Authentication of organic feed by near-infrared spectroscopy combined with chemometrics: a feasibility study

Organic products tend to retail at a higher price than their conventional counterparts, which makes them susceptible to fraud. In this study we evaluate the application of near-infrared spectroscopy (NIRS) as a rapid, cost-effective method to verify the organic identity of feed for laying hens. For this purpose a total of 36 organic and 60 conventional feed samples from The Netherlands were measured by NIRS. A binary classification model (organic vs conventional feed) was developed using partial least squares discriminant analysis. Models were developed using five different data preprocessing techniques, which were externally validated by a stratified random resampling strategy using 1000 realizations. Spectral regions related to the protein and fat content were among the most important ones for the classification model. The models based on data preprocessed using direct orthogonal signal correction (DOSC), standard normal variate (SNV), and first and second derivatives provided the most successful results in terms of median sensitivity (0.91 in external validation) and median specificity (1.00 for external validation of SNV models and 0.94 for DOSC and first and second derivative models). A previously developed model, which was based on fatty acid fingerprinting of the same set of feed samples, provided a higher sensitivity (1.00). This shows that the NIRS-based approach provides a rapid and low-cost screening tool, whereas the fatty acid fingerprinting model can be used for further confirmation of the organic identity of feed samples for laying hens. These methods provide additional assurance to the administrative controls currently conducted in the organic feed sector.     

Major components of honey analysis by near-infrared transflectance spectroscopy

NIR transflectance spectroscopy was used to analyze fructose, glucose, and moisture in honey. A total of 161 honey samples were collected during 1992 (46), 1995 (58), and 1996 (57). Samples were analyzed by instrumental, enzymatic (fructose and glucose), and refractometric (moisture) methods. Initially, different calibrations were performed for each of the 3 years of sampling. Good predictions were obtained for all three components with equations of the particular year. But good predictions were not always obtained when the equations calculated one year were applied to samples from another year. To perform a lasting calibration, unique calibration (121 samples) and validation (40 samples) sets were built; honeys of the 3 years were included in both sets. Good statistics (bias, standard error of validation (SEV), and R(2)) were obtained for all three components of the validation set. No statistically significant differences (p = 0.05) were found between instrumental and reference methods.     

Authentication of the botanical origin of honey by near-infrared spectroscopy

Fourier transform near-infrared spectroscopy (FT-NIR) was evaluated for the authentication of eight unifloral and polyfloral honey types (n = 364 samples) previously classified using traditional methods such as chemical, pollen, and sensory analysis. Chemometric evaluation of the spectra was carried out by applying principal component analysis and linear discriminant analysis. The corresponding error rates were calculated according to Bayes' theorem. NIR spectroscopy enabled a reliable discrimination of acacia, chestnut, and fir honeydew honey from the other unifloral and polyfloral honey types studied. The error rates ranged from <0.1 to 6.3% depending on the honey type. NIR proved also to be useful for the classification of blossom and honeydew honeys. The results demonstrate that near-infrared spectrometry is a valuable, rapid, and nondestructive tool for the authentication of the above-mentioned honeys, but not for all varieties studied.     

Determination of chemical composition of commercial honey by near-infrared spectroscopy.

The feasibility of using near-infrared spectroscopy to determine chemical composition of commercial honey was examined. The influences of various sample presentation methods and regression models on the performance of calibration equations were also studied. Transmittance spectra with 1 mm optical path length produced the best calibration for all constituents examined. The regression model of modified partial least squares (mPLS) was selected for the calibration of all honey constituents except moisture, for which the optimal calibration was developed with PLS. Validation of the established calibration equations with independent samples showed that the spectroscopic technique could accurately determine the contents of moisture, fructose, glucose, sucrose, and maltose with squared correlation coefficients (R(2)) of 1.0, 0.97, 0.91, 0.86, and 0.93 between the predicted values and the reference values. The prediction accuracy for free acid, lactone, and hydroxymethylfurfural (HMF) contents in honey was poor and unreliable. The study indicates that near-infrared spectroscopy can be used for rapid determination of major components in commercial honey.     

Determination of polarimetric parameters of honey by near-infrared transflectance spectroscopy.

