystal size and morphology. Image analysis techniques are combined with discriminant factorial analysis leading to results that allow the computation of the complexity of crystals through a new parameter, the agglomeration degree of crystals. With this methodology it has been possible to distinguish automatically among three different classes of crystals, and within each class as per their complexity. Agreement between manual and automated classification measured in terms of a performance index is 90% on average. The effect of supersaturation and impurity concentration on the type, amount and complexity level of the agglomerates was further determined.

The potential of combined infrared and fluorescence spectroscopies as a method of determination of the geographic origin of Emmental cheeses

Abstract The potential of near-infrared (NIR), mid-infrared (MIR) and front-face fluorescence spectroscopies combined with chemometric methods was investigated for discriminating Emmental cheeses from different European geographic origins. A total of 91 Emmental cheeses produced during winter in Austria ( n = 4 ), Finland ( n = 6 ), Germany ( n = 13 ), France ( n = 30 ) and Switzerland ( n = 38 ) were investigated. Cheese samples were classified by factorial discriminant analysis. From NIR, a good classification of cheeses was observed for 89% and 86.8% of the calibration and validation spectral data sets, respectively. The MIR results were comparable to those observed with NIR. Within the 3000–2800 cm−1 region, 84.1% and 85.7% of the cheeses were correctly classified for the calibration and validation data sets, respectively. The classification obtained with the tryptophan fluorescence spectra was considerably lower (only 67.6% and 41.7% for classification and validation spectra, respectively). But tryptophan fluorescence spectra allowed a good discrimination of Emmental cheeses made from raw milk or from thermised milk. Common components and specific weights analysis were performed on the physico-chemical, infrared and tryptophan fluorescence data sets. These results showed that spectroscopic techniques may provide useful fingerprints and allow the identification of Emmental cheeses according to the geographic origin and production conditions.

Cognitive and SPECT characteristics predict progression of Parkinson?s disease in newly diagnosed patients

To identify features in cognitive functioning and regional cerebral blood flow (rCBF) in newly diagnosed Parkinson's disease (PD) patients and to determine whether these factors are able to predict the progression of the disease in general and the development of cognitive decline in particular. 50 previously treatment-naive PD patients participated in the study. Cognitive assessment and SPECT were performed twice: at the time of diagnosis and then 3 years later. Six patients died or refused to continue. The Mattis dementia rating scale, the WAIS-R digit span test, a word list learning/recall test, a word fluency task and the Stroop word-colour test were used to assess cognitive function. rCBF was measured in 10 pairs of regions of interest. Principal component analysis of the data from the final examination was used to determine which variables allowed the formation of patient subgroups. Thereafter, factorial discriminant analysis (FDA) was performed in order to obtain a predictive model of these final classes. A stepwise procedure enabled the identification of 3 clusters (26, 16 and 2 patients). As the patients in the smallest cluster met the criteria for dementia at the final examination, they were discarded from further analyses. All the cognitive variables contributed to the constitution of the two other clusters. Age, educational level and all the rCBF parameters also contributed but to a lesser extent. Comparison of these groups showed reduced overall cognitive efficiency and an exacerbated subcorticofrontal syndrome in the 16-patient cluster. FDA showed that the best predictive model for the final classes was based on 7 variables: educational level, semantic and alternating word fluency, Stroop interference index and the right medial frontal, left parietal and left lenticular nucleus rCBF findings. Even though both cognitive and rCBF parameters help predict the progression of newly diagnosed PD patients and bearing in mind the limitations of the SPECT method used here, it appears that the contribution of cognitive assessment is greater than that of rCBF measurement.

A Revision of the <I>Cinara</I> Species (Hemiptera: Aphididae) of the United States Pinyon Pines

The Cinara species (Hemiptera: Aphididae: Lachninae) found on Pinus edulis Engelm. and P. monophylla Torr. & Frem. are revised and now include eight species. Included are redescriptions of the viviparous apterae and alatae of C. atra (Gillette & Palmer), C. caliente Hottes, C. edulis (Wilson), C. puerca Hottes, C. tanneri (Knowlton), C. terminalis (Gillette & Palmer), and C. wahtolca Hottes and a description of C. anelia sp. nov. New synonyms are established using discriminant factor analysis and traditional morphological methods: C. poketa Hottes becomes C. atra, C. nitidula Hottes becomes C. terminalis, and C. pinona Hottes, C. metalica Hottes, C. apacheca Hottes, C. pinata Hottes, and C. rustica Hottes all become C. edulis. The subspecies C. wahtolca curtiwahtolca Hottes is also no longer recognized. A key to the viviparae, diagnostic figures, and collection maps are provided.

