Application of partial least-squares regression to near-infrared reflectance spectroscopic determination of shive content in flax

Shive, the nonfiberous core portion of the stem, in flax fiber after retting is related to fiber quality. The objective of this study is to develop a standard calibration model for determining shive content in retted flax by using near-infrared reflectance spectroscopy. Calibration samples were prepared by manually mixing pure, ground shive and pure, ground fiber from flax retted by three different methods (water, dew, and enzyme retting) to provide a wide range of shive content from 0 to 100%. Partial least-squares (PLS) regression was used to generate a calibration model, and spectral data were processed using various pretreatments such as a multiplicative scatter correction (MSC), normalization, derivatives, and Martens' Uncertainty option to improve the calibration model. The calibration model developed with a single sample set resulted in a standard error of 1.8% with one factor. The best algorithm was produced from first-derivative processing of the spectral data. MSC was not effective processing for this model. However, a big bias was observed when independent sample sets were applied to this calibration model to predict shive content in flax fiber. The calibration model developed using a combination sample set showed a slightly higher standard error and number of factors compared to the model for a single sample set, but this model was sufficiently accurate to apply to each sample set. The best algorithm for the combination sample set was generated from second derivatives followed by MSC processing of spectral data and from Martens' Uncertainty option; it resulted in a standard error of 2.3% with 2 factors. The value of the digital second derivative centered at 1674 nm for these spectral data was highly correlated to shive content of flax and could form the basis for a simple, low-cost sensor for the shive or fiber content in retted flax.     

Multivariate calibration standardization across instruments for the determination of glucose by Fourier transform near-infrared spectrometry

The transfer of multivariate calibration models is investigated between a primary (A) and two secondary Fourier transform near-infrared (near-IR) spectrometers (B, C). The application studied in this work is the use of bands in the near-IR combination region of 5000-4000 cm(-)(1) to determine physiological levels of glucose in a buffered aqueous matrix containing varying levels of alanine, ascorbate, lactate, triacetin, and urea. The three spectrometers are used to measure 80 samples produced through a randomized experimental design that minimizes correlations between the component concentrations and between the concentrations of glucose and water. Direct standardization (DS), piecewise direct standardization (PDS), and guided model reoptimization (GMR) are evaluated for use in transferring partial least-squares calibration models developed with the spectra of 64 samples from the primary instrument to the prediction of glucose concentrations in 16 prediction samples measured with each secondary spectrometer. The three algorithms are evaluated as a function of the number of standardization samples used in transferring the calibration models. Performance criteria for judging the success of the calibration transfer are established as the standard error of prediction (SEP) for internal calibration models built with the spectra of the 64 calibration samples collected with each secondary spectrometer. These SEP values are 1.51 and 1.14 mM for spectrometers B and C, respectively. When calibration standardization is applied, the GMR algorithm is observed to outperform DS and PDS. With spectrometer C, the calibration transfer is highly successful, producing an SEP value of 1.07 mM. However, an SEP of 2.96 mM indicates unsuccessful calibration standardization with spectrometer B. This failure is attributed to differences in the variance structure of the spectra collected with spectrometers A and B. Diagnostic procedures are presented for use with the GMR algorithm that forecasts the successful calibration transfer with spectrometer C and the unsatisfactory results with spectrometer B.     

Wavelength interval selection in multicomponent spectral analysis by moving window partial least-squares regression with applications to mid-infrared and near-infrared spectroscopic data

A new wavelength interval selection procedure, moving window partial least-squares regression (MWPLSR), is proposed for multicomponent spectral analysis. This procedure builds a series of PLS models in a window that moves over the whole spectral region and then locates useful spectral intervals in terms of the least complexity of PLS models reaching a desired error level. Based on a proposed theory demonstrating the necessity of wavelength selection, it is shown that MWPLSR provides a viable approach to eliminate the extra variability generated by non-composition-related factors such as the perturbations in experimental conditions and physical properties of samples. A salient advantage of MWPLSR is that the calibration model is very stable against the interference from non-composition-related factors. Moreover, the selection of spectral intervals in terms of the least model complexity enables the reduction of the size of a calibration sample set in calibration modeling. Two strategies are suggested for coupling the MWPLSR procedure with PLS for multicomponent spectral analysis: One is the inclusion of all selected intervals to develop a PLS calibration model, and the other is the combination of the PLS models built separately in each interval. The combination of multiple PLS models offers a novel potential tool for improving the performance of individual models. The proposed procedures are evaluated using two open-path Fourier transform infrared data sets and one near-infrared data set, each having different noise characteristics. The results reveal that the proposed procedures are very promising for vibrational spectroscopy-based multicomponent analyses and give much better prediction than the full-spectrum PLS modeling.     