NIR transflectance spectroscopy was used to determine polarimetric parameters (direct polarization, polarization after inversion, specific rotation in dry matter, and polarization due to nonmonosaccharides) and sucrose in honey. In total, 156 honey samples were collected during 1992 (45 samples), 1995 (56 samples), and 1996 (55 samples). Samples were analyzed by NIR spectroscopy and polarimetric methods. Calibration (118 samples) and validation (38 samples) sets were made up; honeys from the three years were included in both sets. Calibrations were performed by modified partial least-squares regression and scatter correction by standard normal variation and detrend methods. For direct polarization, polarization after inversion, specific rotation in dry matter, and polarization due to nonmonosaccharides, good statistics (bias, SEV, and R(2)) were obtained for the validation set, and no statistically (p = 0.05) significant differences were found between instrumental and polarimetric methods for these parameters. Statistical data for sucrose were not as good as those of the other parameters. Therefore, NIR spectroscopy is not an effective method for quantitative analysis of sucrose in these honey samples. However, NIR spectroscopy may be an acceptable method for semiquantitative evaluation of sucrose for honeys, such as those in our study, containing up to 3% of sucrose. Further work is necessary to validate the uncertainty at higher levels.     

基于近红外光谱技术的蜂蜜掺假识别

为了实现蜂蜜掺假的快速识别,应用近红外光谱结合模式识别方法对蜂蜜掺假现象进行了识别分析。该研究收集了中国不同品种、不同地域的典型天然蜂蜜样品,根据目前市场上常见的蜂蜜掺假手段,掺假物质及相对含量情况配制了掺假蜂蜜样品,利用傅立叶近红外光谱仪采集其透反射近红外光谱,分别采用偏最小二乘判别分析（PLS-DA）,独立软模式法（SIMCA）,误差反向传播神经网络（BP-ANN）和最小二乘支持向量机（LS-SVM）等模式识别方法,进行蜂蜜掺假识别研究。研究结果表明：利用这4种方法在蜂蜜中掺入果葡糖浆和果葡糖水的情况下均能很好地识别出掺假蜂蜜样品,其中对于掺入果葡糖浆的掺假情况,校正集的正确判别率均达到95%以上,验证集的正确判别率均达到87%以上,对于掺入果葡糖水的掺假蜂蜜校正集的正确判别率均达到93%以上,验证集的正确判别率均达到84%以上。通过比较4种不同的识别算法,发现采用LS-SVM时,对两种掺假情况下校正集和验证集的正确判别率均达到了100%,表明基于近红外光谱的蜂蜜掺假快速准确识别是可行的。     

近红外光谱结合化学计量学方法检测蜂蜜产地

为了实现蜂蜜产地的快速判别,应用近红外光谱结合化学计量学方法对蜂蜜产地进行了判别分析。kennard-Stone法划分训练集和预测集。光谱用一阶导数加自归一化预处理后,再用小波变换（WT）进行压缩和滤噪。结合滤波后光谱信息,分别用径向基神经网络（RBFNN）和偏最小二乘-线性判别分析（PLS-LDA）建立了苹果蜜产地和油菜蜜产地的判别模型。对不同小波基和分解尺度进行了讨论。对苹果蜜,WT-RBFNN模型和WT-PLS-LDA模型都是小波基为db1、分解尺度为2时的预测精度较好,都为96.2%。对油菜蜜：WT-RBFNN模型在小波基为db4和分解尺度为1时,预测精度较好,为85.7%;WT-PLS-LDA模型在小波基为db9、分解尺度也为1时,预测精度较好,为90.5%。研究表明：WT结合线性的PLS-LDA建模比WT结合非线性的RBFNN建模更适于蜂蜜产地判别;近红外光谱技术具有快速判别蜂蜜产地的潜力。     

支持向量机在苹果分类的近红外光谱模型中的应用

建立了一套苹果近红外光谱采集装置来减少因苹果的部位差异性而造成的试验误差。采用一种新的机器学习算法——支持向量机(SVM)建立不同产地、不同品种苹果的近红外光谱分类模型。通过选定RBF函数作为核函数，并确定合适的光谱预处理方法和核函数中惩罚系数C、正则化系数γ，使得所建立的不同品种苹果分类模型的回判识别率和预测识别率均达到100%，不同产地苹果分类模型的回判识别率为87%，预测识别率为100%，与传统的判别分析法相比其预测识别精度提高5%左右。结果表明，支持向量机可以建立高精度的苹果近红外光谱分类模型。     

Classification of wine and alcohol vinegar samples based on near-infrared spectroscopy. Feasibility study on the detection of adulterated vinegar samples

Near-infrared (NIR) spectroscopy was used to discriminate between wine vinegar (red or white) and alcohol vinegar. One orthogonal signal correction method (OSC) was applied on a set of 73 vinegar NIR spectra from both origins and artificial blends made in the laboratory in order to remove information unrelated to a specific chemical response (tartaric acid), which was selected due to its high discriminant ability to differentiate between wine vinegar and alcohol vinegar samples. These corrected NIR spectra, as well as raw NIR spectra and 14 physicochemical variables, were used to develop separate classification models using the potential functions method as a class-modeling technique. The aforementioned models were compared to evaluate the suitability of NIR spectroscopy as a rapid method for discriminating between vinegar origins. The transformation of vinegar NIR spectra by means of an orthogonal signal correction method prompted a notable improvement in the specificity of the constructed classification models. The classification model developed was then applied to artificial vinegar blends made in the laboratory to test its capacity to recognize adulterated vinegar samples.     