Monitoring the geographic origin of both experimental French Jura hard cheeses and Swiss Gruyère and L’Etivaz PDO cheeses using mid-infrared and fluorescence spectroscopies: a preliminary investigation

Abstract The potential of mid-infrared and intrinsic fluorescence spectroscopies was investigated for determining the geographic origin of different hard cheeses. Mid-infrared (1700–1500 cm −1 region), fluorescence emission spectra, following excitation at 250 and 290 nm, and fluorescence excitation spectra following emission at 410 nm were recorded directly on cheese samples. Twelve experimental hard cheeses were made using identical and controlled cheese-making conditions with milks originating from three different regions in Jura (France). The results of factorial discriminant analysis (FDA) performed on the fluorescence and mid-infrared spectra of the experimental cheeses showed a good discrimination of the cheeses. In a second step, 25 Gruyere Protected Denomination of Origin (PDO) cheeses manufactured at different altitudes in Switzerland, i.e., 11 Gruyere PDO cheeses from the lowlands (

Identification by fluorescence spectroscopy of lactic acid bacteria isolated from a small-scale facility producing traditional dry sausages

Three different fluorescence spectra were recorded following excitation at 250 nm (aromatic amino acids+nucleic acids, AAA+NA), 316 nm (NADH) and 380 nm (FAD) for 20 type strain collections of lactic acid bacteria (LAB). Evaluation of the data using principal component analysis and factorial discriminant analysis showed a good discrimination of considered LAB at the genus, species and genus-species level. AAA+NA fluorophores showed the highest percentage of good classification. From AAA+NA spectra recorded on LAB isolated from a small-scale facility producing traditional dry sausages, we succeeded to identify 28 of 29 wild strains. This method allowed us to discriminate between Lactobacillus sakei subsp. carnosus and Lactobacillus sakei subsp. sakei. Thus, intrinsic fluorescence is an economical and powerful tool for the identification of wild LAB isolated from meat and meat products.

Solid-Phase Microextraction, Gas Chromatography, and Mass Spectrometry Coupled with Discriminant Factor Analysis and Multilayer Perceptron Neural Network for Detection of Escherichia coli

This study was performed to investigate the ability of using discriminant factor analysis (DFA) and an artificial neural network (ANN) to identify and quantify the number of Escherichia coli (ATCC 25922) in nutrient media from data generated by analysis of E. coli volatile metabolic compounds using solid-phase microextraction (SPME) coupled with gas chromatography (GC) and mass spectrometry (MS). E. coli was grown in super broth and incubated at 37°C for 2 to 12 h. Numbers of E. coli were followed using a colony counting method. An SPME device was used to collect the volatiles from the headspace above the samples, and the volatiles were identified using GC-MS . DFA was used to classify the samples from different incubation times. From DFA, it was possible to differentiate super broth from media containing E. coli when cell numbers were 105 CFU or more. The potential to predict the number of E. coli from the SPME-GC-MS data was investigated using a multilayer perceptron (MLP) neural network with back propagation training. The MLP comprised an input layer, one hidden layer, and an output layer, with a hyperbolic tangent sigmoidal transfer function in the hidden layer and a linear transfer function in the output layer. Good prediction was found as measured by a regression coefficient (R2 = 0.996) between actual and predicted data.

Low-level detection of ethanol and H 2S with temperature-modulated WO 3 nanoparticle gas sensors

Low-level detection of ethanol and H 2S was achieved with thermally modulated WO 3 nanoparticle gas sensors. Nanoparticle WO 3 films, with a mean grain size of ∼5 nm and a thickness of ∼20 μm, were produced by advanced reactive gas evaporation onto alumina substrates. The working temperature of the sensor was periodically modulated between 150 and 250 °C, and the response was analysed by fast Fourier transform (FFT) and discrete wavelet transform (DWT) methods in order to extract characteristic parameters from the sensors’ response transients. After calibration of the sensor for low concentrations of ethanol and H 2S, it was possible to detect as little as 200 ppb of ethanol and 20 ppb of H 2S (both of them dry gases) with good accuracy. Long-term sensor behaviour was assessed. Unsupervised and supervised linear pattern recognition methods, specifically principal component analysis (PCA) and discriminant factor analysis (DFA), were successfully applied to distinguish the investigated gases.