Scattering and absorption effects in the determination of glucose in whole blood by near-infrared spectroscopy

Optical properties of whole bovine blood are examined under conditions of different glucose loadings. A strong dependency is established between the scattering properties of the whole blood matrix and the concentration of glucose. This dependency is explained in terms of variations in the refractive index mismatch between the scattering bodies (predominately red blood cells) and the surrounding plasma. Measurements in the presence of a well-known glucose transport inhibitor indicate that variations in refractive index mismatch are related to the penetration of glucose into the red blood cells and demonstrate that increased scattering involves the uptake of glucose by red blood cells. Finally, multivariate calibration models are presented for the measurement of glucose in a whole blood matrix. These models are based on near-infrared spectral data collected from 80 different samples prepared from a single whole blood matrix. Calibration studies are performed over the combination, first-overtone, and short-wavelength spectral regions. The best calibration model is generated from combination region spectra, providing a standard error of prediction (SEP) of less than 1 mM over the concentration range of 3-30 mM. The model based on the first-overtone region is slightly degraded but still provides acceptable performance (SEP = 1.20 mM). The model based on the short-wavelength region is further degraded (SEP = 2.53 mM). To rationalize these results, an analysis of the selectivity of the calibration models is performed by computing the glucose net analyte signal. It is established that the models based on the combination and first-overtone regions are dominated by glucose absorption information, while the model computed from the short-wavelength region is based primarily on scattering information. This result provides evidence that absorption information is needed in order to obtain a glucose calibration model with acceptable performance.     

Short-wave near-infrared spectroscopy of biological fluids. 1. Quantitative analysis of fat, protein, and lactose in raw milk by partial least-squares regression and band assignment

The present study has aimed at providing new insight into short-wave near-infrared (NIR) spectroscopy of biological fluids. To do that, we analyzed NIR spectra in the 800-1,100-nm region of 100 raw milk samples. The contents of fat, proteins, and lactose were predicted by partial least-squares (PLS) regression and band assignment in that region was investigated based upon PLS loading plots and regression coefficients. For the fat prediction, the whole set of samples was divided into two groups and the fat concentration was predicted for the samples that were not included in the calibration procedures. The correlation coefficient and root-mean-square error of prediction (RM-SEP) in the better prediction run were found to be 0.996 and 0.087 wt %, respectively. Assignment of the bands due to fat was proposed based upon the regression coefficients and PLS loading weights, and the importance of a pretreatment in the prediction was discussed. Milk proteins also yielded sufficient correlation coefficients and RMSEP although the contributions of protein bands to the milk spectra were much smaller than those of the fat bands. The sizes of the calibration models for protein prediction were considered. This is the first time that good correlation coefficients and RMSEP of proteins have ever been obtained for the short-wave NIR spectra of milk. For lactose, noisy regression coefficients with limited prediction ability were obtained. Band assignment was investigated also for bands due to proteins and lactose. We propose the detailed band assignment for the short-wave NIR region useful for various biological fluids. The results presented here demonstrate that the short-wave NIR region is promising for the fast and reliable determination of major components in biological and biomedical fluids.     

Application of latent root regression for calibration in near-infrared spectroscopy. Comparison with principal component regression and partial least squares

Several analytical applications of spectroscopy are based on the assessment of a linear model, linking laboratory values to spectral data. Among various procedures, the following three methods have been used, i.e. principal component regression (PCR), partial least squares (PLS) and latent root regression (LRR). These methods can be applied in order to tackle the high collinearity commonly observed with spectral data. A collection of 99 near-infrared spectra, each including 351 data points, was used for the comparison of the 3 methods. The dependent variable was the specific production of pelleting. The spectral collection was divided into 49 and 50 observations for calibration and validation, respectively. The main elements of comparison were the minimum error observed on the verification set, the number of regressors introduced in the models and the stability of the errors around the minimum values. The minimum errors were 3.29, 3.13 and 3.07 for PCR, PLS and LRR, respectively. LRR required a large number of regressors in order to obtain the minimum error. Nevertheless, it gave very stable results, and the errors were not markedly increased when an arbitrary large number of regressors was introduced into the LRR model.     