Determination by near-infrared spectroscopy of perseitol used as a marker for the botanical origin of avocado (Persea americana Mill.) honey

This paper reports the application of near-infrared (NIR) reflectance spectroscopy to determine the concentration in honey of perseitol, a sugar that is specific to avocado honey. Reference values for perseitol were obtained by high-performance liquid chromatography analysis in 109 honey samples. Although the average concentration of perseitol in honey samples was only 0.48%, accurate prediction equations were successfully developed. The regression model of modified partial least squares was superior to that of principal component regressions. Calibrations based on the first or second derivative of Log(1/R) were equally good (R(2) > 0.95). Using half of the samples for calibration and the second half for validation, the correlation between actual and predicted values of the second half was satisfactory (R(2) = 0.87), the slope did not differ from 1, bias was low (0.005%), and the standard error of prediction was relatively low (0.13%). It was concluded that NIRS analysis may be used to detect to what extent honeybees have harvested avocado nectar but not to authenticate avocado honey as unifloral.     

Determination of acid-detergent fiber and crude protein in forages by near-infrared reflectance spectroscopy: collaborative study

A collaborative study was conducted to determine the standard error of difference among laboratories for near-infrared reflectance spectroscopic (NIRS) determination of acid-detergent fiber (ADF) and crude protein in forages. The 6 participating laboratories were members of the USDA/ARS National Near-Infrared Reflectance Spectroscopy Forage Research Project. The NIRS calibration equations were developed in the Associate Referee's laboratory for crude protein and ADF and were transferred to the instrument in each of the other collaborating laboratories. The calibration set included over 650 diverse forage samples with crude protein and ADF calibration data; the validation set included 94 samples of bermudagrass. Among-laboratory reproducibility for the NIRS method, calculated as the relative standard deviation for reproducibility (RSDR), was 1.14% for ADF and 0.42% for crude protein. The variance component for among-laboratory variation (coefficient of variation) was 2.54% for ADF and 2.89% for crude protein. These results confirm that it is possible to calibrate, validate, and transfer (NIRS) equations and data among laboratories for the accurate determination of ADF and crude protein, and thereby demonstrate that NIRS can be used as a standard method for the analysis of forages. The method has been adopted official first action.     

Floral classification of honey using mid-infrared spectroscopy and surface acoustic wave based z-Nose Sensor

Fourier transform infrared spectroscopy (FTIR) and z-Nose were used as screening tools for the identification and classification of honey from different floral sources. Honey samples were scanned using microattenuated total reflectance spectroscopy in the region of 600-4000 cm(-1). Spectral data were analyzed by principal component analysis, canonical variate analysis, and artificial neural network for classification of the different honey samples from a range of floral sources. Classification accuracy near 100% was achieved for clover (South Dakota), buckwheat (Missouri), basswood (New York), wildflower (Pennsylvania), orange blossom (California), carrot (Louisiana), and alfalfa (California) honey. The same honey samples were also analyzed using a surface acoustic wave based z-Nose technology via a chromatogram and a spectral approach, corrected for time shift and baseline shifts. On the basis of the volatile components of honey, the seven different floral honeys previously mentioned were successfully discriminated using the z-Nose approach. Classification models for FTIR and z-Nose were successfully validated (near 100% correct classification) using 20 samples of unknown honey from various floral sources. The developed FTIR and z-Nose methods were able to detect the floral origin of the seven different honey samples within 2-3 min based on the developed calibrations.     

Geographic classification of spanish and Australian tempranillo red wines by visible and near-infrared spectroscopy combined with multivariate analysis

Visible (vis) and near-infrared (NIR) spectroscopy combined with multivariate analysis was used to classify the geographical origin of commercial Tempranillo wines from Australia and Spain. Wines (n = 63) were scanned in the vis and NIR regions (400-2500 nm) in a monochromator instrument in transmission. Principal component analysis (PCA), discriminant partial least-squares discriminant analysis (PLS-DA) and linear discriminant analysis (LDA) based on PCA scores were used to classify Tempranillo wines according to their geographical origin. Full cross-validation (leave-one-out) was used as validation method when PCA and LDA classification models were developed. PLS-DA models correctly classified 100% and 84.7% of the Australian and Spanish Tempranillo wine samples, respectively. LDA calibration models correctly classified 72% of the Australian wines and 85% of the Spanish wines. These results demonstrate the potential use of vis and NIR spectroscopy, combined with chemometrics as a rapid method to classify Tempranillo wines accordingly to their geographical origin.     