Alteration of raw-milk cheese by Pseudomonas spp.: monitoring the sources of contamination using fluorescence spectroscopy and metabolic profiling

Thirty Pseudomonas spp. strains isolated from milk, water, cheese centre and cheese surface in two traditional workshops manufacturing raw milk St. Nectaire cheese were characterised by fluorescence spectroscopy and Biolog metabolic profiling. Factorial discriminant analysis (FDA) of the two data sets revealed clear linkages between groups of isolates. In the first workshop, milk could be incriminated as the sole source of cheese contamination. In the second one, milk and cheese centre isolates were found similar but surface cheese contaminants could be linked to a secondary contamination originating from water. Thus, it is possible to characterise, differentiate and trace Pseudomonas spp. strains using the fluorescence and metabolic profiling techniques. In addition, the two data sets were found highly correlated by canonical correlation analysis (CCA). Fluorescence spectroscopy however showed several advantages because of its low cost and processing speed.

Determining the geographic origin of Emmental cheeses produced during winter and summer using a technique based on the concatenation of MIR and fluorescence spectroscopic data

A technique that used multivariate data analysis to combine mid-infrared (MIR) spectroscopy with front-face fluorescence spectroscopy was used to discriminate between Emmental cheeses originating from different European countries: Austria (n=12), Finland (n=10), Germany (n=19), France (n=57), and Switzerland (n=65). In total, 163 Emmental cheeses produced in winter (n=91) and summer (n=72) periods were investigated. When Factorial Discriminant Analysis was applied to either the infrared or fluorescence spectral data the classifications were not satisfactory. Therefore, the first twenty principal components (PCs) of the PCA extracted from each data set (MIR and tryptophan fluorescence spectra) were pooled (concatenated) into a single matrix and analysed by Factorial Discriminant Analysis. Correct classifications were obtained for the samples for 89% of the calibration spectra and 76.7% of the validation spectra. The discrimination for cheeses from Finland was excellent, while Austrian, German, French and Swiss cheeses were also discriminated well although a few samples were misclassified. It was concluded that concatenation of the data from the two spectroscopic techniques is an efficient technique for authenticating Emmental cheeses independently of their manufacturing period.

Ecomorphological indicators of feeding behaviour in the bears (Carnivora: Ursidae)

AbstractPatterns of morphological variation in the skulls of extant bears were studied as they relate to diet and feeding behaviour. Measurements of craniodental features were used to compute indices that reflect dietary adaptations of the dentition and biomechanical properties of the skull, jaw and related musculature. Species were classified as either carnivores, omnivores, herbivores or insectivores. Differences among dietary groups were assessed with analysis of variance and discriminant factor analysis. Results demonstrated significant morphological separation among all four groups. Carnivores were distinguished by, among other features, molar size reduction, flexible mandibles and, most surprisingly, relatively small carnassial blades. In contrast, herbivores displayed, among other features, large molar grinding areas, rigid mandibles and large carnassial blades. The insectivorous sloth bear was characterized by extreme reduction of the post‐canine teeth. As expected, omnivores tended to have morphology intermediate between that of carnivorous and herbivorous ursids. Comparison with previous studies revealed that bears exhibit a different set of morphological specializations for diet than other carnivoran groups. Carnivorous ursids, for example, were found to share aspects of craniodental morphology with omnivorous canids. The relatively weak adaptations for carnivory observed in bears may be the result of selection for the ability to cope with temporal fluctuations in dietary components. The giant panda Ailuropoda melanoleuca was found to have a relatively stiff jaw and great mechanical advantage of the jaw‐closing muscles, features previously observed in carnivorous canids and unexpected in this herbivorous bear. Comparison of patterns of morphological variation and patterns of phylogenetic relationships among species revealed surprisingly strong congruence between morphology and phylogenetics.

Validation Studies: Matters Of Dimensionality, Accuracy, And Parsimony With Predictive Discriminant Analysis And Factor Analysis

Two studies were used as examples that examined issues of dimensionality, accuracy, and parsimony in educational research via the use of predictive discriminant analysis and factor analysis. Using a two-group problem, study 1 looked at how accurately group membership could be predicted from subjects’ test scores. Study 2 looked at the dimensionality structure of an instrument and if it developed constructs that would measure theorized domains.

