Bootstrapped Latin Partitions Research Articles

Discrimination of corn variety using Terahertz spectroscopy combined with chemometrics methods

High-oil corn is a high-quality variety of corn possessing higher oil content with greater caloric energy than normal corn. Hence, controlling the purity and authenticity of high-oil corn is of great importance in current crop research. The aim of this study is to develop a novel method for corn variety discrimination using Terahertz (THz) spectroscopy and signal classification analysis. In brief, the method involves feature extraction and variable selection of raw signals from Terahertz time-domain waveforms (THz-TDW) and absorption spectrum (THz-AS), and the use of classifiers on those treated signals to establish the discrimination models. Principle component analysis (PCA) were used for feature extraction with THz-TDW, while three different methods of variable selection were implemented with THz-AS, including uninformative variables elimination (UVE), uninformative variables elimination-successive projections algorithm (UVE-SPA) and competitive adaptive reweighted sampling (CARS). Then, two classification algorithms, Linear discriminant analysis (LDA) and support vector machine (SVM), were employed and compared in the discrimination models. Bootstrapped Latin partitions (BLP) method with 10 bootstraps and 5 Latin-partitions was applied to validate these models. Our modeling results suggest SVM as the better classification algorithm achieving higher identifying accuracy, such that the PCA-SVM model for THz-TDW has achieved 94.7% accuracy. The results also indicate variable selection as an important step to create an accurate and robust discrimination model for THZ-AS. The CARS-SVM model with radial basic function (RBF) has achieved 100% average accuracy in prediction set, while the UVE-SVM and UVE-SPA-SVM have achieved 91.2% and 99.1% accuracy, respectively. These results demonstrate that high-oil corn and normal corn can be identified successfully by using THz spectroscopy with discriminant analysis, suggesting our techniques to provide an efficient and practical reference for classifying crop varieties in agriculture research, while expanding the application of THz spectroscopy in the related field.

Self-Optimizing Support Vector Elastic Net.

Chemometrics is widely used to solve various quantitative and qualitative problems in analytical chemistry. A self-optimizing chemometrics method facilitates scientists to exploit the advantages of chemometrics. In this report, a parameter-free support vector elastic net that self-optimizes two key regularization constants, i.e., λ for L2 regularization and t for L1 regularization, is developed and referred to as self-optimizing support vector elastic net (SOSVEN). Response surface modeling (RSM) and bootstrapped Latin partitions (BLPs) are incorporated for the optimization. Responses at a set of design points over the ranges of the two factors are evaluated with an internal BLP validation using a calibration set. A 2-dimensional interpolation with a cubic spline fits a response surface to determine the best condition that gives the best-estimated response. The SOSVEN with RSM had comparable performances with the one tuned by grid search, while the RSM is more efficient. The developed SOSVEN was compared with two parameter-free chemometrics methods, super partial least-squares regression (sPLSR) and super support vector regression (sSVR) for calibration, and sPLS-discriminant analysis (sPLS-DA) and support vector classification (SVC) for classification. For calibration, the SOSVEN with RSM worked equivalently well or better than the other two self-optimizing methods for the evaluations using meat and hemp oil data sets. For classification, a reference wine data set and mass spectra of different marijuana extracts were used. The three classifiers had similar performances to identify the cultivars of wines with nearly 98% of accuracy. The SOSVEN significantly outperformed sPLS-DA and SVC to classify the mass spectra of marijuana extracts with an overall accuracy of 97%. These results demonstrated excellent abilities of SOSVEN for classification and calibration.

Automatic soft independent modeling for class analogies

Soft independent modeling of class analogy (SIMCA) is an important method for authentication. The key parameters for SIMCA, the number of principal components and the decision threshold, determine the model's performance. In this report, a self-optimizing SIMCA that automatically determines these two parameters is devised and referred to as automatic SIMCA (aSIMCA). An efficient optimization is obtained by incorporating response surface modeling (RSM) and bootstrapped Latin partitions with the model-building dataset. A set of design points over the ranges of the two parameters are evaluated with respect to sensitivity and specificity by using the model-building data from target and non-target classes. Averages of the sensitivity and specificity are used as responses for the design points. A 2-dimensional interpolation and a bivariate cubic polynomial were used to model the response surface. As a control method, a grid search that evaluates all combinations of the two parameters over the same ranges was performed in parallel to determine the best conditions for SIMCA and the modeling performance was compared to aSIMCA with RSM. The developed aSIMCA methods were evaluated by authenticating two botanical extracts sets, i.e., marijuana and hemp, with spectral datasets collected from various spectroscopic techniques, including nuclear magnetic resonance, high-resolution mass, and ultraviolet spectrometry. Results of a paired t-test indicated that the aSIMCA with the RSM had similar performance with the one optimized by the grid search for modeling marijuana and hemp, while the RSM was more computationally efficient. The 2-dimensional interpolation is preferred because the better efficiency and the fit to the response surface is more precise.