Optimization of discriminant partial least squares regression models for the detection of animal by-product meals in compound feedingstuffs by near-infrared spectroscopy

This paper evaluates two multivariate strategies for classifying near-infrared (NIR) spectroscopic data for the detection of animal by-product meals (henceforth generically termed AbP) as an ingredient in compound feedingstuffs. Classification models were developed to discriminate between the presence and absence of animal-origin meals in compound feeds using two forms of discriminant partial least squares (PLS) regression: the algorithms PLS1 and PLS2. The training set comprised 433 commercial feeds, of which 148 contained AbP and the other 285 were stated to be AbP-free. Since the initial set contained unequal numbers of each class, the effect of this imbalance was analyzed by applying the same algorithms to a training set containing equal numbers of AbP-free and AbP-containing samples. The best classification model (97.42% of samples correctly classified), obtained with PLS2, that showed less sensitivity to the use of class-unbalanced sets, was externally validated using a set of 18 samples (10 AbP-containing and 8 AbP-free); all samples were correctly classified, except for one AbP-free sample that was classified as containing AbP (false positive). The results suggest that the application of PLS discriminant analysis to NIR spectroscopic data enables detection of AbP, a feed ingredient banned since the bovine spongiform encephalopathy (BSE) crisis; this confirms the value of NIRS qualitative analysis for product authentication purposes.     

In vivo near-infrared spectroscopy of rat skin tissue with varying blood glucose levels

We have performed in vivo measurements of near-infrared rat skin absorption in the 4000-5000-cm(-1) spectral range (2.0-2.5-microm wavelength) during a glucose clamp experiment in order to identify the presence of glucose-specific spectral information. Spectra were collected during an initial 3-h period where the animal's blood glucose concentration was held at its normal value. The blood glucose level was then increased above 30 mM by venous infusion of glucose and held for 2 h, after which it was allowed to return to normal. Spectra were recorded continuously during the procedure and are analyzed to identify spectral changes associated with changes in glucose concentration. Because the change in absorbance due to an increase in glucose concentration is small compared to changes due to other variations (e.g., the thickness of the skin sample), a simple subtraction of absorbance spectra from the hyperglycemic and euglycemic phases is not instructive. Instead, a set of principal components is established from the euglycemic period where the glucose concentration is constant. We then examine the change in absorbance during the hyperglycemic period that is orthogonal to these principal components. We find that there are significant similarities between these orthogonal variations and the net analyte signal of glucose, which suggests that glucose spectral information is present. The analysis described here provides a procedure by which the analytical significance of a multivariate calibration can be evaluated.     

In vivo noninvasive measurement of blood glucose by near-infrared diffuse-reflectance spectroscopy

This paper reports in situ noninvasive blood glucose monitoring by use of near-infrared (NIR) diffuse-reflectance spectroscopy. The NIR spectra of the human forearm were measured in vivo by using a pair of source and detector optical fibers separated by a distance of 0.65 mm on the skin surface. This optical geometry enables the selective measurement of dermis tissue spectra due to the skin's optical properties and reduces the interference noise arising from the stratum corneum. Oral glucose intake experiments were performed with six subjects (including a single subject with type I diabetes) whose NIR skin spectra were measured at the forearm. Partial least-squares regression (PLSR) analysis was carried out and calibration equations were obtained with each subject individually. Without exception among the six subjects, the regression coefficient vectors of their calibration models were similar to each other and had a positive peak at around 1600 nm, corresponding to the characteristic absorption peak of glucose. This result indicates that there is every possibility of glucose detection in skin tissue using our measurement system. We also found that there was a good correlation between the optically predicted values and the directly measured values of blood samples with individual subjects. The potential of noninvasive blood glucose monitoring using our methodology was demonstrated by the present study.     