Morphological convergence in conifer-dwelling passerines

We tested for morphological convergence in conifer specialists among 88 passerines belonging to seven different phylogenetic lineages by discriminant factor analysis. We found a parallel trend among the seven lineages in body mass and digital pad morphology, whereas no such trends existed for the feeding and flight apparatus. Compared to the control species, the conifer specialists have smaller body masses and higher digital pads with maximal widths lying more distally within each of the seven lineages. These traits are interpreted as adaptations to dwelling among coniferous needles.

Transient State Detection and Prediction of Organic Overload in Anaerobic Digestion Process Using Statistical Tools

Anaerobic wastewater treatment is a very complex system, which involves a great number of process (bioprocess as well as physicochemical process) running in parallel and series. This means that multivariable analysis is needed for state identification of such a process. The objective of this work is to use statistical multivariable tools to detect different overloads (non stable conditions) in an anaerobic digester and select the most influencing variables on state detection. It is also considered to build a predictor for these variables using both linear and non linear techniques. Factorial discriminant analysis (FDA) is a linear methodology used for classifying multivariate data in a less dimension space, for example a two dimension plane. By the other hand, artificial neural networks (ANN) can be trained to classify a group of data with the advantage of being a non linear tool. Both techniques were compared for organic overload classification by the analysis of 7 variables of a wastewater treatment process carried out in an anaerobic reactor. Similar classification capability was found for both techniques. In spite of FDA simplicity, this tool seems to have similar performance than ANN. By the analysis of FDA coefficients and the variables variances, it was possible to determine the main variable for classification. It was found that hydrogen flowrate (QH 2 ) and methane flowrate (QCH 4 ) are the most influencing variables for classification. In order to forecast the state of the process it was considered interesting to by able to predict these two key variables. Time series analysis (a linear model) and neural networks (non linear model) were used for 1 hour prediction of QH 2 and QCH 4 . The validation error for both techniques were very small, 2.5% and 5.5% for times series analysis and neural network, respectively. Linear models seem to give as good results as non linear and more complex models (neural networks) for classification and prediction

Flavor analysis in a pharmaceutical oral solution formulation using an electronic-nose

Flavors are commonly used in pharmaceutical oral solutions and oral suspensions to mask drug bitterness and to make the formulation more palatable. Flavor analysis during product development is typically performed by human organoleptic analysis, which is often expensive and less objective. A novel approach using a metal oxide sensor-based instrument (electronic-nose) for headspace analysis was explored to replace human sensory perception for consistent qualitative and quantitative analysis of flavors in a pharmaceutical formulation. The use of the electronic-nose technique to qualitatively distinguish among six common flavoring agents (raspberry, red berry, strawberry, pineapple, orange, and cherry) in placebo formulations was demonstrated. The instrument was also employed to identify unknown flavors in drug formulation placebos. Raspberry flavor samples from different lots made by the same manufacturer, as well as freshly prepared and aged samples, were also distinguished by electronic-nose. Therefore, the instrument can potentially be used for identity testing of different flavor raw materials and the flavored solution formulations. The electronic-nose was also employed successfully for quantitative analysis of flavors in an oral solution formulation. The quantitative method might be used to assay the flavor concentration during release testing of the oral solution formulation or to monitor flavor shelf-life in the marketed container. It can also be implemented for packaging selection for the formulation in order to ensure the flavor shelf-life. Chemometric methodologies including principal component analysis (PCA), discriminant factorial analysis (DFA), and partial least squares (PLS), were used for data processing and identification.

Discrimination and identification of a refrigerant gas in a humidity controlled atmosphere containing or not carbon dioxide: application to the electronic nose

Atmospheric pollution and green house effects are now one of the major preoccupations to improve the general quality of life. With an electronic nose application we propose a system dedicated to the main detection of refrigerant Forane 134a (R134a) gas leakage in air conditioned systems and atmospheres. For our application based on a metal oxide type gas sensor array (six TGS elements) mainly coupled with discriminant factorial analysis (DFA) for the pattern recognition, we propose to study the effect of the relative humidity and an antagonist gas carbon dioxide as interfering gases on the R134a gas and concentration discrimination. We show that the relative humidity in a wide range (18–85%) has a more important influence on the target gas and concentrations discrimination than carbon dioxide. We present the discrimination results for the gases and afterwards for the R134a concentrations. Then we show the ability of our system to identify correctly a test data set composed of unknown gas samples. For the gases concentrations identification we also show that a correct identification rate can be obtained specially when the relative humidity rate range is reduced.