Quantitative analysis of ternary isomer mixtures of saccharide by terahertz time domain spectroscopy combined with chemometrics

Terahertz time-domain spectroscopy (THz-TDS) provides potential for analysis of sample features which cannot be detected by other commonly used spectroscopic techniques. This paper presented a novel THz-TDS approach combined with chemometrics to analyze ternary saccharide isomer mixtures including d-(−)fructose, d-(+)galactose anhydrous, and d-(+)mannose. The infrared (IR) and THz absorption spectral features of the saccharide isomers were compared and discussed. Support vector regression (SVR) and partial least squares (PLS) were applied for quantitative modelling on the collected THz spectra of the ternary isomer mixtures. The performances of the preprocessing methods of Savitzky–Golay (S-G) smoothing and wavelet transform were investigated for the PLS and SVR models. To achieve a reliable and unbiased estimation, bootstrapped Latin partitions (BLPs) was applied for statistical cross-validation. Multiway ANOVA was applied to find out significant factors and interaction between different factors. Results show that both PLS and SVR are applicable for the three compounds, d-(−)fructose, d-(+)galactose, and d-(+)mannose, in mixtures. For example, the correlation coefficient (R2) of the three saccharide isomers between experimental and predicted values in cross-validation set are 0.969, 0.903, and 0.938, respectively. As a conclusion, the proposed method can be feasible for the analysis of the ternary saccharide isomer mixtures and have potential for the analysis more complex mixtures in nutritional analysis.

Discrimination of brands of gasoline by using DART-MS and chemometrics

In this study, Direct Analysis in Real Time Mass Spectrometry (DART-MS) was used to analyze four different brands of gasoline collected from five local fuel service stations. Chemometric methods including analysis of variance-principal component analysis (ANOVA-PCA) and partial least squares discriminant analysis (PLS-DA) were applied to classify the gasoline samples based on brand and identify characteristic features for gasoline discrimination. To test the robustness of this method, the gasoline samples were collected once per week for eight weeks and they were analyzed by DART-MS within 24 h of collection and again two weeks after the last sample collection. Both full DART-MS spectra data and selected ion spectra data were used to construct PLS-DA models and the average classification rates of both models were 99.9 ± 0.1% for 100 independent evaluations with bootstrapped Latin partitions and were 100% for a new validation set collected two weeks later with no parametric changes. The major characteristic features corresponded to polymeric compounds in fuel additives which supported our hypothesis that proprietary additives blended in gasoline should be the marker components to distinguish gasoline with different brands. The weathered gasoline samples at different extents, i.e. 30%, 50%, 70%, and 90%, were studied and compared with un-weathered samples. The polymeric compound patterns in gasoline were found to be brand dependent and weathering extent dependent and the patterns for the weathered gasoline with different brands were still significantly different. DART-MS is demonstrated as a promising tool for the discrimination of brands of gasoline.

High-Throughput Chemotyping of Cannabis and Hemp Extracts Using an Ultraviolet Microplate Reader and Multivariate Classifiers