Determination of physiological levels of glucose in an aqueous matrix with digitally filtered Fourier transform near-infrared spectra

A procedure is described for the measurement of clinically relevant concentrations of glucose in aqueous solutions with near-infrared (NIR) absorbance spectroscopy. A glucose band centered at 4400 cm-1 is used for this analysis. NIR spectra are collected over the frequency range 5000-4000 cm-1 with a Fourier transform spectrometer. A narrow-band-pass optical interference filter is placed in the optical path of the spectrometer to eliminate light outside this restricted range. This configuration provides a 2.9-fold reduction in spectral noise by utilizing the dynamic range of the detector solely for light transmitted through the filter. In addition, a novel spectral processing scheme is described for extracting glucose concentration information from the resulting absorbance spectra. The key component of this scheme is a digital Fourier filter that removes both high-frequency noise and low-frequency base-line variations from the spectra. Numerical optimization procedures are used to identify the best location and width of a Gaussian-shaped frequency response function for this Fourier filter. A dynamic area calculation, coupled with a simple linear base-line correction, provides an integrated area from the processed spectra that is linearly related to glucose concentrations over the range 1-20 mM. The linear calibration model accurately predicted glucose levels in a series of test solutions with an overall mean percent error of 2.5%. Based on the uncertainty in the parameters defining the calibration model and the variability of the magnitudes of the integrated areas, an overall uncertainty of 7.8% was estimated for predicted glucose concentrations.     

Moisture sorption as a potential condition marker for historic silks: noninvasive determination by near-infrared spectroscopy

Given their ephemeral nature, the preservation of historic silks can be problematic. Rapid, on-site condition monitoring would offer significant benefits to conservators and museum curators concerned with continued access to collections. In this paper, near-infrared spectroscopy (NIR) is investigated as a noninvasive approach to the characterization of silk fabrics and particularly for determining the moisture content of silks as a potential age-related marker. Bands within the NIR spectrum of silk are assigned to contributions from water and the silk fibroin polymer. The water bands may be deconvolved to show separate contributions from bound and structural water. When silk is exposed to deuterium oxide, the water OH NIR bands are rapidly lost. The accompanying changes in the amide-related NIR absorptions reflect differential accessibility of regions within the semi-crystalline fibroin aggregate. NIR spectra were recorded while silk was maintained at a range of relative humidity; complementary gravimetry provided absolute reference data for moisture sorption. A single spectral parameter, the intensity of the water combination band, is sufficient to indicate the relative moisture content of silk and allows distinction of unaged and heat, light, and humidity aged silks. The results confirm that NIR has significant potential for on-site studies at collections in support of the preservation and access of our silk heritage.     

Partial least squares regression calibration for determining wax content in processed flax fiber by near-infrared spectroscopy

The quality of flax fiber in the textile industry is closely related to the wax content remaining on the fiber after the cleaning process. Extraction by organic solvents, which is currently used for determining wax content, is very time consuming and produces chemical waste. In this study, near-infrared (NIR) spectroscopy was used as a rapid analytical technique to develop models for wax content associated with flax fiber. Calibration samples (n=11) were prepared by manually mixing dewaxed fiber and isolated wax to provide a range of wax content from 0 to 5%. A total of fourteen flax fiber samples obtained after a cleaning process were used for prediction. Principal component analysis demonstrated that one principal component is enough to separate the flax fibers by their wax content. The most highly correlated wavelengths were 2312, 2352, 1732, and 1766 nm, in order of significance. Partial least squares models were developed with various chemometric preprocessing approaches to obtain the best model performance. Two models, one using the entire region (1100-2498 nm) and the other using the selected wavelengths, were developed and the accuracies compared. For the model using the entire region, the correlation coefficient (R2) between actual and predicted values was 0.996 and the standard error of prediction (RMSEP) was 0.289%. For the selected-wavelengths model, the R2 was 0.997 and RMSEP was 0.272%. The results suggested that NIR spectroscopy can be used to determine wax content in very clean flax fiber and that development of a low-cost device, using few wavelengths, should be possible.     