Energy from fat determined by near-infrared reflectance spectroscopy

The amount of energy derived from fat in foods is a requirement of U.S. nutrition labeling legislation and a significant factor in diet development by health professionals. Near-infrared (NIR) spectroscopy has been used to predict total utilizable energy in cereal foods for nutrition labeling purposes, and in the current study, was investigated as a method for evaluation of the amount of energy derived from fat. Using NIR reflectance spectra (1104-2494 nm) of ground cereal samples and reference values obtained by calorimetry and by calculation, modified PLS regression models were developed for the prediction of percent energy from fat and energy from fat/g. The models were able to predict the percent of utilizable energy derived from fat with SECV and R(2) of 1.86-1.89% of kcal (n = 51, range 0-43.0) and 0.98, respectively, and SEP and r(2) of 1.74% of kcal (n = 55, range 0-38.0) and 0.98, respectively, when used to predict independent validation samples. Results indicate that NIR spectroscopy provides useful methods for predicting the energy derived from fat in food products.     

Analysis of nitrogen fractions in silage by near-infrared spectroscopy

Total nitrogen, soluble nitrogen (SN), nonprotein nitrogen (NPN), and acid-detergent insoluble nitrogen (ADIN) were analyzed in grass silage by near-infrared (NIR) spectroscopy. A set of 144 samples was used to calibrate the instrument by modified partial least-squares regression, and the following statistical results were achieved: standard error of calibration (SEC) = 0.449 and square correlation coefficient (R (2)) = 0.98 for total nitrogen x 6.25, SEC = 0.425 and R (2) = 0.95 for SN x 6.25, SEC = 0.414 and R (2) = 0.94 for NPN x 6.25, and SEC = 0.139 and R (2) = 0.84 for ADIN x 6.25. To validate the calibration performed, a set of 48 silage samples was used. Standard errors of prediction were 0.76, 0.64, 0.63, and 0.25 for total nitrogen, SN, NPN, and ADIN (all of them multiplied by 6.25), respectively, and R (2) for the regression of measurements by reference method versus NIR analysis were 0.94, 0.92, 0.90, and 0.48 for total nitrogen, SN, NPN, and ADIN, respectively. To compare the results obtained by NIR spectroscopy with those obtained by the reference methods for total nitrogen, SN, and NPN of the validation set, linear regression and paired t tests were applied, and the results were not significantly different (p = 0.05). When mean square prediction error analysis was applied, it could be concluded that for total nitrogen, SN, and NPN, a robust calibration model was obtained and that the main error was unexplained error. Statistical data for ADIN were worse than those of the other parameters; as a result NIR spectroscopy is not an effective method for quantitative analyses of ADIN in silage; nevertheless, it may be an acceptable method for semiquantitative evaluation.     

Authentication of the botanical origin of honey by front-face fluorescence spectroscopy. A preliminary study

The potential of front-face fluorescence spectroscopy for the authentication of unifloral and polyfloral honey types (n = 57 samples) previously classified using traditional methods such as chemical, pollen, and sensory analysis was evaluated. Emission spectra were recorded between 280 and 480 nm (excit: 250 nm), 305 and 500 nm (excit: 290 nm), and 380 and 600 nm (excit: 373 nm) directly on honey samples. In addition, excitation spectra (290-440 nm) were recorded with the emission measured at 450 nm. A total of four different spectral data sets were considered for data analysis. After normalization of the spectra, chemometric evaluation of the spectral data was carried out using principal component analysis (PCA) and linear discriminant analysis (LDA). The rate of correct classification ranged from 36% to 100% by using single spectral data sets (250, 290, 373, 450 nm) and from 73% to 100% by combining these four data sets. For alpine polyfloral honey and the unifloral varieties investigated (acacia, alpine rose, honeydew, chestnut, and rape), correct classification ranged from 96% to 100%. This preliminary study indicates that front-face fluorescence spectroscopy is a promising technique for the authentication of the botanical origin of honey. It is nondestructive, rapid, easy to use, and inexpensive. The use of additional excitation wavelengths between 320 and 440 nm could increase the correct classification of the less characteristic fluorescent varieties.     