As the use of Cannabis products as natural medicines burgeons, it is also appearing as a food ingredient. It is important to screen Cannabis samples as ingredients by profiling their chemical compositions, which is referred to as chemotyping. Two sets of botanical extracts were studied. The first set is referred to as Cannabis contained plant materials from 15 samples of the sativa, indica, and hybrids of the two species. The second set contained 20 extracts from the variety of Cannabis sativa with low tetrahydrocannabinol (THC) concentrations, i.e., below 0.3%, and, henceforth, will be referred to as hemp. An ultraviolet (UV) microplate reader provides a cost-effective and high-throughput method for identifying chemotypes of plant extracts by their spectra. The microplate reader affords rapid measurements of small volumes, e.g., 50 µL, which demonstrates a potential to significantly reduce the analysis time and cost for Cannabis and hemp chemotyping or chemical profiling. Replicate samples were measured on different days to demonstrate the robustness of the method. Projected difference resolution (PDR) maps were used to visualize the separations among the classes. Five multivariate classifiers, fuzzy rule-building expert system (FuRES), super partial least squares-discriminant analysis (sPLS-DA), support vector machine (SVM), and two tree-based support vector machines (SVMtreeG and SVMtreeH) were evaluated. The classifiers were validated with ten bootstrapped Latin partitions (BLPs). For the Cannabis extracts, the SVMtreeG yielded the best performance and the classification accuracy was 99.1 ± 0.4% for spectra collected in the nonlinear absorbance range. For the hemp extracts, the SVM classifier performed the best with a 97.4 ± 0.6% classification accuracy. These results demonstrate that the UV microplate reader coupled with multivariate classifiers can be used as a high-throughput and cost-effective approach for chemotyping Cannabis.

On Splitting Training and Validation Set: A Comparative Study of Cross-Validation, Bootstrap and Systematic Sampling for Estimating the Generalization Performance of Supervised Learning

Model validation is the most important part of building a supervised model. For building a model with good generalization performance one must have a sensible data splitting strategy, and this is crucial for model validation. In this study, we conducted a comparative study on various reported data splitting methods. The MixSim model was employed to generate nine simulated datasets with different probabilities of mis-classification and variable sample sizes. Then partial least squares for discriminant analysis and support vector machines for classification were applied to these datasets. Data splitting methods tested included variants of cross-validation, bootstrapping, bootstrapped Latin partition, Kennard-Stone algorithm (K-S) and sample set partitioning based on joint X–Y distances algorithm (SPXY). These methods were employed to split the data into training and validation sets. The estimated generalization performances from the validation sets were then compared with the ones obtained from the blind test sets which were generated from the same distribution but were unseen by the training/validation procedure used in model construction. The results showed that the size of the data is the deciding factor for the qualities of the generalization performance estimated from the validation set. We found that there was a significant gap between the performance estimated from the validation set and the one from the test set for the all the data splitting methods employed on small datasets. Such disparity decreased when more samples were available for training/validation, and this is because the models were then moving towards approximations of the central limit theory for the simulated datasets used. We also found that having too many or too few samples in the training set had a negative effect on the estimated model performance, suggesting that it is necessary to have a good balance between the sizes of training set and validation set to have a reliable estimation of model performance. We also found that systematic sampling method such as K-S and SPXY generally had very poor estimation of the model performance, most likely due to the fact that they are designed to take the most representative samples first and thus left a rather poorly representative sample set for model performance estimation.

Overoptimism in cross-validation when using partial least squares-discriminant analysis for omics data: a systematic study

Advances in analytical instrumentation have provided the possibility of examining thousands of genes, peptides, or metabolites in parallel. However, the cost and time-consuming data acquisition process causes a generalized lack of samples. From a data analysis perspective, omics data are characterized by high dimensionality and small sample counts. In many scenarios, the analytical aim is to differentiate between two different conditions or classes combining an analytical method plus a tailored qualitative predictive model using available examples collected in a dataset. For this purpose, partial least squares-discriminant analysis (PLS-DA) is frequently employed in omics research. Recently, there has been growing concern about the uncritical use of this method, since it is prone to overfitting and may aggravate problems of false discoveries. In many applications involving a small number of subjects or samples, predictive model performance estimation is only based on cross-validation (CV) results with a strong preference for reporting results using leave one out (LOO). The combination of PLS-DA for high dimensionality data and small sample conditions, together with a weak validation methodology is a recipe for unreliable estimations of model performance. In this work, we present a systematic study about the impact of the dataset size, the dimensionality, and the CV technique used on PLS-DA overoptimism when performance estimation is done in cross-validation. Firstly, by using synthetic data generated from a same probability distribution and with assigned random binary labels, we have obtained a dataset where the true classification rate (CR) is 50%. As expected, our results confirm that internal validation provides overoptimistic estimations of the classification accuracy (i.e., overfitting). We have characterized the CR estimator in terms of bias and variance depending on the internal CV technique used and sample to dimensionality ratio. In small sample conditions, due to the large bias and variance of the estimator, the occurrence of extremely good CRs is common. We have found that overfitting peaks when the sample size in the training subset approaches the feature vector dimensionality minus one. In these conditions, the models are neither under- or overdetermined with a unique solution. This effect is particularly intense for LOO and peaks higher in small sample conditions. Overoptimism is decreased beyond this point where the abundance of noisy produces a regularization effect leading to less complex models. In terms of overfitting, our study ranks CV methods as follows: Bootstrap produces the most accurate estimator of the CR, followed by bootstrapped Latin partitions, random subsampling, K-Fold, and finally, the very popular LOO provides the worst results. Simulation results are further confirmed in real datasets from mass spectrometry and microarrays.