Polarity determination by electron energy-loss spectroscopy: application to ultra-small III-nitride semiconductor nanocolumns

Channeling-enhanced electron energy-loss spectroscopy is applied to determine the polarity of ultra-small nitride semiconductor nanocolumns in transmission electron microscopy. The technique demonstrates some practical advantages in the nanostructure analysis, especially for feature sizes of less than 50 nm. We have studied GaN and (Al, Ga)N nanocolumns grown in a self-assembled way by molecular beam epitaxy directly on bare Si(111) substrates and on AlN buffer layers, respectively. The GaN nanocolumns on Si show an N polarity, while the (Al, Ga)N nanocolumns on an AlN buffer exhibit a Ga polarity. The different polarities of nanocolumns grown in a similar procedure are interpreted in terms of the specific interface bonding configurations. Our investigation contributes to the understanding of polarity control in III-nitride nanocolumn growth.     

Electron paramagnetic resonance spectroscopy studies of oxidative degradation of an active pharmaceutical ingredient and quantitative analysis of the organic radical intermediates using partial least-squares regression

Electron paramagnetic resonance (EPR) spectroscopy was used to study the radical species formed during the oxidation of an active pharmaceutical ingredient in the solid state. It was found that the extent of radical generation correlated to the formation of an oxidative degradation product. Multifrequency EPR and electron nuclear double resonance spectroscopy gave additional information on the identity of the organic radical species involved in the oxidation process, and a mechanism was proposed for the degradation, involving the formation of both carbon-centered and peroxy radicals. The multivariate analysis technique of partial least-squares (PLS) regression was then used to determine the extent of oxidation of the active pharmaceutical ingredient from the EPR spectra. The suitability of this approach was demonstrated from its application to a series of standards. The conventional approach for the quantitative analysis of EPR spectra is to measure the peak height or to perform double integration of the spectral region containing the signal of interest. Both of these methods have intrinsic errors associated with them, particularly for weak EPR signals with a poor signal-to-noise ratio or a sloping background response. The results obtained showed that greatly improved quantitation was obtained using the PLS regression approach.     

Implementation of LOCAL algorithm with near-infrared spectroscopy for compliance assurance in compound feedingstuffs

Seven thousand four hundred and twenty-three compound feed samples were used to develop near-infrared (NIR) calibrations for predicting the percentage of each ingredient used in the manufacture of a given compound feedingstuff. Spectra were collected at 2 nm increments using a FOSS NIRSystems 5000 monochromator. The reference data used for each ingredient percentage were those declared in the formula for each feedingstuff. Two chemometric tools for developing NIRS prediction models were compared: the so-called GLOBAL MPLS (modified partial least squares), traditionally used in developing NIRS applications, and the more recently developed calibration strategy known as LOCAL. The LOCAL procedure is designed to select, from a large database, samples with spectra resembling the sample being analyzed. Selected samples are used as calibration sets to develop specific MPLS equations for predicting each unknown sample. For all predicted ingredients, LOCAL calibrations resulted in a significant improvement in both standard error of prediction (SEP) and bias values compared with GLOBAL calibrations. Determination coefficient values (r(2)) also improved using the LOCAL strategy, exceeding 0.90 for most ingredients. Use of the LOCAL algorithm for calibration thus proved valuable in minimizing the errors in NIRS calibration equations for predicting a parameter as complex as the percentage of each ingredient in compound feedingstuffs.     

Enantiomeric excess determination by fourier transform near-infrared vibrational circular dichroism spectroscopy: simulation of real-time process monitoring

The first use of near-infrared (NIR) Fourier transform vibrational circular dichroism (FT-VCD) to follow changes in the enantiomeric excess (EE) of chiral sample molecules in time using a flow-cell sampling apparatus is reported. Simultaneous changes in the fractional composition and the EE of a mixture of two different chiral molecules were monitored as a function of time. This simulates the progress of the chemical reaction from a chiral reactant to a chiral product where the mole fractions and EE values of both species may change with time. For the molecules studied, alpha-pinene, camphor, and borneol, the accuracy of following EE changes for one species alone is approximately 2%, while for simultaneously following EE changes in two species it is approximately 3% for 30 min sampling periods at 16 cm(-1) spectral resolution. These findings demonstrate the potential for VCD to be used in the NIR region for real-time monitoring of the composition and %EE of chemical reactions involving the synthesis of chiral molecules.     