Method for determining frying oil degradation by near-infrared spectroscopy

A rapid near-infrared (NIR) spectroscopic method for measuring degradation products in frying oils, including total polar materials (TPMs) and free fatty acids (FFAs), has been developed. Calibration models were developed using both forward stepwise multiple linear regression (FSMLR) and partial least-squares (PLS) regression techniques and then tested with two independent sets of validation samples. Derivative treatments had limited usefulness, especially in the longer (1100-2500 nm) wavelength region. When using a wavelength region of 700-1100 nm, PLS models gave improved results compared to FSMLR models. The best correlations (r) between the NIR and wet chemical methods for TPM and FFA were 0.983 and 0.943, respectively. In the longer wavelength region (1100-2500 nm), FSMLR models were as good or better than PLS models. The best correlations for TPM and FFA obtained in this region were 0.999 and 0.983, respectively.     

Quantification of fatty acids in forages by near-infrared reflectance spectroscopy

Near-infrared reflectance spectroscopy (NIRS) was evaluated as a possible alternative to gas chromatography (GC) for the quantitative analysis of fatty acids in forages. Herbage samples from 11 greenhouse-grown forage species (grasses, legumes, and forbs) were collected at three stages of growth. Samples were freeze-dried, ground, and analyzed by GC and NIRS techniques. Half of the 195 samples were used to develop an NIRS calibration file for each of eight fatty acids, with the remaining half used as a validation data set. Spectral data, collected over a wavelength range of 1100-2498 nm, were regressed against GC data to develop calibration equations for lauric (C12:0), myristic (C14:0), palmitic (C16:0), stearic (C18:0), palmitoleic (C16:1), oleic (C18:1), linoleic (C18:2), and alpha-linolenic (C18:3) acids. Calibration equations had high coefficients of determination for calibration (0.93-0.99) and cross-validation (0.89-0.98), and standard errors of calibration and cross-validation were < 20% of the respective means. Simple linear regressions of NIRS results against GC data for the validation data set had r2 values ranging from 0.86 to 0.97. Regression slopes for C12:0, C14:0, C16:0, C18:0, C16:1, C18:2, and C18:3 were not significantly different (P = 0.05) from 1.0. The regression slope for C18:1 was 1.1. The ratio of standard error of prediction to standard deviation was > 3.0 for all fatty acids except C12:0 (2.6) and C14:0 (2.9). Validation statistics indicate that NIRS has high prediction ability for fatty acids in forages. Calibration equations developed using data for all plant materials accurately predicted concentrations of C16:0, C18:2, and C18:3 in individual plant species. Accuracy of prediction was less, but acceptable, for fatty acids (C12:0, C14:0, C18:0, C16:1, and C18:1) that were less prevalent.     

Analysis of fiber content in flax stems by near-infrared spectroscopy

The conventional means of measuring the fiber content of flax is time-consuming and laborious, and the results obtained vary with the analysis technique used. The plant tissues must first be "retted", a process by which the fibers are separated from the rest of the stem, either by indigenous organisms in the soil when the stems are left in the field or by water (anerobic bacteria) or enzymatic retting. The fiber content is then determined by mechanical or manual separation. In this study, fiber content of flax stems was measured rapidly and objectively by near-infrared spectroscopy (NIRS) using whole pieces of stem in a large cell, in reflectance mode. Compared to the conventional method, the standard error of performance of the NIRS method was between 0.96 and 1.45% (dry matter basis), depending on the model and data processing used. NIRS calibrations were generated by hand separation of fiber from water-retted specimens. The water retting procedure takes several days to complete and requires considerable trained labor to complete the hand separation step. The NIRS procedure was conducted on pieces of stem to simulate measurement in the field.     

Initial study of honey adulteration by sugar solutions using midinfrared (MIR) spectroscopy and chemometrics

Fourier transform infrared (FTIR) spectroscopy and attenuated total reflection (ATR) sampling have been used to detect adulteration of honey samples. The sample set comprised 320 spectra of authentic (n = 99) and adulterated (n = 221) honeys. Adulterants used were solutions containing both d-fructose and d-glucose prepared in the following respective weight ratios: 0.7:1.0, 1.2:1.0 (typical of honey composition), and 2.3:1.0. Each adulterant solution was added to individual honeys at levels of 7, 14, and 21% w/w. Spectral data were compressed and analyzed using k-nearest neighbors (kNN) and partial least squares (PLS) regression techniques. A number of data pretreatments were explored. Best classification models were achieved with PLS regression on first derivative spectra giving an overall correct classification rate of 93%, with 99% of samples adulterated at levels of 14% w/w or greater correctly identified. This method shows promise as a rapid screening technique for detection of this type of honey adulteration.