Feature expansion by a continuous restricted Boltzmann machine for near-infrared spectrometric calibration

A modified algorithm for training a restricted Boltzmann machine (RBM) has been devised and demonstrated for improving the results for partial least squares (PLS) calibration of wheat and meat by near-infrared (NIR) spectroscopy. In all cases, the PLS calibrations improved by using the abstract features generated from the RBM so long as the nonlinear mapping increased the dimensionality. The evaluations were validated using bootstrapped Latin partitions (BLPs) with 5 bootstraps and 3-Latin partitions which proved useful because of the statistical learning and random initial conditions of the RBM networks. By using a noise decay parameter, initial large amounts of noise could be used and the benefits of simulated annealing achieved as the noise level is slowly decreased. This paper demonstrates for the first time that using abstract features and enlarging the spectral data can improve the calibration results and exemplifies the Copiosity Principle. Two NIR reference datasets were evaluated. The first set of wheat spectra was calibrated for protein concentration and the second set of meat spectra was calibrated for moisture, fat, and protein concentration. The RBM feature extraction improved the linearity of the models and reduced embedded noise. The RBM also can help eliminate some difficult spectral preprocessing stages such as variable alignment and feature selection. RBMs benefit from derivative preprocessing of the NIR spectra or other preprocessing that enhances the differences among the spectra.

Chemical profiling of floral and chestnut honey using high-performance liquid chromatography-ultraviolet detection

Using the two-way images of phenolic compounds from high-performance liquid chromatography-ultraviolet diode array detection (HPLC-DAD), floral and chestnut honey from Turkey were successfully differentiated. A fuzzy rule-building expert system (FuRES), support vector machine classification tree (SVMTreeG), and super partial least-square discriminant analysis (sPLS-DA) were used to develop classification models. Normalization, retention time alignment, square root transform, and dissimilarity kernel were evaluated as data preprocessing methods. The bootstrapped Latin partition was used with 100 bootstraps and 4 partitions. Classification rates of FuRES and SVMTreeG with a square root transform were 97.6±0.4% and 97.6±0.4% for classifying the type of honey, respectively. The measures of precision are 95% confidence intervals. HPLC-DAD was demonstrated as a reliable analytical method for authentication of honey.

Simultaneous determination of amino acid mixtures in cereal by using terahertz time domain spectroscopy and chemometrics

Terahertz (THz) spectroscopy displays special features comparing to the commonly used spectroscopies like infrared and Raman and provides potential applications in broad areas. In this work, terahertz time-domain spectroscopy (THz-TDS) has been utilized for the qualitative and quantitative analyses of ternary mixtures of L-Glutamic acid, L-Glutamine, and L-Tyrosine which have similar chemical structures and properties. Mixtures of amino acids were prepared with yellow foxtail millet matrix to simulate more natural situation instead of polyethylene (PE). Partial least squares (PLS) and support vector machine (SVM) were compared for quantitative analysis in this work. Preprocessing methods, multiplicative scatter correction (MSC), Savitzky-Golay (S-G) smoothing, first derivative and wavelet transform, were investigated based on bootstrapped Latin partitions and external test. The SVM model with MSC preprocessing yielded a stable model and gave accurate prediction for the ternary amino acid in foxtail millet analyzed by THz-TDS. The THz absorption spectra and corresponding concentration profiles of three amino acids components from mixtures were resolved by multivariate curve resolution alternating least squares (MCR-ALS). The results show that THz spectroscopy could be applicable for analyzing more nutritional compounds in different cereals in the future.