Nondestructive determination of solids and carotenoids in tomato products by near-infrared spectroscopy and multivariate calibration

Tomato is an important player in the agricultural market. It is the second most consumed vegetable in the world and is a source of important micronutrients such as lycopene and beta-carotene. Recent research has demonstrated that these carotenoids can act as free-radical quenchers in the body and prevent aging, tissue damage, heart disease, and certain cancers. Besides these microcomponents, tomato is composed of soluble and insoluble solids. In industry, these solids govern factory yield and play a major role in the tomato trade. Nowadays, standard methods for determining tomato solids and carotenoids are time and labor consuming. In this work, we present the development of a simultaneous and nondestructive method for determining total and soluble solids, as well as lycopene and beta-carotene, in tomato products by near-infrared spectroscopy. PLS-1 was the calibration technique chosen. For spectra preprocessing, MSC and second derivative were applied. As variable selection techniques, the correlogram cutoff, the successive projections algorithm, the dimension wise selection, and spectra splitting approach were applied. Best models presented satisfactory prediction abilities evaluated through its RMSEP and r values: total solids 0.4157, 0.9998; soluble solids 0.6333, 0.9996; lycopene 21.5779, 0.9996; beta-carotene 0.7296, 0.9981.     

Nondestructive direct determination of heroin in seized illicit street drugs by diffuse reflectance near-infrared spectroscopy

A new method has been developed for the fast and nondestructive direct determination of heroin in seized street illicit drugs using partial least-squares regression analysis of diffuse reflectance near-infrared spectra. Data were obtained from untreated samples placed in standard glass chromatography vials. A heterogeneous population of 31 samples, previously analyzed by a reference method, was employed to build the calibration model and to have a separated validation set. Based on the use of zero-order data for a calibration set of 21 samples, after standard normal variate and quadratic linear removed baseline correction (detrending), in the wavelength range from 1111 to 1647 nm, 8 PLS factors were enough to obtain a root-mean-square error of prediction of 1.3% w/w, with a quality coefficient of 10% for the estimation of the accuracy error in the prediction of heroin concentration in unknown samples and a residual predictive deviation of 5.4.     

Multivariate determination of glucose in whole blood by attenuated total reflection infrared spectroscopy

A spectral analysis of whole EDTA blood was undertaken by using attenuated total reflection and Fourier-transform infrared spectroscopy. The concentration of blood glucose was measured by an enzymatic method using glucose dehydrogenase and ranged between 40 and 290 mg/dL with an average concentration of 90.4 mg/dL. Multivariate calibration with the partial least-squares (PLS) algorithm was performed on spectral data between 1500 and 750 cm-1 showing a varying background from different unidentified interfering compounds. Cross validation was carried out for optimizing the PLS model. PRESS was 19.8 mg/dL, which was calculated on the basis of 127 standards, whereas the estimated standard deviation for the calibration fit was computed to be 11.9 mg/dL. Infrared spectroscopy can be used for monitoring glucose levels within the normal physiological range in a complex matrix like whole blood as an alternative to electrochemical sensors.     

Quantitative nondestructive methods for the determination of ticlopidine in tablets using reflectance near-infrared and Fourier transform Raman spectroscopy

Two different nondestructive spectroscopy methods based on near-infrared (NIR) and Fourier transform (FT) Raman spectroscopy were developed for the determination of ticlopidine-hydrochloride (TCL) in pharmaceutical formulations and the results were compared to those obtained by high-performance liquid chromatography (HPLC). An NIR assay was performed by reflectance over the 850-1700 nm region using a partial least squares (PLS) prediction model, while the absolute FT-Raman intensity of TCL's most intense vibration was used for constructing the calibration curve. For both methodologies the spectra were obtained from the as-received film-coated tablets of TCL. The two quantitative techniques were built using five "manual compressed" tablets containing different concentrations and validated by evaluating the calibration model as well as the accuracy and precision. The models were applied to commercial preparations (Ticlid). The results were compared to those obtained from the application of HPLC using the methodology described by "Sanofi Research Department" and were found to be in excellent agreement, proving that NIR, using fiber-optic probes, and FT-Raman spectroscopy can be used for the fast and reliable determination of the major component in pharmaceutical analysis.