Spectral Representation of Proton NMR Spectroscopy for the Pattern Recognition of Complex Materials

Proton nuclear magnetic resonance (NMR) spectroscopy provides a powerful tool for chemical profiling, also known as spectral fingerprinting, because of its inherent reproducibility. NMR is now increasing in use for authentication of complex materials. Typically, the absorbance spectrum is used that is obtained as the phase-corrected real component of the Fourier transform (FT) of the free induction decay (FID). However, the practice discards half the information that is available in the dispersion spectrum obtained as the imaginary component from the FT. For qualitative analysis or quantitative analysis of small sets of absorbance peaks, the symmetric and sharp peaks of the real spectra work well. However, for pattern recognition of entire spectra, trading peak resolution for peak reproducibility is beneficial. The absolute value of the complex spectrum gives the length or magnitude of magnetization vector in the complex plane; therefore, the magnitude relates directly to the signal (i.e., induced magnetization). The magnitude spectrum is obtained as the absolute value from the real and imaginary spectral components after the FT of the FID. By breaking with tradition and using the magnitude spectrum the reproducibility of the spectra and consequent recognition rates can be improved. This study used a 500-MHz 1H NMR instrument to obtain spectra from 4 diverse datasets; 12 tea extracts, 8 liquor samples, 9 hops extracts, and 25 Cannabis extracts. Six classifiers were statistically evaluated using 100 bootstrapped Latin partitions. The classifiers were a fuzzy rule-building expert system (FuRES) tree, support vector machine trees (SVMTreeG and SVMTreeH), a regularized linear discriminant analysis (LDA), super partial least squares discriminant analysis (sPLS-DA), and a one against all support vector machine (SVM). All classifiers gave better or equivalent results for the magnitude spectral representation than for the real spectra, except for one case of the 24 evaluations. In addition, the enhanced reproducibility of the absolute value spectra is demonstrated by comparisons of the pooled within sample standard deviations. For pattern recognition of NMR spectra, the magnitude spectrum is advocated.

Automated support vector regression

Multivariate calibration is an important procedure for analytical chemistry. Automated or self‐configuring methods can be used by scientists who lack expertise, may be embedded into data processing pipelines, and are less prone to user bias; however, the development of such algorithms is often neglected by the chemometrics community. Support vector regression (SVR) is a powerful method for accommodating megavariate data. SVR offers the advantage of fast calibration and flexibility in a variety of loss functions (ie, minimization of the residual error). By embedding bootstrapped Latin partitions (BLPs) into the calibration, the key parameter, the cost C, can be optimized to furnish an automated method. The methods are termed super SVR (sSVR). The BLP predictions of the calibration set accurately model the external prediction error of the entire calibration set. Prediction rates for super partial least squares (sPLS) are compared with sSVRs using three loss functions, Gauss, Laplace, and Huber. For linear data with uniformly distributed noise, sPLS is faster and gave better predictions. However, for data with outliers or a real data set of single‐beam near infrared spectra of bovine plasma, gasoline, and wheat, the sSVRs performed better than sPLS, and generally, the Huber loss function gave the best results.

Support vector machine classification trees based on fuzzy entropy of classification

The support vector machine (SVM) is a powerful classifier that has recently been implemented in a classification tree (SVMTreeG). This classifier partitioned the data by finding gaps in the data space. For large and complex datasets, there may be no gaps in the data space confounding this type of classifier. A novel algorithm was devised that uses fuzzy entropy to find optimal partitions for situations when clusters of data are overlapped in the data space. Also, a kernel version of the fuzzy entropy algorithm was devised. A fast support vector machine implementation is used that has no cost C or slack variables to optimize. Statistical comparisons using bootstrapped Latin partitions among the tree classifiers were made using a synthetic XOR data set and validated with ten prediction sets comprised of 50,000 objects and a data set of NMR spectra obtained from 12 tea sample extracts.

Quantitative measurements of binary amino acids mixtures in yellow foxtail millet by terahertz time domain spectroscopy

Terahertz time domain spectroscopy (THz-TDS) combined with chemometrics has been utilized for the qualitative and quantitative analysis of binary mixtures of l-glutamic acid and l-glutamine which have similar chemical structures and properties. The binary mixtures of amino acids were prepared with yellow foxtail millet matrix, substituted for polyethylene (PE) as previously reported. After proper pretreatment of absorption spectra, quantitative analysis was achieved by partial least squares (PLS) and interval partial least squares (iPLS) regressions. The performance of models was evaluated based on the root mean square error of prediction (RMSEP) and correlation coefficient (R2) of cross-validations with bootstrapped Latin partitions as criterion. The iPLS yielded better results with low RMSEP (0.39±0.02%, 0.39±0.02%), and higher R2 values (0.9904, 0.9906) for glutamine and glutamic acid comparing to the conventional PLS models. Multivariate curve resolution alternating least squares (MCR-ALS) was successfully applied for resolution of pure THz spectra and concentration profiles of two amino acids components from mixtures.

Diagnosis of patients with chronic kidney disease by using two fuzzy classifiers

The feasibility of two in-house fuzzy classifiers, fuzzy rule-building expert system (FuRES) and fuzzy optimal associative memory (FOAM), for diagnosis of patients with chronic kidney disease (CKD) was investigated. A linear classifier, partial least squares discriminant analysis (PLS-DA), was used for comparison. The CKD data used in this work were taken from the UCI Machine Learning Repository. Composite datasets were created by adding different levels of proportional noise to evaluate the robustness of the two fuzzy approaches. Firstly, 11 levels of proportional noises were added to each numeric attribute of the training and prediction sets one after another, and then these simulated training and prediction sets were combined in pairs. Thus, a grid with 121 groups of simulated data was generated, and classification rates for these 121 pairs were compared. Secondly, the performances of two fuzzy classifiers using the simulated datasets, in which 11 levels of noise were randomly distributed to each numeric attribute, were compared and the average prediction rates of FuRES and FOAM were 98.1±0.5% and 97.2±1.2%, respectively, with 200 bootstrap Latin partitions. The PLS-DA can give 94.3±0.8% with the identical evaluation. Confluent datasets comprised of the original and modified datasets were also used to evaluate FuRES, FOAM, and PLS-DA classification models. The average prediction rates of FuRES and FOAM obtained from 200 bootstrapped evaluations were 99.2±0.3% and 99.0±0.3%. PLS-DA yields slightly worse accuracy with 95.9±0.6%. The results demonstrate that both FuRES and FOAM perform well on the identification of CKD patients, while FuRES is more robust than FOAM. These two fuzzy classifiers are useful tools for the diagnosis of CKD patients with satisfactory robustness, and can also be used for other kinds of patients.

Application of terahertz time-domain spectroscopy combined with chemometrics to quantitative analysis of imidacloprid in rice samples

Terahertz time-domain spectroscopy (THz-TDS) has been utilized as an effective tool for quantitative analysis of imidacloprid in rice powder samples. Unlike previous studies, our method for sample preparation was mixing imidacloprid with rice powder instead of polyethylene. Then, terahertz time domain transmission spectra of these mixed samples were measured and the absorption coefficient spectra of the samples with frequency range extending from 0.3 to 1.7THz were obtained. Asymmetric least square (AsLS) method was utilized to correct the slope baselines that are presented in THz absorption coefficient spectra and improve signal-to-noise ratio of THz spectra. Chemometrics methods, including partial least squares (PLS), support vector regression (SVR), interval partial least squares (iPLS), and backward interval partial least squares (biPLS), were used for quantitative model building and prediction. To achieve a reliable and unbiased estimation, bootstrapped Latin partition was chosen as an approach for statistical cross-validation. Results showed that the mean value of root mean square error of prediction (RMSEP) for PLS (0.5%) is smaller than SVR (0.7%), these two methods were based on the whole absorption coefficient spectra. In addition, PLS performed a better performance with a lower RMSEP (0.3%) based on the THz absorption coefficient spectra after AsLS baseline correction. Alternatively, two methods for variable selection, namely iPLS and biPLS, yielded models with improved predictions. Comparing with conventional PLS and SVR, the mean values of RMSEP were 0.4% (iPLS) and 0.3% (biPLS) by selecting the informative frequency ranges. The results demonstrated that an accurate quantitative analysis of imidacloprid in rice powder samples could be achieved by terahertz time-domain transmission spectroscopy combined with chemometrics. Furthermore, these results demonstrate that THz time-domain spectroscopy can be used for quantitative determinations of other pesticides in other agricultural products.

Terahertz time-domain spectroscopy combined with fuzzy rule-building expert system and fuzzy optimal associative memory applied to diagnosis of cervical carcinoma.

Combined with terahertz time-domain spectroscopy, the feasibility of fast and reliable diagnosis of cervical carcinoma by a fuzzy rule-building expert system (FuRES) and a fuzzy optimal associative memory (FOAM) had been studied. The terahertz spectra of 52 specimens of cervix were collected in the work. The original data of samples were preprocessed by Savitzky-Golay first derivative (χderivative), principal component orthogonal signal correction (PC-OSC) and emphatic orthogonal signal correction to improve the performance of FuRES and FOAM models. The effect of the different pretreating methods to improve prediction accuracy was evaluated. The FuRES and FOAM models were validated using bootstrapped Latin-partition method. The obtained results showed that the FuRES and FOAM model optimized with the combination S-G first derivative and PC-OSC method had the better predictive ability with classification rates of 92.9±0.4 and 92.5±0.4%, respectively. The proposed procedure proved that terahertz spectroscopy combined with fuzzy classifiers could supply a technology which has potential for diagnosis of cancerous tissue.

Comparison of three algorithms for the baseline correction of hyphenated data objects.

Three novel two-way baseline correction algorithms, that is, orthogonal basis (OB), fuzzy optimal associative memory (FOAM), and polynomial fitting (PF), were evaluated with high performance liquid chromatography-mass spectrometry (HPLC-MS) and gas chromatography/mass spectrometry (GC/MS) data objects. Among these algorithms, both OB and FOAM are two-way baseline correction algorithms, which reconstruct the entire two-way backgrounds from blank data objects, while the PF algorithm is a pseudo-two-way method, which models each ion chromatogram baseline with a third-order polynomial. The performance of baseline correction methods was first evaluated with respect to the signal-to-noise ratios (SNRs) of 4 major peaks of the HPLC-MS total ion current (TIC) chromatograms of celery seed extracts. Then, the effect of baseline correction on pattern recognition was evaluated by using 42 two-way headspace (HS) solid phase microextraction (SPME) GC/MS data objects of 7 polychlorinated biphenyl (PCB) mixture standard solutions. Two types of classifiers, that is, a fuzzy rule-building expert system (FuRES) and partial least-squares-discriminant analysis (PLS-DA) were evaluated in parallel. Bootstrapped Latin partitions (BLPs) were used to give an unbiased and generalized evaluation of the classification accuracy. Results indicate that SNRs of major peaks of the TIC chromatogram representative of two-way HPLC-MS data objects are increased by baseline correction. In addition, higher prediction accuracies can be obtained by performing baseline correction on the entire GC/MS data set prior to pattern recognition. It is also found that proper data transformation is able to improve the performance of baseline correction. This report is the first of two-way baseline correction methods for hyphenated chromatography/mass spectrometry data objects. Both the orthogonal basis and FOAM baseline correction methods are novel in-house algorithms and proved to be generally effective for two-way baseline correction in the present study. Polynomial fitting is a conventional baseline correction method for one-way data objects and is applied to two-way data objects for the first time. It is applicable when blank data objects are unavailable.

Authentication of organically and conventionally grown basils by gas chromatography/mass spectrometry chemical profiles.

Basil plants cultivated by organic and conventional farming practices were accurately classified by pattern recognition of gas chromatography/mass spectrometry (GC/MS) data. A novel extraction procedure was devised to extract characteristic compounds from ground basil powders. Two in-house fuzzy classifiers, i.e., the fuzzy rule-building expert system (FuRES) and the fuzzy optimal associative memory (FOAM) for the first time, were used to build classification models. Two crisp classifiers, i.e., soft independent modeling by class analogy (SIMCA) and the partial least-squares discriminant analysis (PLS-DA), were used as control methods. Prior to data processing, baseline correction and retention time alignment were performed. Classifiers were built with the two-way data sets, the total ion chromatogram representation of data sets, and the total mass spectrum representation of data sets, separately. Bootstrapped Latin partition (BLP) was used as an unbiased evaluation of the classifiers. By using two-way data sets, average classification rates with FuRES, FOAM, SIMCA, and PLS-DA were 100 ± 0%, 94.4 ± 0.4%, 93.3 ± 0.4%, and 100 ± 0%, respectively, for 100 independent evaluations. The established classifiers were used to classify a new validation set collected 2.5 months later with no parametric changes except that the training set and validation set were individually mean-centered. For the new two-way validation set, classification rates with FuRES, FOAM, SIMCA, and PLS-DA were 100%, 93%, 97%, and 100%, respectively. Thereby, the GC/MS analysis was demonstrated as a viable approach for organic basil authentication. It is the first time that a FOAM has been applied to classification. A novel baseline correction method was used also for the first time. The FuRES and the FOAM are demonstrated as powerful tools for modeling and classifying GC/MS data of complex samples, and the data pretreatments are demonstrated to be useful to improve the performance of classifiers.